Biomaterials Science: An Introduction to Materials in Medicine

basket).

Part 1

Materials Science and Engineering

Section I.1 Properties of Materials

Section I.2 Classes of Materials Used in Medicine

Section I.1

Properties of Materials

Chapter I.1.1. Introduction: Properties of Materials: The Palette of the Biomaterials Engineer

Chapter I.1.2 The Nature of Matter and Materials

Chapter I.1.3 Bulk Properties of Materials

Chapter I.1.4 Finite Element Analysis in Biomechanics

Chapter I.1.5 Surface Properties and Surface Characterization of Biomaterials

Chapter I.1.6 Role of Water in Biomaterials

Section I.1

Properties of Materials

Chapter I.1.1 Introduction: Properties of Materials: The Palette of the Biomaterials Engineer

Jack E. Lemons

University Professor, Schools of Dentistry, Medicine and Engineering, University of Alabama at Birmingham, Birmingham, AL, USA

The platform, or palette, upon which the biomaterials engineer arranges information into parts for subsequent blending, let us say an art of biomaterials science, has expanded and evolved significantly in content over past decades. The depth and breadth of what is now included on this palette goes well beyond expectations expressed by founding members of the Society for Biomaterials in the late 1960s and early 1970s.

The science in biomaterials science has included fundamental aspects of physical, mechanical, chemical, electrical, and biological (compatibility) properties of the synthetic and natural origin biomaterials per se. Also, the methods for measuring and analyzing properties are equally applicable to the structures of the biological host. Following the recognition of the need by founders of the discipline in the 1960s, one focus has been the fundamental structure versus property relationships leading to in vivo biocompatibility. These relationships, and the supporting scientific information, have changed with time and experience, especially as the biological and clinical disciplines have also evolved. For example, considerations for biocompatibility are very different for biomaterials listed within biotolerant (called inert), surface bioactive (intended), and biodegradable categories. This shift of emphasis is reflected in the progression of content of the first three editions of this book. For example, initial considerations focused on materials were based primarly on substances of the metallics, ceramics, and polymerics available within various industrial applications. Thus, the emphasis in the first edition was on materials of synthetic origin and the science leading to biomaterials. The second and third editions represent the transitions from combination products to the new areas of bioactives and biodegradables. Thus, as an integrated, comprehensive, and authoritative text, this third edition reflects a broad range of biomaterials, and therefore basic properties of new generation biomaterials for regenerative medicine.

Considering relationships between biomaterial and biological systems (the interface) and the dynamics of change from nanoseconds to years, we now better understand many mechanisms of interaction at the dimensions and concentrations used to describe interactions of atoms. It is also realized that all biomaterial and host environment interactions play a role in the broader aspects of biocompatibility, especially the functionality and longevity of implant devices. In this regard, Part I on Materials Science and Engineering emphasizes the more basic information on the bulk and surface properties of synthetic and natural origin biomaterials. Critical aspects of constitution (chemistry) and structure (nano-, micro-, and macrodimension) relationships are presented as related to properties of implant systems. These basic considerations from the nature (I.1.2), bulk (I.1.3), and surface (I.1.5) properties are also interrelated to biomechanics (I.1.4).

In this edition, the concepts of property versus structure relationships have been expanded in terms of the role of water in biomaterials (I.1.6). The science of interactions of synthetic and natural substances with water is recognized as one of the key aspects of surgical implant biocompatibilities. This is especially important for the evolution of the discipline to include new generation biomaterials needed for future implant applications.

In summary, the content of this Part I is broadly applicable to all parts of the third edition. Therefore, students are advised to always consider the basic principles as provided in this section. This has been recognized as critical to the education of a specialist in biomaterials science leading to the selection of biomaterials for medical treatments utilizing all types of implant devices.

Chapter I.1.2

The Nature of Matter and Materials

Buddy D. Ratner

Professor, Bioengineering and Chemical Engineering, Director of University of Washington Engineered Biomaterials (UWEB), Seattle, WA, USA

Introduction

Biomaterials are materials.

Biomaterials are materials. What are materials and how are they structured? This is the subject of this chapter, a lead into subsequent chapters with discussions of the bulk properties of materials, mechanics of materials, surface properties, a discussion of water (since all biomaterials function in an aqueous environment, and that environment can alter both the nature of the material and the interaction that occurs with the material), and finally discussions of specific classes of materials relevant to biomaterials (polymers, metals, ceramics, etc.).

Atoms and Molecules

The key to understanding matter is to understand attractive and interactive forces between atoms.

This Nature of Matter section aims to communicate an understanding of the basic structure of materials that will drive their properties – both the mechanical properties important for specific applications (strong, elastic, ductile, permeable, etc.), and the surface properties that will mediate reactions with the external biological environment.

The key to understanding matter is to understand attractive and interactive forces between atoms. Argon is a gas at room temperature – it must be cooled to extremely low temperatures to transition it into liquid form. An argon atom interacts (attracts) very, very weakly with another argon atom – so at room temperature, thermal fluctuations that randomly propel the atoms exceed attractive forces that might result in the coalescence to a solid material. A titanium atom strongly interacts with another titanium atom. Extremely high temperatures are required to vaporize titanium and liberate those atoms from each other. The understanding of matter is an appreciation of interactive forces between atoms.

What holds those atoms and molecules together to make a strong nylon fiber or a cell membrane, or a hard, brittle hydroxyapatite ceramic, or a sheet of gold, or a drop of water? Even in the early 18th century, Isaac Newton was pondering this issue: There are therefore Agents in Nature able to make the Particles of Bodies stick together by very strong Attractions.

Entropy consideration would say these molecules and atoms should fly apart to increase randomness. However, there is an energy term contributing to the stability of the ensemble leading to a negative Gibbs free energy, which, according to the second law of thermodynamics, should make such solids energetically favorable (of course, we intuitively know this). Thus, we must examine this energy term. We know of just four attractive forces in this universe:

• Gravitational

• Weak nuclear

• Strong nuclear

• Electromagnetic.

Gravity holds us to the surface of the planet Earth (a massive body), but the gravitational potential energy of two argon atoms is only about 10−⁵² J, 30 orders of magnitude weaker than is observed for intermolecular forces. The weak nuclear force and the strong nuclear force are only significant over 10–4 nm – but molecular dimensions are 5 × 10−¹ nm. So these forces do not explain what holds atoms together. This leaves, by default, electromagnetic forces (positive charge attracts negative charge). Electromagnetic forces have appropriate magnitudes and distance dependencies to justify why atoms interact. Interactions can be weak, leading to liquids, or stronger, leading to solids.

Electromagnetic forces manifest themselves in a number of ways. The types of interactions usually observed between atoms (all explained by electrostatics) are summarized in Table I.1.2.1. We consider here van der Waals forces (also called induction or dispersion forces), ionic forces, hydrogen bonding, metallic forces, and covalent interactions.

TABLE I.1.2.1 Forces that Hold Atoms Together

Van der Waals or dispersion forces rationalize the interaction of two atoms or molecules, each without a dipole (no plus or minus faces to the molecules). For example, argon atoms can be liquefied at low temperature. Why should this happen? Why should argon atoms want to interact with each other enough to form a liquid? Figure I.1.2.1 explains the origin of dispersive forces. Such forces are important, as they dictate the properties of many materials (for example, some polymers such as polyethylene which has no obvious dipole), but also they explain why the lipids in cell membranes assemble into the bilayer structure described later in this textbook. A typical van der Waals interactive force (for example, CH4 … CH4) is about 9 kJ/mol. The Ar … Ar interactive force is approximately 1 kJ/mol.

FIGURE I.1.2.1 (A) Consider the electron clouds (charge density in space) of two atoms or molecules, both without permanent dipole moments. (B) Electron clouds are continuously in motion and can shift to one side of the atom or molecule; therefore, at any moment, the atoms or molecules can create a fluctuating instantaneous dipole. (C) The fluctuating instantaneous dipole in one molecule electrostatically induces such an instantaneous dipole in the next molecule.

Ionic forces are probably the easiest of the intermolecular forces to understand. Figure I.1.2.2 illustrates a unit cell of a sodium chloride crystal. The + and − charges are arrayed to achieve the closest interaction of opposite charges, and the furthest separation of similar charges. This unit cell can be repeated over and over in space, and the forces that hold it together are the electrostatic interaction of a permanent + charge and a permanent − charge. Typical ionic bond strengths (for example, NaCl) are about 770 kJ/mol.

FIGURE I.1.2.2 The unit cell of a sodium chloride crystal illustrating the plus–minus electrostatic interactions.

Hydrogen bonding interactions are also straightforward to appreciate as electrostatic interactions. An electronegative element such as oxygen (it demands electrons) can distort the binding electron cloud from the hydrogen nucleus leaving the hydrogen (just a proton and electron), with less electron and thus more plus-charged proton. This somewhat positive charge will, in turn, then interact with an electronegative oxygen (Figure I.1.2.3). Typical hydrogen bond strengths (for example, O–H … H)are about 20 kJ/mol.

FIGURE I.1.2.3 A hydrogen bond between two water molecules.

Metallic bonding is explained by a delocalized sea of electrons with positively charged nuclear cores dispersed within it (Figure I.1.2.4). A single metallic bond is rarely discussed. The total interactive strength is realized through the multiplicity of the plus–minus interactions. The strength of this interaction can be expressed by heats of sublimation. For example, at 25°C, aluminum will have a heat of sublimation of 325 kJ/mol, while titanium will be about 475 kJ/mol.

FIGURE I.1.2.4 Metallic bonding in magnesium. The 12 electrons from each Mg atom are shared among positively charged nuclear cores (the single + charge on each magnesium atom in the figure is simply intended to indicate there is some degree of positive charge on each magnesium nuclear core).

Covalent bonds are relatively strong bonds associated with sharing of pairs of electrons between atoms (Figure I.1.2.5). Typical covalent bond strengths (for example, C–C) are about 350 kJ/mol.

FIGURE I.1.2.5 Covalent bonding along a section of polyethylene chain. Carbons share pairs of electrons with each other, and each hydrogen shares an electron pair with carbon.

Molecular Assemblies

Atoms can combine in defined ratios to form molecules (usually they combine with covalent bonds), or they can form cohesive assemblies of atoms (think of gold and metallic interactive forces, for example). Thus, materials can be made of atoms or of molecules (i.e., covalently joined atoms). The difference between the dense, lubricious plastics used in orthopedics, the soft, elastic materials of catheters, and the hard, strong metals of a hip joint is associated with how those atoms and molecules are organized (due to attractive and interactive forces) in materials. Metals used in biomaterials applications can be strong, rigid and brittle, or flexible and ductile. Again, the difference is largely how the atoms making up the metal are organized, and how strongly their atoms interact.

Molecules also organize or assemble. The widely varying properties of polymers are due to molecular organization. The assembly of lipid molecules to make a cell membrane is another example of this organization.

A key concept in appreciating the properties of materials is hierarchical structures. The smallest size scale that we need consider here in materials is atoms, typically about 0.2 nm in diameter. Atoms combine to form molecules with dimensions ranging from 1 nm to 100 nm (some large macromolecules). Molecules may assemble or order to form supramolecular structures with dimensions up to 1000 nm or more. These supramolecular structures may themselves organize in bundles, fibers or larger assemblies with dimensions reaching into the range visible to the human eye. This concept of hierarchical structure is illustrated in Figure I.1.2.6, using collagen protein as the example.

FIGURE I.1.2.6 Collagen fibers make up many structures in the body (tendons, for example). Such anatomical structures as tendons are comprised of collagen fibrils, formed of aligned bundles of collagen triple helices that are themselves made up of single collagen protein chains (α-chains). The α-chains are constructed of joined amino acid units, and the amino acids are molecules of carbon, oxygen, nitrogen, and hydrogen atoms in defined ratios and orders. (Illustration from Becker, W. (2002). The World of the Cell. Reprinted with permission of Pearson Education, Inc.)

The single α-chains comprising the collagen triple helix would break under tension with an application of nanograms of force. On the other hand, the collagen fibers in a hierarchical structure such as a tendon can support many kilograms of force. Such hierarchical structures are noted frequently in both materials science and in biology.

Surfaces

As assemblies of atoms and/or molecules form, within the bulk of the material, each unit is uniformly bathed in a field of attractive forces of the types described in Table I.1.2.1. However, those structural units that are at the surface are pulled upon asymmetrically by just the units beneath them. This asymmetric attraction distorts the electron distributions of the surface atoms or molecules, and gives rise to the phenomenon of surface energy, an excess energy associated with this imbalance. For this reason surfaces always have unique reactivities and properties. This idea will be expanded upon in Chapter I.1.5.

Conclusion

In this section we have reviewed the transition from chemistry to matter. Interacting assemblies of atoms and molecules comprise matter. Without matter, we cannot have biomaterials. Matter exists because of electrostatic forces – positive and negative charges, in all cases, hold atoms together. The strength of those interactions, associated with the magnitude of the charge on each atom, and the environment the atoms are in (water, air, etc.) ultimately dictates the properties of matter (a soft gel, a hard metal, etc.). Now that we have a general idea what matter is, we can take these concepts from physics and chemistry and bring them to a consideration of the mechanical properties of materials, the surface properties of materials, and then into the specifics of polymers, metals, ceramics, and other types of materials.

Bibliography

1. Barton A. States of Matter, States of Mind. Bristol, UK: Institute of Physics Publishing; 1997.

2. Holden A. The Nature of Solids. New York, NY: Dover Publications Inc; 1992.

3. Becker WM. World of the Cell. In: Kleinsmith LJ,, Hardin J, eds. 5th ed. Pearson Education, Inc. Upper Saddle River, NJ 2003.

Chapter I.1.3

Bulk Properties of Materials

Christopher Viney

School of Engineering, University of California at Merced, Merced, CA, USA

Introduction

Bulk Versus Surface

The success or failure of many biomaterials depends on the physical and chemical characteristics of their surface. It is the surface properties that dictate interactions between a material and its environment, and thus whether a permanently implanted material will be tolerated or rejected. In cases where the implanted material is required to degrade at a controlled rate (e.g., when the implant is a temporary support, such as a dissolving suture or a scaffolding for cells that regenerate tissue), then the bulk material must be capable of sustaining those properties continuously as it becomes the new surface.

Regardless of their particular surface properties, biomaterials are usually also required to exhibit certain bulk characteristics, especially including those attributes (mechanical properties) that relate to the ability to carry loads dependably without undue deflection or premature failure. Mechanical properties will be addressed in detail in this chapter, followed by a brief consideration of selected other bulk properties (thermal, optical) that are often significant in the context of biomaterials.

The concepts addressed in this chapter are fundamental. They can be found scattered through standard materials science or mechanical engineering textbooks, and undergraduate materials science or mechanical engineering courses. The aim in this chapter is to achieve consistency of presentation, connectivity of knowledge, and appreciation of relevance.

Microstructure and Properties

The bulk properties of a material depend not only on what types of atoms and molecules it contains (composition), but also on how those atoms and molecules are arranged (microstructure). For example, if the material is a polymer, do the individual molecules follow a random trajectory (and so occupy an approximately spherical region of space), or are they extended (so that they occupy a region of space approximated by a rectangular prism or a cylinder)? Are the molecules locally packed in a regular array? Do the molecules locally all point in the same direction? And how are these various spatial attributes controlled by processing – assuming that they can be controlled? A strong culinary analogy can be made: in addition to specifying a list of ingredients, recipes also contain instructions on how to process the ingredients to obtain the desired dish. Processing is what makes baked potatoes, mashed potatoes, fries, and shepherd’s pie crust so different.

The microstructure of a material is a specification of the structural features at length scales that cannot be discerned with the naked eye. Most materials exhibit microstructural detail at several different length scales, down to the nanoscale. Therefore, the term microstructure encompasses structures in the size range 10−4 m to 10−9 m.

Some bulk properties, such as the stiffness of a metal, depend principally on the type (composition) of metal being characterized, and not on the microstructure. Such properties are referred to as intrinsic properties, and they include density and heat capacity. (Note that the stiffness of a polymer is not an intrinsic property, because the stiffness measured by attempting to stretch the polymer in a particular direction depends on whether or not the molecules are preferentially aligned in that direction.)

Other bulk properties, such as the yield strength of a metal, depend on attributes such as the average grain (crystal) size, as well as the number and distribution of defects in the crystal structure. Yield strength is an example of an extrinsic property – as is the stiffness of polymers. (Both stiffness and yield strength will be defined and discussed in detail in the section Bulk Mechanical Properties Determined from Stress–Strain Plots.)

It is because some properties are intrinsic while others are extrinsic, and because different extrinsic properties are sensitive to changes in different microstructural attributes at different length scales, that it is possible to control and therefore optimize several bulk properties simultaneously. For example, a metal with a fine grain size will have higher yield strength than a coarser-grained metal, but both can have the same stiffness.

Load, Nominal Stress, Extension, and Nominal Strain

When a load (force) is applied to an object, the possible consequences include translation, rotation, and deformation of the object. We are concerned here with deformation. To properly quantify the deformation caused by a load, we need to take into account the dimensions of the object to which the load is being applied. For example, we are often told that spider silk (especially the dragline) is strong, although it is our experience that a strand of spider silk can be broken more easily than a strand of dental floss. The difference is that the dental floss is much thicker than the spider dragline; a bundle of spider dragline having a total thickness equal to that of the dental floss would be significantly more resistant to breaking than the floss. Consider a sample of material represented by the rectangular prism shown in Figure I.1.3.1.

FIGURE I.1.3.1 Examples of a tensile (normal) stress, a shear stress, and a mixed stress with both normal and shear components, illustrating the relationship between load direction and surface orientation in each case. For the example of shear stress, the originally rectangular front face of the sample distorts to a parallelogram as shown, and the corresponding shear strain is defined as tan θ .

Let a force F be applied normal to a pair of opposite faces (in this case the top and bottom faces) of the sample, which initially each have area A1. The actual shape of the sample is not important for the analysis that follows; all that is necessary is that the cross-section perpendicular to the applied force should be uniform. A cylinder (e.g., representing a fiber) would serve just as well. The ratio:

defines the nominal stress being applied to the sample. It is a nominal stress because the actual value of the area (cross-section) to which the load is applied will change as the sample deforms in response to the load; electing to calculate stress with reference to the initial cross-sectional area is a contrived choice. Nominal stress is often synonymously referred to as engineering stress. The stress is tensile if it is a pull, elongating the sample, and it is compressive if it is a push, shortening the sample. The SI units of stress are newtons per square meter (N.m−2) or pascals (Pa).

The elongation of a sample in response to a tensile load will depend on the length of sample that there is to elongate. To visualize this, think of a line of atoms in the direction of the load, with the bonds between atoms being represented by springs. A tensile load causes each bond to stretch; consequently, the more bonds there are in the line, the longer the line will become in response to a given load. For purposes of comparing the responses of different types of material, it is therefore useful to quantify the effect of loads (and their related stresses) by considering the resultant length change per unit length, more commonly known as strain. If the initial length of the sample is l1, and the applied stress causes a change to length l2, then the nominal (or engineering) strain is given by:

where λ denotes the extension ratio, i.e., the ratio of final length to initial length.

It is a nominal strain because electing to calculate strain with reference to the initial length is again a contrived choice. The strain is tensile or compressive in accordance with whether it is caused by a tensile stress or a compressive stress. Strain, like extension ratio, is dimensionless.

The relationship between stress and resultant strain is commonly summarized in a stress–strain plot, where by convention the stress is plotted vertically and the strain is plotted horizontally, with data being collected while the sample is deformed at a constant strain rate.

True Stress and True Strain

A more realistic sense of stress imposed on a material during a stress–strain test is obtained if the applied force is scaled with reference to the actual cross-sectional area of the sample, A. In that case, we refer to the sample being subjected to a true stress defined by:

Unfortunately, because A changes continuously during the course of a stress–strain test, it can be difficult to measure in practice. This is why the contrived, but more expedient, nominal stress is so often used instead of true stress.

The inadequacy of nominal strain as a measure of sample deformation becomes apparent when we consider the cumulative effect of successive strains. Let a sample of initial length l1 be stretched to a length l2, and thence to length l3. Calculated separately, the corresponding nominal strains are:

The nominal strain for the deformation directly from length l1 to length l3 would be calculated as:

To address this limitation, we consider a revised definition of strain, in which the length change of the sample is measured in very small increments dl, with each increment being scaled relative to the length l of the sample immediately prior to that increment. Thus, a small increment in true strain is defined as:

The true strain corresponding to deformation from length l1 to l2 can then be found by integrating the above expression between limits l1 and l2:

If this new interpretation of strain is applied to the case of a sample of initial length l1 being stretched to a length l2, and thence to length l3, the separate nominal strains are now:

The sum of these true strains is:

In other words, true strains behave properly when we try to add them.

Stresses and strains are almost always presented without formal identification as nominal or true, in which case it can (usually) be assumed that they are nominal. True stresses and strains are (usually) identified explicitly as such. However, it is good practice to avoid any possible ambiguity by being explicit and specific when talking about stresses and strains.

It is possible to interconvert nominal and true stresses, and nominal and true strains, if we make the assumption that the sample volume does not change during deformation. The circumstances under which this assumption is valid will be addressed in the section Condition for Incompressibility (Zero Volume Change) below.

If volume is conserved during deformation from condition 1 (characterized by sample length l1 and cross-sectional area A1) to condition 2 (characterized by length l2 and cross-section A2), then:

At that stage in the deformation, the true stress is:

The interconversion between nominal and true strains is even easier to obtain:

Shear Stress and Shear Strain

Referring again to the rectangular prism of material shown in Figure I.1.3.1, consider now the effect of applying the force F parallel to a pair of opposite faces (in this case again the top and bottom faces), which initially each have area A1. For this configuration of applied load, the ratio:

defines the shear stress being applied to the sample.

Seen from the front of the sample, the outline becomes distorted from a rectangle to a more general parallelogram. This shape change can be used to describe the shear strain γ caused by the shear stress:

The use of σ to denote tensile or compressive stress, ε to denote tensile or compressive strain, τ to denote shear stress, and γ to denote shear strain is a widely adopted convention.

If a force is applied obliquely (neither perpendicular nor parallel; see Figure I.1.3.1) to a pair of opposite faces of the sample, the ensuing stress can be resolved into normal and shear components, as can the resulting strain.

Bulk Mechanical Properties Determined from Stress–Strain Plots

Simple stress–strain plots can be used to define and quantify several mechanical properties of a material. Figure I.1.3.2 displays representative plots for a ductile metal (titanium), a ceramic (alumina, Al2O3), and a crystalline polymer (high density polyethylene, HDPE) subjected to a tensile stress.

FIGURE I.1.3.2 Representative nominal stress versus nominal strain plots for three different classes of implantable material: ductile metal (titanium alloy; 6 wt% Al, 4 wt% V); ceramic (alumina); and crystalline polymer (high density polyethylene).

Figure I.1.3.3 repeats the stress–strain plot for titanium schematically, emphasizing features that relate to specific mechanical properties.

FIGURE I.1.3.3 Schematic nominal stress versus nominal strain plot for a ductile metal, emphasizing the features that relate to specific mechanical properties.

Elastic Deformation

Elastic Constants

The relationship between stress and strain is initially linear as the stress is increased from zero. This behavior is equivalent to the load-extension response of a simple spring, as first stated quantitatively by Robert Hooke and embodied in Hooke’s Law. In the case of bulk material, the constant of proportionality between tensile (or compressive) stress and strain is known as the Young’s modulus, E:

The equivalent constant of proportionality in the case of a shear stress being applied is known as the shear modulus, G:

Similarly, we can define a bulk modulus, K, as the constant of proportionality relating pressure to the volume change caused by that pressure:

where a pressure P causes a reduction in volume from an initial volume V1 to a final volume V2. The negative sign is used to obtain a convention in which the values of K are positive.

In all three cases above, the definitions of the different types of elastic modulus assume that the cause of deformation (tensile stress, shear stress or pressure) is sufficiently small to ensure that the response (tensile strain, shear strain or volume change) is proportional to the cause. This condition also ensures that the response is reversible; when the stress or pressure is removed, the material returns to its original dimensions.

There is one additional constant that is commonly encountered in the context of the elastic deformation of materials: Poisson’s ratio, υ (Figure I.1.3.4).

FIGURE I.1.3.4 Longitudinal elastic strain (in this case elongation) parallel to the direction of the applied load is accompanied by a transverse elastic strain (in this case contraction) perpendicular to the direction of the applied load. The negative of the ratio of transverse elastic strain to longitudinal elastic strain defines Poisson’s ratio.

For most materials subjected to a tensile stress, the resulting longitudinal strain (parallel to the direction of the applied load) is accompanied by a transverse strain (perpendicular to the direction of the applied load). In other words, when we pull on a piece of material, it usually becomes thinner as well as longer. Poisson’s ratio is defined as:

The negative sign is used to obtain a convention in which the values of υ are positive for most materials.

A relatively small number of materials – known as auxetic materials – have a negative Poisson’s ratio. Typically, these are low-density, cellular materials, with cell walls that are able to hinge or buckle when loads are applied. Pulling on a piece of auxetic material causes it to become wider as well as longer, while attempting to squash it will cause it to become thinner as well as shorter.

For isotropic materials, i.e., those in which structure and therefore properties do not vary with direction, it can be shown that only two of the four elastic constants are independent. The relationship between the elastic constants is captured in the following two equations:

Given any two of the constants E, G, K, and υ, these equations suffice to find the remaining two. E, G, and K all have the same units as stress, while υ is dimensionless.

In the case of many biological materials, especially those with a fibrous microstructure, E, G, and υ are anisotropic – in other words, the values obtained when these constants are measured depend on the direction along which load is applied to the sample. (The bulk modulus, K, relates to uniform compression of a material, and so, while its value is dependent on whether and to what extent the sample is microstructurally anisotropic, it does not depend on the orientation of the sample during tests.)

The transverse Young’s modulus of fibers can be difficult to measure accurately, because of the practical challenges inherent in obtaining dependable stress versus strain data from fiber samples loaded parallel to a diameter. There are relatively few options for obtaining such data. We can attempt to use atomic force microscopy, applying and measuring force parallel to the fiber diameter on a suitably small target area of sample, and measuring the very small deflections that characterize strain in a thin (equivalent to short) sample. Alternatively, we can attempt to produce and test a significantly thicker sample of the same material while preserving the molecular alignment typical of the fiber. Better yet, we might be able to calculate the transverse Young’s modulus from first principles, building on the methods that will be introduced in the sections Quantitative Prediction of Elastic Behavior in an Atomic Solid and Quantitative Prediction of Elastic Behavior in a Molecular Solid. However, each of these methods has limitations, so that it is not at all unusual for fibers to be subjected to the simplifying – and itself incorrect – assumption that they are elastically isotropic.

Materials that are used as loadbearing implants need to have elastic properties that appropriately match the properties of the tissue that they replace. For example, if a dental filling has a Young’s modulus significantly different from that of the tooth, the compressive stresses that accompany chewing will cause the filling and adjacent natural material to respond with different longitudinal strains, which in turn may cause the interface between the filling and the tooth to fail. Also, the filling and the tooth must have compatible values of Poisson’s ratio. If the transverse strain induced in the filling is smaller than that which would have been induced in the material replaced by the filling, then the remaining tooth will tend to pull away from the filling.

Condition for Incompressibility (Zero Volume Change)

Consider a rectangular block of material, initially having length l1, width b1, and thickness t1. A force applied parallel to the length of the block stretches the sample to a new length l1; correspondingly, the width decreases to b1 and the thickness decreases to t2. Initially, the volume of the block is:

After stretching, the volume of the block is:

from the definition of strain;

from the definition of υ

plus higher order terms in εlongitudinal.

Because we are dealing with small strains, the higher order terms in εlongitudinal can be approximated as zero, so that:

Zero volume change in response to elastic deformation requires:

This condition pertains in mammalian arteries. The walls have to give in response to each pulse of blood delivered by the heart, but the volume of arterial lumen occupied by fluid is necessarily constant. Therefore, the artery itself has to be incompressible.

There is a second consideration involving Poisson’s ratio that is important in the context of arteries. An artery can be approximated as a (variably) pressurized cylinder. Simple elasticity theory shows that a pressurized cylinder made from incompressible material will become wider and shorter, rather than longer and thinner. Joints made to secure an arterial graft must be able to withstand the longitudinal tensile stress that develops during each pressurization cycle.

Quantitative Prediction of Elastic Behavior in an Atomic Solid

Some quite simple models can be used to predict the load versus extension, and hence stress versus strain relationship for the elastic deformation of materials, allowing us to obtain values of elastic modulus from first principles.

The Morse potential is often used to describe the energy of interaction, U(r), which exists between two atoms, in terms of their separation, r:

where U0, a, and r0 are constants.

The positive and negative terms respectively represent the short-range repulsion and long-range attraction between the two atoms.

Given U(r), it is possible to calculate the force F(r) between the atoms, and thence the stiffness E(r) of the bond between them.

Energy and force are related by the definition:

Force and stiffness are related by the differential form of Hooke’s Law:

The initial stiffness that would be exhibited in a stress–strain test can now be predicted by setting r = r0:

This type of analysis lies at the heart of many materials modeling programs.

Quantitative Prediction of Elastic Behavior in a Molecular Solid

Here we consider materials in which the basic unit of structure is a flexible chain of covalently bonded atoms. Flexibility is afforded by the backbone bonds being able to rotate, as illustrated in Figure I.1.3.5.

FIGURE I.1.3.5 Schematic illustration of a linear polymer chain, where vertices denote the location of backbone atoms. The black and blue lines together represent an extended (rod-like) conformation (shape) of the molecule. Rotation around backbone bonds can produce many other conformations, conferring flexibility on the molecule. For example, rotation around bond AB can move the segment denoted in blue to the position denoted in red, leading to a more compact conformation.

Such rotations allow the chains to adopt many different conformations (shapes), and are therefore associated with increased entropy. By quantifying how stretching the material restricts the number of conformations accessible to the chains, it is possible to calculate the accompanying entropy change, and hence the change in free energy. It is then simple to proceed from the relationship between energy and strain to a relationship between force (or stress) and strain, as we saw in the previous example.

We start by fixing one end of a chain at the origin (0,0,0) of a (Cartesian) coordinate system. Let the other end be at position (x,y,z). The probability of this happening depends on the number of possible chain conformations between the ends thus positioned.

We make some assumptions that will simplify the subsequent mathematics without diminishing the lessons that are learned from this example. The assumptions are: (1) the chains are freely-jointed, which means that we do not impose specific valence angles on the covalently bonded atoms; (2) each chain follows a random walk between its ends, which is only strictly true while the material remains unstretched; (3) the chains have length but no width, so that self-intersection is allowed; and (4) the chains are infinitely long.

A Gaussian (statistical) analysis of random walks can be used to show that the number of conformations between the ends of a chain with ends at (0,0,0) and (x,y,z) is proportional to:

Here n is the number of links in the flexible chain, and l is the length of each link. The quantity nl² represents the mean square end-to-end distance of the chains ‹r²›, i.e., the root mean square end-to-end distance √‹r²› is equal to √nl, consistent with the properties of a random walk. The entropy of the chain is therefore:

where W is the number of ways in which the chain can be rearranged between its fixed end points, and k is Boltzmann’s constant:

where C is a constant:

Now consider stretching the chain to extension ratios λx, λy, λz, along axes x, y, and z respectively. After stretching:

Therefore, as a result of stretching:

This change in entropy reflects the decreased number of conformations available to the chain when stretched – in other words, the stretched state is less probable than the relaxed state. Now consider all the chains in a unit volume of the material:

But x² + y² + z² = r², where r is the straight line distance between the ends of each chain. So, because the distribution of chains is spherically symmetrical in the undeformed state:

if N is the number of chains per unit volume

because freely-jointed chains are assumed.

per unit volume.

If we assume that the work done (energy needed) to stretch the chains is only used to decrease entropy, i.e., the chain conformations change, but the bonds are not stretched, then:

per unit volume,

where T denotes absolute temperature.

This result is similar in structure to the formula for stored energy per unit volume during the elastic deformation of a metal, ceramic or glass (see the section on Resilience below):

We therefore refer to:

NkT as the statistical modulus or rubber modulus, and

ΔU as the stored energy function (SEF).

Now consider the bulk deformation of rubber.

Conservation of volume requires that:

Assuming that the force is applied uniaxially along xleads to:

Substituting this result into the SEF gives:

The total work done in deforming the entire volume of the sample is therefore:

This work is related to the necessary force by:

We now have a result that predicts the relationship between nominal stress and nominal strain for a molecular solid that consists of flexible chains of covalently bonded atoms. Such a material is known as an elastomer. The correspondence between theory and experiment is good (perhaps surprisingly, given the number of assumptions in this relatively simple model), up to an extension ratio of about six. At higher extension ratios, the chains are far from the assumed condition of a random trajectory, and the effects of finite chain length become significant. Also, at that stage of sample deformation the molecules are so well-ordered that crystallization can occur, and so the material no longer has the flexibility that is characteristic of an elastomer.

Examples of elastomers include both natural and synthetic rubber, as well as the tissue that comprises arterial walls. The relationship derived above shows that the stress required to maintain a given strain is an increasing function of temperature. We therefore see that arteries are more resistant to dilation a higher temperatures, requiring a corresponding increase in the effort expended by the heart to maintain blood circulation. This can be a contributing factor to heat stroke.

Plastic Deformation

Yield Strength and Ductility

We return now to consider additional mechanical properties that can be gleaned from nominal stress versus nominal strain curves such as those shown in Figures I.1.3.2 and I.1.3.3. As the stress on a material increases, a point may be reached where the response is no longer linear. There is a permanent (irreversible) deformation, i.e., deformation that is not recovered on removing the stress. This means that some rearrangement of the atoms or molecules must have been triggered. Irreversible deformation is known as plastic deformation. It occurs commonly in metals and polymers, and rarely in ceramics.

The amount of plastic deformation associated with a given stress is found by drawing a line from the point of interest on the stress–strain curve, parallel to the initial linear segment of the curve, and marking the point where this constructed line intersects the horizontal (strain) axis. The horizontal distance from the origin to the point of intersection is a measure of the plastic strain. The value of plastic strain required to break the material defines the ductility (Figure I.1.3.3). The stress at which departure from a linear stress–strain relationship occurs is known as the proportional limit (Figure I.1.3.3).

It is not always easy to ascertain the proportional limit accurately, and so a more practical determination of the condition for plastic deformation is provided by the yield strength (Figure I.1.3.3), which is the stress at which noticeable plastic strain occurs. In this context, noticeable is often taken to be a value of 0.002 (0.2%) for metals – although it may be chosen to be much higher in polymers, where elastic behavior persists to strains that may be ten or even a hundred times higher, and where the transition from elastic to plastic behavior may be difficult to pinpoint.

A marked yield drop is often displayed in the stress versus strain plot when the yield stress of a polymer is exceeded (Figure I.1.3.2). The yield drop occurs because the polymer chains become partially aligned during initial deformation, allowing easier relative motion of chains and reducing the stress needed to further deform the material. The chains then straighten and extend as deformation continues, requiring little increase in stress. Eventually, the opportunities for relative motion of chains are exhausted, along with the capacity for additional chain extension, and the stress versus strain plot then rises sharply.

Strength and Failure

Failure of a loadbearing material occurs when the material ceases to perform its loadbearing function. Different interpretations of failure are in common use, and so it is important to be clear about the intended meaning when we speak of materials failure.

For materials that must not undergo in situ permanent deformation, failure is synonymous with the yield stress being exceeded, and so the yield stress represents an estimate of the strength of the material in those cases.

For materials in which permanent deformation (and the attendant shape change) is acceptable, failure may be deemed to occur when a noticeable neck (constriction) develops in the material. The effect of the neck is to concentrate the load on a smaller area; therefore, the load that can be supported by the sample is decreased. On a nominal stress versus nominal strain plot, where stresses are referred to the original loadbearing areas of the sample, the stress required for additional deformation decreases. Therefore, the onset of necking corresponds to a maximum in the nominal stress versus nominal strain plot, defining the ultimate tensile strength (UTS) or simply the tensile strength of the material (Figure I.1.3.3). There is no corresponding maximum on a true stress versus true strain plot, because the stress continues to rise past the onset of necking as values of stress are obtained by dividing the applied load by the actual (small) cross-section of the neck.

On both nominal stress versus nominal strain and true stress versus true strain plots, a breaking strength can be defined at the point where the material actually breaks. However, to reach this point practically, at least in the case of most metals, the tensile strength of the material must first be reached, and it is tensile strength that represents the practically measurable maximum stress that a metal can survive. Ceramics, on the other hand, usually reach their breaking stress before they yield.

Thus, the perceived strength of a material depends in part on the definition of failure that is used (onset of plastic deformation, onset of necking, or actual occurrence of breaking). It also depends on the conditions that are used for testing. The strain rate used in a tensile test can affect the maximum stress that is reached, depending on the ability of the microstructure to undergo rearrangement to accommodate the imposed deformation.

Over long periods of time, even stresses that are below the conventionally measured yield strength of a material may be sufficient to cause gradual elongation and eventual failure via creep. Therefore, creep tests, in which a fixed load is applied while the strain is monitored until the sample breaks, are important for materials that are required to have a long projected service life – as is the case with many implants.

Similarly, materials that will be subjected to cyclic loading patterns during service may undergo fatigue failure at loads that are smaller than those needed to trigger failure in a conventional stress versus strain test. Testing to determine the relationship between average load, maximum and minimum load, and the number of loading cycles that lead to failure is required in such cases. Finally, the catastrophic failure of a material can be triggered by pre-existing flaws that locally concentrate the effect of an applied stress. We will explore this phenomenon in the section Statistical Aspects of Failure.

Stability of Necks

A neck will be stable if its formation is accompanied by sufficient work hardening, i.e., by microstructural changes that increase the resistance of the material to further deformation and thus increase the load that can be carried by the necked region. Referring to Figure I.1.3.6 we see that this condition requires:

FIGURE I.1.3.6 True stress and sample geometry in the vicinity of a neck.

if we neglect the product of infinitesimals.

We need to convert the change in area into a change in length, in order to write this condition in terms of stress and strain. Because the deformation is plastic at the stage where necking occurs, we can assume constant volume:

if we again neglect the product of infinitesimals.

from the definition of εt.

But previously (see section True Stress and True Strain) we saw that εt = ln (εn + 1). Therefore:

for stability

This is Considère’s criterion. Applied graphically (Figure I.1.3.7) it is known as Considère’s construction.

FIGURE I.1.3.7 .

Hardness

The hardness of a material is measured by applying a known load to a small indenter of known geometry (typically pyramidal or spherical) in contact with the surface of the material, for a known period of time. The dimensions of the resulting indentation are measured, and this information, together with the experimental conditions, is used to rate the hardness of the material on a relative scale.

Indents made in a soft material under a given set of conditions are larger than those made in a hard material. Therefore, hardness provides a measure of how successfully a material resists plastic deformation, which in turn is characterized by both yield strength and tensile strength. It is therefore possible to empirically develop calibration charts that can be used to convert hardness measurements into both yield strength and tensile strength values.

Hardness testing is popular for estimating the yield strength and tensile strength of materials because the equipment is relatively simple and inexpensive, and the tests are nondestructive.

Resilience

Resilience is a measure of the elastic energy that can be stored in a unit volume of stressed material. It corresponds to the area underneath a stress versus strain plot, extending from zero strain up to the strain at which the sample yields:

If the stress versus strain relationship is linear up to the yield strain (typical for ceramics, and approximate for many metals and polymers), the corresponding area under the stress versus strain plot is triangular, and the integral in the previous equation simplifies to:

Resilience is an important consideration for materials that need to be springy, i.e., that absorb and return significant amounts of energy. Examples include natural and replacement tendons and ligaments, and materials used in foot arch reconstruction or support.

Toughness and Fracture Toughness

Although these two terms sound superficially similar, they refer to quite different properties.

Toughness

Fracture toughness is a measure of how successfully a material resists the propagation of cracks. Consider a sample of material containing either an internal crack of length 2a or a surface crack of length a, and subjected to a stress σ, as shown in Figure I.1.3.8. We will explore the former case in detail; the results are identical in the second case.

FIGURE I.1.3.8 Sample geometry used in defining the fracture toughness of a material. Examples of both a surface crack and an internal crack are shown. The cracks are assumed to be cylindrical (blue circles) for ease of mathematical description; a flattened profile (red ellipses) would be more realistic.

When the crack grows, elastic energy stored in the material is used to break bonds and create new surface. The energy released in forming a cylindrical crack is:

where t represents sample thickness

if we perform a rigorous analysis on a flattened (more realistic) crack rather than the cylindrical one. The energy required to make the surface of the crack (the new interface between the material and its environment) is:

taking account of the fact that two surfaces are created by interrupting bulk material, and using Γs to denote energy per unit area of surface.

Propagation (growth) of the crack requires the accompanying energy release to exceed the accompanying energy consumption:

or we can identify a critical stress σc for crack propagation:

For a material in which plastic deformation occurs while cracks propagate, the factor Γs is increased accordingly. The previous equation can be rearranged to give:

or, more generally:

where Y is a constant that depends on sample and loading geometry. Since the quantities on the right-hand side of this equation are material-specific properties, it follows that the expression on the left-hand side is also a material-specific property. It is known as the fracture toughness of the material. It quantifies the maximum size of flaw (crack) that can be tolerated if the material is to support a particular stress. Its units are unusual: MPa.m¹/².

The presence of the factor Γs in the equation for fracture toughness emphasizes that the ability of a material to resist fracture depends on the environment in which it is tested. In particular for biomaterials, where long in-service lifetimes are required and expected, it is important for mechanical testing to be performed under environmental conditions that are properly representative of the conditions that will be encountered by the material when in service.

Statistical Aspects of Failure

Microstructural variability between samples, especially the distribution of pre-existing flaws of different sizes, leads to variable results in tests of mechanical properties. From an engineering design point of view, materials selection should be guided by the low values of whatever property is (or properties are) relevant to the product being able to perform properly. Even if a property such as ultimate tensile strength is high on average, it is the probability of encountering a sample with a low value of tensile strength that must be used to inform the choice of safe design limits. Thus, statistical measures of spread in measured data – and hence of a material’s reliability – are needed.

Consider a batch of samples, each of which is tested to determine the ultimate tensile strength. Let F denote the fraction of those samples that have failed at a stress σF therefore represents a cumulative failure probability, and it is an increasing function of σ. It can be shown that, if all the samples have the same size and shape:

where the parameters m and σ0 are characteristic of the material, and m is known as the Weibull modulus. Therefore:

versus lnσ will yield a straight line with slope equal to the Weibull modulus of the material. The term modulus is sometimes incorrectly interpreted here as referring to a type of stiffness, because of its frequent association with elastic constants. However, in all the senses that it is used in this chapter, modulus simply means number, in keeping with its origin in the Latin word modulus, a (small) measure.

Engineering materials that are required to exhibit great in-service reliability, for example the metal alloys used in critical aircraft components, exhibit Weibull moduli well over 100. In contrast, ordinary glass has a Weibull modulus of around 5, and common ceramics typically have Weibull moduli in the range 5 ≤ m ≤ 25. Perhaps surprisingly, the Weibull modulus of natural materials can be quite low. Despite their reputation for high average breaking strength, stiffness, and toughness in conventional tensile tests, natural silk fibers exhibit low Weibull moduli (5.8 for silkworm silk, and 3.4 for spider dragline). The poor reproducibility of breaking strength constrains our ability to incorporate these materials in reliable design, and suggests that their best use might be in composites where applied loads are distributed across multiple fibers.

Other Bulk Properties

The bulk properties of a material are of course not limited to performance in mechanical tests. Other types of bulk properties that may have to be taken into account during the selection of a biomaterial include: thermal properties; optical properties; electrical properties; and magnetic properties. Even cost – financial and environmental – can be considered to be a bulk property, given the dependence of cost on: (1) the type and volume of material that is used; and (2) the complexity of microstructural rearrangement during processing.

Thermal Properties

Thermal conductivity becomes a significant consideration if an implanted material contributes to an unnatural flow of heat through the surrounding tissue. For example, metal rods selected for their combination of stiffness, strength, fracture toughness, and biocompatibility can promote heat loss and cause the patient to feel colder than normal. Heat conduction through a metal dental filling can also be a source of discomfort, and may therefore play a role in the choice between a metal or non-metal filling.

Thermal expansion is a (nominal) strain that occurs when the temperature of a material is changed:

where α is the thermal expansion coefficient; its units are (degrees)−1, with degrees measured on whatever standard scale is used to measure temperature T.

Attempts to match the properties of a dental filling to those of the surrounding tissue must therefore extend beyond achieving compatible elastic constants (see section Elastic Constants) to include consideration of thermal expansion, to avoid premature failure of the interface.

Optical Properties

In the context of biomaterials, the most significant bulk optical properties are color, refractive index, and transparency; all three are important in the selection of materials for intraocular lenses or fluids.

The color of a transparent material is controlled by composition, and therefore demands a high degree of quality control to avoid impurities that could adversely affect color. Long-term stability of material composition is also important: components should not selectively diffuse into the surrounding tissue, and no chemical changes should occur – either by reactions between the components or in response to light.

Refractive index n of a material is defined as:

where c denotes the speed of light in the subscripted medium.

When light crosses the interface between two media (materials), it is deviated from its original path by an angle that is an increasing function of the difference between the refractive indices of the media. Therefore, the effectiveness of a material as a lens is directly related to its refractive index. Refractive index increases along with the content of electron-rich atoms, which is why lead crystal is so useful in both lenses and decorative glassware.

Transparency is a qualitative term that describes the ability of a material to transmit light without attenuating (absorbing or scattering) it. To minimize absorption, the primary bonds in the material must be strongly covalent or ionic (and definitely not metallic). Scattering is minimized if the material is free of internal interfaces (which could reflect light) and compositional differences (which would be associated with refractive index differences that could deviate light from an uninterrupted path through the material). Thus, an optimally transparent material will either be a homogenous single crystal or it will be completely and homogeneously amorphous.

There is no commonly accepted quantitative definition of transparency. Instead, it is usual to consider the complementary property, opacity. This is itself a colloquial term, but it lends itself to a more formal description by way of the mass attenuation coefficient. If light of a given frequency (color) and initial intensity I0 travels a distance x through a material, then the intensity is decreased to:

where μ is the mass attenuation coefficient of the material, and ρ is the density.

Weibull Modulus and Non-Brittle Materials

The Weibull distribution of fracture probability for materials provides a statistical measure of brittleness, via the specification of a Weibull modulus m (Derby et al., 1992). The value of m can in theory range from zero (totally random fracture behavior, where the failure probability is the same at all stresses, equivalent to an ideally brittle material) to infinity (representing a precisely unique, reproducible fracture stress, equivalent to an ideally non-brittle material).

This quantitative approach to defining non-brittleness is more robust than qualitative description, since there is no