Академический Документы
Профессиональный Документы
Культура Документы
Doctor of Philosophy
Written by:
Jake Diablo Cao
Supervised by:
Professor Michael Ferry
&
Dr Kevin James Laws
ORIGINALITY STATEMENT
I hereby declare that this submission is my own work and to the best of my knowledge it
proportions of material which have been accepted for the award of any other degree or
diploma at UNSW or any other educational institution, except where due acknowledgement is
made in the thesis. Any contribution made to the research by others, with whom I have
worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the
intellectual content of this thesis is the product of my own work, except to the extent that
assistance from others in the project's design and conception or in style, presentation and
COPYRIGHT STATEMENT
I hereby grant the University of New South Wales or its agents the right to archive and to make
available my thesis or dissertation in whole or part in the University libraries in all forms of
media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain
all proprietary rights, such as patent rights. I also retain the right to use in future works (such
I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation
Abstract International.
I have either used no substantial portions of copyright material in my thesis or I have obtained
permission to use copyright material; where permission has not been granted I have
applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.
AUTHENTICITY STATEMENT
I certify that the Library deposit digital copy is a direct equivalent of the final officially
approved version of my thesis. No emendation of content has occurred and if there are any
minor variations in formatting, they are the result of the conversion to digital format.
Signed
Date 19/Jun/2013
ACKNOWLEDGEMENTS
I would like to thank all individuals and organisations that assisted and supported me in the
First and foremost I would like to acknowledge my supervisors, Prof. Michael Ferry and Dr
Kevin Laws for their guidance, continual support and encouragement over the course of my
entire candidature.
I would also like to thank my external supervisor, A/Prof. Nick Birbilis of Monash University, for
In addition, I would like to thank Drs Nicholas Kirkland, Penny Martens, Philip Boughton and
Elizabeth Boughton for all the help they have provided over the years, especially in regards to
I would like to acknowledge the funding of this research by the Australian Postgraduate Award
(APA) and the Australian Research Council (ARC) Centre of Excellence for Design in Light
Metals.
PUBLICATIONS
[1] J.D. Cao, N.T. Kirkland, K.J. Laws, N. Birbilis, and M. Ferry: 'Ca-Mg-Zn bulk metallic
[2] J.D. Cao, K.J. Laws, N. Birbilis and M. Ferry: ‘Potentiodynamic polarisation study of
bulk metallic glasses based on the Mg-Zn-Ca ternary system’, Corrosion Engineering,
[3] J.D. Cao, P. Martens, K.J. Laws, P. Boughton and M. Ferry: ‘Quantitative in vitro
[4] K.J. Laws, K.F. Shamlaye, J.D. Cao, J. Scicluna, M. Ferry: ‘Locating new Mg-based bulk
metallic glasses free of rare earth elements’, Journal of Alloys and Compounds, 542
(2012) 105-110.
[5] B. Chow, A. Baume, P. Lok, J.D. Cao, N. Coleman, A. Ruys and P. Boughton:
62
[6] J.D. Cao, K.J. Laws, C. Liu and M. Ferry: ‘Thermoplastic formability of CaMgZn bulk
Technology, submitted.
[7] X. Zhou, K.D. Ralston, K.J. Laws, J.D. Cao, M. Ferry and N. Birbilis, ‘Effect of the degree
[8] K.J. Laws, B. Gun, C. Reddy, J.D. Cao, K.F. Shamlaye and M. Ferry, ‘Magnesium-rich,
PRESENTATIONS
Podium
[9] “Exploring New Ca-Mg-Zn Bulk Metallic Glasses as Bioresorbable Metals”, 9th World
Others
[13] “New materials for bone repair become nutrients, not poison”, The Science Show,
http://www.abc.net.au/radionational/programs/scienceshow/new-materials-for-
bone-repair-become-nutrients2c-not-poison/3709958
[14] “Mending broken bones with glass”, Australia Unlimited, 24 May 2012:
http://www.australiaunlimited.com/science/mending-broken-bones-glass
Poster
[16] Poster presentation at the Australian Research Council, Centre of Excellence, Design
[18] Materials Science & Engineering Research Symposium, University of New South
[19] School of Materials Science & Engineering Poster Competition, University of New
[20] School of Materials Science & Engineering Postgraduate Research Poster Forum,
AWARDS
Prize.
Prize.
bioresorbable metals’
ABSTRACT
Amorphous metals, or bulk metallic glasses (BMGs), based on the Mg-Zn-Ca ternary alloy
system, are an emerging class of material that offer many improvements over conventional
crystalline magnesium alloys. These materials present a very real possibility of replacing
current metallic and polymeric implant materials in a variety of roles. However, there is a
processes. The aim of this thesis is to assess the potential of these newly discovered
The thesis began by focusing on a promising BMG composition, Mg65Zn30Ca5. The evaluation of
its degradation and associated effects on cell viability provided a quantitative baseline for the
subsequent work. However, its poor thermoplastic formability prompted the exploration of
A range of Mg-rich and Ca-rich BMG compositions were systematically explored. Although Ca-
rich BMGs exhibited superior thermoplastic formability compared to the Mg-rich BMGs, they
underwent rapid dissolution in biocorrosion environments. By enriching Ca-rich BMGs with Zn,
their corrosion potential was enobled, but the corrosion rate remained high. Nevertheless, the
amount of hydrogen evolved was low considering the degree of corrosion. Mg-rich BMGs on
the other hand contain significantly less Ca, the most reactive element within the alloy. As such,
Mg-rich BMGs exhibited superior corrosion resistance, and signs of passive corrosion kinetics.
The interaction between the elements Mg, Zn, Ca was found to be very complex. The
monotonic increase in Zn content in both Ca-rich and Mg-rich BMGs did not guarantee an
improvement in their corrosion resistance. This suggested that the corrosion rate was not
solely determined by the Zn content but, rather, the combination of Mg:Zn:Ca ratio.
Overall, the results and findings from this research furthered our understanding on the
amorphous alloy were amongst the first to be reported. This newly attained data and
knowledge may be used to support the technological push to use bioresorbable metals in
TABLE OF CONTENTS
ACKNOWLEDGEMENTS .................................................................................................... IV
PUBLICATIONS ................................................................................................................... V
PRESENTATIONS ............................................................................................................... VI
ABSTRACT ......................................................................................................................... XI
CHAPTER 1 - INTRODUCTION............................................................................................. 1
2.6.6 Tapping................................................................................................................ 39
4.5 Discussion.................................................................................................................... 86
LIST OF FIGURES
Figure 2.1: Mg-Zn-Ca ternary phase diagram, indicating reported BMG systems within this
Figure 2.2: Hydrogen evolution for Mg immersed in SBF with/without phosphates [52]. .......... 8
Figure 2.3: Subcutaneous gas bubbles observed on postoperative radiographs for 4 weeks
during Mg implant degradation in the femora of female Dunkin Hartley guinea pigs [10].
........................................................................................................................................... 9
Figure 2.4: (a) Definition of glass transition temperature [54]. (b) Schematic representation of
the specific volume as a function of temperature for a liquid that can crystallise [64]. 13
Figure 2.5: Molar Gibbs free energy difference (∆Gl-s) between the liquid and crystalline states
as a function of normalised temperature for various BMG compositions. Note lower ∆Gl-
s corresponds to a lower critical cooling rate (Rc) for glass formation, shown in
Figure 2.6: Critical casting thickness Zc and critical cooling rate Rc vs. reduced glass transition
temperature (Tg/Tliq) for various metallic glass-forming alloys. Data is also correlated
Figure 2.7: Schematic diagram of the inverted die casting apparatus developed for injection
Figure 2.8: Images of (a) chill block melt-spinner and (b) planar-flow-casting unit. .................. 18
Figure 2.10: Examples of thermoplastic forming of BMGs – (a) blow moulding [45], (b) hot-
Figure 2.11: Schematic TTT diagram, showing direct casting (solid green line), and the
superplastic forming route (dashed green line) whereby the material is reheated into
the supercooled liquid region, formed into final geometry, and then cooled pass the
Figure 2.12: Super plasticity achieved in Mg65Cu25Y10 BMGs. (a) Unfractured sample showing
an elongation of ~1300% (T = 160 oC and έ = 5 x 10-3 s-1) [91]. (b) As-cast and deformed
samples showing an elongation of over 1000% (T = 160 oC and έ = 10-2 s-1) [92]. .......... 23
Figure 2.13: Animal studies of Mg-based glass in comparison with a crystalline Mg alloy
reference sample. (a, c) Glassy Mg60Zn35Ca5 and (b, d) WZ21 crystalline Mg alloy
implanted in two types of porcine abdominal tissues: (a, b) muscle after 27 days, and (c,
d) subcutis after 91 days. Formation of fibrous capsules are indicated by white arrows
[9]..................................................................................................................................... 26
Figure 2.14: Yield strength (σy) plotted against Young’s modulus (E) for various classes of
Figure 2.16: (a) Asymmetric thread with sharp connections to the screw shaft, (b) asymmetric
Figure 2.17: Relation of screw pull-out strength to bone density in vertebral bodies. Increasing
Figure 2.18: Relationship between experimentally derived pull-out force and that predicted by
the equation relating pull-out strength to shear strength, and screw geometry [111].. 36
Figure 2.19: If the thread angle is kept constant, (a) increasing thread depth results in
increased pitch, and (b) decreasing thread pitch results in decreased thread depth [110].
......................................................................................................................................... 37
Figure 2.20: Common screw heads (a) pan, (b) button, (c) round, (d) truss, (e) flat
Figure 2.21: A common cutting flute used on solid cortical bone screws [110]. ........................ 40
Figure 2.22: Screw tips: (a) blunt, (b) corkscrew, (c) self-tapping with flutes, (d) self-cutting with
trochar. ............................................................................................................................ 40
Figure 2.23: Comparison between (A) non-self-tapping and (B) self-tapping screws. ............... 41
Figure 3.1: (a) A boron nitrate coated graphite crucible and a copper mould and (b) a
schematic illustration of the inverted die casting setup coupled with the induction
furnace. ............................................................................................................................ 47
Figure 3.2: XRD scan of Mg65Zn30Ca5 bulk metallic glass synthesised for this study. ................. 48
Figure 3.3: Differential scanning calorimetry data obtained for Mg65Zn30Ca5 bulk metallic glass,
using a heating rate of 20 °C/min. Arrows indicate the glass transition (Tg) and first
Figure 3.4: Compared to the crystalline pure Mg, the amorphous MgZnCa had a more noble
corrosion potential (Ecorr) and a significantly lower corrosion current (icorr, derived from
Figure 3.5: ICP-OES data on the Mg, Zn and Ca ion levels in the cell media, collected at 24 h
intervals. Samples were immersed in MEM (without cells) in days 1-3, and L929 were
seeded onto the samples on day 4, and cultured until day 6. MgZnCa elevated the
Figure 3.6: Percentage viability at 72 h after seeding. The positive and negative controls
showed the lowest and highest cell counts, respectively. Compared to the negative
control, the silicone glue did not have a significant effect on the cell viability (p > 0.05),
but the final cell count in the presence of MgZnCa and pure Mg were significantly
Figure 3.7: Optical microscope montage image of L929 cells on tissue culture plate, in the
presence of Mg65Zn30Ca5. Three different zones were identified around the immersed
sample by the cell density: (1) minimal cell growth (2) scattered cell growth and (3)
normal cell growth. (Phase contrast microscope images were taken at the end of 72 h,
just before cell counting, and stitched together using Microsoft Image Composite
Editor®) ............................................................................................................................ 59
Figure 3.8: Phase contrast microscope image, showing confluent L929 cells surrounding the
Figure 3.9: Isothermal DSC traces of Mg65Zn30Ca5 carried out at 5 °C increments over its
through isothermal DSC traces between the temperatures of 120-145 °C. ................... 63
Figure 3.11: Stainless steel die used to thermoplastically form the BMG screws. ..................... 64
Figure 3.12: BMG rods were injection casted to dimensions appropriate for subsequent
forming. ........................................................................................................................... 64
Figure 3.13: The thermoplastic forming setup. (Left) the Carver hydraulic hot-press; (right) a
schematic diagram of the prototype die, which would be inserted in between the hot-
plates. .............................................................................................................................. 65
Figure 4.1: Section of the Ca-Mg-Zn phase diagram, according to [50]. Compositions reported
by Senkov et al. [33] are marked by blue circles, and the new high zinc compositions
studied in the current work are marked by red triangles. The alloy with the largest
critical casting size in this study was CaBMG2 which appears to lie directly on a
peritectic triple point between the CaMg2, Ca3Zn and CaZn2 liquidus phase fields. ....... 70
Figure 4.2: XRD traces for the six Ca-rich BMGs synthesised for this study. The critical casting
Figure 4.3: Differential scanning calorimetry data obtained for the six Ca-rich BMGs at a
heating rate of 20 °C/min. Arrows indicate the glass transition (Tg) of each alloy. ........ 77
Figure 4.4: Young’s modulus (black triangles) and hardness (green squares) of the six Ca-BMGs,
Figure 4.5: Weight loss measured after up to 16 days immersion in MEM at 37 °C in humidified
CO2 atmosphere (a) for each sample, and (b) compared to the comparative ratio of
Figure 4.6: Average polarisation curves for the synthesised Ca-BMGs tested in MEM at 37 °C. 80
Figure 4.7: Comparison of the two measurements of corrosion rate, icorr versus weight loss. The
Figure 4.8: Hydrogen evolution from the Ca-BMGs immersed in MEM compared with pure Mg.
......................................................................................................................................... 82
no observable solid metal; SEM micrographs showing (b) a porous surface morphology;
passivation layer; (d) a localised porous patch observed in a cross-section taken parallel
to the sample surface, which suggests that corrosion has progressed in all directions. 84
Figure 4.10: Representative corrosion surface of the Ca-BMGs, showing (A) glassy material, (B)
corroded layer, and (C) precipitated layer. EDS mapping of this region showed a
significant depletion of Ca in the corroded layer (B) and the presence of oxides in the
Figure 4.11: Dilatometry and DSC traces (both carried out at a heating rate of 2 °C/min) of (a)
Mg66Zn30Ca4 (b) Ca65Mg15Zn20 (c) Ca55Mg17.5Zn27.5. Black dotted lines indicate the first
derivative (dL/dT) of the contraction/expansion, or dL/L0. Dashed green and red lines
indicate the first two inflexions in the dilatometry data, which respectively correspond
Figure 4.12: (a) Isothermal DSC traces of Ca55Mg17.5Zn27.5, based on which allowed the
Figure 4.13: Thermoplastic forming of a prototype screw, from an as-cast BMG rod. .............. 96
Figure 5.1: (a) Hydrogen evolution measurements in Mg-based BMGs showing a distinct drop
between the Mg-BMGs, showing active and passive (red arrow) behaviours [9]. ....... 101
Figure 5.2: XRD traces for 10 Mg-rich BMGs synthesised for this study. ................................. 106
Figure 5.3: DSC traces for Mg72-xZn24+xCa4 and Mg70-xZn24+xCa6 (x=0, 2, 4, 6 and 8). ................. 107
Figure 5.4: Representative polarisation curves of Mg-rich, (a) MgZnCa4 (b) MgZnCa6, and (c) Ca-
rich BMGs tested in MEM at 37 °C, in humidified CO2 atmosphere. ............................ 109
Figure 5.5: The sensitivity of alloy corrosion rate to individual elements, (a) Mg (b) Zn (c) Ca,
Figure 6.1: PLD targets prepared from as-cast Mg65Zn30Ca5 (a) wedge and (b) plate. ............. 119
Figure 6.2: XRD scan taken before (black) and after (red) being ablated by laser. Data of the red
curve was recorded using mono-capillary attachment to focus the x-ray beam. ......... 120
Figure 6.3: Secondary electron image viewed at 50° tilt, showing the presence of globules
Figure 6.4: Scanning electron micrographs of (left) globules observed on the MgZnCa film
surface, deposited using 11.5 J/cm2; and (right) globules observed by Krebs and Bremert,
Figure 6.5: TEM foil sample machined using focused ion beam, containing a partial globule. 123
Figure 6.6: Elemental mapping carried out on the TEM foil sample. ....................................... 124
Figure 6.7: TEM image estimating the BMG film thickness to be 100 nm................................ 125
Figure 6.8: Convergent beam electron diffraction of both the globule and thin film showing the
absence of a diffraction pattern, suggesting they are amorphous in nature. .............. 125
Figure 6.9: 3-source configuration for the combinational sputter deposition of ZrCuAl metallic
Figure 6.10: Sputtering assembly used in the sputtering of MgZnCa. ...................................... 129
Figure 6.11: Sputter deposition rates of two identical amorphous targets (square and diamond),
each showing a strong linear correlation with the applied power. .............................. 130
Figure 6.12: Comparison between the coated and uncoated Si (100) substrates clearly showing
Figure 6.13: PCL scaffold successfully coated using PLD after 60 min. ..................................... 132
Figure 6.14: Sputter coated PCL scaffold after 5 min at 45W................................................... 133
LIST OF TABLES
Table 2.2: Summary of the physical and mechanical properties of various implant materials in
Table 2.3: Factors affecting the holding power of bone screws [110]. ...................................... 35
Table 3.1: Calculated ion release rates based on ICP-OES data. ................................................ 55
Table 4.1: Summary of composition, glass transition (Tg), crystallisation (Tx), melting (Tm),
liquidus (Tl) temperatures, and critical casting thickness (Zc) for the six Ca-BMGs. The
Table 5.1: Summary of composition, glass transition (Tg), crystallisation (Tx), melting (Tm),
liquidus (Tl) temperatures, and critical casting thickness (Zc) for the BMGs synthesised
for this study. The thermoplastic forming region (ΔT) is calculated by subtracting Tg
Table 5.2: Summary of icorr and Ecorr of the alloy compositions used in the study (error refers to
LIST OF EQUATIONS
Equation 1 ..................................................................................................................................... 8
Equation 2 ................................................................................................................................... 34
Equation 3 ................................................................................................................................... 35
Equation 4 ................................................................................................................................... 38
GLOSSARY
DC Direct current
HA Hydroxylapatite
PCL Polycaprolactone
CHAPTER 1 - INTRODUCTION
A biomaterial can be defined as any material used to make devices to replace a part of a
function of the body in a safe, reliable, economical, and physiologically acceptable manner.
The materials to be used in-vivo must be approved by the appropriate regulatory body, such as
the Drug Administration (FDA) in the United States, and the Therapeutic Goods Administration
(TGA) in Australia. The approval processes are similar, and involves a series of biocompatibility
tests, which measure the material acceptance by the surrounding tissues and by the body as a
whole. Biocompatible materials must not: (i) irritate the surrounding structures; (ii) provoke an
abnormal inflammatory response; (iii) incite allergic or immunologic reactions, and (iv) cause
cancer.
The efficacy of a biomaterial relies on the ability of its surface to regulate cell adhesion.
based alloys are designed to minimise the body’s innate immune response, essentially bio-
inert. However, when in contact with bone tissues, bio-inert materials can become
encapsulated in connective tissue, preventing its close contact with the bone, requiring
repeated surgeries to amend. In addition, the modulus mismatch between these alloys and
natural bone can lead to stress-shielding, causing further health complications. Furthermore,
once the broken bone has healed, these bio-inert implants should also be removed via a
secondary surgical procedure. Remaining inside the body may cause further complications due
Bioresorbable metals are breaking the current paradigm in biomaterial science to develop only
corrosion resistant bio-inert metals. Rather than designing for longevity, new biomaterials
based on magnesium capitalises on its tendency to corrode. Magnesium and its alloys have
been the focus of research and development for use in orthopaedic applications. These
bioresorbable implants are designed to degrade over time, as the wound heals, such that
secondary surgeries are no longer necessary for its removal. In addition, as metals, they offer
bioabsorbable polymers.
Magnesium is the lightest structural metal, and it has an elastic modulus ~40 GPa, much closer
to that of human bone than stainless steel, CoCr- and Ti-alloys, which reduces the effect of
stress shielding. In addition, the degradation of Mg has been shown to have a beneficial effect
corrosion.
In an effort to reduce alloy corrosion rate, corrosion resistant Mg-alloys have been
investigated. However, these grades were not originally designed for use inside the human
body. As such, they often contain harmful constituents such as aluminium and rare-earth
elements. Concerns for biocompatibility led to the use of biocompatible elements – Ca and Zn,
which are both pre-existing elements in the body. The additions of Ca and Zn have been
reported to have grain refinement effects, strengthening the alloy, hence the focus on the
Mg-Zn-Ca alloy system has generated significant research attention. However, the solid
solubility of Ca and Zn in Mg are low, limited to 0.82 at.% and 2.4 at.% respectively.
Amorphous metals, or metallic glasses, based on the Mg-Zn-Ca ternary alloy system have
shown to exhibit many favourable properties for use as a biomaterial. Unlike crystalline Mg
alloys, the amorphous metal compositions are not limited by the solubility limit of the
elements in the matrix, therefore a wide range of alloy compositions can be synthesised to
form chemically homogeneous, and structurally monolithic alloys. In addition, amorphous Mg-
Zn-Ca has been reported to show improved corrosion resistance (reduced hydrogen evolution),
good biocompatibility, and ability to be shaped like a thermoplastic polymer. This suite of
The advancement in materials science that enabled the synthesis of amorphous metals from
Mg-Zn-Ca was first reported in 2005, and remains relatively recent. As an emerging
bioresorbable material candidate, there is a general lack of, and a strong demand for data on
its properties, especially its degradation process. The motivation for this work is to develop an
in-depth understanding of this new amorphous metal, encompassing critical areas relevant for
use as a biomaterial. The knowledge and potential intellectual properties generated from this
project may be used to critically assess the suitability of this class of material as a biomaterial,
and support the technological push to use bioresorbable metals in biomedical applications.
2.1 Background
based alloys are essentially bio-inert, remaining as permanent fixtures after the tissue has
healed [1]. Although these materials are durable and exhibit good mechanical properties
(Table 2.1), they also contain elements that are toxic to the human body. These bio-inert
materials were designed to minimise the body’s innate immune response [2], but when these
metals degrade, due to corrosion and/or wear, the release of debris can cause significant
concerns [1, 3-6]. Furthermore, bio-inert materials in contact with bone tissues can become
encapsulated in connective tissue, preventing its close contact with the bone, requiring
In recent years, there has been a growing interest in magnesium and its alloys, as a
bioresorbable metal, for use in orthopaedic fixation devices [1, 4, 8, 9]. Rather than designing
for longevity, this bioresorbable material relies on its tendency to corrode whilst releasing
essential minerals into the body. The implant would remain in the body and maintain
mechanical integrity for up to 12–18 weeks while the bone tissue heals, before degradation
and replacement by natural tissue [1, 10]. This approach avoids the necessity to remove the
implants via secondary surgery. The mechanical strength of Mg-alloys are closer to that of
bone, compared with existing bio-inert materials (Table 2.2), thereby reducing the effects of
stress shielding. In addition, when these Mg-alloys degrade, the release of essential minerals
have been reported to have stimulatory effects on the growth of new bone tissues [11] as well
Yield strength
Ultimate tensile Elongation Reduction
(0.2% offset)
strength (MPa) (%) of area (%)
(MPa)
316L annealed [13] >490 >190 >40 -
316L cold-worked [13] >860 >690 >10 -
CoCrMo (cast) [14] >655 >450 >8 >8
CoCrWNi (wrought) [15] >860 >310 >30 -
CoNiCrMo (solution 793-1000 241-448 50.0 65.0
annealed) [16]
CoNiCrMo cold worked: >1000 >650 >20.0 >60.0
medium hard [16]
CoNiCrMo cold worked: >1200 >1000 >10.0 >50.0
hard [16]
CoNiCrMo (cold-worked >1793 >1586 >8.0 >35.0
and aged) [16]
Ti Grade 1 [17] >240 >170 >24 >30
Ti Grade 2 [17] >345 >275 >20 >30
Ti Grade 3 [17] >450 >380 >18 >30
Ti Grade 4 [17] >550 >483 >15 >25
Ti6Al4V [18] >860 >795 8-10 15-25
Ti13Nb13Zr [19] 1030 900 15 45
In order for bioresorbable metals to succeed, they must overcome challenges such as
hydrogen evolution, ion toxicity, and gradual loss of mechanical strength [1, 10, 20]. As such,
corrosion control has been deemed a viable option, as controlled degradation will: (1) reduce
the significance of gaseous hydrogen evolution, (2) reduce the rate at which ions are released
into the body, and (3) control the rate of loss of mechanical strength.
There have been many attempts to moderate both the corrosion and mechanical behaviour of
magnesium through alloying; these include Al, Zn, Ca, Pd, Mn, and rare earth elements (Y, La,
Ce, Nd, Pr, Dy) [10, 21-27]. However, most of these elements are harmful to human health,
and are not safe for in-vivo use. Amongst these alloying elements, Ca and Zn are both pre-
existing elements in the body, and, as such, focus on Mg-Zn-Ca alloy system has resulted in
significant research attention [22, 23, 28]. The additions of Ca and Zn have been reported to
have a grain refining effect, thereby strengthening the alloy [29, 30]. Hanzi et al. [31] found the
optimal composition window for Ca and Zn is 0.2-0.4 wt.% Ca and 5-6 wt.% Zn, respectively, for
enhanced ductility and simultaneously high strength of the material. However, the solid
solubility of Ca and Zn in Mg are limited to 0.82 at.% and 2.4 at.% respectively. Exceeding these
thresholds would cause the formation of intermetallic secondary phases and a higher number
of grain boundaries, as a result of grain refinement. Both of these are the weak points in the
material where preferential and accelerated corrosion is likely to occur. Consequently, it has
also been reported that this improvement in mechanical properties of the alloy comes at a
Advances in materials technology have revealed that it is now possible to synthesise bulk
metallic glasses (BMGs) entirely from Mg, Zn and Ca constituents [9, 32-37]. These amorphous
magnesium alloys are supercooled liquids, and can exceed the traditional solid solubility limit
Zn and 6 at.% Ca have been reported to be successfully alloyed into the system, forming BMGs
[32, 35, 36]. Figure 2.1 shows the Mg-Zn-Ca phase diagram. Mechanical properties of Mg-
based metallic glass are summarised in Table 2.2. The amorphous nature of these materials
allows for improved corrosion characteristics compared to their crystalline counterparts [38-
43], and also enables the use of thermoplastic forming (TPF) processes within the supercooled
liquid region where these material exhibit exceptionally low viscosity [42, 44-46]. In addition,
Mg-Zn-Ca are all pre-existing elements in the body, and in-vitro, in-vivo evaluations carried out
on these BMGs have shown promising results, with no signs of inflammation or adverse
cellular reactions [21, 22, 37, 47-49]. The suite of attractive properties makes Mg-BMGs a
Figure 2.1: Mg-Zn-Ca ternary phase diagram, indicating reported BMG systems within this
2.2 History
Magnesium has the lowest density (1.7 g/cm3) of all structural metals [51]. The first recorded
use of Mg dates back to 1907 by Lambotte, who used a pure Mg plate to secure a lower leg
bone fracture [1]. Mg and alloys were later introduced into orthopaedic and trauma surgery in
the first half of the 20th century by McBride and Verbrugge [10]. Unfortunately, Mg is also the
most active metal in the EMF (electromotive force) series that is used for structural
applications [51], its corrosion is always accompanied by hydrogen evolution [9, 52], Equation
To quantitatively measure the rate Mg corroded, Song et al. [52] immersed 99.96 wt.% Mg
coupons (approximately 1-2 cm3) in 250 mL of simulated body fluid (SBF). At the average
corrosion rate of 19-44 mg/cm2/day, based on the weight loss measurement, it was predicted
Song et al. also found that the rate of hydrogen evolution from corroding Mg in the SBF
solution increased with time (Figure 2.2). The different hydrogen evolution rates in solutions
with and without phosphates suggest that the constituents in the body fluid could play
Figure 2.2: Hydrogen evolution for Mg immersed in SBF with/without phosphates [52].
To illustrate the severity of this hydrogen evolution problem, Witte et al. [10] implanted four
Mg alloys (AZ31, AZ91, WE43 and LAE442) into the femora of female guinea pigs. The post-
operative radiograph (Figure 2.3) shows the formation of a subcutaneous gas bubble around
the guinea pig femora. Although, according to Equation 1, the gas evolved should be hydrogen,
the group was unable to combust the gas collected by puncturing the bubble with a syringe. It
is also worth noting that no adverse reaction was observed in the guinea pigs.
Figure 2.3: Subcutaneous gas bubbles observed on postoperative radiographs for 4 weeks
during Mg implant degradation in the femora of female Dunkin Hartley guinea pigs [10].
Due to the problems with the formation of subcutaneous gas bubbles, associated with the
rapid corrosion of Mg, early attempts using crystalline Mg as a biomaterial was abandoned
when stainless steels were introduced as implant materials [1, 10]. Today, stainless steels are
considered to be ‘the gold standard’ in many ways because of their combination of high
strength, good toughness and bio-stability. However, the detrimental effect of Cr and Ni
(constituents of stainless steel), even in small quantities has made titanium the preferred
choice [53]. But as bio-inert materials, they exhibit poor interfacial bonding between the
Amorphous alloys are non-crystalline metallic solids exhibiting the phenomenon of a glass
transition [54]. Amorphous alloys were first discovered some fifty years ago by Duwez et al.
[55], by rapidly quenching Au80Si20 liquid at a rate of 105 to 106 °C/s. The large rate of heat
extraction requires at least one dimension of the resulting product to be very small (i.e.
thickness). Hence, critical casting diameter (Dc) has been used as an indication to the glass
forming ability (GFA) of these amorphous alloys. It was not until 1974 before the first bulk
metallic glass (BMG) was discovered [56], with the word “bulk” reserved for amorphous metals
with a Dc greater than one millimetre. More BMGs were discovered in the following years, but
1. The alloys are multi-component systems consisting of more than three elements
2. There is a significant difference in atomic size with the size ratios above 12% among
Amorphous Mg-Zn ribbons were first reported in 1977 [58]. Following the aforementioned
empirical rules, in 2005, Gu and Shiflet [32] introduced Ca into the mix. The addition of Ca
atoms formed a more efficiently packed Ca-centred cluster structure, which not only increases
the diffusion distance of the Mg and Zn species to form their respective intermetallics;
addition of Ca atoms also has a jamming effect, whereby the local movement of Mg and Zn
atoms, required in the formation of Mg7Zn3, is made more difficult by its presence, hence
slowing crystal nucleation kinetics [32, 59, 60]. Together, the resultant ternary Mg-Zn-Ca alloys
Ca-Mg-Zn metallic glass was also discovered around the same time as the Mg-Zn-Ca system
[33-35]. Although the two systems contain the same constituent elements, the base material
which strongly affects the system is different. This has divided the research efforts into Mg-
and Ca-based BMGs. Nevertheless, both systems reinforce the potential for metallic glass to be
To retain an amorphous structure and form a vitreous solid from a metallic alloy melt, the
formation of crystalline phases from the supercooled liquid must be supressed. In traditional
crystalline alloys, nucleation and growth are two competing forces during solidification. When
the melt is cooled below its liquidus temperature, there is a significantly increase in the driving
force to nucleate with further decreases in temperature. Grain growth can start immediately
following the nucleation event, until individual grains impinge upon each other.
In amorphous alloys, the thermodynamic driving force for grain nucleation is low, due to their
high packing density and tendency to form short-range order [61, 62]. The low tendency to
nucleation enables the glass forming melt to be sufficiently cooled without crystallisation,
which is accompanied by an increase in its viscosity. By then, the highly viscous melt prevents
further nucleation due to its very limited atomic mobility. The growth of nanocrystallites are
also retarded by low-mobility atoms rejected to the periphery of nascent crystals to such an
extent that nucleation is really the only part of the phase transformation that occurs [63].
Therefore, to avoid crystallisation, the melt must be cooled at a suitable rate, on a timescale
much shorter than that which would permit nucleation and growth of any crystalline phases.
A liquid at temperatures below its melting point is called a supercooled liquid (SCL) [64]. When
a SCL is cooled rapidly, there is very limited time for atoms to rearrange themselves into a
regular crystalline lattice. The resultant material behaves more like a solid than a liquid, and is
termed a glass [64]. At this point, a glass can be defined as a non-crystalline solid
demonstrating the phenomenon of a glass transition and the corresponding physical state is
Despite the discontinuity in the specific heat observed at the glass transition (Figure 2.4a), this
transition from the liquid to glassy state cannot be termed a thermodynamic phase transition
[65]. The value of the glass transition temperature (Tg) is known to be dependent on the
cooling or heating rate. A low cooling rate allows time for atomic motion and, hence, allows
the liquid to maintain its equilibrium structure by lowering its glass transition temperature
(Figure 2.4b). Rather, Tg is a function of the experimental cooling or heating rate during
measurements. As such a heating rate of 20 °C/min has become the standard in differential
(a) (b)
Figure 2.4: (a) Definition of glass transition temperature [54]. (b) Schematic representation
of the specific volume as a function of temperature for a liquid that can crystallise [64].
Since crystallisation is the result of nucleation and growth, which are governed by both the
thermodynamics and kinetics of crystallisation, so too is the formation of metallic glass. From a
thermodynamic standpoint, a driving force exists for nucleation to occur, by definition, the
Gibb’s free energy (ΔG), when the liquid is cooled below its melting point. Whilst the enthalpy
of fusion provides the driving force for nucleation, it is reduced by the energy required to
create a liquid-crystal interface. Therefore, it may be readily appreciated that a lower free
energy difference between the SCL and crystal phases means a lower driving force for
nucleation, which correlates with a lower critical cooling rate needed for glass formation [61]
(Figure 2.5). Similarly, at the same cooling rate, a lower ΔG will correlate with a greater GFA.
Figure 2.5: Molar Gibbs free energy difference (∆Gl-s) between the liquid and crystalline
states as a function of normalised temperature for various BMG compositions. Note lower
∆Gl-s corresponds to a lower critical cooling rate (Rc) for glass formation, shown in
From a kinetics standpoint, the growth of a crystal involves the atomic rearrangement within
the molten metal. A liquid with high viscosity and sluggish kinetics (poor mobility) in the SCL
state can greatly retard the formation of thermodynamically favoured crystalline phases in the
melt. This results in large GFA and high thermal stability of the supercooled liquid [65].
Crystallisation can also be hindered through topological methods [67], i.e.: (i) selecting
elements of a large atomic radius [59, 68] (effectively increasing the diffusion distance
required for structural rearrangement), and (ii) tailoring the alloy composition such that the
resultant SCL structure is densely and efficiently packed, thereby hindering diffusion and aiding
in the retention of an amorphous structure upon cooling [60]. Hence, high GFA may be
achieved when the local neighbourhood around each atomic species of the alloy is efficiently
To form a BMG, the melt must be quenched rapidly, on a timescale much shorter than that
which would permit crystal nucleation and growth. Common techniques used today to
produce BMGs from the molten state include: gravity casting, injection casting, melt-spinning,
which has a SCL region of ~80 K and only requires a critical cooling rate of 10 °C/s or less [71],
can be directly cast into complex shapes. However, Mg-Zn-Ca requires a faster cooling rate,
~102 °C/s, and only has a SCL region of ~20 °C [32, 36]. Therefore, in the production of Mg-Zn-
Ca BMGs, these aforementioned techniques can only produce simple geometries. But like all
BMGs, these can subsequently undergo thermoplastic forming (TPF) to achieve more complex
final geometries.
Figure 2.6: Critical casting thickness Zc and critical cooling rate Rc vs. reduced glass transition
temperature (Tg/Tliq) for various metallic glass-forming alloys. Data is also correlated with
This simple method of casting is identical to that carried out in a common foundry, whereby
the alloy charge material is melted in a suitable crucible and manually poured in air under the
influence of gravity into a removable copper mould to achieve high cooling rates. Simple
sample geometries such as rods, plates, wedges and cones have been produced using this
Unlike simple gravity casting, which often exposes the melt to air, injection casting can be
carried out under inert atmosphere, allowing the production of reactive/flammable alloy
pressure to force the molten alloy into a mould [75]. A schematic representation of the casting
equipment is shown in Figure 2.7, whereby the melt (inside an induction furnace) is injected
upwards (using a positive gas pressure) into a copper mould through a silica tube. This method
has been used in the discovery work by Gu and Shiflet [32], producing Mg-Zn-Ca BMGs with
Figure 2.7: Schematic diagram of the inverted die casting apparatus developed for injection
2.4.3 Melt-spinning
Melt-spinning is mostly used for Al, Fe-based amorphous ribbon production, both of which
have low GFA [76]. It has also been widely used in the synthesis of Mg-Zn-Ca ribbons [41, 43],
for its simplicity, efficiency, and high cooling rate (up to 106 °C/s) [77, 78]. In this process, a jet
of liquid metal is ejected from a nozzle and impinges on the outer surface of a rotating copper
wheel, where a thin layer is formed from a melt puddle and rapidly solidifies as a continuous
into either chill block melt spinning (CBMS) or planar flow casting (PFC) [79, 80] (Figure 2.8). In
the CBMS process, a free jet of liquid metal is ejected from a circular nozzle positioned at a
sufficiently large (more than several millimetres) nozzle-to-wheel gap, while, in the PFC
technique, a slit nozzle with a small nozzle-to-wheel distance (<1 mm) is applied. In the latter
process, the melt puddle is constrained between the nozzle lip and the wheel surface, the
liquid puddle is more stable and tends to wet the wheel surface better than the free jet CBMS.
Therefore, the ribbon obtained by PFC technique is much better than that prepared by the
alternative free jet CBMS, which makes the PFC possible to produce ribbons up to tens of
centimetres in width.
Figure 2.8: Images of (a) chill block melt-spinner and (b) planar-flow-casting unit.
2.4.4 Strip-casting
Despite the remarkable properties of BMGs, there are several drawbacks that prevent their
widespread commercialisation and application, one such factor is the lack of a continuous
production process for bulk amorphous alloys [81]. Twin-roll strip casting aims to fill in this gap,
combining casting and hot-rolling into a single step, it enables continuous fabrication of BMGs
in sheet form (Figure 2.9). After melting, the molten alloy is transferred into a tundish where
the metal is fed through a slot-type ceramic nozzle used to achieve a uniform flow across the
width of the rolls. By using water-cooled Cu alloy as the roll material, cooling rates of 102–
103 °C/s can be achieved [82, 83]. Twin-roll strip casting has been utilised to fabricate alloys
with a wide range of critical cooling rates, such as Fe-, Zr- and Cu-based BMGs [84, 85].
Cu47Ti33Zr11Ni6Sn2Si1, with respective critical cooling rates of ~1 °C/s [71] and ~250 °C/s [86].
Using these methods, the supercooled liquid must be simultaneously quenched and fill the
entirety of the mould, which makes the production of thin sections with high aspect ratio
particularly challenging. Fortunately, the sluggish crystallisation kinetics of BMGs in the SCL
region provide a unique processing opportunity, allowing the as-cast BMGs of simple geometry
whilst maintaining their amorphous nature [44]. TPF techniques include, blow moulding, hot
There are many advantages to shape amorphous metals using TPF routes rather than through
the traditional subtractive manufacturing techniques. One of the biggest advantages for TPF is
the ultra-precisely formability of the amorphous metals, allowing it to produce atomically flat
surfaces [88]. In contrast, machining or cutting not only creates surface defects, it also subjects
the material to high temperature and strain-rate deformation conditions [89]. Combined with
the low thermal conductivity of amorphous metals, machining may result in material oxidation
and crystallisation. Therefore, TPF is the preferred method of choice for shaping amorphous
metals.
(a) (b)
(c)
Figure 2.10: Examples of thermoplastic forming of BMGs – (a) blow moulding [45], (b) hot-
It is common practice to first synthesise samples of simple geometry, such as glassy ingots,
rods or wedges. Subsequent TPF processes may be carried out within the SCL region, which is
above the glass transition temperature (Figure 2.11). The ideal processing window should
maximise the working temperature whilst balancing the crystallisation kinetics of the material
to avoid crystallisation. At temperatures above this optimum processing window, the material
temperatures below the optimal working window, BMGs are extremely hard and have low
Figure 2.11: Schematic TTT diagram, showing direct casting (solid green line), and the
superplastic forming route (dashed green line) whereby the material is reheated into the
supercooled liquid region, formed into final geometry, and then cooled pass the glass
transition temperature.
The thermoplastic formability of a BMG is reflected in its viscosity within its SCL region, as well
as the width of this SCL region. BMGs with large working window are desired, as this minimises
the restriction on temperature, time, and will also have higher tolerance on local temperature
variations that are intrinsic to the forming process itself. Highly viscous flow may also limit the
complexity of TPF processes applicable to the material (Section 4.6). Although it is possible to
decrease viscosity by increasing the processing temperature [90, 91], this also increases the
rate of crystallisation thereby limiting the timeframe available to form these BMGs. Hence, the
Within the SCL region, BMGs can undergo extreme superplasticity and processed by
conventional thermoplastic forming methods. Gun et al. [91, 92] investigated the superplastic
flow behaviour of Mg65Cu25Y10 in the SCL region, whereby the metallic glass samples exhibited
elongations in excess of 1000% without breakage (Figure 2.12). Schroers et al. [45] were able
Figure 2.12: Super plasticity achieved in Mg65Cu25Y10 BMGs. (a) Unfractured sample showing
an elongation of ~1300% (T = 160 oC and έ = 5 x 10-3 s-1) [91]. (b) As-cast and deformed
samples showing an elongation of over 1000% (T = 160 oC and έ = 10-2 s-1) [92].
As biomaterials, metals are preferred over ceramics and polymers for their strength and
toughness. But the modulus mismatch between the existing bio-inert metals and bone can
lead to stress-shielding, causing further health complications. The Young’s modulus of Mg-Zn-
Ca BMGs are much closer to that of bone, and they also offer many other favourable
2.5.1 Biocompatibility
At some point in its functional lifespan, the entirety of the material will be resorbed into the
body. This mandates the use of biocompatible elements. Zn and Ca are both pre-existing
elements in the body, and as such, many studies have been carried out on the Mg-Zn-Ca
There are many limitations with the existing in-vitro cytotoxicity testing methods, nonetheless
this type of tests can effectively assess the material before progressing onto animal trials.
Studies in this area have been carried out on a range of compositions within the Mg-Zn-Ca
ternary system; these include Mg, Zn, Ca contents as high as 99 at.% [27], 35 at.% [9], and 65
at.% [48], respective, and there has yet to be any indication of adverse reactions.
The general consensus amongst these studies is that the alloys tested do support cellular
activities, but there have been contradicting reports to its exact effect on the cells. The study
by Gu et al. [47] on Mg66Zn30Ca4 and Mg70Zn25Ca5 BMGs showed decreasing cell viabilities
when cultured in direct contact. This was attributed to cell sensitivity to the fluctuating
environment; these include ion releases, pH changes, hydrogen evolution, and the release of
There had been reports to the contrary, whereby the presence of the alloy promoted cell
growth. The findings from an extraction study by Li et al. [27] concluded that cell viability was
positively affected by the increasing extract concentrations of Mg-1 wt.%Ca alloy, whereby all
extract concentrations, 10%, 50% and 100%, resulted in L929 cell viabilities exceeding 100% of
the negative control, with 100% extract giving the highest viability, ~180%. Similarly, using an
environmental electron microscope, Gu et al. [47] observed L929 and MG63 cells attached to
suggest that these alloys are safe for in-vivo use. Xie et al. [49] conducted in-vivo evaluation of
four weeks after first implantation, revealed no signs of inflammation, and the authors
concluded that the BMG was safe to use as an implant material. Zberg et al. [9] implanted
Mg60Zn35Ca5 BMG into the abdomen tissues of domestic pigs, to evaluate the tissue reaction to
Mg glass during degradation. Mg60Zn35Ca5 BMG was compared with a reference crystalline
WZ21 Mg alloy (Figure 2.13). All samples showed a typical fibrous capsule (white arrow),
indicative of foreign-body reaction. However, only the crystalline samples showed pronounced
hydrogen evolution (area between implanted discs, indicated by dashed lines, and fibrous
Figure 2.13: Animal studies of Mg-based glass in comparison with a crystalline Mg alloy
reference sample. (a, c) Glassy Mg60Zn35Ca5 and (b, d) WZ21 crystalline Mg alloy implanted in
two types of porcine abdominal tissues: (a, b) muscle after 27 days, and (c, d) subcutis after
In crystalline materials, the observed plastic behaviour is usually controlled by the motion of
dislocations, which are line defects in the crystalline lattice. Above the nanoscale regime, Hall-
Petch relationship can be used to describe the strengthening of a material through the
plastic deformation more difficult [94]. This is termed dislocation-mediated plasticity. However,
BMGs lack the atomically even glide planes to enable easy-slip, its disordered atomic
plasticity, deformation in metallic glasses instead occurs in thin shear bands [95].
Grain boundaries and other crystal defects are considered as structural weak points in
crystalline metals [42]. BMGs lack these microstructural defects, thus resulting in yield
strengths 2-3× larger, high toughness, and an elastic limit of about 2% compared to 0.5% for
their crystalline counterparts [46, 96]. Figure 2.14 summarises the work of Ashby and Greer
[97] who compared the elastic limit and Young’s modulus of BMGs to 1500 other useful
commercial alloys. The shaded zone in this figure shows the fundamental boundary of the
accessible region of the plot, limited by the theoretical strength (σy = E/20), which the BMGs
Figure 2.14: Yield strength (σy) plotted against Young’s modulus (E) for various classes of
The solid solubility limit of Zn and Ca in crystalline Mg-Zn-Ca alloys are 2.4 at.% and 0.82 at.%
respectively [50]. Around these limits, studies have also shown that both Zn and Ca have grain
refinement effects, thereby improving the mechanical properties of the alloy [29, 30].
Exceeding these limits, the formation of second phases reduces the effectiveness of the
strengthening of the alloys, and may also be detrimental to their corrosion properties [22].
Amorphous Mg-Zn-Ca is not bound by these solid solubility limits; it can exceed these values
and still form a monolithic structure. The mechanical properties of amorphous and crystalline
Mg-Zn-Ca alloys are summarised in Table 2.2. The compilation of values covers a range of BMG
compositions; Zn from 15-35 at.%, but Ca content remains between 3-6 at.%. Ca content has
been found to have a significant impact on the mechanical strength of the BMG. Li et al. [99]
concluded that the optimum composition range defined by strength and deformation should
be 3-5 at.% Ca for Mg72-xZn28Cax alloys. Indeed, the initial work by Gu and Shiflet [32] also found
that Ca contents from 2-6 at.% and Mg contents from 65-85 at.% showed significant ductility,
whereby amorphous ribbons of these compositions can be bent more than 180° and pinched
along the bending edge without breaking them; however, with >6 at.% Ca, the amorphous
Table 2.2: Summary of the physical and mechanical properties of various implant materials
Renewed interest in Mg-alloys as a potential bioresorbable implant material has led to many
corrosion studies in this area, but the different testing conditions amongst these studies make
it very difficult to compare the results. In particular, the use of different electrolytes and/or
buffers can have a huge impact on the measured corrosion rates [24, 104, 105]. Hence, a
comparative summary of all the studies is not provided. Rather, based on the merits in these
respective studies, some conclusions can be drawn, and will be discussed herein.
their crystalline counterparts [38-43, 47]. This has been attributed to: (i) chemical
homogeneity, (ii) monolithic structure, and (iii) lack of grain boundaries. Work on Ni- [38] and
Zr-based [39, 40] BMGs have all concluded that the amorphous alloys present superior
corrosion resistance compared to their heat treated crystalline counterparts. Heat treatment is
preferred in these studies to ensure the alloys are of the same chemical composition, and that
the observed differences in corrosion are indeed caused by the difference in their crystallinity.
Similarly, Wang et al. [43] studied the corrosion behaviour of amorphous Mg67Zn28Ca5 ribbon
and its crystalline counterpart, obtained by annealing. It was found that the amorphous ribbon
exhibited more noble corrosion potential, as well as a smaller corrosion current density.
In a separate study, Wang et al. [41] evaluated the corrosion behaviour of Mg67Zn28Ca5, from
sample annealed at 160 °C. It was proposed that the free energy in metastable amorphous
metal is higher than that of its crystalline counterparts, evident through its thermodynamic
tendency to crystallise. As such, those atoms in an amorphous state possess higher potential
energy, thus requiring less activation energy to corrode. The increase in corrosion rate with
further crystallisation was explained by the presence of a large number of grain boundaries,
dislocations and microstructural defects, which readily provide the active sites for the
With increasing crystallinity, the potential energy decreases, and the number of active sites
increases. It was also proposed that there exists a point (% crystallinity), whereby the driving
force for corrosion is the same for each of these two mechanisms which will result in the
lowest corrosion rate [106]. However, the extent of crystallisation at each annealing
temperature was not characterised. In addition, personal communication with the authors
revealed that the active sites were also estimated. Together, the lack of conclusive and factual
In neutral pH, Zn is the least reactive element within the Mg-Zn-Ca ternary system [107]. The
general approach to alloy design had been to enrich the system with Zn, in order to reduce the
corrosion rate. This follows from the findings by Zberg et al. [9], who showed that when the Zn
content of the alloy was above 28 at.%, the formation of a passive and stabilisation zinc- and
oxygen-rich surface layer was observed (using energy dispersive x-ray spectroscopy). Similarly,
work by Gu et al. [47] showed reduced corrosion current density for Mg66Zn30Ca4, compared
with Mg70Zn25Ca5 BMGs. These reduced corrosion rates have also been correlated to a
reduction in hydrogen evolution. The generation of hydrogen gas as subcutaneous gas bubbles
has troubled the wide application of Mg alloys for decades [10, 108]. Through Zn alloying, the
simultaneous effects of BMG composition and surface passivation present a possible solution
Additional Elements
In an effort to further improve the mechanical and corrosion properties of the Mg-Zn-Ca BMGs,
Pd and Y [23, 25] have been added. However, the addition of either of these elements had a
significant negative impact on the GFA of the quaternary alloy, resulting in the formation of
Before going into the detailed design of various types of screws, we must understand the
different bone structures. Bone is not a uniformly solid material. Cortical bone is the compact
outer layer of bone. It is composed of compact bone tissue with a porosity of 5-30% [109]. This
tissue type gives bone their smooth, white and solid appearance and accounts for 80% of the
total mass of an adult skeleton. Occupying the interior space is the cancellous bone. It is
composed of a network of rod- and plate-like elements that allow room for blood vessels and
marrow. It has higher porosity than cortical bone, i.e. 30-90% [109].
2.6.1 Nomenclature
A range of orthopaedic screws are available to select from, based on the locations and type of
the bone to be inserted into. Screws are commonly categorised as self-tapping or non-self-
Subsequently, the outer diameter, the pitch and the length will then uniquely define a
The factors that govern the strength of the screw are the geometry of its cross-sectional area,
the material it is made from, and its surface finish, especially around the junction between the
thread and the root (Figure 2.16) [110]. The rounded corners at the junction of the base of the
thread and the screw shaft reduce the stress concentration, which are typically associated with
sharp corners.
Figure 2.16: (a) Asymmetric thread with sharp connections to the screw shaft, (b)
Logically, a bigger screw will yield greater strength. However, the end-use limits many physical
between the inherent strength of the screw and its ability to gain holding power in bone.
The most important factor in determining the screw pull-out strength is the strength of the
bone into which the screw is placed (Figure 2.17) [110]. Research conducted in this field has
shown that the pull-out strength strongly correlates to the bone density of the patient
𝑆 = 21.6𝜌1.65 Equation 2
Where S = shear strength (MPa) and ρ = apparent density of trabecular bone (g/cm3)
Figure 2.17: Relation of screw pull-out strength to bone density in vertebral bodies.
However, if we focus on a porous material such as cancellous bone, its resistance to pull-out is
dependent on six factors (Table 2.3). Three are related to the geometry of the screw and the
other three relate to preparation of the hole into which it is placed (Equation 3). These
identified factors have been successful in predicting mostly the pull-out strength of non-
tapped screws placed in porous foam, which was used to model cancellous bone (Figure 2.18).
Table 2.3: Factors affecting the holding power of bone screws [110].
Holding power increases with increased screw diameter and length of engagement (increased
surface area in contact with bone), decreased pitch, and smaller minor diameter. Not tapping
is advantageous, especially in softer cancellous bone where there is little concern about screw
breakage and more concern that using a tap removes too much material from the hole
Where
𝐹𝑠 = predicted shear failure force (N)
𝑆 = material ultimate shear strength (MPa)
𝐴𝑠 = thread shear area (mm2)
𝐿 = length (mm)
𝐷𝑚𝑎𝑗𝑜𝑟 = major diameter (mm)
(𝐿 × 𝜋 × 𝐷𝑚𝑎𝑗𝑜𝑟 ) = area of a cylinder of
diameter 𝐷𝑚𝑎𝑗𝑜𝑟 and length 𝐿
𝑇𝑆𝐹 = thread shape factor (dimensionless) =
(0.5 + 0.57735 𝑑 ⁄𝑝)
𝑑 = thread depth (mm) = (𝐷𝑚𝑎𝑗𝑜𝑟 − 𝐷𝑚𝑖𝑛𝑜𝑟 )⁄2
𝐷𝑚𝑖𝑛𝑜𝑟 = minor root diameter (mm)
𝑝 = thread pitch (mm)
Figure 2.18: Relationship between experimentally derived pull-out force and that predicted
by the equation relating pull-out strength to shear strength, and screw geometry [111].
The thread shape factor (TSF) is defined as 0.5 plus a constant, times the ratio of thread depth
to pitch. This means the TSF, and the screw holding strength, will increase with thread depth,
and decreasing pitch. In reality there are practical limitations in altering these parameters. If
the thread angle is kept constant, increasing thread depth will also increase pitch, and
Figure 2.19: If the thread angle is kept constant, (a) increasing thread depth results in
increased pitch, and (b) decreasing thread pitch results in decreased thread depth [110].
There exists a practical upper limit for the TSF value, above which relatively little increase was
found in pull-out strength, this value was found to be 1.0 [110]. Commercially available
In designing a screw, if the outer (Dmajor) and the root (Dminor) diameters are known, the ideal
Where
2.6.4 Pilot-hole
Hughes and Jordan [112] showed that, in a rigid material such as cortical bone, there is a
critical pilot-hole size, relative to the screw diameter, beyond which the pull-out strength of
the screw drops dramatically, but below which there is no effect. In their studies, this critical
pilot-hole size was ~90% of the screw major diameter. In materials stimulating hard cancellous
bone, the critical pilot-hole size was 68-74% of the screw diameter. While for soft cancellous
2.6.5 Head
The under-surface of the screw head is normally spherical, which allows for transmission of
force to a plate even if the screw is not aligned perpendicular to it (Figure 2.20 e, f), which is of
frequent occurrence in surgery. In addition, the work by Nunamaker [114] found that
traditional Philips drive design was inferior to the hexagonal shaped imbus head.
Figure 2.20: Common screw heads (a) pan, (b) button, (c) round, (d) truss, (e) flat
2.6.6 Tapping
Tapping is the process of cutting internal thread into a material. Self-tapping screws are
designed to cut their own thread path during insertion without prior use of a tap (Figure 2.21).
These screws are produced with sharp cutting flutes at the leading end of the threaded portion
of the screw. Self-tapping screws are designed primarily for use in cortical bone [115].
Ansell and Scales [116] found that three flutes extended over three threads required the least
torque for insertion. However, these flutes weaken the pull-out strength of this portion of the
screw because the fluted threads have 17-30% less thread surface.
Figure 2.21: A common cutting flute used on solid cortical bone screws [110].
Non-self-tapping screws do not have flutes and are designed with a blunt tip (Figure 2.22a).
These screws require a predrilled pilot hole and threads cut with a tap. The most important
advantage of non-self-tapping screws is that the axial load and torque applied during tapping
and screw insertion are less than that with self-tapping screws.
a b c d
Figure 2.22: Screw tips: (a) blunt, (b) corkscrew, (c) self-tapping with flutes, (d) self-cutting
with trochar.
In a non-self-tapping screw (Figure 2.23a), the majority of the torque applied to the screw is
used in generating compression. However, in a self-tapping screw (Figure 2.23b), the pilot hole
is larger and the majority of the torque applied to the screw is used in cutting threads and
Figure 2.23: Comparison between (A) non-self-tapping and (B) self-tapping screws.
The screw pull-out strength depends on the material density. The larger hole created by the
tap does not decrease pull-out strength in cortical bone because of its density; in less dense
trabecular or osteopenic bone, the larger hole has a progressively larger effect and can
Bulk metallic glasses (BMGs) based on the Mg-Zn-Ca ternary system are promising candidates
characteristics, and contain no toxic elements, making them suitable for a range of
orthopaedic applications. However, as a new entrant to the biomaterials field, the concept of a
bioresorbable metal is faced with many challenges. It is evident that not much work has been
carried out to assess their biocompatibility and to understand their degradation mechanisms,
both of which are critical for their success. Working on the amorphous Mg-Zn-Ca alloy system,
this thesis aims to make a significant contribution in these areas. The data generated may be
used to support future regulatory approval of this type of materials, or assist to determine its
suitability as a biomaterial.
biology, biomedical engineering and tissue engineering. The scope of this thesis is limited
mainly to the materials science aspect of this amorphous metal, multi-disciplinary work shall
The first part of the thesis (Chapter 3) focuses on a promising alloy composition, Mg65Zn30Ca5,
recently reported, as of 2009. Due to the limited data in the literature, in-vitro
cytocompatibility study was carried out, using a novel protocol, to validate its biocompatibility.
The crystallisation kinetics of this alloy was then mapped out, paving the way for subsequent
thermoplastic forming.
The poor thermoplastic formability in Mg65Zn30Ca5 (Section 3.6) prompted the exploration of
Ca-rich BMGs (Chapter 4). These alloys have been reported to be capable of extreme
deformation when processed within its supercooled liquid region, but controlling its corrosion
remains a challenge. Hence, this chapter focuses on the enhancement of corrosion resistance
thermoplastic formability.
Chapter 5 combines the biocorrosion data collected on the Mg- and Ca-rich BMGs in order to
establish generalised trends. The effect of each elemental addition on the overall system
In conjunction with the foregoing research areas, thin film metallic glasses were explored in
techniques, sputtering and pulsed laser deposition, were utilised in an attempt to deposit
these two deposition processes have been extensively applied to BMGs and very little has
been published in this area. This work will focus on thin film characterisation, in order to make
comparison between the two techniques. This section is followed by the major conclusions of
3.1 Background
Bioresorbable metals, such as Mg-based bulk metallic glasses (BMGs), are breaking the current
paradigm in biomaterials science to develop only corrosion resistant bio-inert metals. Rather
than longevity, bioresorbable metals are designed to corrode at a controlled rate [1, 4, 108].
Orthopaedic components made from this material do not require secondary surgery to be
removed and, in addition, they offer superior load bearing capability compared to the existing
bioabsorbable polymers.
Mg-based BMGs were first reported in 2005. At the commencement of this thesis, in 2009,
there were very limited literature sources available on these BMGs, especially in regards to its
biocompatibility. The only work published was the in-vivo work by Zberg et al. [9], who carried
out animal studies in the abdominal walls and cavities of domestic pigs to evaluate the tissue
observed in Mg60Zn35Ca5 BMG, which led to the conclusion that the material was
biocompatible. However, such a conclusion was based on a histological study of the explanted
porcine abdominal tissues surrounding the implanted samples after 27 and 91 days of
implantation, and very little information is available about the material’s effect on the host
According to the United States Food and Drug Administration (FDA), a material cannot be said
study [9]. A material must be accepted by the surrounding tissues and by the body as a whole,
in order to be categorised as biocompatible [3], which means that the material must not: (i)
irritate the surrounding structures; (ii) provoke an abnormal inflammatory response; (iii) incite
ISO 10993 gives guidance on procedures to follow in the biological assessment of potential
medical devices and materials [117], as part of its overall evaluation and development. Based
on the device type, tissue types it will be in contact with, and the contact duration, a
combination of biological evaluations are required to demonstrate its safety. These initial
• Cytotoxicity
• Sensitisation
• Irritation
• Intracutaneous reactivity
• Systemic toxicity
• Subchronic toxicity
• Genotoxicity
• Implantation
• Haemocompatibility
Zberg’s work on Mg-Zn-Ca BMGs demonstrated its potential for use in biomedical applications.
Although an in-vivo study was conducted, clearly a more detailed cytocompatibility assessment
is also warranted, to satisfy the stringent requirements of the biomaterial regulatory bodies,
It is clear that the Mg-Zn-Ca alloy system presents a combination of very attractive properties
for use a bioresorbable biomaterial (Section 2.5.2). With the intention of fabricating
orthopaedic components from this Mg-rich BMG, Mg65Zn30Ca5 was selected. This composition
has >28 at.% Zn, and was synthesised then characterised for thermoplastic forming
experiments. In addition, its in-vitro cytotoxicity was assessed using L929 murine fibroblasts, to
validate its potential suitability as a bioresorbable material. However, existing protocols for
testing cytotoxicity, ISO 10993-5 [118], which uses cell culture techniques to determine the
biological responses caused by materials and/or their extracts, was not intended for testing
materials designed to degrade. Therefore, a new protocol, based on the knowledge and
the increase in ion concentration in the media (as a result of material degradation), and their
associated effect on cellular activities were both assessed. The outcomes of this study not only
provided insights into the mechanisms of bioactivity and bioresorption, it also generated data
Using an induction-heated, low pressure injection die caster (Figure 3.1), magnesium (99.8
wt.%), zinc (99.995 wt.%) and Mg-Ca master alloy (99.8 wt.%) were introduced into the melt to
form Mg65Zn30Ca5, hereafter referred to as MgZnCa. The charge was heated to 700 °C, cooled
to room temperature, and then reheated to 650 °C, followed by suction casting at 450 °C to
form 1 mm thick plates, using copper moulds. The amorphous nature of the alloys was
Netherlands) with a Cu-Kα x-ray source (λ=1.541 Å). The glass transition, crystallisation, melting
and liquidus temperatures were determined using DSC (204 F1 Phoenix, Netzsch Selb,
protective argon atmosphere. The as-cast BMG plates were cut and ground into discs (ф = 4
then sectioned into 1 mm thick discs. The impurity levels in the magnesium, zinc, and Mg-Ca
Figure 3.1: (a) A boron nitrate coated graphite crucible and a copper mould and (b) a
schematic illustration of the inverted die casting setup coupled with the induction furnace.
corrosion of intermetallic phases or grain boundaries. Hence, XRD was used for detecting the
possible presence of any intermetallic phases, and by elimination, confirm the amorphous
nature of the alloys. Figure 3.2 shows the XRD data for the as-cast MgZnCa. There were no
detectable crystalline diffraction peaks, and scattered signals were observed around 38° and
66°, together confirming the amorphous nature of the sample. Figure 3.3 shows the DSC trace
for MgZnCa obtained at a heating rate of 20 °C/min. It can be seen that Tg and Tx of the BMG
was 132 and 147 °C, respectively, giving a TPF window of ~15 °C. The distinct glass transition
seen in Figure 3.3 further verified the amorphous nature of the samples.
Figure 3.2: XRD scan of Mg65Zn30Ca5 bulk metallic glass synthesised for this study.
Figure 3.3: Differential scanning calorimetry data obtained for Mg65Zn30Ca5 bulk metallic
glass, using a heating rate of 20 °C/min. Arrows indicate the glass transition (Tg) and first
crystallisation (Tx).
Cytotoxicity assay
Four repeats of each sample type were prepared and used in the cytotoxicity assay. The disc-
shaped samples were disinfected with ethanol (80% v/v) washed and then sterilised under
ultra-violet light for 15 min on each side. To minimise disruption to the cells during the media
changes, each sample was fixed onto a 35 mm diameter tissue culture plate (TCP) using
medical-grade silicone glue (Q7-4840, Silastic, Dow Corning Corporation, Midland, USA). 2 mL
of sterile minimum essential medium (MEM M0643, Sigma-Aldrich, Sydney, Australia), with 2.2
g/L of sodium bicarbonate (S5761, Sigma-Aldrich, Sydney, Australia), 10% fetal bovine serum
P4458, Sigma-Aldrich, Sydney, Australia), was added to each culture plate containing the
sample. Sodium bicarbonate buffer was used in the MEM to maintain a physiological pH of 7.4.
In addition, the combination of bicarbonate and CO2 atmosphere can simulate the in-vivo
buffering mechanism [105, 119]. However, it is worth noting that the amount of sodium
bicarbonate added to the MEM follows the preparation instructions given by the supplier, and
The samples were immersed in MEM and stored in an incubator (Heraeus Heracell 150,
Thermo Fisher Scientific, Melbourne, Australia) at 37 °C, with a humidified 5% CO2 atmosphere.
To minimise the variation in pH as the material degraded, the solution was replaced every 24 h,
which also gave a useful timescale for monitoring the ion dissolution.
To establish a baseline for comparing the degradation rates, the samples were immersed in
MEM without cells for 72 h. After 72 h of sample immersion, 2 mL of L929 murine fibroblasts
cell suspension (50,000 cells/mL) was seeded onto each culture plate. Cell-only and latex
controls were used, and as medical-grade silicone glue was used to fix the samples, this was
also used as a control. All samples were stored in the same incubator under the same
conditions for a further 72 h, with media changes every 24 h to ensure the pH remained stable.
Throughout the study, media exchange was performed daily, collecting 2 mL of media from
each sample. 2 mL represents the entire quantity of media in the TCP, such that the ion
concentration in the fresh media is not affected. These exchanged media were separately
frozen and subsequently analysed using ICP-OES (OPTIMA 3000DV, Perkin Elmer, Waltham,
At the conclusion of the study, 72 h after cell seeding or 144 h since immersion in media, the
exchanged media was again collected for ICP-OES analysis. Then, a mixture of MEM and
trypsin, in a 2:1 ratio, was pipetted into the TCP to detach the cells. These cells were
subsequently counted using a cell viability analyser (Vi-CELL XR, Beckman Coulter, Sydney,
Australia), which automatically performs the trypan blue dye exclusion method. Throughout
the cytotoxicity assay, sample plates were periodically monitored and imaged using a phase
Electrochemical testing
The author’s previous biocorrosion studies on BMGs showed a strong correlation between the
corrosion rates determined using potentiodynamic polarisation (PDP) and those determined
by weight loss methods [66] (Section 4.4.3). Although weight loss is an accurate simulation of
the actual corrosion environment, PDP was selected as it yields a greater amount of
information on the corrosion mechanisms, allowing for more thorough comparison between
the materials tested. The corrosion rate determined using PDP was also checked against the
Electrochemical testing was performed using a three electrode flat-cell (K0235, Princeton
Applied Research, Oak Ridge, USA) with MEM (Gibco 61100, Life Technologies, Melbourne,
Australia) used as the electrolyte, 2.2 g/L of sodium bicarbonate was added during preparation,
following the supplier instructions. A list of the key components of the media is provided in
APPENDIX E. Since this test was not performed under sterile conditions, 0.2 g/L sodium azide
(S8032, Sigma-Aldrich, Sydney, Australia) was also added to the media to inhibit scum growth.
The flat-cell contained 300 mL of media and had an exposed working electrode area of 1 cm2.
A separate glass well, immersed in the cell, contained the saturated calomel (SCE) reference
electrode and the counter electrode was platinum wire mesh. All experiments were carried
out inside the CO2 incubator (Sanyo MCO-15AC, Quantum Scientific, Brisbane, Australia).
Henceforth, all potentials quoted herein are in reference to SCE, in MEM at 37 °C and a pH of
7.4. Measurements were made under the control of a potentiostat (VMP-3Z, BioLogic, Claix,
France) using EC-Lab v10.10 software (BioLogic, Claix, France). PDP tests were performed
following an open circuit potential (OCP) period of ~15 min, which is similar to values used in
other studies [24, 120, 121] and a scan rate of 1 mV/s. Each test was carried out from -150 mV
below to 500 mV above the OCP and the maximum current was limited to 2 mA. A minimum of
five separate scans were performed for each alloy to ensure reproducibility. The EC-Lab
software was also used to perform Tafel-type analysis that allowed fits to individual anodic or
cathodic data as well as entire curve portions. This also permitted reproducible and
Statistical analysis
Prism (GraphPad software, La Jolla, USA) was used to perform statistical analysis. Analysis of
the variance (ANOVA) was used to determine statistical significance in the data. Dunnett’s
method [122] was also employed to make multiple comparisons to the control.
PDP scans were performed under simulated body conditions inside an incubator. It was found
that MgZnCa possessed a more noble corrosion potential (-1.27 V) compared to crystalline Mg
(-1.68 V). To negate the scatter in the data, at least five PDP scans were performed for each
sample. The median trace with the most representative corrosion current (icorr) and corrosion
potential (Ecorr) was then selected and plotted, as shown in Figure 3.4. Tafel-slopes were fitted
to the PDP traces and used to determine the average icorr of each sample type.
Figure 3.4: Compared to the crystalline pure Mg, the amorphous MgZnCa had a more noble
corrosion potential (Ecorr) and a significantly lower corrosion current (icorr, derived from Tafel-
slope fitting).
Cell-culture media were exchanged daily, and their metallic ion levels analysed using ICP-OES.
As shown in Figure 3.5a, the increase over existing Mg ion level in the media (20 mg/L), was
significantly greater for crystalline Mg (average 108 mg/L/day) compared with MgZnCa
(average 56 mg/L/day). Also, the presence of Ca and Zn in the amorphous MgZnCa alloy and its
subsequent degradation slightly elevated their respective concentrations in the media (Figure
Figure 3.5: ICP-OES data on the Mg, Zn and Ca ion levels in the cell media, collected at 24 h
intervals. Samples were immersed in MEM (without cells) in days 1-3, and L929 were seeded
onto the samples on day 4, and cultured until day 6. MgZnCa elevated the concentrations of
Mg, Zn and Ca ions. In comparison, pure Mg released significantly more Mg ions than
MgZnCa.
For testing under in-vitro conditions, disc-shaped samples (ф = 4 mm, h = 1 mm) with an
exposed surface area of 25 mm2 were used. Minus the pre-existing ion level in the cell media
(Figure 3.5a), the elevation in ion concentration from sample dissolution were converted to a
new unit (taking into account the exposed surface area), from mg/L/day to μg/cm2/day (Table
3.1).
The aim of this study was to quantitatively measure the increase in ion concentration in the
media, as a result of material degradation. A combination of PDP and ICP-OES were used to
examine the degradation process. The corrosion rate, measured with PDP scans, was taken as
the icorr from a Tafel-type analysis. Ion levels in the exchanged media were analysed using ICP-
OES, and the elevation in levels was determined by comparing with the stock media solution.
Theoretically, the amount of metal ions lost in the corrosion process is expected to be equal to
Attempts were made to correlate between the two methods. The average corrosion current in
pure Mg during testing was 36.0 μA/cm2 (Figure 3.4). Based on this value, the rate of loss of
ions in this material is calculated to be ~392 μg/cm2/day. This value, determined using
determined through the analysis of the exchanged media using ICP-OES (Table 3.1). Whilst the
2+
PDP method assumes the simplest dissolution reaction for Mg (i.e. 𝑀𝑔(𝑠) → 𝑀𝑔(𝑎𝑞) + 2𝑒 − )
and assumes MgZnCa releases ions in the same stoichiometric ratio, ICP-OES is only capable of
detecting concentrations of ions in the solution. As such, surface salts formation, which has
been widely reported to accompany the degradation of MgZnCa and pure Mg [9, 119], reduces
the solution ion content. Considering the limitations of each method, the aforementioned
Direct comparison of the corrosion current (determined using PDP) between MgZnCa (6.9
μA/cm2) was significantly lower than that of pure Mg (36.0 μA/cm2). This is reinforced by the
analysis of exchanged media, which showed MgZnCa released 446 μg/cm2/day of Mg ions, half
that of pure Mg (859 μg/cm2/day). These findings are expected, as it has been widely reported
structure [1, 9]. The incorporation of Zn and Ca in MgZnCa also leads to their dissolution at
36.0 μg/cm2/day and 32.9 μg/cm2/day, respectively. These concentrations were not harmful to
PDP is a short term test that forces the sample to corrode, but it does not take into account
the formation of surface oxides that may affect future corrosion rates. Shown in Figure 3.5,
there was a general upward trend in Mg and Zn concentration released into the media over
the six days, which suggests that neither the presence of L929 murine fibroblasts nor the
corroded surface layers affected the subsequent rate of corrosion. The upward trend was
probably due to the increased surface area caused by a dealloying process, whereby a
corroded nanoporous structure is left behind that further facilitates corrosion by allowing
electrolyte to penetrate into the material bulk [66]. Also shown in Figure 3.5 was that the Ca
ion content was higher in the cell-only control group than in pure Mg samples. This was
attributed to the formation of salts, namely CaP [119, 123], depleting the solution ion content.
Unlike MgZnCa which also releases Ca ions, in pure Mg, the formation of CaP consumes the Ca
ions in the media. Therefore, its analyzed concentration is lower than that of the cell-only
control group.
All samples were immersed in cell media for three days (72 h). On the fourth day, L929 murine
fibroblasts were seeded onto samples and cultured for a further three days. Figure 3.6 shows
the final percentage viability at the conclusion of the study (72 h after seeding, six days since
immersion). As expected, the latex (positive control) and the cell-only (negative control)
samples set the minimum and the maximum cell counts, respectively. The medical-grade
silicone glue did not have a significant effect on the final cell count (p > 0.05). However, the
presence of both MgZnCa and crystalline Mg had an obvious influence on cell viabilities, as
their cell counts were significantly lower than the cell-only control (p < 0.05). Even though
these values were significantly different, viabilities remained relatively high (>67%), thereby
Figure 3.6: Percentage viability at 72 h after seeding. The positive and negative controls
showed the lowest and highest cell counts, respectively. Compared to the negative control,
the silicone glue did not have a significant effect on the cell viability (p > 0.05), but the final
cell count in the presence of MgZnCa and pure Mg were significantly different (p < 0.05).
Throughout this experiment, phase contrast microscopy was used to periodically monitor the
cell morphology and proliferation. Figure 3.7 shows a typical example of the characteristic
zones at various distances away from the immersed sample. As observed in all samples, these
Figure 3.7: Optical microscope montage image of L929 cells on tissue culture plate, in the
presence of Mg65Zn30Ca5. Three different zones were identified around the immersed sample
by the cell density: (1) minimal cell growth (2) scattered cell growth and (3) normal cell
growth. (Phase contrast microscope images were taken at the end of 72 h, just before cell
The use of silicone glue did not contribute to the formation of cell inhibition zones around the
samples. Studies carried out on the glue-only control sample showed that this material was
surrounded by confluent cells attached to its surface (Figure 3.8), and there was no indication
of cell inhibition. Hence, the formation of minimal cell growth zones was attributed to the
direct contact with the samples, as the entire TCP was exposed to the same ion concentration,
and cells were confluent elsewhere. Indeed, this phenomenon has been documented
elsewhere [47, 93], and believed to be caused by cell sensitivity to the fluctuating environment,
Figure 3.8: Phase contrast microscope image, showing confluent L929 cells surrounding the
silicone glue.
The cell viability for both pure Mg and MgZnCa was lower than the cell-only control, but still
considered high for a direct contact experiment. There was no significant difference between
the cell viability cultured on degradable samples held down by silicon glue, and those cultured
on silicon glue only. This suggests the dissolution of Mg, Zn and Ca ions at the rates of 859
μg/cm2/day, 36.0 μg/cm2/day and 32.9 μg/cm2/day, respectively, can support cell growth.
Although MgZnCa had reduced corrosion rate, its associated cell viability was lower compared
to that of pure Mg. This was believed to be caused by the cell sensitivity to fluctuating
products released from MgZnCa, which contain Zn and Ca (not present in pure Mg).
Work by Li et al. [27] found the release of Mg, Zn and Ca ions can stimulate cell growth,
resulting in viabilities to up to 180% of the negative control. In contrast, the current study
found the elevated ion concentrations of did not promote growth beyond the negative control
level. In direct contact, the bioresorbable metal samples caused the formation of different
cellular growth zones, and no cell attachment was observed on the actual sample.
techniques are not appropriate for this class of high strength and brittle material. Instead,
amorphous metals can undergo thermoplastic forming (TPF), allowing it to be formed like
thermoplastic polymers, and plastically deformed to very large strains [90, 91]. Hence, the
a biomaterial.
TPF can only be carried out in the temperature window above the glass transition and below
first crystallisation. The large SCL region of BMG is useful, since it reduces the restriction on
processing temperature and time, and will also have higher tolerance on local temperature
Mg-rich BMGs, including Mg65Zn30Ca5, with high strength and significant ductility was first
reported by Gu and Shiftlet [32], in 2005. These BMGs have a critical casting diameter of up to
3 mm, with mass density ranging from 2.0 to 3.0 g/cm3. The typical measured microhardness is
2.16 GPa, corresponding to a fracture strength of about 700 MPa and specific strength of
around 250-300 MPa cm3/g. With the intention to thermoplastically form this BMG into
orthopaedic components, its crystallisation kinetics must first be determined using differential
Based on the isochronal DSC trace (Figure 3.3), there is a ~15 °C thermoplastic forming window
for the said composition. Isothermal DSC traces were carried out across this range, obtaining
isothermal traces provided information on the onset and offset of crystallisation, which are
the time available to process the material at a given temperature before it begins to crystallise
(Figure 3.10).
Although there is ample time at lower temperatures to perform TPF processes, the viscosity of
the material can decrease dramatically at higher temperatures [90, 91], allowing more
Figure 3.9: Isothermal DSC traces of Mg65Zn30Ca5 carried out at 5 °C increments over its
A prototype die (Figure 3.11), made from stainless steel, was developed to thermoplastically
form as-cast BMG rods into screws. To minimise the time spent at elevated temperatures, and
consequently the likelihood of crystallisation, cast BMG samples were produced as close to the
inner diameter of the die as possible (Figure 3.12). The BMG rod was then snipped to the
appropriate length.
Figure 3.11: Stainless steel die used to thermoplastically form the BMG screws.
Figure 3.12: BMG rods were injection casted to dimensions appropriate for subsequent
forming.
The TPF process was carried out using a hot-press (Model 2697, Carver Laboratory press, with
a Eurotherm 937 temperature controller), at 130 °C, a schematic of the thermoplastic forming
setup is shown in (Figure 3.13). The stainless steel die, soaking in silicon oil bath, was placed on
a hydraulic hot-press. Once the die reached the target temperature and stabilised, BMG rod
was inserted into the die. A tool steel plunger was used to apply force to the BMG, whereby a
Figure 3.13: The thermoplastic forming setup. (Left) the Carver hydraulic hot-press; (right) a
schematic diagram of the prototype die, which would be inserted in between the hot-plates.
According to these data, there is approximately a 10 min window at 130 °C to work on the
BMG before the onset of crystallisation. Numerous attempts at thermoplastic forming were
unsuccessful (Figure 3.14). Evident in the photograph, thread marks were observable, and the
region between the screw head and shaft was well formed in all attempts. The failure was
attributed to the geometry of the die. The small pitch and large thread depth made it difficult
for the applied force to transfer into viscous flow over the entire length of the shaft.
Mg65Zn30Ca5 bulk metallic glass was cast directly from the melt in the form of 1 mm thick plates
corrosion experiments and subsequent chemical analysis of the exchanged media solution, the
amorphous alloy exhibited a slower corrosion rate and released Mg ions into the media at
approximately half the rate of a crystalline Mg reference sample. Although both types of
material had an influence on the L929 cell viabilities over 72 h culture period, the viabilities
remained relatively high, thereby indicating that they are capable of supporting cellular
activities. However, direct contact with the samples created regions of minimal cell growth
around both amorphous and crystalline samples, and no cell attachment was observed.
Attempts were made to form the as-cast Mg65Zn30Ca5 rods into prototype screws. The
application of force through the top plunger was unable to fill the mould. This was believed to
be due to the crystallisation behaviour of the material, whereby the required flow viscosity
could not be reached when working within the amorphous region of the amorphous-crystalline
transformation map.
GLASSES
4.1 Background
In the context of metallic implants, recent years have seen significant attention focused on
magnesium and its alloys for use as bioresorbable orthopaedic fixation devices. The majority of
studies have tested commercially available grades of crystalline alloys, which were not
designed for in-vivo use. Recent research has revealed that it is now possible to synthesise bulk
metallic glasses (BMGs) entirely from Mg, Zn and Ca constituents [32-35], which are pre-
existing elements in the body. As a frozen supercooled liquid, BMG compositions are not
limited by the interstitial solubility limit of the elements in the matrix, hence a wide range of
alloy compositions can be synthesised. The amorphous nature of such BMGs and their ability
potentially allow for superior corrosion resistance and mechanical properties, with the ability
Amongst the bioresorbable Mg-based alloys reported, the most notable is the amorphous Mg-
BMGs based upon Mg-Zn-Ca. The human body has an inherent tolerance to Mg, Zn and Ca. The
adult daily recommended allowances for these elements are 1000 mg/day Ca [124], 420
mg/day Mg [125], and 10 mg/day Zn [126]. In addition, the presence of these ions has been
Staphylococcus aureus and oral anaerobes [127, 128]. Extract and direct contact cytotoxicity
tests using L929 murine fibroblasts and MG63 human osteosarcoma cells have shown
improved cell viabilities than pure Mg [47]. Furthermore, in-vivo assessment in the abdomen
crystalline Mg alloys. However, its glass forming ability (GFA) is poor, with a critical casting
thickness of only ~3 mm reported for Mg66Zn30Ca4 and Mg67Zn28Ca5 (at.%) [9, 32]. In addition,
the thermoplastic forming window of these alloys is only 20 °C. At 135 °C, which is 10 °C below
the first crystallisation reaction, there is only ~160 s of processing time available before the
this context, investigation of Ca-rich BMGs from the Ca-Mg-Zn system would not compromise
the biocompatibility (since the daily dosage limit of Ca is greater than Mg) and, hence, would
~160 s of processing time available for Mg-Zn-Ca BMGs, Laws et al. [102] mapped out the
processing window for Ca65Mg15Zn20. At 130 °C, 10 °C below its first crystallisation,
crystallisation does not commence until ~300 s. Laws et al. were also able to achieve
elongation in excess of 850% in the SCL region of this alloy. Compared to Mg-Zn-Ca alloys, Ca-
BMGs have notably larger critical casting thickness and SCL region up to 50 °C, which is more
than double that of the Mg-BMGs [33, 103], allowing for more complex TPF processes. Wang
50% of Ca65Mg15Zn20 extract had shown no detectable cytotoxic effects on L929 (murine
fibroblast), VSMC (vascular smooth muscle cells) and ECV304 (endothelial) cells. Unfortunately,
the rapid rate of corrosion caused the material to completely disintegrate after no more than 3
h after exposure. In static distilled water at room temperature, Dahlman et al. [129] also
reported destructive corrosion reactions of the same material, decomposing into a multiphase
powder. Such reports indicate that controlling biocorrosion is a major hurdle for the success of
bioresorbable implants.
The ideal bioresorbable BMG will have a combination of satisfactory corrosion resistance on a
fitness for purpose basis, a large TPF window, sufficient mechanical toughness and strength.
Work by Zberg et al. [9] clearly showed that Mg-Ca-Zn BMGs exhibit much more passive
corrosion characteristics when the Zn concentration is greater than ~28 at.% due to the
formation and stabilisation of a zinc- and oxygen-rich surface layer, which they confirmed via
energy dispersive x-ray spectroscopy (EDS). It has also been observed that a higher Zn content
increases the thermal stability of Ca-Mg-Zn BMGs [32, 33], thereby potentially increasing the
In this part of the thesis, a series of six unique compositions were synthesised in the Ca-rich
region of the Ca-Mg-Zn ternary alloy system (Figure 4.1), with higher Zn contents than
previously reported alloys [33]. The tailored compositions were aimed at improving their: (i)
corrosion resistance [48, 129, 130] and (ii) thermal stability for thermoplastic processing [33]
over previously reported Ca-based BMGs. The synthesised material properties were
Figure 4.1: Section of the Ca-Mg-Zn phase diagram, according to [50]. Compositions reported
by Senkov et al. [33] are marked by blue circles, and the new high zinc compositions studied
in the current work are marked by red triangles. The alloy with the largest critical casting size
in this study was CaBMG2 which appears to lie directly on a peritectic triple point between
A mixture of pure elements and master alloys, namely magnesium (99.8 wt.%), zinc (99.995
wt.%), calcium (99.8 wt.%) and Mg-Ca master alloy (99.8 wt.%), were used to form the nominal
compositions indicated in Figure 4.1. The impurity concentrations in the magnesium, zinc, and
Mg/Ca master alloy can be found in APPENDIX A, APPENDIX C and APPENDIX D, respectively.
All materials were mechanically scrubbed and placed within a boron nitride coated graphite
crucible for melting. The melting chamber was vacuum purged with high purity argon gas
(99.997 vol.%) and left for 20 min under circulating argon atmosphere. The charge was
induction heated to 850 °C, cooled to room temperature, reheated to 700 °C, followed by
gravity wedge casting and injection casting into copper moulds at 620 °C. The plate mould has
mould pattern has a width of 40 mm and 100 mm in length with a length to thickness ratio of
10:1. Further details of this casting facility are given elsewhere [75].
High purity magnesium was used as a reference material in the biocorrosion tests. The
spectrographic analysis, using induction coupled plasma atomic emission spectroscopy (ICP-
The critical casting thickness and amorphous nature of each alloy was confirmed using a Philips
X-Pert Materials Research Diffractometer with a Cu-Kα x-ray source (λ=1.541 Å), fitted with a
monocapillary tube with a focal spot size of 0.5 mm by sequentially scanning the through-
section at the centre of the wedge samples. The glass transition, crystallisation, melting and
liquidus temperatures were determined using DSC (Netzsch DSC 204 F1 Phoenix). Scans were
atmosphere. Nanoindentation tests were carried out using UMIS (Fischer-Cripps Laboratories)
with a Berkovich indenter. A series of 100 indents (10 × 10 array, 100 µm spacing) were
performed under force control, with load increasing over 25 increments to the maximum loads,
held for 5s, then unloaded over 20 increments. A range of maximum loads were used, in 1 mN
oxygenated saline solution with a salt (NaCl) content of ~0.9 wt.% at a pH of 7.4 and
and pH that has the largest impact on corrosion behaviour. Therefore, the present work
attempts to maintain a constant and representative electrolyte via the use of buffered
simulated body fluid. It is crucial that any saline electrolyte contains the correct amount of Cl-
as found in the body, as increasing or decreasing the concentration (and or electrolyte pH) will
Therefore, minimum essential medium (MEM, 56416C, Gibco) was used as the electrolyte in
this study since it is close to the physiological conditions due to the addition of a variety of
amino acids and vitamins that are present in human plasma. The media were prepared by
dissolving a powder form of the MEM with distilled water, adding the 2.2 g/L sodium
bicarbonate (NaHCO3, Sigma® S5761) buffer, and stirring at room temperature until
completely dissolved. 0.2 g/L sodium azide (Sigma Aldrich S8032) was also added to the media
to inhibit scum growth. A list of the key components of the media is provided in APPENDIX E.
Mass loss experiments were performed in a CO2 incubator (MCO-15AC, Sanyo) at the human
physiological temperature (37 ± 0.1 °C). The samples were placed into vented containers
containing 50 mL of solution per cm2 of exposed sample surface area [104]. This volume was
chosen as it represented a ratio similar to the approximate amount of blood plasma the body
contains (2.75 L) relative to the approximate surface area of typical bone fixation devices (140
cm2) [8, 133]. The solution was buffered to maintain a physiological pH of 7.4 [134], although
to ensure that the pH remained stable, the solution was also replaced every 48 h, with this
timeframe selected based on the buffering capacity of the solution. Although these
approximations do not exactly replicate the actual environment that a specific implant would
experience in the body, these help to justify the selected in-vitro testing parameters. The mass
of each sample was measured using an XP105 Analytical Balance (Mettler-Toledo) with an
accuracy of 0.001 g. After removal from the solution, corrosion products were removed by
immersing the sample in a 2 M chromic acid solution (200 g/L CrO3, 10 g/L AgNO3) for 5 min at
50 °C. This removal technique is commonly employed in magnesium corrosion studies [135].
Determination of the amount of hydrogen evolution during corrosion was carried out inside
the CO2 incubator at 37 °C. Polished BMG samples were weighed and their exposed surface
area measured prior to being placed into a beaker containing 300 mL of media. Hydrogen
bubbles were collected into an inverted funnel, submerged into the media, covering the
sample. A pipette filled with media was attached to this funnel; the hydrogen generated
displaced the media inside the pipette, and was measured by reading the solution level on the
Electrochemical tests were performed using a three electrode flat-cell (Princeton Applied
Research). The cell contained 300 mL of media and had an exposed working electrode area of
1 cm2. A separate glass well, immersed in the cell, contained the saturated calomel (SCE)
reference electrode and the counter electrode was platinum wire mesh. All experiments were
carried out inside the CO2 incubator. Henceforth, all potentials quoted herein are in reference
to SCE, in MEM at 37 °C and a pH of 7.4. Measurements were under the control of a BioLogic®
VMP-3Z potentiostat using EC-Lab v10.10 software. Potentiodynamic polarisation (PDP) tests
were performed following an open circuit potential (OCP) period of ~15 min, which is similar to
values used in other studies [24, 120, 121] and a scan rate of 1 mV/s. Each test was carried out
from -150 mV below to 500 mV above the OCP. The maximum current was limited to 2 mA. A
minimum of 5 separate scans were performed for each alloy to ensure reproducibility. The EC-
Lab software was also used to perform Tafel-type analysis that allowed fits to individual anodic
or cathodic data as well as entire curve portions. This also permitted reproducible and
4.4 Results
Figure 4.2 shows the XRD data for the as-cast Ca-BMGs. There were no detectable crystalline
diffraction peaks, and the broad diffraction scattering maxima were observed around 33° and
58°, confirming the amorphous nature of the samples. It is also notable that the shape of the
major halo has a slight shoulder on its right-hand side and leans slightly toward higher
scattering angles for higher Zn-concentrations; this may be attributed to the increase in the
number of Zn-Zn atom neighbours (the shortest bond length in the alloy), and with its higher
atomic number resulted in a much higher scattering factor. Figure 4.3 shows the DSC data for
the as-cast Ca-BMGs obtained at a constant heating rate of 20 °C/min. The glass transition (Tg),
crystallisation (Tx) and melting (Tm) behaviour were determined from the DSC traces and these
thermophysical values are summarised in Table 4.1. It can be seen that Tg and Tx of the Ca-
BMGs range from 119 to 129 °C and 138 to 146 °C respectively, giving a thermoplastic forming
Table 4.1: Summary of composition, glass transition (Tg), crystallisation (Tx), melting (Tm),
liquidus (Tl) temperatures, and critical casting thickness (Zc) for the six Ca-BMGs. The
Figure 4.2: XRD traces for the six Ca-rich BMGs synthesised for this study. The critical casting
thickness (Zc) was determined by sequentially scanning the through-section, at the centre of
Figure 4.3: Differential scanning calorimetry data obtained for the six Ca-rich BMGs at a
heating rate of 20 °C/min. Arrows indicate the glass transition (Tg) of each alloy.
The Young’s modulus and hardness values were determined by the Oliver and Pharr method
[136], from the load-displacement curves. As the hardness can be affected by the indentation
depth and applied load [137], the strain gradient independent hardness was calculated using
the method described by Fornell et al. [138]. Modulus was found to be depth independent
Figure 4.4: Young’s modulus (black triangles) and hardness (green squares) of the six Ca-
Analysis of the nanoindentation results showed that these Ca-BMGs have an elastic modulus in
the range 35-46 GPa, and a hardness of 0.7-1.4 GPa (Figure 4.4). The modulus data are
comparable to that of pure Mg (41-45 GPa), and somewhat closer to the Young’s modulus of
bone. Their hardness values are ~1/3 of that observed for Mg65Cu25Y10 BMGs [139], and ~1/2 of
In CO2 buffered MEM at 37 °C, sample weight loss from immersion tests were evaluated. Due
to differences in composition, the less corrosion resistant samples had to be removed earlier
to leave sufficient mass for weight measurements. Taking into account the initial exposed
surface area, the average weight loss rates are shown in Figure 4.5a. In the Ca-Mg-Zn ternary
system, it appears that the Zn content has a notable effect on the biocorrosion rate of the
system. There is a linear correlation between the Zn content and mass loss rate (Figure 4.5b),
as discussed in Section 4.5.2. Compositions with the highest zinc content (30 at.%), smallest
(Ca+Mg):Zn ratio (2.33), correlated to a lower mass loss rate relative to the other compositions.
humidified CO2 atmosphere (a) for each sample, and (b) compared to the comparative ratio
PDP scans yielded mechanistic information regarding the corrosion kinetics of the alloys. It was
found that Ca-BMGs possess more noble corrosion potentials compared to crystalline high
purity Mg, -1.93 VSCE, related to the high level of Zn in the former. To negate the scatter in the
data, at least five PDP scans were performed for each sample. The median trace with the most
representative corrosion current (icorr) and corrosion potential (Ecorr) was then selected and
plotted, as shown in Figure 4.6. The Ecorr of the six Ca-BMGs varied from -1.44 to -1.39 VSCE.
Figure 4.6: Average polarisation curves for the synthesised Ca-BMGs tested in MEM at 37 °C.
Since PDP and weight loss are widely accepted methods for assessing the corrosion behaviour
of alloys, both tests were carried out as they can be complementary in the sense that one is
instantaneous (PDP), yet yields mechanistic information, whereas the other is long term (mass
loss) and, whilst revealing no mechanistic information, yields a ground truth state. As both
testing methods provide information on the corrosion rate, and were conducted in the same
simulated body environment, a correlation can be drawn between them. The corrosion rate
from the PDP scans was taken as the icorr from a Tafel-type analysis. The corrosion rate from
the weight loss test used the data given in Figure 4.5a. When presented in their native units,
the two different test methods for measuring biocorrosion displayed a linear correlation
(Figure 4.7), whereby a large mass loss rate is accompanied by a correspondingly high icorr.
Figure 4.7: Comparison of the two measurements of corrosion rate, icorr versus weight loss.
To assess if the dissolution of the Ca-BMGs led to an appreciable evolution of hydrogen (from
the corresponding cathodic reaction, 2H2O + 2e- 2OH- + H2), hydrogen collection tests were
performed (results given in Figure 4.8). Some hydrogen evolution was expected owing to the
visual presence of very fine gas bubbles emanating from the dissolving surface.
Figure 4.8: Hydrogen evolution from the Ca-BMGs immersed in MEM compared with pure
Mg.
Nominally the driving force for hydrogen evolution diminishes as the corrosion potential
approaches ~-1.2 VSCE (corresponding to ~-1 V versus the normal hydrogen electrode).
However, in this instance the amount of hydrogen evolution remains relatively high (and
comparable to pure Mg which displays a much lower corrosion potential), which is attributed
[140]. As such, this leads to an important realisation in that the supersaturation achieved from
alloying elements in the BMG form cannot only impact the anodic (dissolution) kinetics, but
indeed also the cathodic efficiency of the BMG. This is discussed further in Section 4.5.2.
In Figure 4.8, hydrogen evolution versus the mass loss is displayed to reveal a proportionality
of the results. However, the native units from the two distinct tests have been retained in
order to avoid errors from assuming general corrosion in the conversion to standard units.
behaviour is also not discussed further herein as the aim of the hydrogen collection tests was
to ascertain if hydrogen evolution was a concern. A more detailed discussion regarding the
correlation and analysis of the various in-vitro tests is given in ref. [141].
Although post-immersion microscopy revealed that the samples exhibit similar geometric
dimensions to those prior to the immersion test, close inspection showed that some samples
sample was sectioned, where examination of its cross-section revealed no sign of observable
solid metal (Figure 4.9a). Subsequently, a series of SEM images were taken from the cross-
sectional surfaces, which showed a porous network of corrosion product, as depicted in Figure
4.9b.
revealing no observable solid metal; SEM micrographs showing (b) a porous surface
significant passivation layer; (d) a localised porous patch observed in a cross-section taken
parallel to the sample surface, which suggests that corrosion has progressed in all directions.
Further SEM investigation carried out on more corrosion resistant samples displayed similar
corrosion morphology, albeit to a lesser degree. With no preferential grain boundaries, the
amorphous Ca-BMG corroded from a localised region on the surface, and spread in all
directions. Figure 4.9c shows a cross-section taken through such an area, whereby the
corrosion boundary has progressed uniformly towards the core of the uncorroded BMG.
In certain cases, it was observed that rather than remaining superficial, the corrosion
preferentially entrenched into the material, leaving behind porous remains. A cross-section
taken parallel to the sample surface revealed similar porous corrosion product. As the
corrosion grew in the plane perpendicular to the image, a porous patch amongst a glassy
matrix was observed (Figure 4.9d). This type of corrosion is characteristic of a dealloying
process, whereby a less noble element (in solid solution) is preferentially dissolved. In this
instance, the least noble, or most active element, is Ca. The preferential dissolution of this
element was confirmed via EDS mapping (Figure 10), whereby a significant drop in Ca content
was observed in the corroded layer. SEM imaging and EDS mapping also revealed an oxide
layer above the corroded layer, formed by ion precipitation from the MEM (Figure 4.10), which
Figure 4.10: Representative corrosion surface of the Ca-BMGs, showing (A) glassy material,
(B) corroded layer, and (C) precipitated layer. EDS mapping of this region showed a
significant depletion of Ca in the corroded layer (B) and the presence of oxides in the
4.5 Discussion
With Ca as the major alloying component in the BMGs used in this thesis, it was strategically
decided to develop compositions that maintain a high GFA whilst maximising Zn content – the
least corrosion prone species present in the alloys. Ca contents as low as 50 at.% and Zn
contents as high as 30 at.% were successfully alloyed and cast to produce BMG samples.
Shifting from Ca-rich compositions, moving further from the deep ternary eutectic reaction
(according to reference [50]) the liquidus temperatures and the intervals between the melting
The GFA of the alloys studied herein tended to be inversely related to the liquidus temperature
and the relative volume fraction of the higher melting point constituents, as indicated by the
area below the baseline of the later melting reactions in the DSC results in Figure 4.3.
These results were somewhat expected as it is known that the thermodynamic driving force
for crystallisation and the volume fraction of the higher melting point CaMg2 and CaZn2 phases
increase with increasing Zn and Mg content [142], and an increase in the critical cooling
interval between Tl and Tg effectively decreases the achievable cooling rate of the casting
system [143] resulting in a direct decrease in the critical casting thickness of these alloys. The
alloy with the largest critical casting size in this study was CaBMG2 which appears to lie directly
on a peritectic triple point between the CaMg2, Ca3Zn and CaZn2 liquidus phase fields [50].
According to the DSC results and the ternary liquidus surface, CaBMG2 had significantly lower
liquidus temperature and much smaller interval between its melting and liquidus
often observed in glasses with a high enthalpy state upon casting [144]. This is attributed to
excessive structural free volume, whereby the reduction in free volume during heating is
proportional to the exothermic heat of this structural relaxation prior to the glass transition
[145]. Compared to the glass transition temperatures of reported Ca-based BMGs with higher
Ca-contents [33] ranging from 102 to 107 °C, the glass transition temperatures of these alloys
are slightly higher, ranging from 119 to 129 °C, thereby indicating an increased thermal
stability of the supercooled liquid, which is likely due to the increased number of Ca-Zn, and
Ca-Mg and Mg-Zn bonds in the structure. It can be seen that the addition of Zn has the
greatest effect on increasing the glass transition temperature, which is likely due to the higher
binding potential of the Ca-Zn atomic pair (-22 kJ/mol) compared to Ca-Mg (-6 kJ/mol) and Mg-
Zn (-4 kJ/mol) [146], hence, CaBMG4 has both the highest Zn content and the highest glass
transition temperature. However, due to these higher glass transition temperatures the size of
the supercooled liquid region in these alloys is reduced (16 - 24 °C) compared to those with
higher Ca-content (35 - 50 °C), as the temperature of the primary crystallisation reaction in all
of these Ca-based BMGs appears to be less affected by alloy chemistry, occurring at similar
The Ecorr of the six Ca-BMGs were found to be centred about ~-1.4 VSCE, which is nobler than
the known Ca65Mg15Zn20 BMG and almost comparable with Mg-rich BMGs. Existing
electrochemical data on Ca65Mg15Zn20 by Morrison et al. [130], obtained Ecorr of -1.54 V, whilst
Zberg et al. [9] obtained -1.28 V for Mg66Zn29Ca5. Due to the varying experimental parameters,
Mass loss data (Figure 4.5a) revealed that, depending on composition, mass loss rate varied
from 2.2 to 25.1 g/cm2/day. This is a rather wide window, which suggests that the rate of
correlation between the corrosion rate obtained by mass loss and PDP tests (Figure 4.7). Both
tests were considered essential, since one yields a ground truth of the physical damage,
whereas the other reveals the electrochemical characteristics that indicate the mechanistic
The Ca-BMGs herein were designed to contain high Zn content in order to increase their
corrosion resistance, and this has been largely successful. Compared with Ca65Mg15Zn20, which
disintegrated within 3h after exposure to simulated body conditions [48], we have been able
to maintain a Ca-rich BMG content, with significantly improved corrosion resistance, surviving
pertinent to note that the corrosion behaviour of Ca and Mg vary rather significantly, with Ca
being much more reactive than Mg [24]. A correlation between the Zn content and mass loss
rate was observed, whereby reduced corrosion rate is accompanied by a higher Zn ratio
(Figure 4.5). However, amongst the alloys studied, for a given Zn concentration, the Ca content
varied by as much as 5 at.%, giving some scatter of the data which was useful in highlighting
the impact of the Ca:Mg ratio. CaBMG1 (Ca57.5Mg15Zn27.5) and CaBMG2 (Ca55Mg17.5Zn27.5), both
with 27.5 at.% Zn, revealed a notable difference in mass loss rate (Figure 4.5).
To better understand the mechanistic aspects associated with the rapid dissolution rates of the
alloys, PDP testing provided an interesting insight. Whilst Ca-BMGs reveal corrosion potentials
in the range of ~-1.4 VSCE, the currents realised in the vicinity of Ecorr are high. Such high
dissolution rates for Ca-rich alloys have also been observed in crystalline Mg-Ca alloys by
Kirkland et al. [24]. Compared to Mg, Ca is capable of undergoing very high rates of anodic
dissolution resulting in rapid corrosion. Conversely, Zn additions also raise the ability to sustain
cathodic reactions, since Zn is a more potent cathode than Mg. As such, the elevated Ca and Zn
A close inspection of the polarisation curves reveals that the cathodic reaction kinetics are
comparable for the various Ca-BMGs. However, it was revealed that while the Ca-BMGs exhibit
significantly higher rates of corrosion compared with pure crystalline Mg, they all evolve
hydrogen to a similar extent. This is believed to be a result of the cathodic reaction at the Ecorr
of the Ca-BMGs that includes both the oxygen and water reduction, with hydrogen only
The inset of Figure 4.6 reveals that the main difference between alloys is the anodic reaction
kinetics. It is clear in this figure that high Ca content results in enhanced anodic dissolution. As
such, the overall customisation of dissolution to lower levels can be achieved by decreasing
and increasing the Ca and Zn content, respectively. However, this strategy necessitates the use
of Zn, which results in higher cathodic kinetics than Mg – effectively increasing the corrosion
rate of the Ca-BMGs. The overall corrosion rate is a combination of anodic and cathodic
reactions, such that whilst biocorrosion rates can be customised via compositional variations,
there may be a practical limit on its implementation. It is very interesting to note, however,
that Ca-BMGs represent a unique niche as far as biocorrosion of metals is concerned. These
alloys support high rates of anodic dissolution (owing to the Ca), but evolve relatively low
levels of hydrogen (owing to the relatively noble potential versus crystalline Ca/Mg/Zn alloys).
Such alloys may be suitable as implants that require only very short functional lifespan (even
shorter than those of crystalline Mg) without the hydrogen evolution penalty associated with
The corrosion morphology of these unique Ca-BMGs exhibit similarities to that of dealloying.
The Ca-Mg-Zn BMG contains reactive Ca and highly noble Zn, a prerequisite in the dealloying
mechanism [147]. The samples removed from the immersion experiments revealed a network
of porous corrosion product (Figure 4.9b). Furthermore, EDS mapping showed a significant
depletion of Ca in the porous corroded layer (Figure 4.10), leaving remnants of Mg and Zn. This
is synonymous to the dealloying process reported by Erlebacher et al. [148], in an Ag-Au alloy,
whereby the more reactive element (Ag) dissolves leaving a nanoporous sponge made almost
entirely out of the nobler element (Au). Unlike a crystalline alloy, Ca-BMGs are monolithic and
single-phase with no grain boundaries for preferential corrosion to take place. Instead, the
dealloying process occurs at the surface of the sample, and spreads in all directions by uniform
corrosion (Figure 4.9c). The formation of a nanoporous sponge facilitated further dealloying by
allowing electrolyte to penetrate into the material, eventually transforming the Ca-BMG from
Above a critical concentration, Zn has been reported to be responsible for the formation of a
passive layer on the corrosion surface of Mg-Zn-Ca metallic glasses, thereby retarding further
corrosion [9]. However, the interfacial SEM analysis on the present Ca-rich alloys did not reveal
the presence of a passive layer. Hence, it appears that the critical Zn threshold reported by
Zberg and co-workers is specific to the Mg-rich BMG system and does not translate to Ca-rich
BMGs.
Compared with crystalline magnesium alloys and the Ca65Mg15Zn20 BMG, which are both under
investigation as bioresorbable biomaterials, the Ca-rich BMGs with high-Zn content reported in
this thesis showed reduced hydrogen evolution (compared to crystalline Mg and Ca65Mg15Zn20)
and increased corrosion resistance (compared to Ca65Mg15Zn20). This indicates that the Ca-rich
BMGs show promise in regard to their suitability as bioresorbable materials for implants.
However, this study has assumed the biocompatibility of these Ca-rich BMGs based on the
reported cytotoxicity studies on alloys within the Ca-Mg-Zn ternary system. Alloys with higher
(i) Ca content [48], (ii) Mg content [47] and (iii) Zn content [9] have all reported no signs of
inflammation or adverse cellular reactions. Nevertheless, the dissolution rate of these unique
alloys differ from those reported, and this could lead to variations in ion levels in solution, pH,
hydrogen evolution, and corrosion products, which are known parameters the cells are
sensitive to [93]. Further in-vitro and in-vivo studies are needed for determining the
microstructural effects that may affect the cell adhesion to the material surface.
The viscosity of the BMG in its SCL can limit the complexity of the TPF processes applicable.
Low viscosity BMGs can be formed into intricate shapes with ease. Highly viscous BMGs are
[90, 91]. However, as discussed in Section 3.6, this shortens the incubation period and
thermoplastic formability.
Using a dilatometer, an insignificant load (15 cN) was applied to the sample over a range of
expansion/contraction data as the BMG is heated (Figure 4.11). Black dotted lines indicate the
first derivative (dL/dT) of the contraction/expansion, or the rate of change of sample length
with respect to temperature. The maxima and minima in the derivative plot correspond to the
inflexions in the dL/L0 plot, as well as highlight the instances where the sample begins to soften
or harden. The first inflexion in the dilatometry curves indicate the beginning of a viscosity
decrease as a result of material softening (Figure 4.11, dotted green lines), and these lines
correlate very well with the corresponding Tg of the materials, determined using DSC. As the
material softens, the application of a constant load consequently results in a contraction in the
material. The second inflexion in the dilatometry curves indicates the beginning of an increase
in viscosity (Figure 4.11, dotted red lines), which is caused by crystallisation of the material.
Unlike the first inflexion, there is an observed delay after Tx, before the second inflexion is
detected. This has been attributed to two possible causes: (1) very small crystallite
percentages have insignificant effect on the viscosity of the system, and (2) the application of a
Ca-based BMGs [102]), whereby the strengthening response of crystallisation has been
Figure 4.11: Dilatometry and DSC traces (both carried out at a heating rate of 2 °C/min) of (a)
Mg66Zn30Ca4 (b) Ca65Mg15Zn20 (c) Ca55Mg17.5Zn27.5. Black dotted lines indicate the first
derivative (dL/dT) of the contraction/expansion, or dL/L0. Dashed green and red lines
indicate the first two inflexions in the dilatometry data, which respectively correspond with
As the same stress is applied to all samples tested, the measured rate of contraction indirectly
reflects on the viscosity of the material, and hence can be used to compare their TPF ability.
The maximum rates of contraction for Mg66Zn30Ca4 and Ca65Mg15Zn20 in the SCL region were -
0.011 %/°C and -0.030 %/°C, respectively. The highest rate of contraction recorded was -
0.178 %/°C for Ca55Mg17.5Zn27.5, indicating a significantly lower viscosity compared to the other
Isothermal traces were performed within the SCL window of Ca55Mg17.5Zn27.5 the SCL, between
125-140 °C, at 2.5 °C increments (Figure 4.12a). The isothermal data obtained was analysed
using Netzsch Proteus - Thermal Analysis v5.2.1 to determine the onset and offset of
crystallisation. Subsequently, these data were plotted in Figure 4.12b, to generate the
At lower temperatures, the DSC traces exhibit an incubation period prior to the onset of
temperature increases. TPF of this material in the mid-temperature range, between 130-
135 °C, would give a reasonable timeframe to carry out the TPF processes before the onset of
crystallisation.
Figure 4.12: (a) Isothermal DSC traces of Ca55Mg17.5Zn27.5, based on which allowed the
After some initial setback forming prototype screws using Mg-rich BMG compositions, Ca-rich
superplastic formability (Section 4.6). The die used in thermoplastic forming was the same as
that used for the initial Mg-rich BMG trial. A full description of the processing facility can be
TPF was carried out at 135 °C, which according to Figure 4.12b has a 6 min window before the
onset of crystallisation. A maximum load of 1 ton was applied, turning the as-cast BMG rod
into a screw (Figure 4.13). In contrast with the previous attempt, the material has shown good
mould replication, evident by the threads which were clearly visible and uniform.
Figure 4.13: Thermoplastic forming of a prototype screw, from an as-cast BMG rod.
The present work demonstrated the potential capability of these Ca-rich BMGs, with high Zn
contents. In order to develop this idea further, a great deal of investment and modifications
would be needed to perfect the screw design and tooling, but such detailed experiments are
beyond the scope of the current thesis. Nevertheless, some recommendations are given below.
TPF of the screw body is relatively straight forward, whereas forming the head would
encounter many obstacles, for it must endure high stresses during the insertion of the screw. A
cross-pattern turned using a screw driver is unsuitable, as the force used will most likely chip
the head itself. An alternative is to utilise a hex-head. This capitalises on the high strength
nature of the BMG, allowing more force to be transferred to the body of the screw.
There are also design modifications to be made to the prototype die, currently made from
stainless steel. Tool-steel would prove to be more durable. Use of the plunger shall be
abandoned, in favour of a three-jaw chuck (with centering capability), which would be more
Aimed at improving formability whilst improving the corrosion resistance over existing Ca-
BMGs, a series of six unique Ca-Mg-Zn BMGs containing high Zn content were successfully
produced by low pressure die casting. These high Zn content Ca-BMGs demonstrated
formed into screws using the same procedure as that applied to the Mg65Zn30Ca5 (Section 3.6).
These Ca-based BMGs also facilitated investigation into how the corrosion rate varies with Zn
concentration. It was observed that minor changes to the composition of the alloys did not
temperature and SCL region of ~20 °C were found for each alloy. However, both the critical
casting size and the rate of corrosion of the alloys were dependent on composition. All BMGs
displayed significantly higher rates of corrosion compared with their Mg-counterparts (as
benchmarked with high purity crystalline Mg) and, in spite of the more noble corrosion
potentials of the Ca-BMGs, some hydrogen evolution still occurred during dissolution.
However, the amount was synonymous with the much lower corrosion rates in crystalline
more reactive than Mg). This may limit the specific applications of such alloys as
biomaterials to those where either high rates of dissolution are required or when the
suitable as implants that require only very short functional lifespan (even shorter than
those of crystalline Mg) without the hydrogen evolution penalty associated with such
rapid dissolution.
increasing the relative proportion of Zn in the alloy. Generally, the cathodic reaction
kinetics were very similar for all alloys, although the anodic reaction kinetics were
• The corrosion morphology on the Ca-BMGs was not typical of that observed in
crystalline alloys where localised pitting and microstructural related (i.e. intergranular
ASSESSMENT
5.1 Background
The development of a bioresorbable metal faces many challenges, mostly associated with its
rapid degradation process. Hence, much research attention has focused on controlling the
corrosion rate of the alloys, as improved corrosion characteristics can (1) reduce the level of
ion toxicity, (2) reduce the amount of hydrogen gas, and (3) control the loss of mechanical
Bioresorbable components are designed to corrode. This mandates the use of elements that
are not toxic to the body and excludes the commercially available magnesium alloy grades,
which contain aluminium or rare earth elements; neither of which are safe for in-vivo use. In a
bid to increase the corrosion resistance of the material without compromising its
system. These alloys contain only elements pre-existing inside the human body and
reactions [9, 47, 48] (Section 3.5.3). Additions of Zn and Ca to the crystalline Mg, has been
reported to modulate the corrosion resistance and mechanical properties of the alloy [22]. But
their solid solubility in the Mg matrix is limited to ~2.4 at.% and 0.82 at.% respectively [50].
material, but only a limited number of compositions have been studied to date. Gu et al. [47]
assessed the corrosion behaviour of Mg66Zn30Ca4 and Mg70Zn25Ca5 BMGs against biomedical
Mg alloys. In comparison, the BMGs showed a reduced rate of corrosion that was attributed to:
(1) chemical homogeneity, (2) absence of a second phase, and (3) Zn content exceeding its
solid solubility in the Mg matrix. These attributes highlight the advantages of amorphous alloys
Working with a range of BMG compositions, Mg60+xZn35-xCa5 (0 ≤ x ≤ 7), Zberg et al. [9]
observed that Mg72Zn23Ca5 showed active corrosion behaviour. However, with increasing Zn
content, >28 at.%, Mg60Zn35Ca5 and Mg66Zn29Ca5 exhibited passive corrosion characteristics.
This was attributed to the formation and stabilisation of a zinc- and oxygen-rich surface layer,
which they confirmed via energy dispersive x-ray spectroscopy (EDS). It was also observed that
such high Zn content resulted in significant decrease in hydrogen evolution (Figure 5.1a) and
Figure 5.1: (a) Hydrogen evolution measurements in Mg-based BMGs showing a distinct drop
between the Mg-BMGs, showing active and passive (red arrow) behaviours [9].
In the Mg-rich system, Ca content ranging from 4-6 at.% have been reported to be glass
formers [47, 99], yet there is a severe lack of corrosion data on these systems. In the present
work, ten Mg-rich BMG compositions were synthesised from the Mg-Zn-Ca ternary alloy
system (Table 5.1). The compositions were selected for generating data aimed at studying the
compositional effect on corrosion. Subsequently, the data collected on these alloys were
combined with those of the Ca-rich BMGs (Chapter 4), allowing for a comprehensive corrosion
A mixture of pure elements and master alloys, namely magnesium (99.85 wt.%), zinc (99.995
wt.%), calcium (99.8 wt.%) and Mg-Ca master alloy (99.8 wt.%) were used to form the nominal
compositions listed in Table 5.1. The impurity concentrations in the magnesium, zinc, and
Mg/Ca master alloy can be found in APPENDIX A, APPENDIX C and APPENDIX D, respectively.
Ca-rich and Mg-rich BMGs were produced by melting the constituent elements in an induction
furnace under high purity argon atmosphere (99.997 vol.%), inside a boron nitride coated
graphic crucible, followed by gravity wedge casting and injection casting into copper moulds.
The amorphous nature of the alloys was confirmed by X-ray diffraction (XRD, Philips X-Pert
Materials Research Diffractometer) with a Cu-Kα x-ray source (λ=1.541 Å). The glass transition,
crystallisation, melting and liquidus temperatures were determined using DSC (Netzsch DSC
204 F1 Phoenix). Scans were performed at a heating rate of 20 °C/min in an alumina crucible in
Table 5.1: Summary of composition, glass transition (Tg), crystallisation (Tx), melting (Tm),
liquidus (Tl) temperatures, and critical casting thickness (Zc) for the BMGs synthesised for this
study. The thermoplastic forming region (ΔT) is calculated by subtracting Tg from Tx (all
# Composition Tg Tx Tm Tl ΔT Zc (mm)
1 Mg72Zn24Ca4 84 115 335 420 31 1.1
2 Mg70Zn26Ca4 86 126 335 398 39 2.8
3 Mg68Zn28Ca4 84 135 331 393 51 4.0
4 Mg66Zn30Ca4 89 140 336 391 52 3.5
5 Mg64Zn32Ca4 89 142 335 436 53 0.7
6 Mg70Zn24Ca6 120 141 352 442 21 3.0
7 Mg68Zn26Ca6 128 151 336 406 23 3.3
8 Mg66Zn28Ca6 129 155 334 410 26 2.8
9 Mg64Zn30Ca6 131 160 335 412 29 2.7
10 Mg62Zn32Ca6 138 169 338 424 30 1.5
*Tg values quoted here are the onset temperature of the glass transition.
To allow for comparison between the Ca-rich and Mg-rich BMGs, the same electrolyte,
preparation procedure, and corrosion test protocol were used. These details can be found in
(Section 4.3).
Variation in alloy composition will not only impact corrosion, but it also affects the glass
forming ability (GFA), which effectively limits the sample size attainable. As such, the exposed
surface areas of the samples used in PDP experiment were not uniform, and this necessitated
an area correction for the specimens in order to present data as current density. The exposed
surface areas were calculated by digital image processing using ImageJ software, allowing
BMG samples were cold mounted in epoxy, then polished using 1200 grit SiC paper to obtain a
uniformly smooth surface. These polished BMG samples, with an average surface area of 26
mm2, were used as the working electrode. All experiments were carried out inside the CO2
incubator. Henceforth, all potentials quoted herein are in reference to SCE, in MEM at 37 °C
and a pH of 7.4. A minimum of five separate scans were performed for each alloy to ensure
reproducibility.
intermetallic phases may occur, resulting in non-uniform corrosion. Hence, XRD was again
used to detect the possible presence of crystalline phases. Figure 5.2 shows XRD traces of the
Mg-rich BMGs. There were no detectable crystalline diffraction peaks and scattered signals
were observed around 38° and 66°. These scattering angles are in agreement with other
studies carried out on similar compositions [32]. Together, the lack of crystalline peaks and the
signal scatter confirmed the amorphous nature of the samples. The glassy nature of these
samples was further verified using DSC (Figure 5.3), where the glass transition phenomenon
was observed. The glass transition temperature (Tg), along with other thermophysical values
In this composition range, the glass transition temperature (Tg) and the first crystallisation
temperature (Tx) both increased with decreasing Mg (increasing Zn) content. Although the Ca
content is low, 4 at.% and 6 at.% respectively, it has a pronounced effect on the critical casting
thickness (Zc) of the samples. This is in strong agreement with the literature, whereby the
addition of Ca is said to form Ca-centred clusters, enlarging the atomic size distribution range
and creating a uniform atomic size separation [32]. Both of these would increase atomic
packing efficiency, thus improving the GFA of these alloys. Hence, small variations in alloy
Figure 5.2: XRD traces for 10 Mg-rich BMGs synthesised for this study.
Figure 5.3: DSC traces for Mg72-xZn24+xCa4 and Mg70-xZn24+xCa6 (x=0, 2, 4, 6 and 8).
PDP scans yielded mechanistic information on the corrosion characteristics of the samples (i.e.
relative rates of anodic or cathodic kinetics), and distinct differences were observed between
the Ca-rich and Mg-rich BMGs. The Ecorr of the Ca-rich BMGs (>50 at.% Ca) ranged from -1.44 to
-1.39 VSCE (Figure 5.4c), which are notably less noble than the Mg-rich BMGs (-1.34 to -1.19 VSCE)
(Figure 5.4a, b). The variation in corrosion potential was attributed to the significant difference
in the Ca content, which is the most reactive species in the alloys [24], as shall be discussed
Analysis of the PDP curves revealed signs of passive corrosion kinetics in some alloys (most
evident by the kink in the current response which can be interpreted as a breakdown-type
event). The passivation and subsequent breakdown behaviour was observed in the Group I
alloys containing lowest Ca content, 4 at.% (Figure 5.4a), and Group II alloys containing 6 at.%
Ca (Figure 5.4b). However, none of the Group III alloys, with > 50 at.% Ca, displayed any signs
Figure 5.4: Representative polarisation curves of Mg-rich, (a) MgZnCa4 (b) MgZnCa6, and (c)
Highest Zn content did not necessarily result in the most notable passive corrosion behaviour,
which again reinforces the complex role that each element plays in the amorphous metallic
glass. Although Group II (6 at.% Ca) revealed the only passivation/breakdown evidence on the
highest Zn containing alloys, 30-32 at.%, this trend was not preserved for Group I alloys. With
only 4 at.% Ca, passive behaviour was more pronounced in systems containing less Zn content,
namely Sample#1 (with 24 at.% Zn). Then, with monotonic increases in Zn, the passivation
region nearly disappeared at 30 at.% Zn, but a greater addition appears to have brought it back
(at a great expense in enhanced cathodic kinetics and hence increased icorr).
In this study, a much wider compositional window was analysed, encompassing both Mg-rich
(Chapter 5) and Ca-rich BMGs (Chapter 4). The author’s previous corrosion study encompassed
both PDP and weight loss methods for measuring corrosion rates, whereby a strong correlation
was observed between the two methods (Figure 4.7). Hence, in the present study, the icorr
values, which are determined from Tafel-type analysis, were used to represent the dissolution
Table 5.2: Summary of icorr and Ecorr of the alloy compositions used in the study (error refers
Bioresorbable BMGs are designed to corrode. In neutral pH conditions, Zn is the only element
in the Mg-Zn-Ca ternary system that does not dissolve vigorously [107], and the general
strategy has been to supersaturate such alloys with Zn, in order to retard dissolution rates.
Figure 5.5 shows the effect of each of the three elements on the corrosion rate of the alloys. It
was observed that the degradation rate of the samples tested had a significant sensitivity to
both the Ca and Mg content. The 4 at.% Ca samples corroded at a rate as low as ~1.1 μA/cm2,
whereas the minimum icorr for the Ca-rich samples was ~30 μA/cm2. The corrosion rate
increased significantly with increasing Ca content. Between 24-32 at.%, the variation in Zn
content was relatively consistent compared with the variations in Mg and Ca. As such, a
trend of Mg effect on corrosion was the inverse to that of Ca – significantly reduced the
corrosion rate.
It was found that the exact role of Zn was unclear and largely negligible in reducing the rate of
corrosion. Nevertheless, its presence was concomitant with some beneficial performance. A
comparison within each alloy group showed that a monotonic increase in Zn did not warrant a
concluded that Zn content >28 at.% exhibited low dissolution rates. Indeed, in alloy group I, a
reduction in icorr was observed with increased Zn content in the alloys, from 24-30 at.%.
However, contrary to this trend, in alloy groups I and II, Sample#5 and Sample#10 both
contained 32 at.% Zn, but the icorr of these samples were amongst the highest in their
respective groups (Table 5.2). The overall dissolution rate is a combination of cathodic and
anodic reactions, both of which are affected by the alloy composition. Although the effect of
Zn addition was more pronounced on the anodic reactions (Figure 5.4) showing evidence of
slowing dissolution, it also affected the cathodic efficiency, effectively increasing the corrosion
rate whereby the cathodic efficiency of Zn is greater than that of Mg or Ca [140]. The net result
of such increases in cathodic kinetics was counter to any passivating effect. Figure 5.5
indicated that Zn has a beneficial impact, but not as dominant as Mg. This figure highlights the
Figure 5.5: The sensitivity of alloy corrosion rate to individual elements, (a) Mg (b) Zn (c) Ca,
Mg-rich and Ca-rich BMGs based on the Mg-Zn-Ca ternary system were successfully produced
by die casting. Potentiodynamic polarisation tests were carried out on these alloys, in
simulated body environment, aimed at exploring the corrosion behaviour of these two classes
of promising bioresorbable implant materials. The major findings of this study include:
• The concentration of the most reactive species, Ca, had a key impact on the corrosion
rate of the alloy, resulting in enhanced dissolution. The trend in Mg was the inverse to
• Zn played a beneficial role in moderating the corrosion resistance of the alloy, but the
variations in Zn content affected both the anodic and cathodic reaction kinetics, and
moderate the corrosion rate of this class of alloys. This passivation behaviour was only
observed in Mg-rich but not Ca-rich BMGs. In neutral pH, Zn is the least reactive
element in the alloy. However the experimentally observed passive region in the BMGs
studied was not solely determined by the Zn concentration. This highlights the
complex interactions that exist between the elements in this compositional window.
• It is appreciated that PDP tests and Tafel analysis alone do not allow for a
comprehensive in-vitro screening to occur. However, such tests have allowed some
key mechanistic aspects to become apparent whilst allowing for screening of multiple
6.1 Background
Coatings are commonly applied to substrates to modify its surface properties such as - wear
applications, coatings can also mediate the cellular interaction between the host and the
biomaterial. Hydroxylapatite (HA) is commonly used to coat dental and orthopaedic implants.
It encourages the ingrowth of natural bone, which acts as a mean of implant fixation. By
applying HA as a coating onto a substrate, one can obtain the mechanical integrity of the
Bioresorbable alloys based on the Mg-Zn-Ca ternary system have been tested in-vitro and in-
vivo, without signs of inflammation or adverse cellular reactions [9, 49]. The release of
essential ions have also been shown to aid in the process of tissue regeneration, wound
healing, and inhibit bacterial growth [1, 4, 12]. A potential bioresorbable MgZnCa coating will
enjoy all of these advantages. Its controlled degradation can be utilised to temporarily modify
the surface properties and mechanical properties of the substrate, as well as potentially
In a collaborative effort to combat chronic ulcers, it has been proposed that a three-
growth while encouraging cell migration, extracellular matrix formation and revascularisation
in a manner that accelerates wound healing. This would be especially beneficial to the chronic
ulcer patients, who have limited wound healing ability, making them extremely susceptible to
infections. Central to this idea are the successful deposition of a bioresorbable coherent
metallic glass coating, which would modulate the release of antibiotics and temporarily modify
There are numerous reports of BMG coatings deposited by vacuum plasma spraying (VPS).
However, the deposited film was not completely amorphous. The crystallisation during the VPS
process was deemed to be a result of the plasma gas composition and the in-flight particle
melting state [151, 152]. In the search for alternative deposition processes, two physical
vapour deposition (PVD) techniques are of interest to depositing BMGs. These are sputtering
and pulse laser deposition (PLD). Sputtering functions by the momentum exchange between
the colliding ion and the target, depositing materials mechanically. PLD utilises high energy
photon pulses to cause localised vaporisation in the target, creating a plume, directed at the
substrate. One of the biggest advantages of PLD is that it can deposit films of the same
stoichiometric ratio as the target, which is an important aspect in the deposition of BMGs.
Alloys based on the Mg-Zn-Ca ternary system offer great promise as a bioactive, bioresorbable
biomaterial. Considering the findings from the previous chapters, on Mg-rich BMGs (Section
3.7) and Ca-rich BMGs (Section 4.8), especially concerning their corrosion performance,
Mg65Zn30Ca5 was selected to be the composition of choice to explore thin film metallic glasses
(TFMG). Not only did Mg65Zn30Ca5 release half the amount of Mg ions as crystalline Mg
(Section 3.5.2), cytotoxicity assessment also found it to support cellular activities (Section
3.5.3). A known cellular response weighed favourably here, as the resultant coating aims to be
implanted in-vivo.
As a member of the multi-disciplinary team, specialising in the materials synthesis, the aim of
this part of the thesis was to explore various thin film deposition techniques, then characterise
and compare the resultant films. Based on these results, preferred deposition technique can
be selected and utilised in the controlled film deposition onto the antibiotics infused
polycaprolactone (PCL) scaffolds. Once coated, the subsequent characterisation of the system
(e.g. the effect of the coating and sterilisation technique on scaffold properties such as pore
size, solubility, and mechanical strength) are beyond the scope of the present thesis, and shall
Pulsed laser deposition (PLD) is a physical vapour deposition process widely used in the thin
film deposition of insulators, semiconductors, metals, polymers, and even biological materials
[153]. Unlike sputtering, which bombards the target surface with ions, PLD utilises high energy
laser to cause localised vaporisation in the target, creating a plume, directed at the substrate,
followed by film growth on a substrate. These steps are repeated thousands of times in a
typical deposition run. One of the biggest advantages of PLD is that it retains stoichiometric
ratio, which is of great importance in the deposition of BMGs. This is the result of the non-
equilibrium nature of the ablation process itself, absorbing high laser energy density in a small
PLD has been used to create pure crystalline HA coating for biomedical purposes. Methods
such as plasma spraying, sputtering have also been explored in the deposition of HA, but the
resulting coating is often a mixture of HA and other phases. Cotell et al. [154] used PLD to
deposit thin films of hydroxylapatite on Ti-6Al-4V substrate, a common titanium alloy used for
surgical implant applications., and was able to obtain a thin film of pure crystalline HA,
The Mg65Zn30Ca5 plate target (Figure 6.1b) was ablated using a focused beam of KrF laser
(345 mJ, 3,410 mbar, 27.6 kV), delivering 40,000 laser pulses at 10 Hz, under ~5×10-8 torr
chamber pressure. The laser spot size was approximately 3 mm × 1 mm, giving a laser fluence
of 11.5 J/cm2. Since ejection of material from the target during the process of laser ablation is
directed primarily perpendicular to the target surface, the target was irradiated with the laser
beam incident at an angle of 45°. The substrate and target surface were kept parallel to each
other, separation between the two being 11 cm. This geometry minimises the extent of spatial
overlap between the incident laser beam and the plume emanating from the target surface,
The aim was to deposit a nano-layer of MgZnCa film in a controlled manner onto PCL scaffold.
However, PLD of BMGs thus far remains unexplored, possibly due to the high cost associated
with it. Hence, the more important questions concerning the author were: (1) the crystallinity
of the film, (2) chemical composition, and (3) surface morphology. To assess these parameters,
In these experiments, as-cast Mg65Zn30Ca5 BMG was used as the target for depositing on a
variety of substrate materials. The as-cast BMG in both wedge and plate (3 mm thick)
geometries were experimented as targets. The wedge was sectioned and cold mounted in two
parts epoxy (Φ = 26 mm), providing a parallel surface with the target stage (Figure 6.1a),
whereas the BMG plate was ground into Φ = 26 mm disc, again with parallel top and bottom
Figure 6.1: PLD targets prepared from as-cast Mg65Zn30Ca5 (a) wedge and (b) plate.
Preliminary trials using the two targets showed that the epoxy mount is undesirable. The
incident laser inputs a significant amount of heat into the target, in order to vaporise it. The
epoxy, in which the target (wedge) is embedded, severely hindered its ability to dissipate this
heat away. A typical deposition spans over 30 min, and during this time sufficient heat is
inputted into the amorphous target causing partial crystallisation (Figure 6.2). Hence, based on
the preliminary trial, only the target made from the as-cast plate was used for further
exploratory experiments.
Figure 6.2: XRD scan taken before (black) and after (red) being ablated by laser. Data of the
red curve was recorded using mono-capillary attachment to focus the x-ray beam.
After deposition, the thin film was analysed by scanning electron microscopy (SEM, S3400,
Hitachi). Figure 6.3 shows the globules observed on the substrate surface. Subsequent energy
dispersive x-ray spectroscopy (EDS) data reveals that the substrate surface is covered
uniformly by a combination of Mg, Zn and Ca, and that these globules were also composed of
these elements.
Figure 6.3: Secondary electron image viewed at 50° tilt, showing the presence of globules
Although poor beam quality can be the cause of undesirable droplet formation [149], this was
not believed to be the case in this work. The formation is in principle unavoidable during
deposition in ultra-high vacuum (UHV) from solid targets (hydrodynamic sputtering), because
cone-like structures pointing in the direction of the laser beam gradually develop on the target
surface when illuminated under a constant angle [153]. These surface globules were also
observed by Krebs and Bremert (Figure 6.4), who found the number of globules to increase
with increasing laser fluence, and is also dependent on the target material, its surface
roughness and strongly increases with the number of pulses on the same target location [157].
In addition, the BMG itself has high viscosity, hence a high surface angle is also expected
compared to other metals if held in the liquid state for any ongoing period of time.
Figure 6.4: Scanning electron micrographs of (left) globules observed on the MgZnCa film
surface, deposited using 11.5 J/cm2; and (right) globules observed by Krebs and Bremert,
To obtain a completely droplet-free film, a mechanical velocity filters may be used between
the target and substrate, which allows the fast atoms and ions to reach the substrate but
removes the slower droplets with velocities of <200 m/s during their transit to the substrate.
whereby the two colliding laser ablation plumes interact with each other and the gaseous
species are redirected toward a substrate placed outside the directional path of either plumes.
The heavier macro-particles continue in their original trajectories and consequently will not
reach the substrate. However, it should be noted that both solutions will further reduce the
deposition rate.
Further investigation was carried out using transmission emission microscopy (TEM). A dual
beam (XT Nova Nanolab 200), combining a high resolution focussed ion beam (Ga FIB) and a
high resolution scanning electron microscope, was used to prepare the TEM foil specimens.
The TEM specimen was selected to contain a partial globule (Figure 6.5), aimed at yielding
Figure 6.5: TEM foil sample machined using focused ion beam, containing a partial globule.
Elemental Mapping
The standard procedure for ion milling firstly involves applying a gold (Au) coating to the entire
sample, aiding in ion conduction. Secondly, a platinum (Pt) coating is applied to the localised
area to be milled by gallium (Ga) ions, thus protecting the underlying features. Subsequent
TEM analysis of the film showed these Au and Pt coatings were still intact (Figure 6.6). This
suggested that ion milling did not damage the underlying film, and the relative size of the film
compared to the globule observed was unaffected by the preparation and/or imaging process.
EDS carried out on the TEM foil showed the elemental composition in the globule is
homogeneous, and identical to the film (Figure 6.6). Possibly due to the high laser fluence used
during PLD deposition [157], the formation of globules was energetically favoured, resulting in
Figure 6.6: Elemental mapping carried out on the TEM foil sample.
Deposition Rate
To determine the deposition rate, film thickness was measured by TEM, and subsequently
calculated based on the deposition duration. Analysis of the TEM image (Figure 6.7) estimates
the film thickness to be ~100 nm, which took 60 min to prepare, under the deposition
Figure 6.7: TEM image estimating the BMG film thickness to be 100 nm.
Crystallinity
The film crystallinity was investigated using convergent beam electron diffraction on the TEM
(Figure 6.8). Both the globule and the film generated a lack of diffraction pattern, indicative of
Figure 6.8: Convergent beam electron diffraction of both the globule and thin film showing
Sputtering is another PVD technique widely used to modify the surface properties of a material.
It is commonly used to deposit gold, chromium and carbon coatings in electron microscopy
sputtering can be used to deposit TiN, hydroxyapatite and bioactive glass coatings to improve
wear resistance, surface biocompatibility and surface activity of the biomaterials respectively.
Sputtering is unique for its mechanism. It uses inert gas, typically argon, at low pressure. The
ionised Ar+ ions accelerate towards the cathode, bombarding the target surface, transforming
the target into the vapour phase by mechanical (momentum exchange) rather than chemical
and/or thermal processes [158, 159]. This is called direct current (DC) sputtering. It
the sputtering target. Virtually all BMGs may be deposited using this technique.
to trap the charged particles close to the target surface, increasing the ionisation efficiency.
Detailed explanation on the force exerted by the magnetic field on a particle with an arbitrary
BMGs contain multiple elements (multi-component). To deposit a metallic glass film of the
same composition imposes great technical challenges on the sputtering target. This is because
ion bombardment of a single multi-component target can lead to surface enrichment of the
component having a lower sputtering yield [161, 162]. Therefore, to achieve the desired film
composition, some research groups have explored combinatory deposition (multiple targets)
[163, 164] (Figure 6.9), others attempted to use a rotating target disc with constituent
elements occupying an area that reflects on the final composition [165, 166].
Figure 6.9: 3-source configuration for the combinational sputter deposition of ZrCuAl
crystallinity of the sputtering target appears to have little effect. Kondoh et al. [167] deposited
ZrAlNiCu thin film utilising RF sputtering method using amorphous and crystallised
Zr55Al10Ni5Cu30 targets, and an elemental composite target composed of Zr, Al, Ni chips and Cu
plate. XRD analysis showed that the films were amorphous. Both films deposited by
amorphous and crystallised targets contain higher fractions of Al and Ni than that of the
respective targets. This was accredited to the variation in sputtering yield and preferential
An in-house DC magnetron sputtering assembly was used to deposit thin film metallic glass
• Pressure transducer (Baratron type 127, MKS Instruments, USA) and digital readout
• Thermocouple vacuum gauge (MDC, USA) and gauge controller (senTorr BA2, Varian,
USA)
• Thermal conductivity gauge (Pirani PR25-K, Edwards, UK) and controller (Pirani 14,
Edwards, UK)
• Mass-flow controller (Type 1179, MKS Instruments, USA) and digital flow meter (Type
A BMG target, with composition of Mg65Zn30Ca5, was cast into a plate mould (thickness = 3 mm)
using the technique described in Section 3.3. The BMG plate was subsequently ground into a
25.4 mm (1”) diameter disc, the size of the corresponding magnetron gun.
All depositions took place under highly controlled conditions, after pre-sputtering of the target
to remove surface oxides. The chamber pressure remained at 2 Pa, with a constant Ar flow
rate of 2 SCCM (standard cubic centimetres per minute). These parameters were held constant,
and used in all subsequent depositions, because when the Ar pressure increases, the
corresponding number of Ar+ ions increases, thereby increasing the rate of sputtering [168]. A
range of power options were explored to map out its effect on deposition rate. However, due
to the small target size, the maximum applied power was limited to 50W, to prevent target
overheating.
The aim was to deposit a MgZnCa film in a controlled manner onto PCL scaffold. Based on the
existing literature on BMG sputtering, the present work focused on: (1) crystallinity of the film
and, (2) deposition rate. To assess these parameters, a range of substrate materials were used,
It is known that the sputtered film thickness is dependent on the sputtering power, duration,
and colliding gas (Ar) pressure. Hence, a series of depositions were carried out using two
identical targets, and the subsequent film thickness measured using stylus profiler (Dektak 2A,
Bruker, Germany). The deposition rates from both targets showed strong linear correlation
with the applied power (Figure 6.11). Nevertheless, small discrepancy was observed between
the deposition rate of these two targets, and this was believed to be due to the variability in
Figure 6.11: Sputter deposition rates of two identical amorphous targets (square and
diamond), each showing a strong linear correlation with the applied power.
Post-deposition examination of the film surface (data not shown), indicated that depositions
carried out at higher sputtering power resulted in smoother film surface. Having established
the power dependent deposition rate, the film thickness could also be estimated based on the
known parameters.
Films deposited via sputtering were significantly thicker compared to that of PLD. As such,
obtain a peak-less XRD trace by depositing MgZnCa film on a fused quartz, it is not conclusive
enough to indicate the amorphous halo was solely due to the film. Therefore (100) Si chips
were used, which yield characteristic peaks. Its XRD trace before and after MgZnCa coating can
be observed in comparison (Figure 6.12). The figure clearly showed the amorphous halo,
Figure 6.12: Comparison between the coated and uncoated Si (100) substrates clearly
The PLD deposition chamber operates in ultra-high vacuum (<10-7 Torr), as such a loading dock
chamber is in place, whereby the sample is first pumped down from atmospheric pressure
(roughing). The sample, together with its backing plate, is then transferred into the main
deposition chamber, by use of a loading arm, remaining under vacuum the entire time. This
loading mechanism (specific to the existing PLD system) for substrate insertion poses a
Nevertheless, after a lengthy 1 h deposition, using the deposition parameters stated in Section
6.3.1, a film of ~100 nm was successfully applied to the PCL scaffold (Figure 6.13).
Figure 6.13: PCL scaffold successfully coated using PLD after 60 min.
In contrast, a sputter-coated sample has much larger size tolerances, and can be coated at a
much faster rate. The vacuum achieved in sputtering is in the order of 10-2 Pa, which is lower
than PLD. However, this also means the use of a loading dock can be avoided, lifting the
sample size restriction. In addition, having direct access to the deposition chamber allows
efficiency.
Sputtering is also a much faster deposition method. Within 5 min, a ~1 μm film can be
achieved (Figure 6.14). Conversely, with such a rapid deposition rate, relatively speaking, the
film thickness is more difficult to control compared to PLD. In theory it is possible to reduce the
sputtering power, such that a lower deposition rate would render the final film thickness
easier to control. Unfortunately, attempts to do so have led to poor film surface finish, and it is
For the purpose of coating substrates with MgZnCa metallic glass thin films, sputtering proved
to be both faster and more economical than PLD. PLD operates under ultra-high vacuum (UHV)
and, hence, has a longer pump down cycle, and samples can only be inserted into the
deposition chamber via a loading dock. This limits the maximum thickness to ~5 mm. In
contrast, magnetron sputtering can achieve deposition rates three orders of magnitude higher
than PLD. In addition, having direct access to the deposition chamber allows the substrates to
be manually placed, maximising the number of samples coated per pump down cycle.
In the field of bioresorbable metals, the enthusiasm is high, and knowledge is low. Although
regulatory requirements on bioabsorbable polymers and bioinert metals are clear, the
regulations and standards for testing bioresorbable metals are almost non-existent. To avoid
biomaterials, one can only expect more stringent scrutiny from all stakeholders before the
The aim of this thesis was to assess the potential of the newly discovered amorphous Mg-Zn-
material that satisfies all of the above requirements, which is the motivation behind the
current work. Mg-rich and Ca-rich regions within the Mg-Zn-Ca ternary system were
systematically explored, aimed at enhancing their physio-chemical properties. This led to the
rise in solution ion levels, progressively as the material degraded. This enabled the study and
comparison of data collected across various research groups, who may use different protocols,
and may not be otherwise comparable. The cytocompatibility study showed that the
amorphous Mg65Zn30Ca5 exhibits far superior corrosion characteristics than its crystalline
counterpart, with dissolution rate of Mg ions in the BMG being 446 μg/cm2/day, approximately
A cytotoxicity study was also carried out in an identical biocorrosion environment to assess the
effects of metal degradation on the cell viability. The study revealed that over 72 h culture
period, the cell viabilities remained relatively high for both the amorphous and crystalline
materials, indicating that they are capable of supporting cellular activities. However, direct
contact with the samples created regions of minimal cell growth around both types of samples,
Over the course of this thesis, numerous reports on the biocompatibility of the Mg-Zn-Ca
ternary alloy system, or a subset of, have been published. Alloys with higher Mg content, Zn
content, and Ca content, which also corrode at a faster rate than the amorphous Mg-Zn-Ca
metal, have all been shown to be safe for in-vivo use. Therefore, it is logical to conclude that
the levels of Mg, Zn and Ca ions from amorphous Mg-Zn-Ca metal would not pose a problem
for the body based on its low corrosion rate, which led to ions released at a much slower rate.
Ca-rich BMGs
In an effort to resolve the poor thermoplastic formability exhibited in Mg-rich BMGs, a series
of six unique Ca-rich BMGs with higher Zn content were synthesised to maintain superior
thermoplastic formability whilst improving the corrosion resistance over traditional Ca-BMGs.
The Ca-rich BMGs, with high Zn content, had a critical casting thickness of up to 4.5 mm.
Mechanical and thermophysical testing revealed a Young’s modulus (stiffness) of ~40 GPa.
Glass transition temperatures ranged from 119 to 129 °C, and the supercooled liquid region
In-vitro biocorrosion testing using a combination of polarisation and mass loss techniques
revealed that these alloys exhibit significantly higher rates of corrosion compared with their
Mg-counterparts (as benchmarked with high purity crystalline Mg). In spite of the much more
noble corrosion potentials of the Ca-BMGs (-1.4 VSCE) compared to the benchmarked crystalline
Mg (-1.9 VSCE), some hydrogen evolution still occurred during dissolution. However, the
amount was synonymous with the much lower corrosion rates in crystalline Mg.
Thermoplastic Formability
The thermoplastic formability of the unique Ca-rich BMGs with higher Zn content
(Ca55Mg17.5Zn27.5) was compared to the Mg-rich (Mg66Zn30Ca4) and Ca-rich (Ca65Mg15Zn20) BMGs.
techniques, it was found to have significantly lower viscosity than the others when processed
within its SCL region, which allows Ca55Mg17.5Zn27.5 to undergo complex TPF operations.
Corrosion Characteristics
Systematic investigation of both the Mg-rich and Ca-rich BMGs paved the way to understand
polarisation, a complex interaction between the elements in this compositional window was
reducing the rate of corrosion, its effect was not dominant. In addition, a form of passivation
was observed in the Mg-rich BMGs, but not in the Ca-rich BMGs, although it appears that such
In the development of bioresorbable BMGs, the general strategy has been to supersaturate
such alloys with Zn, in order to retard their dissolution rate. The results of this work could
change this approach, since it was shown that Zn did not play a dominant role in modifying the
dissolution rate, and that the passive corrosion behaviour was not determined solely by the Zn
concentration.
Generally, the cathodic reaction kinetics were very similar for all Ca-rich BMGs, although the
anodic reaction kinetics were sensitive to variations in the Ca:Mg:Zn ratio. But even with
higher Zn content, these Ca-rich BMGs underwent rapid rates of dissolution in biocorrosion
environments. This is due to the inherently reactive nature of Ca (even more reactive than Mg).
This resulted in a corrosion morphology that consistently revealed a damage form that can be
Capitalising on the bioresorbable nature and its superior corrosion resistance, Mg65Zn30Ca5 was
deposited as thin film metallic glass using sputtering and pulsed laser deposition (PLD). PLD
operates under ultra-high vacuum, meaning it takes longer to obtain satisfactory vacuum for
deposition. Additionally, with the current system setup, the samples were inverted, and can
only be inserted into the deposition chamber via a loading dock. This greatly limits the sample
geometry to a maximum thickness of 5 mm, and increases the overall duration for each cycle.
In contrast, magnetron sputtering proved to be faster, and more economical than PLD. It is
capable of achieving deposition rates three orders of magnitude higher than PLD and grants
the user direct access to the deposition chamber. This allowed the substrates to be manually
placed, maximising the number of samples coated per pump down cycle.
Over the course of this thesis, many novel discoveries with respect to these new bioresorbable
biomaterials were reported through work from this study and that of other research groups.
Currently, at the end of this 3.5 years project, the biocompatibility, processability, and
degradation mechanisms of the amorphous Mg-Zn-Ca metals are much better understood.
Although the corrosion mechanisms in amorphous Mg-Zn-Ca alloys were found to be very
unlikely to be improved further by alloy composition. The existing Ca-rich and Mg-rich BMG
from these compositions would see decreases in glass forming ability and thermal stability,
Due to the prohibitive high cost of tooling, the thermoplastic forming of screw prototypes in
this thesis served as a proof of concept. Significant progress can be made in this front, given
the facilities. The author believes that further discoveries are likely to come from the design,
With new tools, implants such as screws can be designed in consideration of the amorphous
metal properties. The implementation of flat or Philips screw drive designs should be avoided,
in favour of a hex-cap. This would capitalise on the high strength of the amorphous metals,
whilst avoiding stressing the material beyond its elastic limit. It is also recommended that the
implant designs be validated using artificial bone material such as Sawbones®, which mimics
REFERENCES
[1] Staiger MP, Pietak AM, Huadmai J, Dias G. Magnesium and its alloys as orthopedic
[2] Hench LL, Polak JM. Third-generation biomedical materials. Science 2002;295:1014-6.
[3] Wong JY, Bronzino JD. Biomaterials. Boca Raton: CRC Press; 2007.
[4] Witte F, Hort N, Vogt C, Cohen S, Kainer KU, Willumeit R, et al. Degradable biomaterials
based on magnesium corrosion. Current Opinion in Solid State and Materials Science
2008;12:63-72.
[5] Puleo DA, Huh WW. Acute toxicity of metal ions in cultures of osteogenic cells derived from
[6] Granchi D, Ciapetti G, Stea S, Savarino L, Filippini F, Sudanese A, et al. Cytokine release in
[7] Stroganova EE, Mikhailenko NY, Moroz OA. Glass-based biomaterials: presnet and future (a
[8] Kirkland NT, Staiger MP, Nisbet D, Davies CHJ, Birbilis N. Performance-driven design of
biocompatible Mg alloys. JOM Journal of the Minerals, Metals and Materials Society
2011;63:28-34.
[9] Zberg B, Uggowitzer PJ, Loffler JF. MgZnCa glasses without clinically observable hydrogen
[10] Witte F, Kaese V, Haferkamp H, Switzer E, Meyer-Lindenberg A, Wirth CJ, et al. In vivo
corrosion of four magnesium alloys and the associated bone response. Biomaterials
2005;26:3557-63.
[11] Revell PA, Damien E, Zhang XS, Evans P, Howlett CR. The effect of magnesium ions on
bone bonding to hydroxyapatite coating on titanium alloy implants. Key Engineering Materials
2004;254-256:447-50.
[12] Lansdown ABG. Zinc in the healing wound. The Lancet 1996;347:706-7.
[13] Standard specification for wrought 18Cr-14Ni-2.5Mo stainless steel sheet and strip for
surgical implants. Medical Devices and Services. Philadelphia: ASTM; 2003. p. 70-3.
[14] Standard specification for Co-28Cr-6Mo alloy castings and casting alloy for surgical
[15] Standard specification for wrought Co-20Cr-15W-10Ni alloy for surgical implant
applications. Medical Devices and Services. Philadelphia: ASTM International; 2001. p. 54-7.
[16] Standard specification for wrought 35Co-35Ni-20Cr-10Mo alloy for surgical implant
[17] Standard specification for unalloyed titanium, for surgical implant applications. Medical
[18] Standard specification for wrought Ti-6Al-4V extra low interstitial alloy for surgical implant
resistant Ti-13Nb-13Zr alloy for total hip-arthroplasty. Bio-medical materials and engineering
1994;4:231-43.
[20] McBride ED. Absorbable metal in bone surgery: A further report on the use of magnesium
[21] Zhang S, Zhang X, Zhao C, Li J, Song Y, Xie C, et al. Research on an Mg-Zn alloy as a
[23] Wang JF, Wei YY, Guo SF, Huang S, Zhou XE, Pan FS. The Y-doped MgZnCa alloys with
ultrahigh specific strength and good corrosion resistance in simulated body fluid. Materials
Letters 2012;81:112-4.
[24] Kirkland NT, Birbilis N, Walker J, Woodfield T, Dias GJ, Staiger MP. In-vitro dissolution of
[25] Gonzalez S, Pellicer E, Fornell J, Blanquer A, Barrios L, Ibanez E, et al. Improved mechanical
performance and delayed corrosion phenomena in biodegradable Mg-Zn-Ca alloys through Pd-
2007;49:1696-701.
[27] Li Z, Gu X, Lou S, Zheng Y. The development of binary Mg-Ca alloys for use as
[28] Tong LB, Zheng MY, Chang H, Hu XS, Wu K, Xu SW, et al. Microstructure and mechanical
properties of Mg-Zn-Ca alloy processed by equal channel angular pressing. Materials Science
[29] Boehlert CJ, Knittel K. The microstructure, tensile properties, and creep behavior of Mg–
Zn alloys containing 0–4.4 wt.% Zn. Materials Science and Engineering: A 2006;417:315-21.
[30] Chen XB, Kirkland NT, Krebs H, Thiriat MA, Virtanen S, Nisbet D, et al. In vitro corrosion
survey of Mg-xCa and Mg-3Zn-yCa alloys with and without calcium phosphate conversion
[31] Hänzi AC, Sologubenko AS, Gunde P, Schinhammer M, Uggowitzer PJ. Design
considerations for achieving simultaneously high-strength and highly ductile magnesium alloys.
[32] Gu X, Shiflet GJ. Mg-Ca-Zn bulk metallic glasses with high strength and significant ductility.
[33] Senkov ON, Scott JM. Glass forming ability and thermal stability of ternary Ca-Mg-Zn bulk
[34] Park ES, Kim DH. Formation of Ca-Mg-Zn bulk glassy alloy by casting into cone-shaped
[35] Laws KJ, Shamlaye KF, Gun B, Wong K, Ferry M. The prediction of glass-forming
compositions in metallic systems - the development of new bulk metallic glasses. Materials
[36] Cao JD, Laws KJ, Birbilis N, Ferry M. Potentiodynamic polarisation study of bulk metallic
glasses based on the Mg-Zn-Ca ternary system. Corrosion Engineering, Science and Technology
2012;47:329-34.
[37] Cao JD, Martens P, Laws KJ, Boughton P, Ferry M. Quantitative in vitro asessment of
Mg65Zn30Ca5 degradation and its effect on cell viability. Journal of Biomedical Materials
[38] Liu S, Huang L, Pang S-j, Zhang T. Effects of crystallization on corrosion behaviours of a Ni-
[39] Li WH, Chan KC, Xia L, Liu L, He YZ. Thermodynamic, corrosion and mechanical properties
of Zr-based bulk metallic glasses in relation to heterogeneous structures. Materials Science and
Engineering: A 2012;534:157-62.
[40] Peter WH, Buchanan RA, Liu CT, Liaw PK, Morrison ML, Horton JA, et al. Localized
corrosion behavior of a zirconium-based bulk metallic glass relative to its crystalline state.
Intermetallics 2002;10:1157-62.
[41] Wang Y, Tan MJ, Pang J, Wang Z, Jarfors AWE. In vitro corrosion behaviors of
2012;134:1079-87.
[42] Andreas K, Jörg L, Florian Dalla T. Rapid Solidification and Bulk Metallic Glasses -
Processing and Properties. Materials Processing Handbook: CRC Press; 2007. p. 17-1--44.
[44] Schroers J, Paton N. Amorphous metal alloys form like plastics. Advanced Materials &
Processes 2006;164:61-3.
[45] Schroers J, Pham Q, Peker A, Paton N, Curtis RV. Blow molding of bulk metallic glass.
[47] Gu X, Zheng Y, Zhong S, Xi T, Wang J, Wang W. Corrosion of, and cellular responses to Mg-
[48] Wang YB, Xie XH, Li HF, Wang XL, Zhao MZ, Zhang EW, et al. Biodegradable CaMgZn bulk
[49] Xie X, Wang X, Wang Y, Zhang G, He Y, Zheng Y, et al. Ca-Mg-Zn metallic glass as
[50] Villars P, Prince A, Okamoto H. Handbook of ternary alloy phase diagrams. Materials Park,
[51] Revie RW, Uhlig HH. Corrosion and corrosion control. 4th edition ed. Hoboken New Jersey:
[53] Biomaterials Science: an introduction to materials in medicine: 2nd edition. second ed.
[54] Zarzycki J, Scott WD, Massart C. Glasses and the Vitreous State: Cambridge University
Press; 1991.
[55] Jun. WK, Willens RH, Duwez P. Non-crystalline structure in solidified gold-silicon alloys.
Nature 1960;187:869-70.
[56] Chen HS. Thermodynamic considerations on the formation and stability of metallic glasses.
[57] Inoue A. Stabilization of metallic supercooled liquid and bulk amorphous alloys. Acta
Materialia 2000;48:279-306.
[58] Calka A, Madhava M, Polk DE, Giessen BC, Matyja H, Vander Sande J. A transition-metal-
[59] Egami T. Universal criterion for metallic glass formation. Materials Science and
Engineering: A 1997;226–228:261-7.
[60] Miracle DB, Senkov ON, Sanders WS, Kendig KL. Structure-forming principles for
[61] Busch R. The thermophysical properties of bulk metallic glass-forming liquids. JOM
2000;52:39-42.
[62] Greer AL. Metallic glasses…on the threshold. Materials Today 2009;12:14-22.
[64] Ediger MD, Angell CA, Nagel SR. Supercooled liquids and glasses. The Journal of Physical
Chemistry 1996;100:13200-12.
[65] Wang WH, Dong C, Shek CH. Bulk metallic glasses. Materials Science and Engineering
Reports 2004;44:45-89.
[66] Cao JD, Kirkland NT, Laws KJ, Birbilis N, Ferry M. Ca–Mg–Zn bulk metallic glasses as
[67] Laws KJ, Shamlaye KF, Cao JD, Scicluna JP, Ferry M. Locating new Mg-based bulk metallic
glasses free of rare earth elements. Journal of Alloys and Compounds 2012;542:105-10.
[69] Miracle DB. Efficient local packing in metallic glasses. Journal of Non-Crystalline Solids
2004;342:89-96.
[70] Laws KJ, Shamlaye KF, Wong K, Gun B, Ferry M. Prediction of glass-forming compositions
in metallic systems: copper-based bulk metallic glasses in the Cu-Mg-Ca system. Metallurgical
[72] Pryds NH, Eldrup M, Ohnuma M, Pedersen AS, Hattel J, Linderoth S. Preparation and
[73] Pryds NH. Bulk amorphous Mg-based alloys. Materials Science and Engineering: A
2004;375–377:186-93.
[74] Park ES, Lee JY, Kim DH. Effect of Ag addition on the improvement of glass-forming ability
and plasticity of Mg-Cu-Gd bulk metallic glass. Journal of Materials Research 2005;20:2379-85.
[75] Laws KJ, Gun B, Ferry M. Large-scale production of Ca65Mg15Zn20 bulk metallic glass
[76] Yang HW, Wang RC, Yuan XG, Li RD, Wang JQ, Tan MJ. Importance of chamber gas
[77] Inoue A, Ohtera K, Tsai A-P, Masumoto T. New amorphous alloys with good ductility in Al-
Y-M and Al-La-M (M=Fe, Co, Ni or Cu) systems. Japanese Journal of Applied Physics
1988;27:L280-L2.
and physical properties of Fe-, Co- and Ni-rich melt-quenched ribbons containing Zr and Hf.
[79] Budurov S, Lazarova M, Stephani G, Fiedler H. Influence of planar flow casting conditions
on the thermal stability and Curie points of amorphous Fe79B14Si7 alloys. Materials Science
[80] Narasimhan MC. Continuous casting method and apparatus for structurally defined
metallic strips. In: USPTO, editor. US: Allied Chemical Corporation; 1980.
[81] Lee JG, Park SS, Lee SB, Chung H-T, Kim NJ. Sheet fabrication of bulk amorphous alloys by
[82] Park S, Park Y, Kim N. Microstructure and properties of strip cast AZ91 Mg alloy. Metals
[83] Park YS, Lee SB, Kim NJ. Microstructure and mechanical properties of strip cast Al-Mg-Si-X
[84] Lee JG, Lee H, Oh YS, Lee S, Kim NJ. Continuous fabrication of bulk amorphous alloy sheets
[85] Oh YS, Lee JG, Kim CP, Kim NJ. Solidification behavior of Fe-base amorphous alloys during
[86] Park ES, Lim HK, Kim WT, Kim DH. The effect of Sn addition on the glass-forming ability of
[87] Ferry M, Institute of Materials M, Mining. Direct strip casting of metals and alloys:
processing, microstructure and properties: Woodhead Pub. and Maney Pub. on behalf of The
[88] Kumar G, Staffier PA, Blawzdziewicz J, Schwarz UD, Schroers J. Atomically smooth surfaces
[89] Bakkal M, Shih AJ, Scattergood RO, Liu CT. Machining of a Zr–Ti–Al–Cu–Ni metallic glass.
[90] Nieh TG, Wadsworth J. Homogeneous deformation of bulk metallic glasses. Scripta
Materialia 2006;54:387-92.
[91] Gun B, Laws KJ, Ferry M. Superplastic flow of a Mg-based bulk metallic glass in the
[92] Gun B, Laws KJ, Ferry M. Elevated temperature flow behaviour of a Mg-based bulk
[94] Schiøtz J, Jacobsen KW. A Maximum in the Strength of Nanocrystalline Copper. Science
2003;301:1357-9.
[95] Yavari AR, Lewandowski JJ, Eckert J. Mechanical properties of bulk metallic glasses. MRS
Bulletin 2007;32:635-8.
[96] Schuh CA, Hufnagel TC, Ramamurty U. Mechanical behavior of amorphous alloys. Acta
Materialia 2007;55:4067-109.
[97] Ashby MF, Greer AL. Metallic glasses as structural materials. Scripta Materialia
2006;54:321-6.
[99] Li QF, Weng HR, Suo ZY, Ren YL, Yuan XG, Qiu KQ. Microstructure and mechanical
properties of bulk Mg-Zn-Ca amorphous alloys and amorphous matrix composites. Materials
metallic glasses: flaw sensitivity and Weibull statistics. Scripta Materialia 2008;58:496-9.
[101] Celarek A, Kraus T, Tschegg EK, Fischerauer SF, Stanzl-Tschegg S, Uggowitzer PJ, et al.
PHB, crystalline and amorphous magnesium alloys: Promising candidates for bioresorbable
[102] Laws KJ, Gun B, Ferry M. Mechanical stability of Ca65Mg15Zn20 bulk metallic glass
during deformation in the supercooled liquid region. Materials Science and Engineering: A
2008;480:198-204.
[103] Senkov ON, Scott JM, Miracle DB. Development and characterization of Ca-Mg-Zn-Cu
[104] Yamamoto A, Hiromoto S. Effect of inorganic salts, amino acids and proteins on the
68.
[105] Yang L, Hort N, Willumeit R, Feyerabend F. Effects of corrosion environment and proteins
[106] Witte F, Fischer J, Nellesen J, Crostack H-A, Kaese V, Pisch A, et al. In vitro and in vivo
[107] Zhang XG. Corrosion and electrochemistry of zinc. New York: Plenum Press; 1996.
[108] Witte F. The history of biodegradable magnesium implants: a review. Acta Biomaterialia
2010;6:1680-92.
[110] Tencer AF, Asnis SE, Harrington RM, Chapman JR. Biomechanics of cannulated and
noncannulated screws. In: AS. E. Asnis RFK, editor. Cannulated Screw Fixation: Principles and
[111] Chapman JR, Harrington RM, Lee KM, Anderson PA, Tencer AF, Kowalski D. Factors
affecting the pullout strength of cancellous bone screws. Journal of Biomechanical Engineering
1996;118:391-8.
[112] Hughes AN, Jordan BA. The mechanical properties of surgical bone screws and some
[113] Finlay FB, Harada I, Bourne RB, Rorabeck CH, Hardie R, Scott MA. Analysis of the pull-out
strength of screws and pegs used to secure tibial componets following total knee arthroplasty.
[114] Nunamaker DM, Perren SM. Force measurements in screw fixation. Journal of
Biomechanics 1976;9:669-75.
[115] Mazzocca AD, Caputo AE, Browner BD, Mast JW, Mendes MW. Principles of Internal
Fixation In: Bruce D. Browner AML, Jesse . Jupiter, Peter G. Trafton, editor. Skeletal Trauma -
[116] Ansell RH, Scales JT. A study of some factors which affect the strength of screws and
[118] ISO 10993-5: 2009, Biological evaluation of medical devices - Part 5: tests for in vitro
cytotoxicty.
[119] Feyerabend F, Drucker H, Laipple D, Vogt C, Stekker M, Hort N, et al. Ion release from
[120] Lunder O, Lein JE, Hesjevik SM, Aune TK, Nisancioglu K. Corrosion morphologies on
magnesium alloy AZ 91. Materials and Corrosion / Werkstoffe und Korrosion 1994;45:331-40.
[121] Zeng RC, Chen J, Dietzel W, Hort N, Kainer KU. Electrochemical behavior of magnesium
2007;17:s166-s70.
[122] Devore J, Farnum N. Applied statistics for engineers and scientists. 2nd ed. USA: Curt
Hinrichs; 2005.
Biomaterialia 2011;7:2704-15.
[124] Food and Nutrition Board IoM. Dietary reference intakes for calcium, phosphorus,
magnesium, vitamin D, and fluoride. Washington, D.C.: National Academies Press; 1997. p.
190-249.
2003;24:27-37.
[126] Food and Nutrition Board IoM. Dietary reference intakes for vitamin A, vitamin K, arsenic,
boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium,
[127] Robinson DA, Griffith RW, Shechtman D, Evans RB, Conzemius MG. In vitro antibacterial
[128] Sheng J, Nguyen PTM, Marquis RE. Multi-target antimicrobial actions of zinc against oral
[129] Dahlman J, Senkov ON, Scott JM, Miracle DB. Corrosion properties of Ca based bulk
[130] Morrison ML, Buchanan RA, Senkov ON, Miracle DB, Liaw PK. Electrochemical behavior
45.
[131] Mudali UK, Sridhar TM, Raj B. Corrosion of bio implants. Sādhanā 2003;28:601-37.
[132] Hassel T, Bach FW, Krause C. Influence of the alloy composition on the mechanical and
different chloride concentrations. In: Kainer KU, editor. 7th International Conference on
[133] Sugawara M, Maeda N. Hemorheology and Blood Flow. Tokyo: Corona Publishing Co.;
2003.
[134] Helgason CD, Miller CL. Basic cell culture protocols. Methods in Molecular Biology. 3rd
[135] Friedrich HE, Mordike BL. Magnesium technology : metallurgy, design data, applications.
[136] Oliver WC, Pharr GM. An improved technique for determining hardness and elastic
modulus using load and displacement sensing indentation experiments. Journal of Materials
Research 1992;7:1564-83.
[137] Nix WD, Gao H. Indentation size effects in crystalline materials: a law for strain gradient
[138] Fornell J, Concustell A, Surinach S, Li WH, Cuadrado N, Gebert A, et al. Yielding and
2009;25:1540-59.
[139] Castellero A, Moser B, Uhlenhaut DI, Torre FHD, Loffler JF. Room-temperature creep and
[140] Sudholz AD, Kirkland NT, Buchheit RG, Birbilis N. Electrochemical properties of
intermatallic phases and common impurity elements in magnesium alloys. Electrochemical and
[141] Kirkland NT, Birbilis N, Staiger MP. Assessing the corrosion of biodegradable magnesium
implants: A critical review of current methodologies and their limitations. Acta Biomaterialia
2011;8:925-36.
[142] Gorsse S, Orveillon G, Senkov ON, Miracle DB. Thermodynamic analysis of glass-forming
ability in a Ca-Mg-Zn ternary alloy system. Physical Review B, Condensed Matter and Materials
Physics 2006;73:224202.
[143] Laws KJ, Gun B, Ferry M. Influence of casting parameters on the critical casting size of
[144] Jiang WH, Jiang F, Green BA, Liu FX, Liaw PK. Electrochemical corrosion behavior of a Zr-
[145] Slipenyuk A, Eckert J. Correlation between enthalpy change and free volume reduction
44.
[146] Takeuchi A, Inoue A. Classification of bulk metallic glasses by atomic size difference, heat
of mixing and period of constituent elements and its application to characterization of the
[147] Aburada T, Fitz-Gerald JM, Scully JR. Pitting and dealloying of solute-rich Al-Cu-Mg-based
amorphous alloys: effect of alloying with minor concentrations of nickel. Journal of the
[149] Chrisey DB, Hubler GK. Pulsed laser deposition of thin films. New York: John Wiley & Sons,
Inc.; 1994.
[150] Wong LHY. Design of a sustained antibiotic releasing scaffold for wound application
[151] Choi H, Kim J, Lee C, Lee KH. Critical factors affecting the amorphous phase formation of
NiTiZrSiSn bulk amorphous feedstock in vacuum plasma spray. Journal of Materials Science
2005;40:3873-5.
[152] Kim J, Lee C, Choi H, Hyungho J, Hwijun K. Ni-Ti-Zr-Si-Sn bulk metallic glass particle
deposition and coating formation in vacuum plasma spraying. Materials Science and
Engineering: A 2007;449-451:858-62.
[153] Eason R. Pulsed laser deposition of thin films: applications-led growth of functional
[154] Cotell CM, Chrisey DB, Grabowski KS, Sprague JA. Pulsed laser deposition of
[155] Mele A, Guidoni AG, Kelly R, Miotello A, Orlando S, Teghil R, et al. Angular distribution
and expansion of laser ablation plumes measured by fast intensified charge coupled device
[156] Chrisey DB, Horwitz JS, Leuchtner RE. Excimer laser ablation of a YBa2Cu3O7−δ target in
a vacuum: characterization of the mass and energy of ejected material. Thin Solid Films
1991;206:111-5.
[157] Krebs H-U, Bremert O. Pulsed laser deposition of thin metallic alloys. Applied Physics
Letters 1993;62:2341.
2001.
Publication; 1980.
[161] Kowalski ZW. Review - Ion sputtering and its application to biomaterials. Journal of
[162] Thomann AL, Pavius M, Brault P, Gillon P, Sauvage T, Andreazza P, et al. Plasma
sputtering of an alloyed target for the synthesis of Zr-based metallic glass thin films. Applied
[163] Deng YP, Guan YF, Fowlkes JD, Wen SQ, Liu FX, Pharr GM, et al. A combinatorial thin film
sputtering approach for synthesizing and characterizing ternary ZrCuAl metallic glasses.
Intermetallics 2007;15:1208-16.
[164] Jing Q, Xu Y, Zhang X-Y, Li G, Li L-X, Xu Z, et al. Zr-Cu Amorphous Films Prepared by
Magnetron Co-sputtering Deposition of Pure Zr and Cu. Chinese Physics Letters 2009;26.
[165] Chu CW, Jang JSC, Chen GJ, Chiu SM. Characteristic studies on the Zr-based metallic glass
thin film fabricated by magnetron sputtering process. Surface & Coatings Technology
2008;202:5564-6.
[166] Chu CW, Jang JSC, Chiu SM, Chu JP. Study of the characteristics and corrosion behavior
for the Zr-based metallic glass thin film fabricated by pulse magnetron sputtering process. Thin
Crystallization of Metallic Glass Sputtered Films by Using Zr System Target. Research Letters in
[168] Ozeki K, Yuhta T, Fukui Y, Aoki H. Phase composition of sputtered films from a
[169] Warrell DA, Cox TM, Firth JD, Jr EJB. Oxford textbook of medicine. 4th ed. New York:
Due to the lack of information provided on the Mg ingot purity level (supplier certificate
attached), inductively coupled plasma – atomic emission spectroscopy (ICP-AES) was carried
ICP-AES analysis of the impurity levels in the Mg used in the BMG synthesis. Analysed by
Rabeya Akter (r.akter@unsw.edu.au), ICP Laboratory, Solid State & Elemental Analysis Unit,
Element Al Zn Mn Fe Cu Ni Si Ce Y
wt.% 0.006 0.008 0.01 <0.005 <0.005 <0.005 0.01 <0.005 <0.005
By deduction, the purity of Mg in the analysed ingot is >99.94 wt.%.
ICP-AES analysis of the impurity levels in the high purity Mg used in this study.
Element Al Zn Mn Fe Cu Ni Si Ce Y
wt.% 0.01 <0.01 0.01 0.004 0.002 <0.001 0.01 <0.001 <0.001
APPENDIX C – ZINC
Details of electrolytes contained in the minimum essential medium (MEM), used in the
corrosion studies.
Minimum
Human
Component Essential Medium
Plasma
(MEM)a