Chapter 1: Metallic Implants for Biomedical Applications

For over 100 years, metals and their alloys have played a central role in the advancement of medical implant applications across virtually all medical specialties. Early metal implants took some advantage of the highly useful combination of properties pertinent to this class of materials, including their high mechanical strength and excellent toughness, essential for load bearing applications, durability and resistance to wear, good thermal and electrical conductivity, ductility needed for their processing into various shapes, and chemical/biological compatibility. By 1895, scientific reports had been published on the use of metal fixation devices, e.g. bone plates, to maintain the correct apposition of fractured bone fragments in the extremities.¹  Still, early designs were fairly simple, suffered from insufficient strength and limited corrosion resistance, and were used to treat only a limited set of conditions. With advances in metallurgy, particularly the development of stainless steel and Co–Cr–Mo alloys, and concomitant progress in aseptic surgical procedures, the corrosion and mechanical performance of these implants and patient outcomes continued to improve.² 

With each passing year, novel alloys, device architectures and functions continued to emerge, a trend that continues today, as reflected by the broad range of functions exhibited by contemporary metallic biomaterials (see Table 1.1), and a great deal of reports that have been published in the scientific literature proposing novel biomedical materials and devices based on metals. However, only a fraction of these devices reach patients. This is because of a stringent regulatory framework that prescribes the kind of benchmarks a device has to achieve in order to be allowed on to the market.

For metallic biomaterials, the critical benchmarks include outstanding biocompatibility and mechanical properties, controlled corrosion profile (high chemical inertness and corrosion resistance for permanent implants, and a controlled degradation profile for resorbable platforms), appropriately high resistance to wear (especially for articulating parts), and osteointegration (for implants interfacing with bone tissues). The definitions for these benchmarks are constantly evolving. For example, the descriptors of biocompatibility have evolved to include lack of toxicity, chemical inertness, lack of foreign body or inflammatory response, non-sensitising, non-allergenic and non-tumorigenic nature, and many others. However, with the discovery of various feedback loops within materials–host systems, a more appropriate definition of biocompatibility may be the ability of a material to perform its function with an appropriate host response in a specific application.³ 

The length and complexity of the testing and approval process very much depends on the intended use, with the guidelines defined by the appropriate regulatory body. For example, the US Food and Drug Administration controls three classes of devices used for medical diagnosis and treatment, which are defined by the safety risks their use may pose. Whereas low-risk Class I devices (e.g. orthodontic wires) see only general controls, high-risk Class III devices (e.g. wire sutures for neurosurgery) have to go through mandatory performance standards, surveillance, and scientific review.⁴  For any device, the first step involves lab-based testing to establish materials and device (e.g. mechanical or electrical) performance, the propensity to corrode and release debris under physiological conditions, wear performance, and compatibility with different cell lines, to name but a few aspects. Implants that successfully pass this stage may proceed to testing using an animal model, and eventually clinical studies.

Subsequent to their market release, data on the implant performance continue to be collected. This is because the long-term use of metallic implants can bring about the development of local and systemic adverse events, the nature and extent of which, and their effect on host health and implant failure, cannot be predicted at the time of testing and may arise within months to years post implantation. Furthermore, implants approved through a ‘fast-track’ approval route (i.e. on the basis of being substantially equivalent in safety and efficacy to a legally marketed device) may not have been subject to the same level of clinical testing when compared to devices that have gone through pre-market approval, and thus may be missing critically important performance data. For example, over the decades, the US Food and Drug Administration have used the ‘fast-track’ 510(k) pathway to approve the use of larger-size femoral components and shortened trunnions that connect the head to the stem in metal-on-metal hip replacements with minimal to no clinical testing. Years after their market release, these design modifications have been found to contribute to much greater rates of mechanical failure, increased production of metal ion debris and increased wear.

With respect to metallic implants, the examples of such adverse events include metallosis in response to the release of wear and corrosion particles from total hip arthroplasty systems with metal-on-metal bearings, and the release of soluble metals from coiled wire-based permanent birth control implants made of Ni–Ti and stainless steel. Predicting such adverse events is difficult because in many cases the immune or inflammatory response is only clinically significant in particular groups of predisposed individuals. Nevertheless, the magnitude of these events for the affected individuals can be significant, and include the development of pseudotumours, neurological events, and bone loss and subsequent failure of the prosthesis in the case of the hip replacement system, and systemic sensitivity events in the case of the birth control implant.⁵  As in most cases the presence of clinically significant events would necessitate device removal and replacement, the economic and healthcare burden associated with these events is also significant.

The post-market release data collection is also what drives the ongoing device development and optimisation. In spite of the lengthy, multi-stage, and expensive assessment of implant compliance, a growing number of devices are getting cleared and approved for clinical use each year across all application areas. The US Food and Drug Administration has reported a steady growth in approvals of metallic devices over the past decade (see Table 1.2).⁶ 

Key to metallic implant advancement is the development of new alloys, since a mixture of several different metals and other elements may not only overcome the limitations of single-component materials by taking on the desirable properties of its constituents, but also exhibit unique properties that are distinct from those of the individual components comprising the alloy. Amongst the most common reasons for alloying is to improve the strength, processability or corrosion resistance of a metal. For example, the addition of even a small amount of alloying agents in the metal matrix introduces atoms of sizes different to that of the primary metal; distributed uniformly throughout the crystalline grains (as either substitutional or interstitial elements), where the alloying atoms act as barriers to slip, thus improving strength. Alloying elements can also strengthen the metal through precipitation hardening, where barriers to dislocation are formed by impurity-rich particles of tens or hundreds of atomic diameters in size.

When choosing suitable alloying elements, the individual and cumulative toxicity of elements within an alloy should be low, with a preference given to the essential trace elements Co and Fe (as these elements are already present in heathy tissues of living organisms), and highly corrosion resistant active elements, such as Ti. The majority of the presently used metallic biomaterials are based on Fe-based alloys (e.g. stainless steels), Co–Cr alloys, and Ti and its alloys. Where the strength, stiffness and wear resistance of Co–Cr alloys is the greatest amongst the aforementioned alloys; stainless steels offer a favourable combination of cost, cyclic twist strength and ductility; and Ti alloys are considered most biocompatible, with the greatest corrosion resistance and specific strength. This statement is of course a generalisation, as the properties of alloys within each group can differ considerably based on their composition, fabrication method and post-processing treatment, and the environment within which they operate. Biomaterials based on Ta, Nb, Mg and Fe are also becoming increasingly popular due to their bioactivity (in the case of Ta, Mg and Fe), chemical inertness (Nb) or controlled degradation under physiological conditions (Mg and Fe). For some time, Ni was one of the most popular alloying elements due to its ability to enhance the formability, weldability, ductility, and, in some instances, corrosion resistance of stainless steels and Co–Cr alloys. The subsequent discovery of the potential acute and chronic adverse effects of exposure to Ni that leached out from the implants, including delayed hypersensitivity, headaches, vertigo, vision changes, altered iron metabolism and homeostasis of Ca, Mg, Mn and Zn, has led to the reduction in the use of Ni in biomedical alloys. The use of another popular alloying element, Al, has been suspected to cause osteomalacia, hepatic dysfunction, anaemia and neurological effects. While Ni and Al are non-essential metals for the human body, even for essential metals that the body requires for the proper functioning of many of its proteins and enzymes, their quantities are tightly controlled via homeostatic regulation, with the presence of excessive amounts of these metals often causing local and systemic toxicity.⁷  For example, although vanadium is considered essential for proteins that participate in phosphate and lipid metabolism, and insulin enhancement, the presence of excess vanadium may result in the development of headaches, weakness, tremor and gastrointestinal issues in patients. This holds true for excess Mo, which can cause urinary tract stones, acute renal failure, and myositis, as well as for Co, Cu, Fe, Mn, Zn and Cr essential metals. Undesirable physiological effects have also been reported for other non-essential metals frequently found in biomedical implants, including Ti, Ag, Au, Pt, Pd, Sn, Hf, Zr, W and Ir. Nb, Ta and Zr are often considered among the least toxic alloying elements, yet are expensive. Also expensive, precious metals, including Au and Au alloys with Pt and Pd, Ag and its alloys, and Pt and Pt alloys are used in restorative dentistry, in antimicrobial treatments (Ag) and as electrodes for electrically active devices.

The bulk properties of metals are defined by their chemical composition, crystalline structure and bond strength. The 3D arrangement of closely packed atoms in metals, where the nuclei of positively charged ions are immersed in a delocalised cloud of freely moving valence electrons, is responsible for the excellent electrical and thermal conductivities of metals, and their high density (in the case of most metals). Greater amounts of delocalised electrons correspond to stronger metallic bonds, i.e. stronger electrostatic interactions between the electron cloud and the ions. The strength of the metallic bonds defines the tensile strength of the material, whereas the non-directional nature of metallic bonds is responsible for the malleability and ductility of metals, which allows them to undergo a significant amount of plastic deformation (plastic strain) without failure (cleaving). The strength of the metallic bonds is greater for metals made of atoms of greater nuclear charge, which have a greater number of electrons they can release into the cloud, and of smaller atom size; the greater strength of the metallic bonds is also responsible for the higher melting temperature and hardness of such metals. As such, atoms with half-filled valence shells (such as W and other transition metals with valence-level d electrons) make solids that are stronger than those made of elements with nearly empty or nearly full valence subshells. While the Young's modulus, ultimate tensile strength and toughness define the application of metallic biomaterials,⁸  their hardness and melting temperature prescribe the kind of techniques that can be used for their processing into implants and devices.

The atoms in most pure metals tend to naturally assemble into either body-centred cubic (bcc), face-centred cubic (fcc) or hexagonal close-packed (hcp) structures. Bcc corresponds to a crystal organisation in which eight atoms occupy the corners of a cube, with one atom located in the cube centre. In other words, one atom is bonded to eight others, giving a coordination number of eight, with two atoms per unit cell. In a fcc structure, eight atoms occupy the cube corners, and six atoms are in the centre of each face (a coordination number of 12, four atoms per unit cell). In a hcp structure, the atoms are organised like stacked hexagons, where six atoms occupy the corners of a hexagon with two atoms located at their centres (6 atoms per cell, a coordination number of 12). Examples of metals that adopt bcc crystal structures include Fe, V, Nb, Cr; fcc structures are adopted by Al, Ni, Ag, Cu and Au; and hexagonal structures by Ti, Zn, Mg, and Cd. The introduction of an alloying element results in the creation of internal stresses within the crystal structure due to the size differences between the atoms; atoms of a larger size apply a compressive force on adjacent atoms, whereas the force the smaller-sized atoms exert is tensile. These internal stresses are often beneficial, enhancing the desirable properties of the material, e.g. resistance to deformation, shear and tensile strength, toughness, hardness and ductility. Depending on the choice of the alloying element and its concentration, the bulk density, reactivity, Young's modulus and other physical properties may remain largely unchanged or change greatly from that of the base metal. The crystalline structures of both pure metals and their alloys can also be altered by physical treatment, e.g. work hardening. For example, defects can be introduced by subjecting the material to severe plastic deformation, when very large strains (by means of a complex stress state or high shear) are applied to the material to achieve a high density of defects that may act as nucleation sites for precipitates, and change the grain structure to equiaxed ultrafine (d < 500 nm) or nanocrystalline (d < 100 nm, with some extreme cases of refinement of d < 30 nm).⁹  These changes will remain permanently within the lattice unless the material is heated and then allowed to re-crystallise. These changes in the structure and texture of the alloy may notably affect its mechanical (strength, ductility) and service characteristics (corrosion resistance). Where grain refinement leads to strengthening of the alloy by inhibiting the dislocation glide, the concomitant transformation of the texture in certain alloys (e.g., Mg1.0%Zn0.3%Ca) may counteract the former strengthening effects by forming an inclined basal texture that promotes basal slip and may also lead to tensile–compression asymmetry.¹⁰ 

Certain heat treatment and cooling regimes can also be used to strengthen the alloys and relieve stress that may have accumulated in the material during fabrication. For example, Ti alloys can be subjected to forging at 595–705 °C for 1–2 h, and then air cooled to relieve undesirable residual stresses from non-uniform hot forging, deformation from cold forming and straightening, asymmetric machining, welding and cooling of castings. Single and multistage annealing, and solution treatment and ageing can be used to improve creep resistance and fracture toughness, and attain greater strength, respectively. The latter treatment involves heating the alloy to a specified temperature (defined by the nature of the alloy), followed by quenching at a controlled rate in either oil, air or water, and then reheating to a temperature between 425–650 °C for 2 h.

A combination of the aforementioned techniques may be used to achieve the desired level of fracture toughness, i.e. the level of resistance to crack propagation and failure under load, and fatigue strength. The latter is an important determinant of implant performance and lifetime because humans tend to engage in a wide range of repetitive physical activities. For example, hip and knee prostheses, fixation plates and wires will be subjected to cyclic loading during walking and running, dental implants will experience cyclic loading during chewing, and electrodes will be subjected to myocardial activity. Joints will experience cyclic compression and bending with a loading strength of ∼50 MPa and ∼200 MPa (loading frequency 1 Hz) respectively, with the total number of loading cycles over the lifetime of an implant of 10⁷.¹¹  This type of loading will result in the weakening of the material via progressive and localised structural damage and the growth of cracks that start at the point where stress concentration occurs, eventually resulting in implant failure by fatigue. Importantly, failure by fatigue can take place at normal operating loads far lower than the tensile strength of the material.

The microstructure of the material is an important determinant of the maximum intensity of cyclic tension it can withstand prior to failure (see Figure 1.1). Other contributing factors include the nature of the stress (cyclic frequency, load vectors), the corrosivity of the environment, and the surface finish and wear. Fatigue life can be extended through surface modification of the implant material. For example, grit-blasting of dental implants made of commercially pure titanium can produce a surface layer of compressive residual stresses that protects them from failure under cyclical tension-compression loading.

Among the key drivers behind the development of new alloys is the growing understanding of how different properties of materials affect their behaviour in vivo, in particular the interactions between materials and cells. For example, the nature of a cellular response is not only determined by the surface chemistry of the material, but also its surface topography and surface and bulk mechanical properties, including the material's Young's modulus, its specific strength (expressed as the ratio between its strength and ductility) and others. Most of the alloys currently used in bone tissue reconstruction have a Young's modulus that is considerably greater than that of natural bone tissue (see Table 1.3). This is in part because in order to perform their load-bearing function, they need to have a suitable level of stiffness. The consequences of this difference include non-homogeneity in the implant–bone load transfer and reduced stress stimulation of the bone, resulting in bone loss, limited bone remodelling and sub-optimal implant integration, which affect implant performance and lifetime in vivo, and the benefits they deliver to the patient.^12–14  A review of 23 longitudinal studies in implant dentistry that performed a follow-up of 10 or more years to evaluate the survival of ∼7700 osseointegrated implants showed a mean survival rate of 94.6% ± 5.97% and a mean marginal bone loss around implants of 1.3 mm ± 0.84 mm, with 14 studies also reporting a cumulative mean success value of 89.7% ± 10.2%.¹²  However, in a number of these studies, the probing depth for bone resorption was significantly less than 5 mm, possibly resulting in missed incidences of peri-implant complications. Furthermore, there is a lack of consensus in the field as to the amount of bone resorption considered appropriate for the number of years in the function, health, and success of the implant. Importantly, the measurement of the distribution of stress and strain in dental implants and in bones when the implant is loaded under normal occlusal loading conditions is difficult to conduct in vivo, and thus such information often comes from modelling studies¹⁴  or those involving human cadavers.¹³ 

While Ti and its alloys have relatively low Young's moduli (110 GPa for Ti6Al4V) when compared to 316L stainless steel and Co–Cr alloys (200–210 GPa), they still greatly exceed the Young's moduli of bones (∼ 10–30 GPa). Among the Ti alloys, those that have β-type phase constitution, i.e. metals with a bcc structure, are characterised by significantly lower Young's moduli when compared to α-type alloys, i.e. those with a close-packed hexagonal crystal structure and (α + β)-type alloys. The phase constitution can be controlled through the selection of alloying elements, with Nb and Ta frequently used to stabilise the β-phase, with some Zr introduced due to its ability to dissolve in the β-phase and increase the strength of the material. As the design of the implant has to take into account economic considerations, some effort has gone into finding suitable affordable alternatives for the use of Nb, Ta, Mo, and Zr, with reports of producing low modulus alloys using Fe (as in Ti–Mn–Fe), Cr (Ti12Cr and Ti–Cr–Sn–Zr), Mn (Ti–Mn), Sn (as in Ti–Cr–Sn and Ti–Mn–Sn), and Al (Ti10CrAl and Ti–Mn–Al).^15–17 

In addition to changing the composition of an alloy, it is possible to change its bulk and surface properties through a careful combination of fabrication and post-processing techniques. Figure 1.2 shows the various stages of metallic implant development where technology-driven innovation can take place.

In parallel to more traditional processing, shaping and forming techniques (the effects of which on the structure and properties of commonly used metals and alloys are discussed in the relevant sections), 3D scanning and printing has recently emerged as a promising means by which to facilitate personalised medicine. Firstly, it allows for the rapid prototyping of implants that fit the unique anatomical features of individual patients, thus improving implant mechanical performance and reducing pain and discomfort for the patient in orthopaedic and dental reconstruction, and improving appearance in craniofacial reconstruction. Because of the use of high-resolution data acquisition tools (e.g. computer tomography) in conjunction with computer-aided modelling and design, it is possible to produce more complex implants and correct defects with improved therapeutic outcome. Thus, the time to recovery, incidence of revision surgeries and the associated radiation burden may also be reduced. More importantly, however, 3D printing allows for the reproduction of the complex structures of the bone the implant is designed to replace, thus facilitating harmonisation of the structural and mechanical properties of the implant to that of the bone, and in doing so, ensuring appropriate load transfer, promoting tissue healing and implant integration. This is because 3D printing allows for the fabrication, and in some instances in situ modification, of complex multi-material structures that vary in terms of their chemical composition, density, porosity, and other features, thus reproducing not only the macroscopic features of the tissue it aims to replace, but also the micro- and nano-scale features.^18–21  In metallic implants, this generally means introducing pores while controlling their shape, dimension, orientation and connectivity to promote bone tissue infiltration, blood vessel formation and nutrient exchange, and reduce stress shielding. The latter arises when there is a poor match between the mechanical properties of the implant and the bone, resulting in the typical stress being removed from the bone by the implant. Lack of appropriate levels of stress leads to osteopenia, where the bone density is reduced, increasing the probability of bone fractures and aseptic bone loosening, and compromising the stress transfer between the bone and the implant. In addition to varying the properties of a single material, or of materials of the same type, it may also be possible to use 3D printing to integrate multiple types of materials during implant fabrication. The resulting implant would thus integrate an anatomically-precise metallic scaffold to support mechanical load, polymer and ceramic particles to promote cell attachment and bone tissue formation, active molecules, proteins and growth factors to stimulate cell differentiation and growth, and precursor cells derived from the host from which the tissue would form.

Although the clinical use of 3D printing is certainly growing, owing to the steadily decreasing prices of printers and increasing availability of computer topography and magnetic resonance imaging at hospitals, at present it is limited to the fabrication of patient-specific implants with an otherwise conventional structure and properties. This is because the in vitro, in vivo and clinical performance of these implant designs has been studied extensively. In contrast, multi-material scaffolds impregnated with biologically active molecules and cells largely remain the subject of laboratory studies,^22,23  with translation hindered by the immense complexity of fabricating a structure that would perform all of the desired functions in a predictable manner. For example, 3D printing that relies on the use of elevated temperatures, potentially reactive chemicals or high-intensity irradiation, e.g. electron beam or selective laser melting of metals, will not be an appropriate choice for the fabrication of scaffolds in situ seeded with living cells or biologically active molecules. However, the impregnation of such scaffolds with cells and molecules post fabrication is also not trivial. Once cells are within the scaffold, it is necessary to ensure an adequate level of nutrient and gas exchange, however integrating a pathway for vascular network formation into a scaffold is challenging from both technological and biological perspectives. Yet, without an effective means of supply and the removal of molecules and gases, any attempt at tissue development will be futile, whereas dying cells may in fact create significant problems by releasing inflammation-stimulating factors and changing the chemistry of the micro-environment in the pores within the bulk of the implant. The metallic scaffold must also be biocompatible, with a controlled degradation profile and the ability to support cell attachment, migration and the formation of complex tissue architectures. Yet, the increased surface-to-volume ratio and poor circulation of fluids within the pores may create a favourable environment for accelerated corrosion and metal ion release.²⁴  It should be noted that for certain applications, increased biodegradation may in fact be desirable, as is the case for Fe-based craniofacial implants.²⁵ 

In addition, one should be able to print the scaffold with the high resolution of the desired structure, which the scaffold will then maintain over time. For example, it is not uncommon for the pores to collapse or for the pore networks to be disconnected during scaffold assembly, or, in the case of solid implants, for the pores to be introduced into the structure due to gas entrapment, changing the mechanical properties of the resultant implant.²⁶  Furthermore, for methods that rely on the use and subsequent removal of the polymer binder, such as the inkjet printing of metal powders, the appearance of defects, shrinkage (and associated delamination and cracking), and difficulty in the removal of unbound powders from the depths of the bulk implant are also of concern.^25,27,28  Examples of defects include spherical voids due to the collapse of vapour cavities, elongated voids due to insufficient fusion, accumulation of partially molten particles, and the balling phenomenon, with the latter two being the result of e.g. incomplete melting of powder particles when using insufficiently high power during selective laser melting.²⁹  This may render the surface less hospitable to cell infiltration and colonisation, as well as affecting the mechanical and chemical properties of the implant. Nevertheless, using electron beam melting and selective laser melting, it is possible to attain a broad range of characteristic crystallographic phases, e.g. α (hcp), β (bcc), α′ (hcp martensite) and α″ (fc-orthorhombic) and phase microstructures and variable hardness in Ti6Al4V, which would otherwise require complex thermo–mechanical processing if conventional wrought and cast billets were used.³⁰ 

To be a viable option for clinics, the printing process itself should be simple, reliable and affordable. Currently, selective laser sintering, direct metal laser sintering and electron beam melting are used in the printing of metal and alloy powders, including those of titanium, titanium alloys, cobalt chrome, and stainless steel, with fused deposition modelling used for the printing of 3D patterns from eutectic metals (see Figure 1.3).²¹  While it is possible to fabricate scaffolds with very good fidelity and resolution of fine features, selective laser sintering is lengthy (∼ 15 h), requires elaborate and expensive infrastructure, e.g. CO₂ lasers, and may expose operators to potentially hazardous dust and nanoparticle condensates (see Table 1.4). Furthermore, as-fabricated implants generally require post-processing to address the surface roughness (e.g. via polishing or sandblasting) and accumulated internal stresses (e.g. via heat treatment).²⁰  Unless addressed, roughness may promote wear in the articulating components of load-bearing systems. Although implants produced using electron beam melting do not generally require thermal post-processing to remove stresses (with beams delivering an energy density of 200 J mm⁻³), the infrastructure is similarly very expensive, while the resolution of features is generally lower when compared to selective laser sintering.

While the cost of ink-jet printers is lower, the resolution of the features is also lower when compared to the aforementioned approaches due to the spreading of the liquid and undesirable rearrangement of the powder particles due to differences in their wettability.³¹  The process is also fairly slow, as the deposition of the next 2D layer of the pattern can only take place once the underlying layer has solidified, with the mechanical properties of the resultant scaffolds being dependent on the chemistry of the binder, and the surface properties and size of the individual metal particles, due to their effect on the distribution of pore sizes and scaffold assembly (see Table 1.5). Electro-hydrodynamic inkjet printing produces printed materials with better resolution and speed, as the surface tension effects are overcome by the use of an electric field, and a rapid evaporating solvent is used. Inkjet printing and its modifications may be best for the fabrication of smaller implants, including implantable electrodes and sensors. Although the fabrication of the structure takes place at room temperature, it is followed by high temperature sintering to attain the necessary level of mechanical performance, which will generally lead to implant shrinkage and anisotropy with respect to implant properties, e.g. density, and thus load-bearing capacity. With respect to health and safety, issues associated with the handling of nano- and micro-scale powders and the possibility of electrical shorts due to the close proximity of the conductive particles to the circuitry in the printer head warrant consideration.¹⁸ 

Surface modification is also frequently employed to control the corrosion dynamics and cell–surface interactions between the implant and surrounding issues (see Figure 1.4).³²  Although many metals and alloys will naturally form an oxide layer on their surface, the chemistry and stability of this layer will be determined by the environment to which the implant was exposed at the time of oxide formation. For example, Ca, P, and S may be present in the oxide layer formed on the biomaterial submerged into a biofluid, and be absent when the layer is formed in ambient air. Besides adding new elements, the oxide layer may experience a change in the native oxide composition; for example, a passivation layer grown on Cr–Co–Mo in biofluid was found to be largely composed of Cr oxide, with only a limited amount of Mo oxide.³³  The uniformity, porosity and strength of adhesion to the underlying metal may also be affected. A greater level of control over the final surface properties can be attained via implant encapsulation and coating with a protective or bioactive layer, ion implantation to increase wear and corrosion resistance by modifying the top-most layer of the bulk implant, etching and structuring to impart desirable surface topography, and functionalisation to change the surface energy and wettability.

Modification of surfaces by mechanical means, for example using machining, grinding, polishing and blasting is often used to impart specific macro- and micro-scale roughness and texture onto a surface. The fabrication of more intricate patterns and textures, with multi-tier complex topography and well-defined micro- and nano-scale features typically demands the use of more expensive highly-controlled environments and high-energy particles, or the use of self-assembly techniques.³⁴  The bombardment of surfaces with high energy ions generated in plasmas can be used to manufacture surfaces with high aspect ratio features, including tall thin pillars and deep narrow trenches and pits,³⁵  with the possibility of fabricating features with aspect ratios of up to 100, deep features with vertical dimensions of >100 µm and radii of <10 µm, or high-aspect ratio features with dimensions below 100 nm. Here, by manipulating the gas composition, and electric and magnetic fields, one can control the intensity, density and direction of the ion flux towards the target surface, thus controlling the shape, density and orientation of the surface features.³⁶  In addition to etching, plasmas can be used for the growth of patterns of features or smooth continuous films (e.g. plasma polymerisation),^35,37  modification of surface chemistry (e.g. oxygen plasma treatment),³⁸  and surface hardening by driving ions into the top layer of the material (e.g. plasma nitriding).³⁹  Both plasma etching and femtosecond laser ablation can also be used to produce surface patterns with high resolution, such as those resembling patterns on natural surfaces capable of repelling bacteria or killing microbial cells on contact.^40–42  These surface patterns may be able to help host cells beat bacteria in their race to the surface, preventing surface biofilm formation and providing host cells with the chemical, physical and mechanical cues that promote tissue formation and implant osseointegration.⁴³  Exposure of the surfaces to acids, hydrogen peroxide, alkalis, anodic oxidation, sol–gel and chemical vapour deposition will also result in changes to the surface chemistry, and in some instances, surface topography, as in chemical mechanical polishing to obtain ultra-smooth surfaces or the hydrothermal etching of Ti using 1 M KOH to attain a nanostructured surface with strong bioactivity.⁴³  Electrochemical anodic oxidation has also been used to induce the formation of ordered oxide layers with nano-scale porosities on Ti, Al, Nb, Ta and Zr with the aim of promoting osseointegration;^44–46  this is because such porous oxides show improved mechanical interlocking with hydroxyapatite, a highly osseoconductive crystalline complex of calcium and phosphate that is present abundantly in the osteoid matrix.

Examples of physical surface modification techniques include thermal spraying, physical vapour deposition, ion implantation and sputtering, and pulsed laser and evaporative deposition methods. It is not uncommon to combine multiple processing techniques, and also follow them with biological activation of surfaces through the immobilisation of proteins, attractor molecules and factors that repel microorganisms and/or promote desirable host cell responses. Successful immobilisation is one in which biomolecules can attach in a controlled fashion with respect to density, surface organisation, and 3D conformation, since these biomolecules are the sites that will be later recognised by proteins and cells in the biofluid, and will eventually determine cell attachment preferences. This is generally done by first chemically modifying the surface with e.g. carboxylated poly(ethylene glycol), and then using –COOH to graft cell-recognisable biomolecules, including ethane-1,1,2-triphosphonic acid, methylenediphosphonic acid, morphogenetic protein-4, polydopamine, selenocystamine, and the cell–adhesive peptide Arg–Gly–Asp–Cys.⁴⁷  On gold and other noble metals, poly(ethylene glycol) and other molecules can be tethered via thiol (–SH or –SS–) or amino (–NH₂) groups due to their high affinity for these metals, whereas stainless steel surfaces may need to be first functionalised with a silane-coupling agent and activated by e.g. plasma treatment before they can be used for the UV-induced graft polymerisation of poly(ethylene glycol). For metals that form a native oxide, e.g. Ti alloys, poly(l-lysine)-g-poly(ethylene glycol) co-polymers can be used as they readily adsorb onto the surfaces of the negatively charged TiO₂, Si_0.4Ti_0.6O₂, and Nb₂O₅ via the positively charged primary amine groups of the poly(l-lysine), with the hydrophilic, uncharged poly(ethylene glycol) side chains pointing away from the surface.⁴⁸  Other strategies for the surface immobilisation of poly(ethylene glycol) and biomolecules include the use of highly-adhesive 3,4-dihydroxyphenylalanine (e.g. via conjugation to a single 3,4-dihydroxyphenylalanine residue or to the N-terminus of Ala–Lys–Pro–Ser–Tyr–Hyp–Hyp–Thr–DOPA–Lys);⁴⁹  glycose-containing diazirines (e.g. via the insertion of singlet carbenes into the H–O bonds of protective oxides, leading to glycosidation);⁵⁰  and photoreactive poly(ethylene glycol) (via nucleophile-reactive dehydroazepines created by the ring expansion of the photolysed intermediates of aryl azides).⁵¹ 

The development of 18/8 (also known as type 304) stainless steel in the 1920s was an early step towards improving the corrosion resistance and mechanical strength of metallic implants, extending the useful lifetime of implants and reducing the incidence of post-operative complications and metallosis (see Figure 1.5). 18 wt% Cr is sufficient for alloy passivation, i.e. the rapid formation of a self-healing, chemically inert, protective Cr oxide layer which prevents the diffusion of oxygen to the underlying material, thus effectively providing the alloy with corrosion resistance in reducing acids and against pitting attack in solutions containing Cl⁻ ions, e.g. biofluids. Together with Cr, the 8 wt% Ni allows the alloy to take on an austenitic fcc crystal microstructure, increasing the density, hardness, and abrasive resistance of the steel, thus improving its resistance to oxidation, corrosion and heat. Nevertheless, while it is more resistant to uniform corrosion than carbon steels, 304 steel is susceptible to pitting corrosion when exposed to fluids rich in Cl⁻ ions and dissolved oxygen (i.e. conditions commonly encountered by an implant).

316L steel provides a greater level of corrosion resistance by reducing the fraction of carbon to <0.03 wt%, increasing the proportion of Ni to 10–12 wt% and adding 2–3 wt% Mo. The latter element is frequently introduced into steel to reduce the susceptibility of the alloys to pitting corrosion, whereas the fraction of carbon is intentionally minimised, as it tends to reduce the fraction of Cr available for passivation via the formation of Cr₃C₂. 316L steel is also austenitic, with a γ–fcc crystallographic structure. Martensitic and ferritic are the other two possible crystal structures of stainless steel, with the former characterised by a distorted bcc organisation of atoms, and the latter characterised by an α–bcc structure. Martensitic steels can be produced by transforming the fcc structure of the austenite into a body-centred tetragonal crystal structure using quenching. The process of cooling needs to be very rapid to minimise the number of C atoms that can diffuse out of the crystal structure, producing a highly strained, carbon-supersaturated, body-centred tetragonal structure that is very hard due to a significant fraction of dislocations that arise from the shear deformations.⁵²  The microstructure of martensitic steels makes them more brittle and less corrosion resistant when compared to austenitic steels, and heat treatment can effectively destroy the martensite by relieving the stresses within the crystal structure, making the steel softer. Cold working (rather than heat treatment) is also used to improve strength and toughness in austenitic steel. Although intermittent application of heat may be used during implant manufacturing in order to be able to shape steel into the desired product, excessive heat (e.g. during welding) may promote the formation of Cr₃C₂ and reduce the quality of the oxide layer, and as a consequence reduce the corrosion resistance of the alloy. It is often necessary to remove thus-formed suboptimal highly porous coatings using mechanical polishing and/or chemical means.

Efforts have been made to replace potentially toxic Ni with other alloying elements, e.g. nitrogen or manganese in Fe–Cr–Mn–Mo–N steels, while preserving the austenitic crystal structure and corrosion resistance of steel. N stabilises austenite and prevents the formation of ferrite, which would render the material ferromagnetic and reduce its ductility. Adding N to steel also strengthens the alloy, since interstitial N atoms produce large lattice distortions and interact with dislocations by means of electrostatic attractions; dislocation-nitrogen complexes may also drag the dislocations.⁵³  The changes in the dislocation structure due to the changes in the stacking fault energy, and specifically the planar dislocation arrangement in N-rich Ni-free steels that favours planar dislocation slips over dislocation cross slips, are responsible for the improved fatigue behaviour of these alloys. With respect to corrosion resistance, the formation of NH₄⁺ or NH₃ (through the reaction between N and H⁺) in pits and crevices may prevent local acidification and thus promote repassivation, whereas the formation of NO₂⁻ and NO₃⁻ can inhibit the anion attack due to N enrichment at the interface and the formation of protective Cr₂Mo₃N during passivation.⁵³  A significant challenge in using higher fractions of N is the resulting increase in the ductile-to-brittle transition temperature, which is proportionate to the amount of N used. This is because the ductile-to-brittle transition temperature is one of the key determinants of the fracture behaviour of the material (along with its impact strength); when the temperature at which steel is used is below the transition temperature, the steel loses its toughness and instead of undergoing plastic deformation it fractures in a brittle mode at the tips of cracks or flaws.⁵⁴ 

Despite the aforementioned issues with corrosion, due to their significantly lower cost when compared to either Ti or Co–Cr alloys, steels continue to be widely used in medicine, particularly in applications where they are either used for a short time or where the dissolved oxygen-rich conditions prevent rapid corrosion (e.g. vascular stents).

Co-based alloys owe their superior corrosion performance in vivo to the formation of a self-regenerating Cr₂O₃ passivation layer. Compared to steels, Co–Cr alloys are characterised by their better resistance to fatigue, wear and corrosion, and have comparable Young's moduli. They are also sufficiently ductile, for shaping and contouring of the implant, with 8% elongation performance. Mo is often added to reduce grain size and enhance strength by forming a solid solution where the local non-uniformity in the lattice impedes dislocation motion and thus plastic deformation. Depending on their composition (see Figure 1.6), implants made from Co–Cr alloys can be prepared by means of casting (e.g. dental Co–Cr–Mo implants), or hot working (e.g. wrought Co–Ni–Cr–Mo–Fe and Co–Cr–W–Ni), for example by hot forging (e.g. Co–Ni–Cr–Mo load-bearing stems for joint replacement). Wrought alloys display greater fracture toughness and fatigue strength, and are thus capable of withstanding large cyclic loads typical of orthopaedic applications. Yet, their ability to withstand crevice and pitting corrosion and mechanical wear is somewhat lower when compared to cast alloys. Although cold working can in principle be used to improve the strength of forged Co–Ni–Cr–Mo alloys, it is challenging, particularly when implants of a larger size are required. Suboptimal resistance to friction wear limits the use of forged Co–Ni–Cr–Mo as load bearing surfaces in devices.

Cast alloys, however, generally have a larger grain size, and distinct boundary separation, with such defects as holes and cavities present within the body of the alloy. Adding Ni improves the processability of cast alloys, whereas adding a small amount of carbon (∼ 0.25 wt%, e.g. Co29Cr6Mo0.23C) makes the alloy easier to cast by reducing its melting temperature and increasing its fluidity. The addition of C also facilitates the precipitation of carbides along the grain boundary and in the interdendritic regions, increasing the hardness (and making the alloy more amenable to work hardening), yield stress, and resistance to wear of the alloy at the expense of its ductility. Solution annealing followed by ageing can increase the ductility by facilitating the dissolution and a more uniform re-precipitation of carbides within the matrix, bringing the austenite/martensite ratio to the desired level.⁵⁵ 

The attractiveness of Ti as a metal for implants stems from its biocompatibility, and in particular, mechanical biocompatibility (due to a lower Young's modulus), lower density (at 4.5 g cm⁻³ compared to > 8 g cm⁻³ for steels and Co–Cr alloys), and low corrosion under physiological conditions. There are four categories of commercially pure Ti defined by the level of impurities, with grade I having the highest purity of the four. Although 0.7% is the maximum allowable level of impurities for grade IV Ti, the fractions of certain elements is tightly regulated due to their significant effect on the key mechanical properties of Ti. Thus, the maximum allowed content of O, N and Fe should not exceed 0.4, 0.05 and 0.5 wt%, respectively. However, while the fractions of O and Fe should be sufficiently low to attain the desired fracture toughness, creep and stress rupture characteristics, respectively, the fraction of O should also be sufficiently high to achieve the required level of tensile strength. In its pure form, Ti is allotropic, with a close-packed hexagonal crystal structure (α-phase), and transitions to a bcc structure (β-phase) when heated above 885 °C up to 1668 °C. When an alloying element is introduced into Ti, it tends to stabilise either the α- or β-phase. For example, O, Al, N, Sn and C stabilise the α-phase, shifting the temperature at which α-to-β transformation takes place higher. Mn, Cr, Fe, Mo, V and Nb are examples of β-phase stabilising elements, and at sufficiently high quantities may allow the alloy to retain some β-phase when cooled to ambient temperature.

The chemical composition and resulting phase structure will prescribe the types of processing that these alloys can be subjected to in order to tune their properties (see Figure 1.7). Alloys with an α or near-α phase structure, e.g. commercially pure Ti, have superior weldability and heat and oxidation resistance, but lower workability and strength. Heat treatment can be effectively used to relieve internal stresses in α-type alloys but will not strengthen the alloy. This is because strengthening often relies on the precipitation of one of the phases within a multi–phase system. For β-phase alloys, exposure to certain temperatures will lead to the loss of this metastable state, and, consequently, strengthening of the alloy. Heat treatment is also used for stress relieving and ageing. Alloys with an α–β two-phase structure, e.g. Ti6Al14V, are widely used in medical applications. They contain α-stabilising Al (5.5–6.5 wt%) and β-stabilising V (3.5–4.5 wt%), producing a structure with a β phase dispersed within an α phase. When the alloy is annealed, and then quenched and reheated again, the metastable β phase precipitates as particles, enhancing the strength of the alloy. At greater fractions of V (as in Ti13V11Cr3Al alloys), the β type microstructure is even more clearly pronounced, with annealing leading to a substantial strengthening and at the same time reduction in the ductility of the alloy.

In addition to mechanical properties, adding elements may change the biocompatibility of the alloy by changing the chemical and physical properties of its oxide. For example, adding Zr (as in e.g. Ti15Nb4Ta4Zr alloys) not only enhances mechanical strength and corrosion resistance, but also improves bone cell compatibility and bone mineralisation when compared to Ti6Al4V. While the oxide on pure Ti or Ti6Al4V alloy largely consists of amorphous and slightly crystalline TiO₂, with small quantities of Ti₂O₃ and TiO, and an insignificant fraction of Al₂O₃ and hydroxyl in the case of Ti6Al4V, oxides on Ti–Zr alloys have a fraction of zirconium oxides that reflects its fraction in the bulk alloy. Oxidation treatment of Nb-containing alloys, e.g. Ti9Nb13Ta4.6Zr, improves their wear resistance due to the formation of Nb₂O₅ particles on the diffusion hardened surface.⁵⁶  While in the case of Zr and Nb, the effect is beneficial, as thicker and more stable coatings offer greater protection, and the inclusion of even small quantities of NiO and metallic Ni into oxides on the surface of Ti–Ni alloy may negatively affect its corrosion resistance and biocompatibility.

While the specific properties will be dependent on the composition and processing, commercially pure Ti used in medical applications typically has tensile strength values of 240–550 MPa, a yield strength of 170–485 MPa, and an elongation of 15–24%. The tensile strength and yield strengths are significantly greater for Ti6Al4V alloys, at 860 MPa and 758–795 MPa, respectively, whereas the elongation is notably lower, at 8–10%. Interestingly, while the tensile strength is the same whether the alloy is wrought or cast, the elongation and yield strength are higher for wrought alloys. Some efforts have been made to design Ti alloys with lower Young's moduli to bring these values lower, closer to that of natural bones, yet it may render the alloy more susceptible to fatigue failure. The resistance of Ti alloys to failure through sheer is also comparatively low, as is its resistance to the combined effects of corrosion and mechanical wear, the conditions typically experienced by articulating load-bearing implant systems. Efforts have been made to design more robust alloys, such as those capable of more rapid passivation under fretting and sliding conditions (e.g. Ti29Nb13Ta4.6Zr⁵⁶ ), martensitic Ti6Al4V ELI alloys that are more resistant to adhesive/abrasive wear, and alloys with more refined grain micro- and nano-structures.⁹  In parallel, attempts to enhance the surface hardness of alloys through coating with hydrogenated amorphous carbon or nitriding to form a hard TiN/Ti₂N coating,^36,39  however there is a possibility of coating delamination and failure under higher loads.

Noble metals such as Au and its alloys have been used in restorative and conservative dentistry and orthodontics for over 4000 years, because of a favourable combination of corrosion resistance, durability and ease of processing. In its pure form, gold is very soft, with a large elongation (∼ 45%) and a low proof stress of ∼30 MPa. For this reason, Au and its alloys can be easily cold worked or cast to fill a cavity, although the softness of Au and some Au alloys (with >85% of Au) makes them unsuitable for load-bearing applications (e.g. tooth crowns or cups) as an implant made of such a material would not be able to withstand typical masticatory forces. Alloying Au with small quantities of Cu or Pt of less than 4% can provide additional strength to the alloy, without negatively affecting its corrosion profile. Since in restorative dentistry appearance plays a role, Ag may be added to the alloy to alter its colour. The melting point of Au alloys increases with the addition of greater levels of Pt and decreases when a small amount of Zn is used. Depending on the alloy composition, oxides containing Cu and/or Ag will form spontaneously on the surface of Ag–Pd–Cu–Au, Ag–Cu–Au, and Pd–Ag–In–Sn alloys. Pd-based alloys, such as Pd–Ag and Pd–Ag–Au, tend to have greater elongation compared to Au–Pd and Au–Ag–Pd alloys, whereas their mechanical yield strength is similar.

An alternative to Au tooth filling is an amalgam due to its lower cost, ease of application, strength, and durability. Dental amalgam is produced by mixing liquid Hg with an Ag-based alloy containing Sn (< 29 wt%) and solid particles of Cu (<6 wt%), and small quantities of Zn (<2 wt%), Hg (<3 wt%) and sometimes other metals, following the amalgamation reaction Ag₃Sn + Hg ↔ Ag₃Sn + Ag₂Hg₃ + Sn₇Hg. The resultant plastic material is highly deformable, and thus can be used in restorations of larger size or complexity, e.g. where the margins of the cavity reside within dentine or cement. After approximately 1 h, the amalgam hardens to 25% of its final hardness level achieved after >24 h. A protective oxide containing SnO₂ encapsulates the fully hardened alloy. Over the years, there has been a lot of debate about the potential toxic effects of Hg, including reports of neurotoxicity (Alzheimer's or Parkinson's diseases), nephrotoxicity and chronic fatigue, yet large scale studies have proven to be inconclusive and this material is still being used today.⁵⁷ 

In recent years, shape memory effects and pseudoelasticity in some alloys (e.g. NiTi alloys) have been recognised as valuable properties that can advance a number of surgical procedures. Pseudoelasticity allows the material to undergo large deformation when a load is applied, subsequently recovering is original form once the load is removed. The shape memory effect also allows the material to undergo significant deformation at temperatures below its phase transformation temperature, recovering its original form after the temperature is increased. This is because the application of stress or temperature to an equiatomic (or nearly equiatomic) NiTi alloy leads to a reversible phase transformation between symmetrical B2 austenite and less symmetrical monoclinic B19′ martensite, with the martensite existing in either a stress–induced or stress–free conformation (see Figure 1.8). The diffusionless martensitic phase transformation can be induced by cooling the alloy to below the martensite transformation temperature, T_M, or, at higher temperatures, by applying a load since the application of stress causes an upshift in the value of T_M. Because the lattice size and symmetry of martensite is very different to that of austenite, the phase transformation induced by quenching from high temperature results in the introduction of internal strain energy in the martensitic phase, which is reduced by twinning, with no macroscopic shape deformation taking place. Deforming the twinned martensite to a desired shape will result in the detwinning of the structures, as twins provide a favourable path to accommodate the dislocations when compared to slip, and the material will hold this shape as long as it remains below the temperature for austenite phase transformation, T_A start. Subjecting the material to temperatures above the T_A finish will recover the B2 structure very rapidly, and subsequent cooling to below T_M finish returns the material to the twinned or detwinned structure, corresponding to the shape of pre-deformed or deformed martensite, respectively. Both irreversible and reversible shape memory effects have their advantages. For example, a bone plate that contracts at body temperature may be used to exert a compression force onto the bone and promote healing; similarly, NiTi rods can be used to correct the relative position of spine vertebrae as a means to treating scoliosis. However, cycling between distinct shapes may be beneficial for active devices that are designed to respond to changes in their environment. Subjecting the latter type of material to repeated cycles of temperatures or to temperatures above a certain level (determined by the nature of the alloy) will lead to a partial or complete loss of memory.

In pseudoelasticity, stress from the load, rather than the temperature, drives the reversible phase transformation of the symmetrical austenite to twinned martensite and its subsequent detwinning at temperatures well above T_A finish, and the recovery of the original austenite crystal symmetry and macroscopic shape after the load is removed. The stress-induced deformation is highly non-linear since the austenite phase does not lend itself to deformation as well as martensite, due to the lower elastic modulus of the latter phase. Pseudoelastic behaviour is useful for the deployment of stents, since a stent can be severely deformed to facilitate implantation, then, once the conduit is removed, will regain its form. The austenite form is made to be a little greater in diameter than the blood vessel so that there is a force on the vessel wall to resist compression due to, for example, muscle contractions. Unlike the case with stainless steel, the stents made of NiTi do not need to be overexpanded to prevent elastic recoil. The forces produced as a result of constrained recovery are also used in orthodontics to treat teeth overcrowding, and in dentistry, where NiTi is used for cleaning and shaping of the root canal.

When used as implants and surgical tools, NiTi and other shape memory alloys are designed so that the temperature at which the material finishes the martensite to austenite transition is well below that of the human body. The alloy composition will have a direct effect on the T_A start, T_A finish, T_M start and T_M finish temperatures, the maximum value of strain that can be thermally recovered, and the hysteresis characteristics, with T_M start and T_M finish being particularly affected by the nature and relative amounts of the alloying elements. For example, adding Cu (10 wt%) to NiTi reduces transformation and pseudoelastic hysteresis, and limits the sensitivity of T_M to the elemental composition of the alloy; in contrast, the addition of Nb widens thermal hysteresis, and the introduction of Pt, Pd, Hf, Au or Zr shifts the transformation temperatures significantly higher compared to TiNi. The addition of Ag may provide antibacterial activity against common implant-associated pathogens, while also slightly increasing the corrosion resistance, elongation and tensile strength of the alloy, whereas Hf implantation may render TiNi more wear resistant owing to the presence of HfO₂ in the protective oxide layer. Increasing the fraction of Hf may positively contribute towards fatigue resistance. Corrosion resistance may also improve with the addition of Mo.

Apart from NiTi (developed in the 1960s), shape memory behaviour has also been observed in AgCd and AuCd (developed in 1930s), Cu-based alloys (e.g. CuAlNi, CuAlBe, CuSn, CuZr), Fe-based FePt and FePd, and others, e.g. InTl, NiAl, MnCu, FeMnSi, and magnetic NiMnGa, etc.^58,59  However, the manipulation of alloy chemistry is limited by the need for the material to remain biocompatible and resistant to corrosion. The alloy microstructure will also affect both its transformation temperatures and key properties. NiTi alloys with a small grain size (< 100 nm) have been shown to have lower transformation temperatures, with a grain size below 50 nm effectively suppressing transformation to martensite due to the effect of nanoscale grains on B19′ martensite morphology.⁶⁰  Crystallinity also affects the shape of the stress–strain response, with polycrystalline coarse-grained NiTi showing a plateau, as opposed to a sharp rise that is evident in the nanocrystalline NiTi response. When compared to other metallic implants, the plateau stress of NiTi is on par with their yield strength, but it is more corrosion resistant and biocompatible than stainless steel. In addition to the use of solid NiTi implants as rods, nails, plates, and vertebrae spacers, scaffolds made of porous shape memory alloys are being investigated for their ability to promote bone tissue remodelling and vascularisation while being strong and lightweight.

In addition to cardiovascular and orthopaedic applications, NiTi alloys are also being used in neurosurgery (as coils, stents and guidewires), in the treatment of urethral, oesophageal, rectosigmoidal, and prostatic constrictions, removal of bile duct obstructions, kidney stones and foreign materials, and for fast closing of wounds during surgery. When permanent NiTi implants are subjected to cyclic loads, fatigue may lead to premature failure, even more so than in stainless streel, Ti- and Co–Cr-based alloys. This type of failure has been observed in NiTi stents implanted into peripheral arteries, e.g. superficial femoral and femoropopliteal arteries, where they are subject to loads associated with systolic/diastolic pressure cycles and limb movement.⁶¹  The nonlinearity of the pseudoelastic response makes fatigue assessment and alloy optimisation challenging. Care should also be taken when considering NiTi for load-bearing implants with architectures containing multiple fine elements, since the possibility of the structural failure of these devices may be greater. The possibility of the leaching of Ni and associated toxicity needs to be considered for applications that would increase the possibility of implant wear and corrosion.

In the case of alloys made of Co–Cr, Ti and stainless steel, efforts have gone into rendering them to be as corrosion resistant as possible, Mg- and Fe-based alloys are intended to undergo a controlled break down under physiological conditions. The use of these materials is primarily aimed at applications where tissue healing and regeneration is expected, and can in fact be promoted by gradually removing the mechanical support provided by the implant. Among the obvious examples are plates and screws that hold a bone together after fracture, and stents that provide radial support to a blood vessel during its remodelling. Once the healing process is complete, the presence of the implant can be detrimental for many reasons. Examples range from providing a suitable platform for biofilm formation and interfering with load transfer (e.g. bone plate), to platelet attachment and thrombus formation (e.g. stent), to releasing potentially allergenic and cytotoxic products of corrosion, or being a source of chronic inflammation, physical discomfort and pain for the patient.

The biocompatibility of Mg, Fe and their alloys stems from their lower density and Young's modulus, which makes them more mechanically compatible with bone, and the ability to form apatites on their surface, which promotes healthy bone formation. However, pure Mg is prone to rapid degradation under in vivo conditions, particularly in environments that are rich in Cl⁻ ions, with the rate of degradation affected by the properties of the implant and that of the environment. Although the products of corrosion (e.g. Mg^{2 +}) are safely metabolised by the body, the concomitant generation of H₂ and alkalisation of the peri-implant milieu may prevent direct tissue contact, whereas significant loss of material may lead to a rapid reduction in the mechanical strength and implant failure. The principle means to match the degradation rate with the tissue regeneration requirements include varying the alloy composition (see Figure 1.9), controlling the grain structure or surface modification. For example, adding a small amount of individual rare-earth elements (e.g. Gd, Nd or Y), or their combination (e.g. Nd, Ce, Dy, La and Pr) can considerably reduce the corrosion rate of MgZnAl alloys. Here, Zn and rare-earth metals are integrated within a protective oxide, increasing their stability when in contact with biofluids by limiting hydration and magnesium egress of the inner layer. In turn, Al may stabilise the hydroxides on the surface of the alloys under chloride-rich conditions, effectively slowing their corrosion. Surface-immobilised amorphous calcium phosphates may further reduce the rate of corrosion.

As a non-essential metal, exposure to Al particles may result in undesirable physiological response, although low doses of Al may be well tolerated. It is worth noting that humans are exposed to potentially large doses of Al and other non-essential elements in their daily lives; for example, Ti and Al are frequent ingredients in sunscreens and antiperspirant deodorants. However, the possible effects of prolonged exposure to rare-earth elements have yet to be fully explored, but La and Ce may be cytotoxic, with Gd and Dy offering the best combination of toxicity and solubility. Some efforts have been made to reduce or fully replace rare-earth elements in Mg alloys with, for example, Ca or Zr.

As an alternative to alloying, a variety of surface treatments have been trialled to either stimulate the spontaneous formation of protective oxides (e.g. compact Mg(OH)₂ + Mg₂Al(OH)₇ via hydrothermal treatment), or to apply a slowly degrading polymer or amorphous SiC coating. To function properly, the coating should be well-attached (including when the implant is subjected to deformation), defect-free and not contain any substances that may be harmful to cells at any point in their degradation.

Whether they are used for orthopaedic, dental, or cardiovascular applications, most metallic biomaterials and device components are expected to withstand corrosion, with the exception of biodegradable materials, where controlled corrosion of the metal is part of their function.⁶²  Corrosion is a coupled oxidation–reduction reaction, whereby the oxidising agent gains an electron and the reducing agent donates an electron. In aqueous environments, e.g. biofluids, the process of corrosion involves the establishment of electrochemical corrosion cells and is governed by thermodynamic and kinetic factors that determine the corrosion tendencies and the corrosion rate, respectively.⁶³ 

Traditional inert materials, including base metals and alloys based on cobalt–chromium, titanium, and stainless steel materials, tend to donate electrons in solution, and as such, are all likely to corrode when exposed to fluids in the peri-implant milieu, an inherently corrosive environment rich in oxygen-containing chemical species. As a result of these corrosion reactions, metals undergo dissolution, and metallic ions are produced in the process. The rate and chemistry of these reactions is dependent on the chemistry of the material and the specific characteristics of the environment (see Table 1.6). For example, when commercially pure titanium and Ti6Al4V alloy are exposed in vitro to the presence of fluoride F⁻ ions at concentrations typical of the oral cavity, the former is subject to pitting corrosion whereas the latter is affected by general corrosion with characteristic surface micro-cracks.⁶⁴ 

The initial instance of surface corrosion upon first exposure to an in vitro or in vivo environment is generally followed by the spontaneous formation of a thin barrier film, often an oxide or hydroxide if corrosion takes place in biofluids, which subsequently protects the underlying bulk metal from further corrosion and degradation. Failure to form such a layer, e.g. due to unfavourable environmental conditions, or its subsequent distortion, e.g. mechanical damage, would render the material vulnerable to corrosion and related failure. The balance between the destruction and reforming of the protective passive layer determines the predisposition of the implanted metal to a specific type of corrosion (see Figure 1.10), e.g. pitting, stress corrosion, corrosion fatigue, and crevice, intergranular and fretting corrosion.⁶³  Given the complexity of chemical and physical processes that can lead to degradation in vivo, it is not unusual for biomaterials to show signs of more than one type of corrosion, with factors that affect the outcome varying from those specific to the electrochemistry of the biofluid, to the type of mechanical loading the implant may experience, to the surface finish and geometry of the implant itself.⁶³ 

General corrosion is a process that describes the uniform loss of material across its entire surface-environment interface. In metallic materials used in vivo, general corrosion takes the form of passive corrosion, the rate of which is defined as the difference between the rates of passivation versus that of passive layer dissolution. Typically, for implants, this rate is low, often not exceeding 1 pm per year, and is unlikely to affect the mechanical properties of the implant. Nevertheless, even at this rate, biochemically non-trivial quantities of potentially toxic Cr, Co and Ni ions may be leached into the peri-implant milieu, affecting tissue health and regeneration. Furthermore, this corrosion rate can increase in the presence of certain chemical species. For example, the corrosion of Ti is greatly accelerated when fluoride ions or hydrogen peroxide (both used in dental care) are added to the solution.^64,65 

A much greater rate of corrosion can occur when the passive layer is removed, e.g. due to physical damage or changes in the local pH, resulting in the localised corrosion of affected areas. Certain chemical sites in the passivation layer can make it more susceptible to pitting corrosion due to increased levels of dissolution at these sites, especially when in contact with liquids containing halide ions. For example, impurities in the metal serve as nucleation sites at which pits are formed. Under mild conditions, these formed pits are likely to be repassivated, even though they can propagate in a metastable fashion, remain small (10–100 nm) and do not cause macroscopic damage (see Figure 1.11).⁶⁶  Once the conditions reach a certain threshold, termed as the pitting potential, stable pitting corrosion may ensue. It should be noted that the nucleation of pits can take place under a wide range of conditions below the pitting potential. Furthermore, the presence of metastable (repassivated) pits plays a role in stress corrosion cracking, which will be discussed later.

The formation of metal hydroxides and localised drops in pH can be expressed as M^{n +} + nH₂O → M(OH)_n + nH⁺. Local acidification of fluids within the pit is the main driver of corrosion and pit growth, while the geometry of the pit prevents the effective influx of dissolved oxygen from outside of the pit. As a result, cathodic processes that take place on the unaffected passivation layer are physically separated from the anodic processes within the pit. The considerable levels of metal cations generated as a result of fast corrosion within the pit cannot effectively diffuse out of the pit, and anions from the surrounding environment, e.g. ions of Cl or F, are drawn into the pit instead to maintain electroneutrality. For chloride ions, the relevant reaction can be expressed as M^{n +} Cl_n⁻ + nH₂O → M(OH)_n + nHCl.

The resulting inter-pit environment is highly conducive to corrosion processes, and the pits themselves can provide a source of metal ions at potentially toxic concentrations. When considering the impact of pitting corrosion of an implant, it is important to consider the susceptibility of the material to depassivation as well as the frequency of pit formation over time. For example, where stainless steel is more prone to depassivation compared to Ti and as such has a greater frequency of pit nucleation, over time this frequency reduces to zero as the repeated nucleation and metastable propagation eliminate the sites at which pits tend to form, e.g. sulfide inclusions on the surface of stainless steel. On the surface of Ti, however, the pit nucleation remains at a low frequency over time, suggesting that for this metal, nucleation is not driven by inclusions but rather by the inherent properties of the titanium oxide layer, and events such as explosive microscopic rupture of the layer perpetrated by the accumulated chloride at its interface.^66–68  The frequency of pit nucleation and the probability of pit propagation turning from metastable to stable is also dependent on the characteristics of the ambient environment, particularly the concentration of corrosion-promoting ions, reduced levels of dissolved oxygen, the great oxidising potential of the reactive species, and increased temperature and reduced pH. Operating conditions under which implants are exposed to cyclic loads may increase their susceptibility to pitting corrosion, such as is the case with Co–Cr-containing metallic alloys that undergo pitting corrosion under these conditions but not under static load. Materials processing, for example certain types of cold working, may also render the alloy more susceptible to pitting corrosion. The introduction of other metals into an alloy may change its susceptibility to pitting corrosion. For example, a small amount of Mo added to steel renders (as in 316L) makes them more resistant to pitting, substantially reducing the number of pitting-related failures. In contrast, adding elements to Ti (as in e.g. Ti6A14V and Ni–Ti shape memory alloys) makes it subject to pitting corrosion under certain medically-relevant conditions, as it creates potential sites of weakness in the passivation layer. Where the pitting is generally superficial in the case of Ti6A14V, the consequences of pitting corrosion in the case of Ni–Ti are more concerning and may result in significant levels of ion release. It should be noted that predicting how the implant will corrode in vivo is not trivial, since the outcomes of in vitro studies may not always correlate with those later observed in vivo in an animal model or in a patient. For example, while type 304 stainless steel undergoes extensive pitting corrosion after days of incubation in 0.9% NaCl (due to localised acidification following the hydrolysis reaction Fe^{2 +} + 2H₂O + 2Cl⁻ → Fe(OH)₂ + 2HCl), a Sherman plate fabricated from the same alloy and retrieved from a patient after 40 years of implantation showed no visible signs of corrosion.⁶⁹  At the same time, other studies have reported significant levels of in vivo corrosion in stainless steel, some leading to severe implant failure due to a combination of pitting attack, fatigue striations, and microcracks (see Figure 1.12).^70–72 

Another type of localised corrosion, i.e. crevice corrosion, takes place when the inflow and outflow of chemical species is reduced due to the geometry of the space created by two surfaces coming into close proximity with each other, for example at the interface between the head of fixing screws and bone or plate (see Figure 1.13). This is because the formation of the protective oxide relies on the availability of dissolved oxygen, whereas the dissolution processes are promoted by increased concentrations of halide species. Just as in the case of pitting corrosion, the narrow dimensions of the crevice (typically in the order of 3 mm) prevents the convective removal of metallic ions and effective delivery of oxygen, with cathodic and anodic processes spatially restricted to the oxide surface and inner part of the crevice, respectively, since the rate of diffusion-driven mixing is not sufficiently high enough to be effective. In physiological fluids, this typically results in an influx of Cl⁻ and similar anions into the crevice, and a significant local drop in the pH. The latter will further promote corrosion, and consequently increase the number of metallic ions released into the crevice that will draw even greater quantities of anions into the pit to maintain electroneutrality, reducing the pH even further. Since the chemical mechanism of crevice propagation is very similar to that of pitting corrosion, similar characteristics of the environment will work to promote or hinder it.

In addition to the effects already discussed, certain types of porous implants may also display increased susceptibility to crevice corrosion. For example, even though titanium is generally considered to be immune to crevice corrosion under physiological conditions, oxide films that form on porous titanium have inferior stability compared to those on solid Ti, with the point of initial failure lying within the depths of the pores. In addition to reduced convectional flow of ions and dissolved oxygen, galvanic coupling to the outer surface promotes a fast dissolution of the material within the pores.⁷³  The geometry and size of the pore, and the degree of porosity and connectivity between the pores also has a notable effect on the corrosion behaviour of cobalt alloys.⁷⁴  Furthermore, even materials considered to be relatively immune from pitting corrosion under normal conditions, e.g. 316L, 316LVM and austenitic microstructured steel and Co–Cr–Mo, may nonetheless remain prone to crevice corrosion to some degree.⁷⁵  When investigating their potential for crevice corrosion, it is important to consider the time over which testing takes place, specifically, the observation period should be sufficiently long for the outcomes of slow processes to come to light, as well as to consider the dynamic nature of the in vivo environment – the latter may change over the course of implantation, and vary between patients.

When coupled with mechanical stress, the above described corrosion reactions can lead to crack propagation and eventual failure of the material, a process termed as stress corrosion cracking or corrosion fatigue depending on whether the loading is static or cyclic in nature. The main issue with both types of failure is that they may proceed slowly (10⁻⁹ to 10⁻⁶ ms⁻¹) with little visible evidence up to the point when fracturing occurs. The action of the mechanical and corrosive factors is synergistic in that when applied independently neither of the factors would be significant enough to induce failure. That is, mechanical loading of the pre-corroded metal or alloy at corrosion-generated or manufacturing flaws, at the same level of stress as in stress corrosion cracking, but in the inert environment will not lead to crack propagation. In contrast, the same loading under a corrosion-inducing environment would lead to time-dependent crack propagation. It should be noted that the level of stress concentrated at these surface flaws is below the critical value for stress-intensity factor at which macroscopic mechanical fracture of the material would occur under inert conditions.⁷⁶  Both external and residual tensile stresses can cause crack initiation and propagation, although residual stresses that are compressive in nature may provide some degree of protection from this type of failure. The environmental factors that contribute to crack initiation in stress corrosion cracking and corrosion fatigue may differ. Nevertheless, the dynamics of the evolution from localised corrosion to cracking are affected by electrochemical processes that take place at the site of corrosion, the presence of specific species that tend to promote corrosion in a given metallic system, the properties of the material (including its processing, e.g. heat treatment), and the nature of the stress–strain system. Superior tensile and fatigue properties in air do not necessarily correspond to reduced susceptibility to stress corrosion cracking or corrosion fatigue. For example, a high-strength low-alloy Mg alloy, MgZn1Ca0.3 (ZX10), showed improved mechanical properties in air when extruded at 325 °C compared to when extruded at 400 °C due to grain-boundary strengthening, yet both materials were found to be susceptible to crack propagation under slow strain tensile and cyclic tension–compression loadings in simulated body fluid.⁷⁷ 

It should be noted that corrosion processes that take place on a cathode, e.g. evolution of hydrogen, absorption, diffusion, and embrittlement, as well as those on an anode, e.g. active dissolution and elimination of metal from the crack tip, can participate in crack initiation. Crack propagation relies on interatomic bonds to be broken, which can result from chemical processes (e.g. dissolution) as well as ductile and brittle mechanical fracture (of metals and oxides, respectively) induced by such events as the adsorption of chemical species from the peri-implant milieu, reactions at the surface and in the metal ahead of the crack tip, and others.⁷⁶  Corrosion sites and flaws, e.g. discontinuities, where the tip of the crack is sharp are more likely to undergo crack propagation as stresses can be concentrated at the tip, as in the case of metastable pitting where rapid repassivation ensures that crack tips remain sharp (see Figure 1.14). In contrast, blunt cracks from general corrosion or stable pitting are less likely to result in stress corrosion cracking. For this reason, it is often metals that resist general or active pitting corrosion that succumb to stress corrosion cracking and corrosion fatigue.

In general, cyclic loads may lead to crack propagation at levels below that of static loads. Furthermore, there is no cut-off point, i.e. fatigue limit, that needs to be breached for the failure to occur, and the probability of failure due to corrosion fatigue is greater at lower cycle frequencies, which is the case for many biomedical implants, such as those used to restore hip and knee joints. The length of the implantation makes failure due to corrosion fatigue more likely, as it extends the amount of time the stressed material is exposed to the corrosive action of bodily fluids. For some materials, such as steel, reducing the probability of pitting corrosion may have a considerable effect on improving their corrosion fatigue resistance.⁷⁸  In addition to alloy chemistry, phase composition and shape also play an important role in defining corrosion resistance.⁷⁹  The extent to which corrosion fatigue and stress corrosion cracking present a real problem for implants remains a subject of debate, with many authors finding no evidence that either of these mechanisms of failure contribute significantly towards loss of performance in real-life implants.⁷⁹ 

Under conditions that favour increased hydrogen evolution at the surface of the intact passivation layer, e.g. when the metal in question is galvanically coupled to another metal with greater activity, hydrogen atoms can interact with the metal lattice, leading to the formation of brittle metal hydrides that are prone to cracking under applied or residual stress. Cobalt-based alloys, e.g. Zimaloy used in orthopaedic applications, showed a notable loss in ductility when incubated in a saline solution under a cathodic potential that can be attained in a crevice sample after 60 days of incubation.⁸⁰  In vivo hydrogen embrittlement was suggested to be the cause of severe corrosion attack on the mating interfaces of retrieved hip-implants with Ti6Al4V/Ti6Al4V modular taper interfaces in the stem. The surfaces of the modular connections showed signs of etching, pitting, delamination, and surface cracking, as well as the precipitation of brittle hydrides as a result of crevice and fretting corrosion, where the oxide layer was mechanically damaged.⁸¹  Naturally, Ti and other materials that rely on the protective effects of an oxide layer are subject to a considerable reduction in fatigue strength under fretting conditions, with the nature of the environment playing a less important role when the damage from fretting is more severe.⁸²  Increasing the fretting resistance, for example by implanting nitrogen into the surface of Ti6Al4V, and changes to the alloy grain structure through, for example, heat treatment, may reduce the susceptibility of some materials to fretting corrosion.⁷⁸  This is because cracks are observed to propagate along the boundary lines between the acicular α and β phases.

From a chemical standpoint, unlike pitting or crevice corrosion where the removal of the protective oxide is typically brought on and then accompanied by considerable changes in chemistry of the fluid in the immediate proximity of the site, infrequent removal of the oxide by mechanical means only results in a minimum level of corrosion needed to accomplish repassivation. Once the instances of mechanical damage become recurrent due to wear or fretting, the repassivation may fail to occur fast enough, resulting in a considerable loss of material due to corrosion (see Figure 1.15). The mechanical damage may originate from fluid–surface interactions under certain flow conditions, similar to the way flowing rivers erode bedrock, as well as from physical contact between solid surfaces. Where erosion corrosion may in principle take place in biomedical devices used to control the flow of biofluids, fretting corrosion is more likely to be encountered in multi-component load-bearing devices, e.g. orthopaedic implants. It is quite possible that in these devices, more than one type of corrosion takes place. For example, in joint implants, a fine layer of fluid typically lubricates the closely fitting surfaces, creating the potential for crevice corrosion events to co-exist with fretting events. The release of metallic particles and ions can be significant under these conditions, affecting the health of the surrounding tissues, with the potential to induce systemic damage to the patient. The loss of material may also compromise the mechanical integrity of the site where the implant is fixed to the bone, leading to implant loosening and compromising the correct load transfer across the interfaces and the bulk of the implant. Thus-stressed materials can be more susceptible to failure due to corrosion fatigue processes. The debris generated in the process of fretting corrosion, e.g. oxide particles of various dimensions, tend to accumulate in the peri-implant milieu, thereby increasing stress.⁸³  If trapped in between the two rubbing surfaces, these particles may enhance damage via abrasion.

Although increasing the hardness of the material reduces the damage it is likely to sustain from fretting wear, most Ti and Co–Cr alloys, and stainless steels are subject to fretting due to shearing micro-movements at the implant–cement or implant–bone interface.^84–86  The chemistry and physical characteristics of the environment may also affect the likelihood of fretting corrosion, although the exact effects may differ between different types of materials. For example, when subject to sliding wear, fretting corrosion resistant alloys based on Ti or Co–Cr (namely Ti6Al7Nb and Co28Cr6Mo) show little change when the pH is varied from 2.5 to 6, whereas the fretting corrosion resistance of high-nitrogen Fe22Cr10NiN and AISI 316L stainless steel reduces with decreasing pH.⁸⁷  Interestingly, when a 316L SS/Ti6Al4V material with a thin layer of Ringer's solution was submitted to fretting conditions, stainless steel was shown to endure a lower level of fretting damage when compared to Ti6Al4V.⁸⁸  The level of damage at each surface is determined by the friction conditions at the contact and the energy that is dissipated at each cycle, as well as the lubricating and corrosive properties of the fluid that separates the two surfaces (see Figure 1.16).⁸⁸  Cathodic protection may significantly lessen the extent of damage related to fretting, and prevent the transition between fretting regimes. Indeed, the electrochemical potential plays a critical role in defining the fretting corrosion in Ti-based alloys, as was demonstrated using a Ti6Al4V sample in contact with an alumina ball at an amplitude of micro-motions of 50 and 100 μm in a 0.9% NaCl solution.⁸⁹  The effect of the applied potential on the wear was particularly pronounced at potentials above −0.2 V Ag/AgCl due to the enhanced oxidation of the third body that is produced as a result of plastic deformation in the metal at higher anodic potentials, requiring less mechanical energy to induce wear damage. As with other types of corrosion, the presence of certain alloying elements may render the material more resistant to chemical dissolution, yet their effect on the mechanical properties of the material may not necessarily be positive, calling for a trade-off. Mixed phase α/β alloys, such as Ti6Al4V and Ti6Al7Nb, may display better characteristics with respect to resistance to corrosion and wear, whereas commercially pure titanium and the near-β Ti13Nb13Zr and β Ti15Mo alloys may show superior corrosion resistance among the group.⁹⁰ 

In some applications, e.g. in restorative dentistry, it may be necessary to place two dissimilar metallic materials into direct electrical contact. In the presence of a corrosive fluid, e.g. saliva rich in acids, salts, proteins and enzymes, galvanic corrosion may take place. This type of corrosion is characterised by changes in the rate of the corrosion on these materials, with the corrosion being promoted on a less immune material and suppressed on a more noble material. This is because the difference in the corrosion potential of these materials drives a flow of electric current through the metal/metal junction, as well as through fluids in the peri-implant milieu and in tissues. This can lead to the formation of a galvanic cell that promotes corrosion on the less noble metal. The reactions involved include cathodic reactions that reduce dissolved oxygen in the electrolyte (i.e. O₂ + 4e⁻ + 2H₂O → 4OH⁻ and O₂ + 4H⁺ + 4e⁻ → 2H₂O) and anodic reactions that dissolve the metal following M → M^{n +} + ne⁻.

Galvanic coupling is held responsible for the intergranular corrosion of dental amalgams, i.e. the multiphase alloys commonly used in restorative dentistry. In amalgams, the different phases have distinct electrochemical properties, with corrosion resistance decreasing in the order of Ag₂Hg₃, Ag₃Sn, Ag₃Cu₂, Cu₃Sn, Cu₆Sn₅, and Sn_7–8Hg.⁹¹  As the most susceptible to corrosion, the latter phase (also known as the γ₂ phase) contributes most towards the observed corrosion in a low-copper amalgam system in an oral environment, whereas the γ and γ₁ phases show greater corrosion resistance. Of most concern is the release of potentially cytotoxic mercury as a result of the reaction Sn_7–8 + 0.5O₂ + H₂O + Cl⁻ → Sn₄(OH)₆Cl₂ + Hg. The free Hg may also interact with the γ phase to generate additional γ₁ and γ₂ phases. In the case of high copper alloys, γ₂ is not present in the final amalgam. However, while the ή phase that is formed instead shows better corrosion resistance than γ₂ and its breakdown does not generate Hg, it is still the most susceptible to corrosion among the phases, interacting with saliva in the manner of Cu₆Sn₅ + 0.5O₂ + H₂O + Cl⁻ → CuCl₂ · 3Cu(OH)₂ + SnO.

Outside of amalgams, the passivation of metals greatly reduces the probability and practical consequences of galvanic coupling, since the corrosion potential of the passivation layer on the less noble metal would be that of a more noble metal.⁸³  Having said that, physical or chemical damage to the protective layer would expose the underlying metal, thereby increasing the probability of galvanic corrosion.⁹²  For example, when galvanic coupling of spontaneously passivating CoCr, CoCr-c, NiCrTi, gold–palladium and Ti6Al4V dental alloys to titanium grade 2 implants was investigated, the passive domain of the NiCrTi alloy was found to be too narrow in the presence of fluorides to afford the needed level of protection, releasing potentially dangerous levels of Ni ions into the milieu.⁹³  This study also highlighted the importance of considering multiple environmental factors when evaluating the probability of galvanic corrosion in metallic couples. For example, lowering the pH of artificial saliva from 6.5 to 3 did not result in significant changes in the electrochemical behaviour of TiG2, Ti6Al4V, and CoCr alloys, and their respective couples, with Ti/Ti6Al4V and Ti/CoCr showing the most stable behaviour. However, a combination of low pH (< 3.5) with high fluoride content (1000 ppm) triggered the active dissolution of TiG2 and Ti6Al4V, with galvanic corrosion observed in CoCr–c/TiG2 couples.⁹³  The processing of a material may affect its purity, geometry, grain structure, surface morphology and internal stresses, thus potentially affecting its susceptibility to galvanic corrosion. For example, in the case of cast and machined Ti-coupled oral implants, the ion release is 40 and 25 ng ml⁻¹ for machined and cast samples, respectively, after 560 h of incubation in saliva.⁹⁴  Even though both materials release ions, it is at a much lower concentration when compared to Cr alloy (76.5% Ni, 14.3% Cr, 5.6% Al, 2.3% Mo, 0.8% Si y 0.5% Fe), at 200 ng ml⁻¹ after only 10 h of immersion. Ti6Al4V/CoCr and Ti6Al4V/CrNi show similar performance when incubated in Ringer's solution.⁹⁵  Significant levels of potentially cytotoxic Cu are also released from coupled palladium (73.7% Pd, 14.9% Cu, 11.4% Ga) and gold alloys (50.0% Au, 31.5% Cu, 13.0% Ag, 5.4% Zn) in saliva,⁹⁴  whereas a Ti6Al4V/Au couple is likely to resist galvanic corrosion.⁹⁵  The galvanic cell phenomenon is often compounded by localised crevice corrosion that is due to the geometry of the assembly.⁹⁶ 

Among the implant designs and features that make them more susceptible to corrosion, modularity ranks highly with regard to its potential to induce clinically-relevant negative outcomes. While many of the early implant components were monolithic, contemporary implants for total hip or total knee arthroplasty often feature modular components. This is because modular components such as acetabular liners and femoral necks better support the unique anatomical structures of the patient, optimising the biomechanics and addressing limb length discrepancies, and enabling revision surgeries that only replace damaged components while retaining well-fixed components within the patient. As such, implant modularity can deliver substantial benefits to the patient with respect to implant performance and post-surgery recovery. However, modular junctions create additional crevices where corrosion can take place, not only for materials with dissimilar properties, e.g. when Ti contacts Co–Cr alloys, but even in instances where the same material with different surface finishes are used. The junction can vary in shape and size, angles at which they are tapered, surface roughness, tolerances, presence of such features as skirts and so forth. For example, a modular head–neck assembly in femoral total hip replacement prosthesis typically has a tapered conical junction between two metals, or between a metal and ceramic, where the head and stem have a tapered bore and cone, respectively. Stems in these assemblies are frequently made of Ti6Al4V or Co–C–Mo and heads are made of Al₂O₃, ZrO₂, or Co–Cr–Mo.⁹⁷  Head–neck taper corrosion can lead to increased levels of Co and Cr in the peri-implant tissues (11.2 and 2.2 ppb, respectively), with patients complaining of adverse local tissue reactions ranging in intensity from tissue discoloration to collection of fluid and lymphocytic infiltrate, soft tissue damage and the appearance of pseudotumours,⁹⁸  with some cases requiring surgical revision.⁹⁹ 

Modularity-related failure is complex and may comprise multiple processes. For example, in assemblies that use titanium alloy stems with Ti alloy or Co–Cr modular necks, both fatigue failure and fretting corrosion processes can be detected (see Figure 1.17).^100,101  The use of neck-preserving stems has been shown to reduce fatigue and fretting corrosion by limiting torsional and bending loads, and thus cumulative stress to which modular taper junctions are subjected.¹⁰²  Certain material combinations (e.g. using Co–Cr neck adapters in place of Ti when using Ti stems) and assemblies that reduce micro-motion at the stem–neck interface during cyclic loading, as well as strategies that prevent interface contamination, may also reduce the incidence of fretting and corrosion.¹⁰³  Improving the tolerances of the interface and application of coatings onto interfacing materials may also positively affect junction stability and corrosion resistance.¹⁰⁴ 

For the effective protection of the bulk metallic implant, it is imperative that the passivation layer is continuous, non-porous and defect-free, and attaches well to the material so as to prevent diffusion of solutions towards the material bulk. The Pilling–Bedworth ratio (R_PB) is frequently used to indicate the probability that the metal will form a sufficiently protective oxide. Defined as the volumetric ratio of the elementary cell of a metal oxide to that of the corresponding metal from which the oxide is formed, a ratio of less than 1 suggests the formation of a thin, discontinuous and/or porous coating with limited protective performance, such as in the case of MgO (R_PB of ∼0.8).¹⁰⁵  Above an R_PB value of 1, the oxide layer is uniform, affording the required protection from subsequent oxidation to metals such as Cr, Ti, and Al. Above an R_PB value of 2, the oxide film once again provides limited protection as the mechanical stresses and strains within the coating result in poor adhesion and a tendency of the oxide to flake off the surface.¹⁰⁵  While a useful indicator for certain types of conditions, the behaviour indicated by the Pilling–Bedworth ratio does not always correspond to that under in vivo and in vitro conditions. Furthermore, for alloys, the sequence of oxide formation is likely to be more complex, and metals that play dominant roles in this process may not necessarily be the ones with the greatest fraction. For example, when the R_PB values were calculated for Ni₃Al, NiAl and NiAl₃, that all form Al₂O₃ on their surface, they were significantly greater than that for pure Al, at 1.71–1.88, 1.64–1.78, 1.48–1.57, and 1.29, respectively.¹⁰⁵ 

The balance between formation and dissolution, and the quality of the passivation layer, depends on the environment. Relatively low levels of dissolved oxygen in bodily fluids are often insufficient to enable efficient repassivation of the metal surface with an oxide. For example, in vivo tests of AISI 316, 316L stainless steels and COP1 alloy exposed to an environment with a typical partial pressure, pO₂, of 28–78 mmHg (corresponding to the values experimentally observed in the venous and arterial blood collected from femoral blood vessels in rabbits) show substantial reduction in the chromium oxide passivation and fatigue durability of these metals, and concomitant increases in the release of ionic iron into the milieu.¹⁰⁶ 

Stainless steels rely on their Cr content for the formation of the protective oxide; at the same time, many steels also have a small fraction of carbon introduced with the aim of improving the mechanical properties of the alloy. The presence of both of these elements creates a challenge when the alloy is subjected to thermal treatment, since higher temperatures promote the formation of a chromium-rich M₂₃C₆ carbide at the grain boundaries, thus depleting these areas of free Cr, creating local weaknesses in the protective oxide layer and facilitating intergranular stress corrosion (see Figure 1.18).¹⁰⁷  Thermally-treated austenitic stainless steel AISI 321 stabilised with Ti has also been shown to be corroded around TiNs, as these sites participate in galvanic coupling with their surrounding continuous phases, with corrosion enhanced due to threshold chromium concentrations being low enough to form a sufficiently strong protective film.¹⁰⁷ 

Chloride ions that are also present in bodily fluids tend to prevent the formation of protective oxides, thereby interfering with this key method of corrosion protection on which many implantable metals and alloys rely. This is because monovalent chloride stabilises suspensoid sols,¹⁰⁸  thus preventing the formation of a protective corrosion product layer on metals. Indeed, the presence of dissolved salts, particularly chloride ions and other halides, in extracellular and intracellular fluids influences implant corrosion in vivo. The exact effects and the products that are formed as a result of these interactions are determined by the chemistry of the metal and of the salts, and their concentration. For example, in the case of magnesium-based materials tested in vitro and in vivo, the presence of CaCl₂ leads to the formation of a protective Ca–PO₄ layer on the metal surface, whereas MgSO₄ affects the topography of the degradation interface, and NaHCO₃ leads to the formation of MgCO₃ as a by-product.¹⁰⁹  In the case of Ti6Al4V, the presence of chloride in the solution promotes pitting corrosion, with the degradation process following the initial formation and consequent breakdown of the protective passive layer, subsequent development of unstable and metastable pits and their repassivation (see Figure 1.19), and the eventual development of stable pits of greater size and their evolution into other forms of localised corrosion.¹¹⁰  The high corrosion rate within sites where repassivation has failed to occur is in part sustained by the cathodic electrochemical reactions that take place on the surface of intact oxide films, with the latter typically showing n-type semiconducting rather than insulating behaviour. The availability of complexing ions, particularly Cl⁻, in the environment has also been shown to significantly reduce the passive zone of Cr, thereby reducing the in vivo performance of stainless steel, since Cr plays a central role in the development of a protective oxide layer on these implants.

The corrosion reaction pathways and their rate are directly linked to the temperature and pH of the environment. In general, an increase in temperature leads to an increase in the rate of chemical reactions. Changes in temperature also affect the equilibrium constant and balance of neutrality, e.g. at 25 °C, [H⁺] = [OH⁻] at a pH value of 7, whereas at 37 °C, neutrality is attained at a pH of 6.81.¹¹¹  Even though the human body has several mechanisms to respond to changes in pH, specifically through fast-acting chemical buffers and slower-acting but more potent physiological buffers that involve complex respiratory and urinary response systems,¹¹²  the very act of implantation may perturb pH balance for days and even several weeks. This is because implantation or pathological conditions are often associated with transient changes in the electrochemical state of the environment due to physical trauma (e.g. hematomas), inflammation, disruption to tissue blood supply, and thus the level of oxygenation and pH in the peri-implant milieu.⁶⁹  For example, the accumulation of hematomas, particularly in patients on anticoagulation or antiplatelet treatment,^113,114  can lead to a significant local acidification of the peri-implant environment (down to a pH of 4.0), resulting in an increased rate of corrosion of the material and severe local damage.¹¹⁵  Once a deep pit or crevice is formed on the surface, it may be difficult to re-passivate due to the restricted access of oxygen to the bottom of the pit, with a recorded pH of as low as 0.5–2.0 at the root of the pit regardless of the level of oxygenation or pH in the surrounding environment.¹¹⁶  This is because Ti^{3 +} dissolved and trapped within the crevice functions as an anode, producing H⁺ via hydrolysis and reducing the level of pH within the compartment. In parallel, to preserve electrical neutrality, Cl⁻ ions migrate into the crevice. Once the pH is sufficiently low, depassivation and active dissolution begins, exacerbated by the presence of chloride ions, and crevice corrosion ensues.

Pourbaix diagrams are frequently used to describe the electrochemical stabilities of materials. These plots show stable phases as a function of electrode potential and hydrogen ion activity of the solution, i.e. its pH, with the solution typically being water at 25 °C.¹¹⁷  The electrode potential, E_H, refers to the voltage potential from the oxidation of molecular hydrogen to solvated protons with respect to the standard hydrogen electrode, as calculated from the standard Gibbs free energies of formation using the Nernst equation. By convention, the standard potential of a reversible hydrogen reaction in a solution, 2H⁺(aq) + 2e⁻ → H₂(g), is set to zero, i.e. chemical potentials for protons and hydrogen at unit activity and fugacity equal zero. Gold and other noble metals have higher standard potentials compared to base metals such as magnesium, and thus are far more stable when placed in an acidic solution when compared to base metals, which will corrode unless they are encased in a protective oxide layer. The lines in the plot represent the equilibrium conditions where the activities for the species on each side of that line are equal (see Figure 1.20); and the regions that these lines separate denote the stable species and at times may include such indicators as immunity, corrosion and passivity, each corresponding to the ability of the material to remain stable, corrode or form a protective layer (oxide or another salt) on its surface.

Although Pourbaix diagrams offer some value when one wants to, for example, estimate the possible effects of changing the composition of an alloy on phase stability, they are generally produced for a standard set of conditions, e.g. a concentration of soluble species of 10⁻⁶ M, a gas pressure of 1 atm, at room temperature, and ignore contributions from kinetic effects. As such, their usefulness in predicting the in vivo degradation of metallic implants is limited. For example, while the fraction where gold and other noble metals show immunity corresponds to their real-life corrosion resistance in water, the region of immunity for common metallic biomaterials, e.g. Zn and Ti,^118,119  lies below the water stability zone, suggesting that these metals would be readily oxidised when placed in water and therefore should not be used in applications where they may come into contact with water. Yet, in real life, the rapid formation of the protective passivation layer ensures protection of the bulk metal, and only in cases where this layer fails to form do we see rapid corrosion of these metals in water. For Ti, this takes place under extreme reducing and extreme oxidising conditions, as these hinder the formation of the oxide and increase the solubility of hydroxides, respectively. Even though, according to the Pourbaix diagrams, these conditions do not occur within the water stability zone, if we consider the kinetic effects of oxidation/reduction of H₂O₂ at the metal electrode, which in the case of Ti are slow, a lack of dissolved oxygen or presence of H₂O₂ may force the reactions into the negative and positive corrosion domains, respectively.¹¹⁷ 

The importance of considering the kinetic effects is particularly evident when one compares the theoretical and actual electrochemical behaviour of Co and Ni. The apparent similarity of their Pourbaix plots suggests that both should corrode readily over a wide range of conditions. Yet, as Ni tends to acquire a passivation layer more rapidly than Co due to relatively slow kinetics of the dissolution reactions of the former, the conditions under which Ni corrodes are somewhat less extensive.¹¹⁷  Furthermore, the reported Pourbaix diagrams do not always correlate well with electrochemical observations, possibly due to the inaccurate estimation of experimental free energies of formation (Δ_fG), an issue that may be at least in part addressed by using more advanced approaches to estimating Δ_fG, e.g. those based on density functional theory (see Figure 1.20).¹²⁰ 

In vitro, biomaterials are commonly tested using artificial biofluids that attempt to reproduce the most critical factors that influence degradation in vivo. While the recipes for simpler solutions such as phosphate-buffered saline or Ringer's solution are standard, when it comes to replicating more complex fluids, such as saliva, there is a much wider variety of recipes (see Table 1.7). Yet, none can truly reflect the complex and dynamic nature of saliva that not only differs between patients, but also may differ in one patient as they progress through their life, have diseases and receive medical and physical therapies. This challenge holds true for other biological fluids, too. As a result, the chemistry of simulated body fluids is typically restricted to matching the concentration of inorganic ions encountered in physiological fluids, but rarely do they contain proteins, amino acids, or vitamins that one would expect to find in natural fluids.

As an example, diabetic patients have different glucose levels in their biofluids, and may also undergo antibiotic treatment due to their greater susceptibility to infections. The presence of sugars at different concentrations can affect corrosion. For example, when the stability of AISI 316L stainless steel was evaluated in the presence of a diabetic-serum pool with different concentrations of glucose and an approximately constant concentration of proteins (6.8 g L⁻¹) and a temperature of 37 °C, the pitting potential of the material decreased with increasing glucose levels.¹²¹  In contrast, addition of the antibiotics ampicillin or cefradine did not affect the pitting potential. Nevertheless, antibiotics often form part of the therapy, and are also integrated in many of the modern implant coatings and bone cements, e.g. SmartSets HV from DePuy International contains gentamicin sulfate at 4.22% w/w. The latter compound has been shown to affect the dynamics of fretting corrosion.¹²²  Interestingly, 316L stainless steel, commercial purity titanium and CrCo alloy show little change in their corrosion potential when incubated in saline containing one of the antibiotics commonly used for the treatment of orthopaedic infections, i.e. tetracycline, tobramycin, clindamycin, cefamandole, bacitracin and chloramphenicol.¹²³  However, the addition of oxytetracycline leads to anodic polarisation and a subsequent reduction in the corrosion rate of the three types of metallic substrate. This can be attributed to the interactions between the complex ring structures of the oxytetracycline with the metals that could have led to sequestration effects and/or the deposition of an inhibition layer. It should be noted that the corrosion protective vs. promoting effect may be dependent on the concentration of the antibiotic in the peri-implant milieu, the presence of other chemical and physical factors, and the chemistry and surface properties of the metal.¹²³  For example, in another study, the related antibiotics doxycycline and the dihydrate and hydrochloride salts of oxytetracycline were found to promote the corrosion of CoCr alloy, while inhibiting the corrosion of as-received titanium.¹²⁴  The corrosion of abraded titanium was inhibited by doxycycline, while for stainless steel inhibition was induced by oxytetracycline dihydrate.

The presence of amino acids that rapidly adsorb onto the surface of the metal can significantly change their polarisation resistance and current densities, with the corrosion outcomes dependent on the composition of the biomaterial. For example, Mg–Ca, AZ31 and AZ91 possess greater corrosion stability in Dulbecco's modified eagle medium (DMEM), a medium containing inorganic minerals and amino acids, than in Hank's solution (which is a mixture of water, inorganic salts and glucose).¹²⁵  This can be attributed not only to changes in electrolyte composition (higher NaHCO₃ and lower NaCl in DMEM), but also to the presence of alanine, glycine and leucine, which reduce the current density and increase the polarisation resistance of the surface. The protective effect of amino acid adsorption has also been shown for steel and aluminium.^126,127  For reducing pitting corrosion in Al, the order of effectiveness of the inhibitors was found to be arginine > histidine > glutamine > asparagine > alanine > glycine.¹²⁸ 

The response was less uniform when foetal bovine serum was introduced into the mix. The rate of corrosion increased for Mg–Ca alloy when foetal bovine serum protein was added to DMEM, while the rate was found to decrease for the AZ91 alloy, and follow an initial increase followed by a decrease for the AZ31 alloy.¹²⁵  A study on the corrosion resistance of Ti alloys in the presence of whole serum also showed an improved resistance in the Ti6Al4V alloy, and reduced resistance in Ti13Nb13Zr and Ti6Al7Nb when compared to phosphate-buffered saline.¹²⁹ 

The nature of protein–surface interface reactions is complex and is governed by van der Waals, hydrophobic and electrostatic forces, and hydrogen bonding. In the case of AZ91, the adsorbed protein layer likely changes the electric field on the metal surface, and reduces the effect of changing pH on the corrosion processes on its surface. When compared to the Mg–Ca alloy, it is possible that the proteins adsorb more easily on the Al₂O₃ layer present on the surface of AZ31 and AZ91 alloys and on coarse particles of partially divorced eutectic Mg17Al12 in AZ91, forming a protective layer that inhibits anodic reactions (see Figure 1.21).¹²⁵ 

In addition to the chemistry of the solution itself, it is important to consider the chemistry of bone cements used in many orthopaedic and dental applications, and the products of degradation that they may release into the peri-implant milieu. For example, an in vitro study on the fretting corrosion of cemented Ultima TPS™ femoral stems showed that the chemistry of the poly(methylmethacrylate) cement affected the corrosion regimes at the stem–cement interface.¹²²  The use of cements containing the BaSO₄ radiopacifier increased the corrosion current under fretting wear, with a greater rate of release of metallic ions when compared to that when ZrO₂-containing cements were used. The corrosion current was further increased under these conditions when a cement loaded with gentamicin sulfate was used. It is possible that BaSO₄ and gentamicin sulfate increase the dissolution rate of CoCrMo and 316L stainless steel, as the presence of sulfite ions is known to render the material more susceptible to local corrosion. Although sulfate ions are thought to inhibit the initiation of pitting corrosion (promoted by Cl⁻), they are known to accelerate the propagation of localised crevice corrosion after the initiation events have taken place, by reducing the critical pitting temperature, hindering CoCrMo repassivation, and increasing the rate at which H⁺ is reduced at the implant surface. The latter may increase the rate at which excess metallic valence electrons are liberated from the surface, and consequently from within the bulk of the metallic matrix. A combination of crevice geometry and elevated temperature at the metal–cement interface, as well as the presence of tiny pores and cracks in the cement, results in a high local concentration of antibiotic-derived sulfates. It possible that barium sulfate particles contribute sulfate ions into the peri-implant milieu through leaching and subsequent dissociation. Their presence may also change the electrochemistry of the environment through other mechanisms, as well as contribute to fretting through abrasion.¹²² 

In addition to the ever changing chemistry of natural fluids, it is important to consider the mechanics of biological flows and how they may affect and be affected by the presence of the implant and the physiological processes in healthy patients and those that may be suffering from relevant medical conditions. Nevertheless, most tests used to evaluate the behaviour of implants in vitro do so under static conditions, due to their relative simplicity and reproducibility. For this reason, the results of static studies often do not represent those of the events that take place once the implant is inserted into a patient. Interestingly, in vitro studies can both over- and underestimate the in vivo performance of a biomaterial. For example, the corrosion rates of gravity-cast AZ91D and LAE442 magnesium alloys were lower when these samples were implanted into guinea pig femura than when tested under standardised immersion and electrochemical tests, with a difference in rate of up to four orders of magnitude.¹³⁰  Interestingly, opposite tendencies in the corrosion rates were observed under in vivo and in vitro conditions.

The importance of selecting relevant testing conditions was highlighted in the study on the degradation behaviour of Mg-based pins under static immersion and interstitial flow (with and without cyclical loading) conditions, using murine (subcutaneous) or canine (tibia) animal models.¹³¹  Subjecting as-drawn pure Mg, as-cast Mg–Zn–Mn, and extruded Mg–Zn–Mn samples to interstitial flow and cyclical loading were both shown to significantly enhance the rate of corrosion, by up to 4–5 and 10 times, respectively, leading to notable reduction in the mechanical strength.¹³¹  The results are similar to those obtained for bioresorbable stents made of AZ31 magnesium alloy and tested under dynamic conditions similar to those experienced by stents in blood vessels, where notable stress- and flow-induced degradation and comparatively high degradation rate are observed.¹³²  Interstitial flow enables the simultaneous delivery of fresh fluid to, and removal of corrosion products from, the implant site, and promotes the formation of Ca/P complexes on implant surfaces, whereas cyclical loading promotes the detachment of corrosion products, and provides applied stress responsible for crack initiation and propagation. While the effects of flow are fast acting in the case of as-drawn pure Mg and as-cast Mg–Zn–Mn, it takes over two weeks for the corrosion rate of the extruded Mg–Zn–Mn to reflect their contribution, due to a relatively fine-grain boundary of the latter (see Figure 1.22).¹³¹ 

As-cast Mg–Zn–Mn pins show considerably different patterns of pitting corrosion when incubated under static immersion and when subcutaneously implanted into mice. Under static conditions, the corrosion begins with the formation of small pits, and then after two weeks evolves into fast localised corrosion due to the multi-phase nature of this material, which leads to microgalvanic coupling. In contrast, under in vivo conditions using a mouse model, the corrosion is uniformly localised and similar to that obtained under in vitro interstitial flow conditions. When implanted in tibia, the Mg–Zn–Mn pins show more localised degradation as a result of mechanical loading. It should be noted that while the study attempted to use a simulated body fluid containing biomolecules in the in vitro tests, DMEM with 10% fetal bovine serum, potential contributions from cells could not be measured, e.g. bone and inflammatory cells that could have contributed to the degradation kinetics of the material.

Potentially more accurate degradation data may be obtained if the dynamic immersion tests are conducted not only under constant, but also variable flow rates. This would more accurately reflect the variability in flow rates observed in the human body, e.g. those generated in human cancellous bone under the conditions of different physiological activities. Bone homeostasis is affected by physical activity, as the mechanical loading provides pressure differentials that force the fluid to flow at different rates.¹³³  The magnitude of the loading, and thus the strain in the bone that is induced, varies according to the activity, with stumbling producing approximately three times the loading on hip joints compared to that of walking.^134,135  This results in significantly different pressure differences and marrow flow rates in the bone that can vary from 0.0072 to 1.67 ml min⁻¹. These differences in flow may have considerable implications for the degradation of metallic materials used to replace cancellous bone, particularly for porous scaffolds. For example, the nature of the degradation products that are formed and their accumulation in the pores of pure Mg scaffolds is dependent on the flow rate (see Figure 1.23).¹³⁶ 

Where the formation of needle-like clusters containing MgCl₂ and Mg(OH)₂ took place at different flow rates, the formation of a pore-clogging layer of spherical deposits consisting of oxygen, calcium, and phosphorus (at Ca : P of 1.67) and rod-like structures occurred at a flow rate of 0.025 ml min⁻¹, and of rod clump-, flower- and rod-plate-like structures made of MgCO₃ and CaCO₃ at flow rates of 0.4 and 0.8 ml min⁻¹.¹³⁶  Over the first 24 h of observation, the weight loss increases with the flow rate. However after 48 h of incubation, samples with a porosity of 30% and 41% show increased corrosion at a flow rate of 0.4 ml min⁻¹ compared to that at 0.8 ml min⁻¹, with this trend remaining stable as the time of incubation is increased to 72 h. In contrast, the sample with 55% porosity experiences an increased weight loss at a flow rate of 0.025 ml min⁻¹ after 48 h, yet after 72 h, the weight loss increases in line with the flow rate. Table 1.8 shows a comparison between the degradation rates obtained under flow conditions to that under static immersion, highlighting the considerably higher rates reported for the former. This example highlights the importance of studying degradation over a wide range of relevant physiological parameters, with sufficient study length and sampling rate needed to capture the data points necessary to build a comprehensive picture of the likely degradation dynamics under in vivo conditions.

When trying to untangle the complex nature of implant corrosion and compare the results of different studies, we rely on the assumption that artificial fluids are inherently stable with respect to their pH and ion concentration, both during their storage and in vitro testing at physiologically relevant temperatures. Yet, there is a lack of consensus in the literature whether this is indeed the case.^137–140  A recent study on the stability of artificial urine, two artificial salivas and six simulated body fluid samples of various composition, under typical storage (5 °C) and experimental (21 °C and 36.6 °C) conditions, found significant changes in ion concentration and pH, and the formation of precipitate after extended storage (see Figure 1.24).¹³⁹  The formation of precipitate in artificial saliva was moderate at 5 °C and substantial at 36 °C, providing further evidence to corroborate the likely change in the ion chemistry of the solution. All fluids had a substantial change in the concentration of calcium ions (decreasing to at least 70% in all tested fluids with the exception of i- and nl-type simulated body fluids), whereas the change was relatively minor in the case of sodium ions (Ca^{2 +} and Na⁺ were the only ions tested in the study). Changes in the pH were also statistically significant, with pH increasing over time for all of the fluids. These results suggest that artificial fluids may neither be good mimics of in vivo fluids (where the concentration of ions and pH may change dynamically), nor can they be considered truly static (where changes in solution chemistry are related solely to the interaction of the solution with the implant). The immersion of the implant into such a solution may further change the nature of the processes that would take place in the absence of the implant, and this effect may be different when the implant is a solid macroscopic material when compared to e.g. a highly porous, large surface area structure, a thin film, or a collection of micro- and nanoscale particles.

Inflammation can also result from surface corrosion and the consequent release of Ni, Co, Cr, V, and Al species that may induce cellular or systemic toxicity, and cause local allergic reactions or sensitisation as well as changes to the chemistry of the peri-implant milieu. Corrosion may also affect the interface between tissues and implants, leading to implant loosening and consequent changes in how the load is transferred by load-bearing implants. This can not only lead to the mechanical failure of the implant itself, but also affect bone remodelling processes, hindering osseosynthesis and promoting osteoclast activity in the instances of inappropriate bone loading. Tissue acidosis is one of the key hallmarks of inflammatory processes, with substantially increased hydrogen ion concentrations detected in inflamed tissues (pH as low as 5.4), in fracture-related hematomas (pH as low as 4.7), and in malignancies, attributed to a combination of cell injury and death, and infiltration by hyperactive inflammatory cells.^141,142  For example, the presence of HLA-DR + /CD14 + /CD68 + macrophages, CD3 + T cells, CD20 + B cells, plasma cells and macrophage granulomas has been detected in periprosthetic connective tissue and femoral heads in response to Co–Cr wear particles released from metal-on-metal hip implants.¹⁴³ 

Inflammatory cells can release significant quantities of reactive oxygen species, e.g. hydrogen peroxide (H₂O₂) and superoxide (O₂⁻), and these species can subsequently interact with implant surfaces as well as with metal ions that are released from these surfaces (see Figure 1.25).¹⁴⁴  These interactions are likely to occur through a Fenton reaction, whereby a metal ion (M^{n +}) interacts with H₂O₂ to form a hydroxyl radical (˙OH), an OH⁻ ion and a metal ion M^{(n + 1) +}. These reactions can take place either on the surface of the implant, or in the surrounding fluid. The thus-released hydroxyl radicals can induce further damage in cells through lipid peroxidation and by triggering a number of important biochemical pathways.¹⁴⁵  The extent to which metals partake in hydroxyl radical production is dependent on their chemistry as well as surface topography. For example, in vitro studies on Ti, TiO₂, Zr, Fe, Cu, Cr, Ni, Au and Al incubated in the presence of H₂O₂ have shown that metals such as Fe and Cu are associated with greater levels of ˙OH radical formation compared to Ti, Zr, Au and Al over the pH range of 5.0–8.0.¹⁴⁶  It has been proposed that in the case of Ti, the formation of a hydrated, gel-like TiO–OH matrix on the surface leads to the trapping and oxidation of superoxide radicals, thereby limiting the generation of free hydroxyl radicals through a superoxide dismutase-mediated reaction, O₂⁻ + O₂⁻ + 2H⁺ → H₂O₂ + O₂. This phenomenon may, at least in part, account for the favourable biocompatibility of Ti. It should also be noted that the presence of H₂O₂ in solution has also been shown to lead to the formation of a thicker oxide bi-layer, suggesting decreased corrosion resistance under these conditions.¹⁴⁷  The dense inner layer may be responsible for the excellent corrosion resistance of Ti implants, whereas the outer porous layer may promote the precipitation and nucleation of hydroxy–carbonated apatite, and enhance bone cell infiltration and osseointegration by serving as a transition layer between stiff Ti and softer bone tissue.^148–150  Implant surface topography may also play a role in the level of oxygen species produced by cells. It has previously been shown that osteoblasts produce fewer reactive oxygen species, including H₂O₂ and oxygen superoxide, when in contact with machine-prepared smooth titanium surfaces compared to acid-etch-created rough counterparts.¹⁵¹  Interestingly, roughened Ti surfaces have been associated with lower levels of cytotoxicity and improved biocompatibility, suggesting that the role of reactive oxygen species in wound healing and osseointegration may be more complex.¹⁵² 

In vivo and clinical studies have related the levels of short- and long-term pathological changes related to biomaterial implantation to the clinical status of the patient, skill of the surgeon, and the physical and chemical properties of the implant. These include the size, shape, and biochemical activity of its surface, and any degradation products that may be generated at the initial stages of corrosion (i.e. prior to oxide layer formation) and subsequently during normal implant operation (i.e. associated with normal wear and ageing of the material).

It is important to note that the environment-mediated increase in the degradation rate of the implant material need not be uniform, as the properties of tissues, degree of trauma,¹⁵³  inflammation, and thus pH and temperature of the environment, as well as mechanical load may differ over the surface of the implant. This corrosion process and the resultant generation of abnormal electrical currents can effectively transform the metallic implant into an electrode within an electrochemical cell, potentially interfering with the normal functioning of adjacent cells and tissues as the latter receive these extreme electrical signals,¹⁵⁴  as well as affecting the corrosion process itself. Therefore, when judging implant compatibility with tissues, it is important to not only consider the corrosion products, but also the currents of electrons that will be exchanged between tissues and implants, and the reaction products of the ensuing reduction–oxidation processes in which cellular products may be involved.¹⁵⁵ 

Metallic biomaterials are frequently used for the fabrication of devices that deliver electrical stimulation to muscle, cardiac, and neural tissues. For example, with heart failure growing to epidemic levels in industrialised countries, an increasing number of patients are implanted with defibrillators and pacemakers every year.¹⁵⁶  In the United States alone, 5 million patients live with a diagnosed heart failure, with 0.4 million new patients diagnosed annually, costing in excess of $50 billion in hospital care alone.¹⁵⁷  Once implanted, electrical stimulators pass an electric current through an electrode placed in the proximity of the excitable target tissue, resulting in tissue activation and its improved function. The electrodes are typically made of highly corrosion resistant materials, for example a Co–Cr–Ni super-alloy of Co (39–41%), Cr (19–21%), Ni (14–16%), Fe (11.3–20.5%), Mo (6–8%) and Mn (1.5–2.5%), sold under the trade name Elgiloy^®. They can also be made from stainless steel, titanium nitride, iridium oxide, or Pt and Pt–Ir alloys, with the latter being the most frequently used electrode materials. The delivery of electrical stimulation can involve Faradaic as well as non-Faradaic processes. In the former, the electrons are transferred between the electrode and biofluid, whereas the latter relies on a rapid change in the concentration of ions at the electrode surface, in the absence of charge transfer between the electrode and biofluid. The Faradaic processes can therefore promote metal oxidation and corrosion, with the subsequent release of soluble metals into the peri-implant environment. By comparison, since the chemistry of tissues and biofluids does not change as a direct consequence of non-Faradaic stimulation, the effect of the stimulation on metal oxidation, corrosion and ion release is significantly lower. Yet, the stimuli that can be delivered using non-Faradaic processes are often not sufficient to attain a clinically significant result, and Faradaic processes are required to deliver the needed level of charge transfer.

Even when corrosion-resistant, chemically inert materials are used, Faradaic corrosion has been reported in vitro, in vivo and in clinical cases. For example, an X-ray fluorescence microscopy study, using the Australian Synchrotron, on tissues in prolonged contact with multi-channel cochlear implant electrode arrays showed the presence of platinum flakes within macrophages, and particularly in the fibrous tissue surrounding some of the arrays, and the diffusion of finer platinum particles further into the tissues as far as the spiral ligament.¹⁵⁸  The study suggested that the flakes may have been generated as part of the manufacturing process, whereas smaller particles were more likely to originate as a result of electrolysis due to charge imbalance with the bipolar pulses from the implant, as well as corrosion of the electrode material. The mechanism of corrosion that has been proposed, based on the experimental evidence gathered to date, suggests the generation and release of both ions and insoluble particles from the surface, with the extent of degradation dependent on the amplitude and frequency of the stimulation current, and the length of stimulation pulse and stimulation treatment (see Figure 1.26).⁶  As with any other type of corrosion, the surface and bulk properties of the material, the geometry of the implant, and the chemistry of the biofluid will also play a role in determining the rate of corrosion and material release. As such, ion and particle release due to corrosion may become a more significant issue as the size of the electrodes decreases, which effectively increases the charge density applied to the electrode.¹⁵⁹ 

In addition to currents that arise due to differences in materials properties or as a result of corrosion processes, external electrical stimulation is often used in clinical practice to reduce pain and inflammation, and promote the healing of injured tissue¹⁶⁰  by activating osseosynthesis processes in the affected bones,^161,162  particularly in the instances of non-union or delayed union of bone fragments, pseudarthrosis and in patients with a history of pathological fractures. For example, direct current, capacitive coupling, and inductive coupling stimulations (see Figure 1.27) have been reported to improve the rate of arthrodesis in patients suffering from non-union complications following spinal fusion surgeries, with the improvement being clinically significant.¹⁶³  This is because external electrical stimulation replaces the electro–mechanical effects that arise naturally during the initial stages of bone healing, whereby electrical potentials are generated from a combination of the piezoelectric effect and streaming potential in the deformed bone under applied load.¹⁶³  The resulting action currents flowing through the tissues stimulate the enrichment of the facture site with nutrients and bone-forming minerals, and the induction of biochemical processes associated with union activation. In addition to promoting osteoblastic activity, weak electrical stimulation may prevent bacterial colonisation of implant surfaces. For example, in studies conducted in vitro and in vivo, anodisation of silver or silver-coated Ti electrodes for several days using 8 μA direct current resulted in a significant reduction in the pathogen load of Staphylococcus aureus, which lasted for several weeks afterwards,¹⁶⁴  with the inhibitory concentrations of electrically generated silver ions being up to 100 times lower than they are for the clinically approved antibiotic silver sulfadiazine.¹⁶⁵  Subsequent reversal of the poles led to pronounced bone-forming activity. Similar results were observed in a rabbit model, where passing a 200 μA electrical current though a stainless steel implant was more effective in preventing the colonisation of S. epidermidis than standard treatment with doxycycline,¹⁶⁶  and in a goat model, where a low direct current of 100 μA was sufficient enough to hinder the development of clinical infection in the proximity of stainless steel percutaneous pins.¹⁶⁷  The coupling of 15–30 V with anodised and non-anodised nanotubular titanium was also reported to result in a significant reduction in S. aureus cell attachment and biofilm formation after 2 days of culture in vitro.¹⁶⁸ 

The use of external stimulation may have significant implications for implant corrosion, with the outcome dependent on the nature of the stimulation and the properties of the protective layer on the implant. For example, invasive application of a direct current in pulsed mode led to a loss in mass in Cr–Ni–Mo AISI 316L steel implants, even when the latter were encapsulated in a passive diamond coating, with anode screws and cathodes being subject to corrosion damage and surface-limited etching, respectively. Importantly, the corrosion products from the corroded anode were driven into the tissue region, both along the direction as well as in the opposite direction of the current, with measurable levels of Cr, Fe, Mo and Ni detected in the compact bone region.¹⁶¹  Enhanced corrosion and the infiltration of the products into tissues may compromise the mechanical performance of implants and induce a significant tissue response in surrounding tissues. The latter can result in cell death and severe chronic inflammation, eventually leading to the formation of a thick fibrous tissue capsule around the implant, the role of which is to separate the body from necrotic cells and tissue debris.¹⁶⁹  It should be noted that in the aforementioned study on Cr–Ni–Mo AISI 316L steel implants in tissues subjected to electrical stimulation, the delivery of alternating (sinusoidal) current or direct current in pulse mode using non-invasive and semi-invasive methods was associated with minimal corrosion damage to the implants.¹⁶¹ 

In addition to electrical stimulation, light irradiation, heat treatment and atmospheric pressure plasma treatment may be used to promote wound healing, stimulate tissue regeneration and immune function, prevent pathogenic colonisation and treat malignant growth. Among these, plasma treatment is particularly interesting as it delivers a potent cocktail of UV photons, mild heating, physical agitation, charged and neutral particles, and chemically reactive species to cells and surfaces, changing their behaviour.^170–172  For cells, the treatment may induce a variety of biochemical responses, from activating the differentiation and migration of host cells, to selectively inducing programmed cell death in cancer stem cells and pathogenic microorganisms.^173,174  For biomolecules, plasma treatment may affect their chemistry and biochemical reactivity.¹⁷⁵  For abiotic surfaces, such as those of implants, the treatment may change their chemistry and topography, thus potentially changing the manner in which these surfaces interact with biological agents, e.g. proteins and cells, and the environment. The changes in surface properties can be those of material addition (e.g. nucleation and growth of films and complex ordered nanostructures on plasma-treated surfaces), material removal (e.g. etching of the plasma-treated surface to uniformly remove the top oxide layers, selective etching of a pattern with or without a template, or increasing roughness), or material modification (e.g. surface functionalisation to increase hydrophilicity, introduction of defects into nanostructured materials, or ion implantation to increase hardness and wear resistance of the top-most layer of the implant without affecting the mechanical properties of the bulk material).^38,39,176  These addition or removal processes may significantly affect the surface geometry, whereas the modification processes may not.

While many studies have been carried out to investigate the effect of the intentional plasma modification of implants (prior to implantation),^36,37,177  little is known about the consequences of possible unintentional surface modification when plasma is used to treat the peri-implant milieu. In addition to directly modifying the chemistry of the surface, plasma treatment is known to significantly affect the chemistry of treated fluids by introducing large quantities of reactive oxygen and nitrogen species, which then react with molecules within the fluid to produce other chemically active species, including ions of Cl and H₂O₂, and to reduce the pH.^178,179  Mild heating and photons emitted by the plasma may further promote these and other chemical reactions in fluids. Cells, too, respond to plasma-generated hydroxyl radicals (˙OH), hydrogen peroxide (H₂O₂), ozone (O₃), superoxide (O^{2 −}), nitric oxide (NO˙) and anionic (OONO⁻) and protonated (ONOOH) forms of peroxynitrite by producing micromolar levels of H₂O₂.^180,181  The breakdown of H₂O₂ generated in cells in response to plasma treatment is then catalysed by iron-containing proteins into cytotoxic ˙OH radicals that can induce cell injury and apoptotic and necrotic cell death, with the outcome dependent on the treatment dose.^180,181  The thus-released H₂O₂ and other cytokines may also trigger lymphocyte activation and immunogenic cell death, although it is difficult to untangle the roles that individual mechanisms may play when plasmas are used to treat animals or humans.¹⁸²  The aforementioned changes in the chemistry of the peri-implant milieu may in principle promote material corrosion via the previously discussed mechanisms. The treatment may also render cells more susceptible to the uptake and accumulation of particles produced in the process of implant degradation,¹⁸³  exacerbating the oxidative stress that cells experience as a result of their interactions with the metallic debris.

Most metallic implants are highly susceptible to bacterial attachment and colonisation. As it settles on the surface, a bacterial cell may entrap ions and small molecules between the surface of the bacterial layer and that of the metallic implant, inducing a localised change in the osmotic concentration and pH of the aqueous phase trapped within the confined microenvironment.¹⁸⁴  Subsequently, the presence of infectious agents on the surface of implants can further stimulate pro-inflammatory responses in the patient, as well as alter local pH through pathogens releasing their products of metabolism into the peri-implant milieu. The latter can acidify or alkalise the environment, change its electrochemical potential, enrich it with corrosion-promoting or corrosion-inhibiting chemical species, and directly interact with the protective layer and products of corrosive degradation, with the individual effects depending on the pathogen in question.¹⁸⁵  Before we go on to discuss possible mechanisms and specific examples of microbe-promoted corrosion or corrosion inhibition, it is important to once again stress that in vitro findings may not necessarily reflect the events that will take place in vivo. For example, if we are to consider the oral cavity, we will find a highly complex and dynamic population of species with great genetic and phenotypic diversity. Furthermore, the chemistry and biochemistry of the oral environment will change with food consumption, oral hygiene practices, pathophysiological processes and so forth (see Figure 1.28).

In general, most reports in the literature focus on the corrosion-promoting properties of bacteria,¹⁸⁶  especially when it comes to dental applications. As one of the major players in dental plaque formation, several studies have investigated the role of Streptococcus sanguis and S. mutans in metals and alloys, showing that the former bacterium promotes the corrosion of Nd₂Fe₁₄B and other magnetic materials used as retainers, and the latter increases the rate of corrosion in commercially pure Ti.^187,188  Yet even though the presence of a S. mutans biofilm on the surface reduces the corrosion resistance of Ti, it does not prevent the formation of a compact passive film similar to that formed under the same conditions in the absence of bacteria. Furthermore, no pitting corrosion was detected in any of the aforementioned studies, thus challenging previously established associations between biofilms and the development of pits (see Figure 1.29).¹⁸⁹  It is possible that discrepancies in these results arises from differences in experimental protocol, including the length of time the biofilm is allowed to grow on the surface, and the presence of specific environmental factors that may the drive the development of biofilms of certain composition, architecture and continuity, or change the properties of the metal.^189–191 

When implants are first exposed to bodily fluids of a reasonably high ionic strength, their surfaces are likely to be subject to the rapid adsorption of macromolecules, e.g. proteins, even in the case where the charge is similar for the macromolecule and the surface.¹⁹²  Protein–metal interactions may involve attachment via non-specific electrostatic interactions (i.e. physisorption) as well as site-specific attachment (i.e. irreversible chemisorption). The chemistry, charge and size of the protein, its relative concentration within the fluid, and the characteristics of the implant, in particular its surface chemistry, topography and properties of the oxide layer, will determine the probability of adsorption on any given surface. The physical and chemical properties of the aqueous phase in the peri-implant milieu will also influence the probability of protein adsorption and the eventual desorption of complexes that proteins form with metals, with pH, ionic strength, physical agitation and shear rates shown to play a role.¹⁹³  Therefore, such events as inflammation or the release of oxidative species due to cell injury or death may significantly affect the protein binding and detachment processes.

Once adsorbed, proteins may first act in a manner similar to a protective coating to limit corrosion-inducing cathodic reactions and lubricate surfaces to reduce the effects of fretting or friction. Depending on their chemistry and surface conformation, they may also promote or inhibit the attachment of host cells and microorganisms. Over time, however, the protective effects of the protein layer may be overcome by corrosion-promoting factors (see Figure 1.30). These include the release of metal ions from the implant surface, mediated by protein complexation and ligand-binding processes, driven by the concentration-dependent exchange of the protein between the surface and biofluid, and the displacement of smaller surface adsorbed proteins, e.g. albumin, by larger proteins with greater binding affinity from the biofluid, e.g. immunoglobulin G and fibronectin. It should be noted that metal loss due to the above described Vroman effect is low and thus difficult to capture accurately within a short observation period. Nevertheless, the metal–protein complexation-induced release of Fe, Cr, and Ni due to the Vroman effect was experimentally investigated in both 316L and 303 stainless steels exposed to fluids containing albumin and fibrinogen.¹⁹⁴  The exact manner in which metals are released from the surface may be dependent on the properties of the surface oxide, e.g. composition, microstructure, corrosion susceptibility, and the presence of defects, as these may affect the relative strength of the protein–metal bond when compared to metal–oxide or metal–hydroxide bonds. However, these dependencies are not fully understood. A study of protein-mediated metal loss processes in 304, 310, and 316L austenitic stainless steels, 430 nickel-free ferritic stainless steel, 2205 duplex stainless steel and pure iron found that the complexation ability of the surface oxides differed between these materials.¹⁹⁵  Within 2 h of exposure to albumin-containing phosphate buffer, iron atom–albumin complexes formed on the surfaces of the oxide layer of 304 steel, resulting in a notable enrichment in Cr in the surface oxide. This then acted as a passivation-like coating that hindered the rate of release of metal and increased the polarisation resistance of this alloy when compared to other alloys in the study. The changes in the polarisation resistance of alloys were attributed to the preferential complexation to Fe over Cr atoms. The complexation and release of Mn was found to be significantly more dependent on the pH of the biofluid compared to Fe, Ni, or Cr, whereas its dependence on complexation mechanisms was the lowest amongst these atoms. While poor protective performance of the iron oxide layer meant that pure Fe had a significantly greater rate of Fe loss when compared to the tested alloys, the addition of albumin to the buffer resulted in a significant decrease in the quantities of Fe lost from its surface, since the binding of the protein acted to partially block sites where the reduction of oxygen may otherwise take place.

In addition to the availability of particular elements on the surface, the microstructure of the outermost surface, surface-to-volume ratio (in the case of powders), crystallinity and presence of a large number of defects will also affect the rate of metal release.¹⁹⁶  It is interesting to note that while the greater crystallinity of oxides is linked to greater rates of metal oxidation during corrosion, the rate of protein-mediated metal loss is greater in the case of amorphous oxides. This is possibly because of the increased probability of metal complexation and complex detachment due to weaker metal–oxide bonds in the latter oxide structures. The presence of reactive oxidative species, particularly H₂O₂, can significantly increase the susceptibility of the metallic implant to protein-mediated metal loss; this phenomenon can have significant clinical ramifications, since increased levels of H₂O₂ are often detected at sites of infection, inflammation or injury. An in vitro study of metal loss from a corrosion-resistant TiAlV alloy when exposed to simulated physiological fluids with varying concentrations of H₂O₂ and albumin showed significant changes in thickness and composition of the surface oxide, with the layer becoming depleted of V and thicker and enriched in Al as a result of H₂O₂ exposure alone, and depleted of Al when exposed to a combination of H₂O₂ and albumin.¹⁹⁷  Interestingly, accelerated corrosion studies over the same period of 24 h did not show any significant changes, with the albumin-catalysed dissolution of the corrosion product layer and consequent formation of a thinner oxide layer detected when the study was run for 120 h.¹⁹⁸  An even longer in vivo investigation of the same alloy implanted in rat tibia, which would expose the surfaces to similar concentrations of H₂O₂ and albumin, reported the eventual stabilisation of the level of metals in the peri-implant tissue after 1.5–6 months. These results highlight the significance of considering both corrosion and protein-mediated degradation mechanisms and selecting appropriate timelines when testing the stability of implant materials.

In addition to H₂O₂, conditions that damage the protective oxide or prevent its repair, e.g. fretting, are likely to facilitate direct protein–metal interactions, which may result in an increased rate of material removal due to the synergistic effects of dissolution and corrosion. Fretting corrosion rates of stainless steel plates and screws in fluids containing proteins of different charge, and adjusted to a pH between 3 and 5, were found to be similar when the alloys were exposed to an albumin-containing solution with a pH below the isoelectric point of the albumin.¹⁹⁹  Interestingly, Ni was preferentially released in protein-rich fluids when compared to the metal release in saline. In contrast, the rate of corrosion was decreased when the alloys were exposed to negatively charged proteins in solutions with pH above their isoelectric points, confirming that protein charge plays an important role in defining its effect on corrosion. The charge of the adsorbed proteins may also affect the nature of the charged species these proteins can attract to the surface, with net-positively-charged lysozymes attracting negatively charged Cl⁻ ions, and net-negatively-charged albumin attracting H⁺ ions. The surface-localised changes in ion concentration may further destabilise the protective oxide layer, increasing the local rate of corrosion.

Besides the aforementioned chemical effects of protein adsorption, the layer may entrap wear debris and metal ions within its matrix, positively contributing towards material deterioration via friction and wear.²⁰⁰  The wear debris may also interfere with protein–protein interactions, leading to protein aggregation around the wear particle, preventing its degradation and potentially leading to increased inflammation,²⁰¹  thus creating favourable conditions for cell damage and increased material degradation via other mechanisms of corrosion.

Whether it is produced through wear or corrosion, metallic debris may significantly undermine the function of the implant system and pose a significant threat to the health of tissues in the peri-implant milieu, and to host systems in general (see Table 1.9). The probability of metallic debris generation is particularly high in systems containing articulating components, e.g. orthopaedic implants. Yet, the use of these devices is on the rise, with the annual number of total hip arthroplasty surgeries exceeding 1.4 million worldwide, with 0.6 million in Europe alone, and is expected to continue to grow.¹⁹  Total hip replacement surgeries are also on the increase, with 332 000 procedures conducted annually in the United States alone.⁵  In part, this is driven by a rapidly ageing population that wants to remain active for longer, as well as by the needs of younger patients suffering from painful and often debilitating joint conditions, including, but not limited to, osteoarthritis, rheumatoid and traumatic arthritis, avascular necrosis, acetabulum defects, bone fractures, noncancerous and malignant abnormal bone growths.²⁰²  Greater levels of physical activity, however, are often associated with a greater incidence of short-term failures of these implants, necessitating replacement surgery. Adverse reactions to metal debris, also known as metallosis, pseudotumour or aseptic lymphocytic vasculitis-associated lesions, are one of the leading causes of replacement surgery. A recent retrospective study using data from the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man showed that of 873 188 primary hip replacements implanted over the observation period, 0.38% required a revision surgery due to adverse reactions to metal debris.⁵  While metal-on-metal hip implants were associated with the greatest risk of revision due to metallosis (3.7%), other implant designs were not immune, with metal-on-polyethylene, ceramic-on-ceramic, ceramic-on-polyethylene, ceramic-on-metal, and metal-on-ceramic implants having non-trivial levels of risks of 0.024%, 0.055%, 0.023%, 0.69%, and 1.97%, respectively. Importantly, while the mechanisms underpinning metal particle generation in metal-on-metal implants are somewhat understood, the case is less clear with the other types of hip joint prosthesis. In metal-on-metal implants, multiple mechanisms of wear particle generation are considered to be at play, including corrosion at modular implant junctions, specifically at the junction between the femoral head and neck and between femoral neck and stem, and articulating mixing of alloys that make up metal femoral heads and titanium femoral necks.^97,203,204  The body responds to these particles with a T lymphocyte-mediated type IV hypersensitivity reaction and associated tissue damage, which is often but not always found to be dose-dependent. The rate and pattern of wear are design-dependent, with greater wear rates reported for metal-on-metal implants with significant edge loading (e.g. in shallow acetabular cups or those implanted at abnormal inclination angles). Metallic particle generation due to increased trunnion wear in ceramic-on-ceramic and metal-on-polyethylene bearings has also been reported, with certain design features, such as the greater size of the head in metal-on-polyethylene bearings, associated with greater risk of wear and corrosion in modular junctions due to greater transmitted torques.^205,206  In addition to material- and design-specific factors that affect debris generation and the pathological events that ensue, the installation of the implant and skill of the surgeon, the goodness of the fit to the anatomy of the patient, and the patient's level of physical activity will all play a role. For example, the likelihood of adverse implant reactions and implant failure due to metallosis is typically greater for patients with a pre-existing metal sensitivity, although some patients will develop sensitivity as a result of implantation, making data collection and analysis challenging.²⁰⁷  Furthermore, patients with certain genetic conditions, chronic diseases, and diet and lifestyles may be more susceptible to premature implant wear and failure.

When a sufficient number of wear particles enter the peri-implant milieu, they trigger a cell-mediated inflammatory response in the tissues in contact with the implant, leading to bone loss (i.e. periprosthetic osteolysis), and subsequent loss of implant integration (i.e. aseptic loosening). The latter will affect the transfer of mechanical stresses from the implant to the bone, as well as between the implant components, eventually leading to implant failure. Even in the absence of obvious mechanical failure of the implant, the performance of loosened implants will be sub-optimal, with local inflammation, pain and reduced function. Understanding the mechanisms of debris-initiated inflammatory processes and bone lesion formation is not trivial due to the complexity and interconnectivity of biochemical and biophysical processes that these particles can trigger, both locally and at the systemic level. We have already discussed some of the challenges in understanding the kinetics of material failure due to chemical and physical events, and the role biomolecules and cells may play in these processes.

It should be noted that in addition to metal particles, chemical degradation and wear of load-bearing implants, especially those with articulating surfaces, is generally accompanied by a release of fragments of bone cements used to fix the implant to the bone, and significant quantities of polymer particles, including poly(methyl methacrylate) and polyethylene debris with ultrahigh molecular weight, as these polymers are commonly used in polymer-on-metal and polymer-on-ceramic bearings. These particles may also induce adverse cell responses, inflammation, bone loss and changes in the chemistry of the peri-implant milieu, thereby affecting the corrosion rate of metals. The loss of cement may also facilitate the ingress of biofluids from the milieu into the bone–metal interface, where the physical confinement creates favourable conditions for the corrosion of metallic components. These particles may also get entrapped between interfaces that are articulating or subject to micromotion, promoting surface wear. Generally, the generation of these particles has been found to be reduced when using polymer-on-ceramic bearings, highly crosslinked polymers in polymer-on-metal bearings, and cement-free metal-on-metal and ceramic-on-ceramic systems.

When compared to polymer particles generated by joint implants, the metallic debris are more numerous, and generally have a smaller dimension and more uniform size distribution. While the volume of the wear particles produced by metal-on-ultra high molecular weight polyethylene polymer implants is estimated to be 40–100 greater than that by metal-on-metal implants, the number of released polymer particles is notably lower, at an average of 5 × 10¹¹ particles per year, compared to an average of 6.7 × 10¹²–2.5 × 10¹⁴ metal particles per year released by metal-on-metal implants.²⁰⁸  Their size allows metallic debris to migrate more easily away from the peri-implant milieu when compared to large molecular weight polymer debris that accumulates within peri-implant tissues. In doing so, metallic particles can reach distant organs, and activate a systemic response. There have also been reports on metallic particles becoming attached to such proteins as albumin, alpha-1-antitrypsin, and apolipoprotein, acting as a hapten and eliciting an adaptive immune response by activating specific cell surface receptors and initiating the inflammatory cascade.²⁰⁹  With a size in nanometres (often, few tens of nanometres for individual particles, and few hundred nanometres to few hundred micrometres for agglomerates) and a large surface-to-volume ratio, metallic particles are also more biochemically reactive compared to larger polymer particles (e.g. ultra-high and high molecular weight polyethylene) or less reactive ceramic particles (e.g. alumina), being increasingly prone to rapid corrosion and releasing considerable quantities of ions into physiological fluids. As a result, the release of the inflammation-related factors TNF-α, IL-6 and IL-1β may be greater when cells are exposed to metallic particles when compared to ceramic particles, as demonstrated by exposing primary macrophages and THP-1 monocytes to Ti and TiO₂ rutile particles,²¹⁰  even when the rutile particles are significantly smaller (0.45 ± 0.26 μm for TiO₂) than Ti particles (3.32 ± 2.39 μm). Having said that, end-stage inflammatory responses and associated bone resorption have also been reported in response to metal oxide particles, e.g. alumina. Furthermore, some studies suggest that polymer particles with a small size (0.3 to <5–10 μm) and irregular shape may elicit greater proinflammatory responses than either metallic or metal oxide particles, yet this concept remains a subject of debate.^209,211  Nevertheless, it is generally accepted that smaller particles (< 1 μm) tend to simulate a greater inflammatory response, with particle concentration, shape, surface area and surface energy controlling the nature and extent of the histological response.

It is worth noting that although debris generation is generally discussed with reference to load-bearing articulating implants, metals and alloys used in, for example, dental implants and cranial anchorage devices, will also deteriorate over time. While their degradation is more commonly associated with the release of metallic ions, loss of integrity in the protective passivation coating due to micro-motion associated wear and/or corrosion may also result in the release of particles. For example, commercially pure cranial anchorage devices used in bone-anchored hearing aids, and thus not subjected to macroscopic wear processes, have been shown to release Ti metallic and oxide particles and ions into the surrounding soft tissue, leading to peri-implant tissue inflammation.²¹²  Where orthopaedic implants are generally fabricated from titanium grade V alloys, it is not unusual for titanium grades II–IV to be used for the fabrication of dental and anchorage devices. The differences in interstitial elements and processing renders these grades different with respect to the mechanical strength, chemical reactivity, and biocompatibility of the bulk materials and products of their degradation. When exposed to suitable conditions, e.g. acidic pH levels or increased concentrations of H₂O₂ in the fluid, these materials can undergo significant corrosive degradation at the interface between metallic components of the devices, even under minimal load-bearing and wear conditions.

In vivo, a response to an adverse stimulus, such as the presence of a metallic particle, involves resident and recruited cells of different types, including macrophages and giant cells, bone cells (osteoblasts and osteoclasts), cells of the extracellular matrix (fibroblasts) and lymphocytes. This response is mediated by a large number of signalling molecules that cells may express as a result of direct contact with metallic particles or after being activated as part of the inflammation cascade, including chemokines, growth factors, cytokines, eicosanoids, degradative enzymes, and reactive oxygen species; their release may promote or suppress inflammation and trigger tissue loss and regeneration.²⁰⁸  When the contact event between the cells in the peri-implant milieu and debris is transient, the recruitment and stimulation of inflammatory cells and acute inflammation-related bone loss is followed by increased osteoblastic activity, tissue regeneration and healthy bone formation. In contrast, when cells in the peri-implant are persistently challenged by particles and ions, the inflammation becomes chronic, and the tissue regeneration stage is often compromised due to increased activity of osteoclasts and a concomitant decrease in the activity of osteoblasts.²¹¹  Unless addressed, this can lead to a considerable loss of bone tissue in the immediate proximity of the implant, and the formation of fibrous pseudo membranes at the interface between the bone and the implant, i.e. fibrosis. The membrane effectively prevents the in-growth of bone tissues into the implant (in the case of porous implants) or the formation of a strong bond between the bone and the implant. This lack of proper integration leads to increased micro-motion of the implant against the tissues, causing tissue injury and further stimulating inflammation.²⁰³ 

Among the aforementioned cells involved in debris-induced inflammation, inflammatory events mediated by macrophages are among the best studied, in part due to the ability of these cells to migrate towards areas where the concentration of metallic debris is high. Once at the site, macrophages will attempt to remove the particles through endocytotic or pinocytotic uptake (in the case of smaller particles, <150 nm) or phagocytosis for particles with a size of 150 nm–10 μm.²¹³  It should be noted that some of the particles will also be removed from the peri-implant milieu by other cell types, e.g. osteoblasts, fibroblasts, and endothelial cells. Due to their ability to stimulate the inflammatory response via, for example, the release of signalling molecules and cell-to-cell stimulation of monocytes, the activity of macrophages generally widens the zone of soft-tissue damage and inflammation.

The signalling chemicals released by mononuclear macrophages upon particle ingestion can also trigger mononuclear phagocytes to fuse and subsequently differentiate into cells capable of degrading larger objects. The biochemical stimuli include factors TNF-α, IL-1β, IL-6, PGE2 and GM–CSF. For example, debris from Co–Cr–Mo alloys, including soluble Co, Cr, Mo, and Ni ions and particulates, have been shown to activate the inflammasome-mediated pathway in macrophages, causing caspase-1-induced cleavage of intracellular pro-IL-1β into its mature form, which in turn causes the secretion of highly pro-inflammatory IL-1β and induction of a broader proinflammatory response.²¹⁴ 

In the case of debris-releasing bone implants, the fusion of cells from the monocyte-macrophage lineage is likely to have two outcomes: chronic inflammation and the presence of foreign particles will trigger the programme for macrophage fusion and differentiation into multinucleated giant cells, whereas events at the bone will trigger the programme for macrophage fusion and differentiation into osteoclasts. The multinucleate giant cells are capable of ingesting foreign fragments with a size greater than 10 μm, whereas osteoclasts resorb the bone tissue. Interestingly, the mechanism by which the degradation of a material is achieved is very similar between these two cell types. Both rely on the cells attaching strongly to the substrate (an oversized debris particle, the surface of the bone or that of the implant), and then releasing lysosomal enzymes and protons into the thus-formed sealed space to degrade the material.²¹⁵ 

The internalisation of metallic debris (both particles and ions) can induce DNA damage, with the extent of damage dependent on the cell type and properties and concentration of the particles. For example, at low concentrations of <0.01 mM, the in vitro exposure of osteoblasts, fibroblasts, and lymphocytes to fluids containing ions of Al, Co, Cr, Fe, Mo, Ni, and V has little effect on viability. At greater concentrations of up to 1 mM, exposure to ions of Co, Ni, and V negatively affects cell proliferation, viability and morphology, whereas it is necessary to increase the concentration of ions >5 mM for fluids containing Al, Cr, Fe, and Mo to obtain an effect of similar magnitude. At lower concentrations of below 1 mM, exposure to ions of Al, Cr, Fe, and Mo induces changes in the shape of the cell, and loss of filopodia and lamellipodia, suggesting that the migration ability of the cells may also be affected across the three cell types.²¹⁶  An important conclusion from this study was that soluble Co and V were able to induce notable toxicity in all three cell types at concentrations below that detected in the peri-implant milieu in animal and clinical models that use Co- and Ti-based alloys. Metal ions have also been shown to be more effective in inducing allergies to metallic implants when compared to metallic particles, as they are effective at both stimulating macrophage secretion of pro-inflammatory cytokines IL-1β, IL-6, and TNFα, similar to particles, as well as upregulating CD80, CD86, and ICAM-1 T cell costimulatory molecules in human monocytes and macrophages.²¹⁷  Ions of Co have also been shown to induce cellular injury through nitric oxide-induced protein nitration to form nitrotyrosine, and apoptosis in macrophages through activation of the caspase-3 pathway.²¹⁸ 

Chemical reactions between metallic particles and biomolecules present in biofluids can produce significant quantities of reactive oxygen and nitrogen species. These molecules play important roles as signalling molecules, capable of activating macrophages without the cell needing to ingest the metal particle. Due to their considerable chemical reactivity, they can also cause oxidative damage to cellular components. For example, Ti debris can generate radicals capable of inducing the peroxidation of linoleic acid in the plasma membrane to malondialdehyde, and to induce the NFkB signalling pathway by triggering NSmase activation via the hydrolysation of sphingomyelin in the membrane.²¹⁹  It is important to note that all cells are susceptible to radical-mediated damage from the exposure to metallic debris, and oxidative stress can arise even from the exposure to nanoscale particles of noble metals. For example, the inhalation of gold particles by rodents causes metabolic and structural changes in the lung tissue, with animals suffering from interstitial pneumonia, fibrosis, chronic inflammatory cell infiltrates of small lymphocytes, congested and dilated blood vessels, scattered dense extravasation of red blood cells, and foci of hemosiderin granules.²²⁰  The extent of the damage suggests that particles not only induces oxidative stress, but also interferes with the antioxidant machinery of the exposed cells through interactions with relevant protein structures.

Unlike pathogenic microorganisms, metallic particles cannot be fully digested by macrophages, and return to the extracellular environment once the macrophage dies where they can once again interact with other cells. Interestingly, experimental evidence suggests that, at least in vitro, phagocytosis by macrophages renders Ti particles unable to stimulate TNF-alpha secretion by macrophages without changing particle size or shape, most likely through changing the chemical composition of their surfaces.²²¹  As previously mentioned, the surface properties of nanoparticles, e.g. their chemistry, free surface energy, topography, are important determinants of their biochemical reactivity and toxicity. Such a mechanism may offer some protection in the case of a one-off release of particles, however, the continued release of particles due to corrosion or wear may limit its efficacy.

The presence of certain types of molecules attached to the surfaces of metallic nanoparticles may increase their cytotoxicity. These molecules can originate from activated or lysed cells, where cells can be host cells, e.g. alarmins or danger signals from injured or dying macrophages or osteoblasts, or they can be microbial cells, e.g. extracellular polysaccharides and virulence factors, including liposaccharides, peptidoglycan, lipoteichoic acid, and teichoic acid. In principle, particles can also harbour attached bacterial cells or even their biofilm communities. Microbial cells and particles that attach to wear particles may originate from infected implants, with microbial patterns remaining even on implants that have been sterilised, or be present in the peri-implant milieu.²²²  The human body is well equipped to deal with a bacterial threat, and once recognised by immune cells, particle-bound bacterial fragments and microbial-associated molecular patterns may trigger strong pro-inflammatory responses. It should be noted that such a response may be mounted by the host body in the absence of clinical infection,²²²  although it is possible that conventional diagnostic methods simply fail to detect microbial populations. Importantly, the response to low quantities of bacterial fragments bound to particle surfaces may far exceed that to the particles themselves. This has been shown experimentally in vitro and in vivo by using Ti particles with and without bacterial fragments on their surface. The inflammatory response that particles free of endotoxins are able to elicit in bone marrow cells from mice and human monocytes is significantly lower than that induced by particles with endotoxins. The exposure to the former particles is also associated with a much lower incidence of osteolysis (by 50–70%).²²³  When endotoxins are re-introduced into the fluid containing the endotoxin-free nanoparticles, the endotoxins rapidly attach to the particle surface, with the resulting opsonised particles regaining their ability to evoke cytokine release and stimulate osteoclast differentiation. This is because endotoxins have strong affinity for metallic biomaterials, and this results in it also being difficult to remove endotoxins using conventional methods of cleaning. The significance of endotoxins in the ability of wear particles to induce inflammation was also confirmed by using an antibiotic capable of inactivating bacterial lipopolysaccharides, where the use of polymyxin B suppressed the pro-inflammatory activity of Ti particles coated with bacterial lipopolysaccharides.²²⁴  The presence of bacterial toxins may also explain the pro-inflammatory response induced by wear particles of inert materials, as these endotoxins ligate TLR2 and TLR4 toll-like receptors, the pattern-recognition receptors on the cells of the innate immune system responsible for the detection of the lipopolysaccharides of Gram-negative bacteria and the lipoteichoic acid and peptidoglycan of Gram-positive bacteria, respectively, stimulating macrophage infiltration of the peri-implant space. TLR2 and TLR4 also respond to endogenous ligands derived from host tissues or cells, including fragments of cells, induced gene products, extracellular matrix components, e.g. fibronectin, heparan sulfate, biglycan, fibrinogen and hyaluronan breakdown fragments.²²⁵ 

Many of the materials-based strategies to prevent inflammation focus on the prevention of wear debris generation, e.g. through the development of wear-resistant coatings or selecting wear-reducing combinations of materials for interfaces that are subject to articulation, and on mitigating bacterial attachment and removal of implant-attached bacterial motifs. Other approaches include immunomodulation strategies, e.g. through interrupting macrophage activation mediated by toll-like receptors (TLRs) and factor-kappa B transcription factor, inhibiting macrophage migration, or by polarising pro-inflammatory macrophages via interleukin 4 treatment to shift them towards an anti-inflammatory, pro-tissue repair macrophage phenotype (see Figure 1.31).²¹¹ 

In the previous section, the ability of wear debris to activate macrophages and trigger their fusion and subsequent differentiation into bone-resorbing osteoclasts was mentioned. However, in addition to stimulating osteoclast activity, wear particles and soluble metals have been reported to depress the differentiation and bone-reforming activity of osteoblasts, hindering implant integration at the initial stages of implantation, and subsequently compromising the continuing repair of the periprosthetic bed needed for the implant to perform its function properly over extended periods of time.²⁰⁹  This is because metallic particles interfere with mesenchymal stem-cell differentiation into functional osteoblasts, reducing their ability to express genes associated with collagen production and proliferate; depending on the particle concentration, they may also initiate programmed cell death in osteoblasts. For example, exposure of Saos-2 osteoblasts to particles from the degradation of Ti and its alloys affects their attachment apparatus by altering their cytoskeletal structure, specifically by interfering with the production and arrangement of ventral stress fibres, β-tubulin and acetylated α-tubulin fibres, and inducing changes in the phosphorylation states of both FAK and Pyk2 tyrosine kinases at focal contacts.²²⁶  The stage of osteoblast maturation and the properties of the particle, e.g. their size, significantly affects the type of events and the extent of damage in response to the debris, with more mature cells showing lower susceptibility to particle-induced apoptosis, and larger particles stimulating the activity of enzymes responsible for the degradation of extracellular matrix proteins during bone formation, e.g. matrix metalloproteinases (MMP)-2 and -9.²²⁷ 

Mitigating the damaging effects of wear particles on bone formation requires better understanding of the events and processes triggered by particle exposure. Yet, the design of in vitro models that would reflect the complexity of the peri-implant milieu, and capture the rich variety of feedback mechanisms and timing of signalling events that take place between cells, is not a trivial matter. Furthermore, peri-implant inflammation and osteolysis may arise due to reasons other than the effects of the wear debris, and these events may be difficult to decouple. For example, both bone resorption and formation are affected by the mechanical pressure that bone cells experience. Osteocytes, the star-shaped cells typically found in mature bone, continuously experience mechanical cues from the surrounding bone (lacunae) and neighbouring osteocytes, osteoblasts and osteoclasts which they contact by means of long cytoplasmic extensions, i.e. processes that reside in pericellular interstitial fluid-filled canals termed canaliculi. Introduction of the implant changes the nature of the peri-implant environment, affecting the flow of fluid within the lacuno-canalicular porosity, and the pressure that this network of cells experience. For example, the micro-motion of the implant may be sufficient to drive the joint fluid into the implant–bone interface, and thus increase the pressure in the peri-implant space. In response to these flow and pressure changes, osteocytes release a host of chemical signals, including nitric oxide, which alter osteocyte apoptosis, and osteoblast and osteoclast activity. The effect of mechanical instability on osteoclast differentiation and bone loss have been shown to depend on the type and magnitude of mechanical stimulation, with supraphysiological loadings of MLO-Y4-osteocytes enhancing osteoclast differentiation by 1.9-fold when compared to physiological loading, with the stimulatory effect on the membrane-bound receptor activator of NF-kappa B ligand (RANKL) and nitric oxide production also being greater in the case of supraphysiological loading when compared to stress shielding.²²⁸  These results suggest that wear particle-induced inflammation and bone resorption may proceed in parallel or be secondary to mechanical instability-induced events.

Despite tremendous progress in the development and application of metallic biomaterials, and the significant benefits that their use brings to patients, there are still gaps that prevent us from taking full advantage of this class of materials. Our knowledge of how these materials truly behave in vivo is still incomplete, which limits intelligent design and optimisation, as well as our ability to predict and respond to the short- and long-term consequences of their clinical use. We still lack in vitro models and characterisation tools that would allow us to capture the complexity of events that take place in the body after implantation, and the intricate links that may exist between different mechanisms of corrosion, wear, and host response. Data from implant clinical use are a result of multiple factors, including those arising from the choice of material, device design, the method of implantation, skill of the surgeon, treatment regime, and a great diversity of patient-related factors. Furthermore, there is great variability with respect to how in vitro, in vivo and clinical studies are designed, performed and interpreted, making comparison challenging. Ideally, a systematic approach based on a unified experimental framework and set definitions would certainly help, however there is a lack of consensus as to what such a framework should look like.