Vous êtes sur la page 1sur 2562


Franco Bassani, Scuola Normale Superiore, Pisa, Italy

Gerald L Liedl, Purdue University, West Lafayette, IN, USA

Peter Wyder, Grenoble High Magnetic Field Laboratory, Grenoble, France

Editorial Advisory Board
Vladimir Agranovich, Russian Academy of Sciences, Moscow, Russia

Angelo Bifone, GlaxoSmithKline Research Centre, Verona, Italy

Riccardo Broglia, Universita degli Studi di Milano, Milano, Italy

Kikuo Cho, Osaka University, Osaka, Japan

Ge! rard Chouteau, CNRS and MPI-FKF, Grenoble, France

Roberto Colella, Purdue University, West Lafayette, IN, USA

Pulak Dutta, Northwestern University, Evanston, IL, USA

Leo Esaki, Shibaura Institute of Technology, Japan

Jaap Franse, Universiteit van Amsterdam, Amsterdam, The Netherlands

Alexander Gerber, Tel Aviv University, Tel Aviv, Israel

Ron Gibala, University of Michigan, Ann Arbor, MI, USA

Guiseppe Grosso, Universita" di Pisa, Pisa, Italy

Jurgen M Honig, Purdue University, West Lafayette, IN, USA

Massimo Inguscio, Dipartmento di Fisica e L.E.N.S., Firenze, Italy

A G M Jansen, Institut Max Planck, Grenoble, France

Th W J Janssen, Katholieke Universiteit Nijmegen, Nijmegen, The Netherlands

Giorgio Margaritondo, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland

Emmanuel Rimini, Universita di Catania, Catania, Italy

Robin D Rogers, The University of Alabama, Tuscaloosa, AL, USA

John Singleton, Los Alamos National Laboratory, Los Alamos, NM, USA

Carl H Zweben, Devon, PA, USA


Physics is the paradigm of all scientific knowledge. Over the centuries it has evolved to a complexity that has
resulted in a separation into various subfields, always connected with one another and very difficult to single
out. Freeman Dyson, in his beautiful book ‘Infinite in All Directions’, distinguishes two aspects of physics and
two types of physicists: the unifiers and the diversifiers. The unifiers look for the most general laws of nature,
like the universal attraction between masses and electric charges, the laws of motion, relativity principles, the
simplest elementary particles, the unification of all forces, symmetry violation and so on. The diversifiers
consider the immense variety of natural phenomena, infinite in their extension, try to explain them on the basis
of known general principles, and generate new phenomena and devices that do not exist in nature. Even at the
beginning of modern science Galileo Galilei, besides studying the laws of motion and laying down the principle
of relativity, was interested in the phenomenon of fluorescence and disproved the theories put forward at his
time. He was both a unifier and a diversifier. The full explanation of fluorescence had to await the advent of
quantum mechanics, as did the explanation of other basic phenomena like electrical conductivity and
The past century witnessed an explosive expansion in both aspects of physics. Relativity and quantum
mechanics were discovered and the greatest of the unifiers, Albert Einstein, became convinced that all reality
could be comprehended with a simple set of equations. On the other hand a wide range of complex phenomena
was explained and numerous new phenomena were discovered. One of the great diversifiers, John Bardeen,
explained superconductivity and invented the transistor.
In physics today we encounter complex phenomena in the behavior of both natural and artificial complex
systems, in matter constituted by many particles such as interacting atoms, in crystals, in classical and quantum
fluids as well as in semiconductors and nanostructured materials. Furthermore, the complexity of biological
matter and biological phenomena are now major areas of study as well as climate prediction on a global scale.
All of this has evolved into what we now call ‘‘condensed matter physics’’. This is a more comprehensive term
than ‘‘solid state physics’’ from which, when the electronic properties of crystals began to be understood in the
thirties, it originated in some way. Condensed matter physics also includes aspects of atomic physics,
particularly when the atoms are manipulated, as in Bose–Einstein condensation. It is now the largest part of
physics and it is where the greatest number of physicists work. Furthermore, it is enhanced through its
connections with technology and industry. In condensed matter physics new phenomena, new devices, and new
principles, such as the quantum Hall effect, are constantly emerging. For this reason we think that condensed
matter is now the liveliest subfield of physics, and have decided to address it in the present Encyclopedia. Our
focus is to provide some definitive articles for graduate students who need a guide through this impenetrable
forest, researchers who want a broader view into subjects related to their own, engineers who are interested in
emerging and new technologies together with biologists who require a deeper insight into this fascinating and
complex field that augments theirs.
In this Encyclopedia we have selected key topics in the field of condensed matter physics, provided historical
background to some of the major areas and directed the reader, through detailed references, to further reading
resources. Authors were sought from those who have made major contributions and worked actively in the

area of the topic. We are aware that completeness in such an infinite domain is an unattainable dream and have
decided to limit our effort to a six-volume work covering only the main aspects of the field, not all of them in
comparable depth.
A significant part of the Encyclopedia is devoted to the basic methods of quantum mechanics, as applied to
crystals and other condensed matter. Semiconductors in particular are extensively described because of their
importance in the modern information highways. Nanostructured materials are included because the ability to
produce substances which do not exist in nature offers intriguing opportunities, not least because their
properties can be tailored to obtain specific devices like microcavities for light concentration, special lasers, or
photonic band gap materials. For the same reasons optical properties are given special attention. We have not,
however, neglected foundation aspects of the field (such as mechanical properties) that are basic for all material
applications, microscopy which now allows one to see and to manipulate individual atoms, and materials
processing which is necessary to produce new devices and components. Attention is also devoted to the ever-
expanding role of organic materials, in particular polymers. Specific effort has been made to include biological
materials, which after the discovery of DNA and its properties are now being understood in physical terms.
Neuroscience is also included, in conjunction with biological phenomena and other areas of the field.
Computational physics and mathematical methods are included owing to their expanding role in all of
condensed matter physics and their potential in numerous areas of study including applications in the study of
proteins and drug design. Many articles deal with the description of specific devices like electron and positron
sources, radiation sources, optoelectronic devices, micro and nanoelectronics. Also, articles covering essential
techniques such as optical and electron microscopy, a variety of spectroscopes, x-ray and electron scattering
and nuclear and electron spin resonance have been included to provide a foundation for the characterization
aspect of condensed matter physics.
We are aware of the wealth of topics that have been incompletely treated or left out, but we hope that by
concentrating on the foundation and emerging aspects of the infinite extension of condensed matter physics
these volumes will be generally useful.
We wish to acknowledge the fruitful collaboration of the members of the scientific editorial board and of the
Elsevier editorial staff.
Special thanks are due to Giuseppe Grosso, Giuseppe La Rocca, Keith Bowman, Jurgen Honig, Roberto
Colella, Michael McElfresh, Jaap Franse, and Louis Jansen for their generous help.

Franco Bassani, Peter Wyder, and Gerald L Liedl

Permission Acknowledgments
The following material is reproduced with kind permission of Nature Publishing Group

Figure 12 of Ionic and Mixed Conductivity in Condensed Phases

Figure 1 of Porous Silicon

Figure 5 of Quasicrystals, Electronic Structure of

Figure 2 of Superconductivity: Flux Quantization


The following material is reproduced with kind permission of the American Association for the
Advancement of Science

Figure 6 of Biomolecules, Scanning Probe Microscopy of

Figure 4 of Excitons: Theory

Figures 5, 6 and 7 of Genetic Algorithms for Biological Systems

Figures 1 and 3 of Rhodopsin and the First Step in Vision

Figures 4 and 5 of Scattering, Inelastic: Electron


The following material is reproduced with kind permission of Taylor & Francis Ltd

Figure 5 of Ionic and Mixed Conductivity in Condensed Phases

Figure 2 of Semiconductor Nanostructures


The following material is reproduced with kind permission of Oxford University Press

Table 1 of Crystal Tensors: Applications

Figure 2 of Cyclotron Resonance: Metals

Figure 4 of Magnetoresistance Techniques Applied to Fermi Surfaces

Acoustics: Physical Principles and Applications to Condensed
Matter Physics
J D Maynard, The Pennsylvania State University, line of liquid helium. For the field of supercon-
University Park, PA, USA ductivity, a paper on acoustic attenuation was in-
& 2005, Elsevier Ltd. All Rights Reserved. cluded as one of the relatively small number of
selected reprints on superconductivity published by
the American Institute of Physics. Acoustic measure-
ments are among the first performed whenever a
Introduction material involving novel physics is discovered. Mod-
ern acoustic techniques, discussed below, can probe
For many condensed matter systems, including liq- the properties of samples only a few hundred microns
uids as well as solids, acoustic measurements provide in size and nanoscale thin films, and may be utilized
a crucial probe of important and fundamental phys- in practical applications such as micro-electro-me-
ics of the system. In the case of solids, one of the first chanical systems (MEMS). Acoustic measurements
fundamental properties to be determined would be provide significant information about condensed
the atomic structure, defined by the minimum in the matter systems, and their accurate and precise meas-
free energy with respect to the positions of the atoms. urement is certainly important.
The next fundamental characteristic of interest might
be the curvature of the free energy in the vicinity of
the minimum, and this would be manifest in the
Acoustics in Solids
elastic constants for the material. As derivatives of
the free energy, elastic constants are closely connect- For solids, acoustic phenomena reflect the elastic
ed to thermodynamic properties of the material; they properties of the material. Interest in elasticity dates
can be related to specific heat, Debye temperature, back to Galileo and other philosophers in the seven-
the Gruneisen parameter, and they can be used to teenth century, who were interested in the static
check theoretical models. Extensive quantitative con- equilibrium of bending beams. With the basic physics
nections may be made if the elastic constants are introduced by Hooke in 1660, the development of
known as functions of temperature and pressure. the theory of elasticity followed the development of
Acoustic measurements not only probe lattice prop- the necessary mathematics, with contributions from
erties, they are also sensitive probes of the environm- Euler, Lagrange, Poisson, Green, etc., and the result-
ent in which all interactions take place, and may be ing theory was summarized in the treatise by A E H
used to study electronic and magnetic properties Love in 1927.
(e.g., through magnetostriction effects). As will be Acoustic and elastic properties of solids are quan-
discussed later, acoustic measurements involve tensor tified in a set of elastic constants. These constants are
quantities, and thus can probe anisotropic properties like spring constants, relating forces, and displace-
of crystals. The damping of elastic waves provides ments, and they may be measured with a static tech-
information on anharmonicity and on coupling with nique, in which a displacement is measured as a
electrons and other relaxation mechanisms. One of linear response to a small applied force. However, it
the most important features of acoustic measure- was long ago learned that a better method is to
ments is that they provide a sensitive probe of phase measure an elastic vibration, as found, for example,
transitions and critical phenomena; important exam- in a propagating sound wave. Most existing com-
ples, in addition to the obvious example of structural plete sets of elastic constants for materials have been
transitions in solids, include the superconducting determined by measuring the time of flight of sound
transition and the superfluid transition. Indeed, one pulses. Recently, a relatively new method, resonant
of the most impressive successes in critical phenom- ultrasound spectroscopy (RUS), is being used. In the
ena has been the use of acoustics to study the lambda RUS method, rather than measuring sound velocities,
2 Acoustics: Physical Principles and Applications to Condensed Matter Physics

one measures the natural frequencies of elastic vib- in the i-direction is @sij =@xj , and Newton’s law may
ration for a number of normal modes of a sample, be written as
and processes these in a computer, along with the
shape and mass of the sample. With a proper con- @sij @2c
¼ r 2i ½3
figuration, a single measurement yields enough fre- @xj @t
quencies to determine all of the elastic constants for a
material (as many as 21 for a crystal with low sym- where r is the mass density.
metry). Samples may be prepared in a wide variety of The symmetric nature of the definitions, and the
shapes, including rectangular, spherical, etc., and it is assumption that the elastic energy must be quadratic
not necessary to orient crystalline samples. A com- in the strains, reduces the number of independent
pelling reason for using RUS has to do with the na- elements of cijkl from 81 to 21. A basic symmetry has
ture of samples of new materials. Whenever a new cijkl invariant if the indices are exchanged in the first
material is developed, initial single crystal samples pair or second pair of the four subscripts (cjikl ¼ cijkl ,
are often relatively small, perhaps on the order of a etc.); thus a reduced system of indices may be used:
fraction of a millimeter in size. Also, with new 11-1, 22-2, 33-3, 23-4, 13-5, 12-6, so that
developments in nanotechnology and the possibility cijkl -cmn . The reduced system is used when tabulat-
of applications in the microelectronics industry, there ing values of elastic constants; however, the full four-
is a great interest in systems which are very small in index tensor must be used in calculations. Additional
one or more dimensions, such as thin films and one- symmetries of a particular crystal group will reduce
dimensional wires. For such small systems, pulse the number of independent elastic constants further
measurements are difficult, if not impossible, but below 21; for example, orthorhombic crystals have
RUS methods may be readily used. nine independent elastic constants, cubic crystals
have three, and isotropic solids have only two.
Physical Principles for Acoustics in Solids
Later, the case of an isotropic elastic solid will be
To begin a theory for acoustics in solids, one may useful for the purposes of illustration. In this case,
imagine a spring, extended with some initial tension, one has c11 ¼ c22 ¼ c33 , c44 ¼ c55 ¼ c66 , c12 ¼ c13 ¼
and consider two points at positions x, and x þ dx. If c23 ¼ c11  2c44 , and all other elements of the
one applies an additional local tension, or stress, s, elastic tensor are zero. The two independent elastic
then the spring stretches and the two points are dis- constants may be taken as c11 and c44, but other
placed by c(x), and cðx þ dxÞ respectively. The sep- combinations, such as Young’s modulus Y ¼
aration between the two points will have changed by c44 ð3c11  4c44 Þ=ðc11  c44 Þ and the bulk modulus
dc, and the fractional change in the separation, de- B ¼ c11  4c44 =3 are also used. The bulk modulus
fined as the strain, is e  dc=dx. Hooke’s law for the appears in an important thermodynamic identity
spring takes the form s ¼ ce, where c is a one-dimen- involving g, the ratio of the specific heat at constant
sional elastic constant. For a three-dimensional elastic pressure cp to that at constant volume, cv:
solid, one may use indices (i, j, etc.) which can take on
the values 1, 2, and 3, referring to the x, y, and z cp Tb2 B
coordinate directions, and generalize the strain to g¼ ¼1þ ½4
cv rcp
1 @ci @cj
eij  þ ½1 Here T is the temperature and b is the thermal ex-
2 @xj @xi
pansion coefficient (TEC).
The symmetric form of eij avoids pure rotations, Anharmonic Effects
which do not involve stress. The stress is generalized
to sij, a force per unit area acting on a surface ele- The basic formulation of acoustics in solids involves
ment, where the first index refers to the coordinate the expansion of energy minima about equilibrium to
direction of a component of the force, and the second second order, or equivalently, assuming a harmonic
index refers to the coordinate direction of the unit potential, quadratic in strain. However, there are a
normal to the surface element. Hooke’s law becomes number of effects which require going beyond second
order. Some effects are related to exceeding ‘‘small
sij ¼ cijkl ekl ½2 displacements’’ from equilibrium, such as in quan-
tum solids with large zero-point motion, and solids
where cijkl is the 3  3  3  3 (81 element) elastic at high temperatures (near melting) where thermal
tensor, and where a summation over repeated indices motions are large. Other effects occur in equilibrium
is implied. For a small volume element, the net force at normal or low temperatures; these include thermal
Acoustics: Physical Principles and Applications to Condensed Matter Physics 3

expansion, lattice thermal conductivity, and acoustic wave solutions can superimpose and pass through
dissipation. one another with no effect. An anharmonic potential
The relationship between anharmonic effects and allows sound waves (lattice vibrations) to interact
thermal expansion is worth discussing. As tempera- and scatter from one another, permitting the transfer
ture is increased, the amplitude of atomic oscillations of energy from an ordered to a disordered form
increases, or equivalently the occupation of higher (acoustic dissipation), and allowing a change in dis-
quantized energy levels increases. If the potential tribution functions in passing from one location to
energy was exactly quadratic in displacements, then another (lattice thermal conductivity).
the center of oscillation, or the expectation value of Anharmonic effects may be probed with acoustic
displacement, would remain the same. With the same experiments by measuring the changes in the elastic
average positions for the atoms, the system would constants as the sample is subjected to increasing
not expand with increasing temperature, and the uniaxial or uniform hydrostatic pressure. The coef-
TEC b would be zero. From the thermodynamic ficients which relate the changes to the pressure are
identity in eqn [4], one would also have cp ¼ cv . On referred to as ‘‘third-order elastic constants.’’ How
the other hand, when potentials are anharmonic, one elastic constants themselves are determined with
may have (in the typical case) a stiffer repulsion at acoustic measurements is discussed next.
short interatomic distances, and weaker attractive
forces at larger distances. The result is that at higher
energy levels, the ‘‘center’’ position between classical Determining Elastic Properties Experimentally
turning points moves to larger distances, and the To determine the nature of sound propagation in sol-
system thermally expands. This situation is illustrat- ids, one must solve eqns [1] through [3] with some
ed in Figure 1. specified boundary conditions. Because of the tensor
Anharmonicity and thermal expansion can also be nature of the equations, the relation between particle
readily related to nonlinear acoustics in fluids. For displacement and the direction of wave propagation
fluid acoustics, nonlinear effects are proportional to a is quite complicated. To tackle the complexity and
dimensionless second order parameter ðr=vÞð@v=@rÞ, make a connection between ultrasound measure-
where v is the sound speed, and the derivative is at ments and the elastic constants, two approaches may
constant entropy. For gases, this parameter is (g  1), be taken. The first approach, used in conventional
which by the thermodynamic identity in eqn [4], is pulse ultrasound, is to note that if one had a sample
proportional to the thermal expansion coefficient b. with a large (infinite) plane surface which is perpen-
Thus the absence of nonlinear acoustic effects coin- dicular to one of the principle axes of the elastic ten-
cides with a vanishing thermal expansion. sor, and if a plane wave could be launched from that
That an anharmonic potential results in acoustic surface, then the tensor equations would uncouple,
dissipation and lattice thermal conductivity may be and a longitudinal wave or one of two transverse
understood by noting that with a harmonic potential, waves could propagate independently. In this case, for
one gets a linear second-order wave equation, whose each wave, the relationship between the sound velo-
city and the independent elastic constants is fairly
straightforward. While the determination of the prin-
ciple axes and the relationships between the three
sound speeds and the relevant elastic constants may
be done analytically, the manipulations are compli-
cated and must be done on a case-by-case basis; there
Potential energy

Atomic position is no elucidating general formula. The simplest case

of an isotropic elastic solid will be presented here for
purposes of illustration. In this case, Newton’s law
may be written in terms of the two independent elas-
tic constants, c11 and c44:

d2 w
r ¼ c11 r2 w  c44 =  ð=  wÞ ½5

¼ c44 r2 w þ ðc11  c44 Þ=ð= . wÞ ½6

Figure 1 Illustration of the relationship between a nonquadratic
potential energy curve (and nonlinear acoustics) and the phe- where the two equations are related by an identity for
nomenon of thermal expansion. the = operator. If one has =  w ¼ 0, then the first
4 Acoustics: Physical Principles and Applications to Condensed Matter Physics

equation becomes a simple wave equation

ffi for a Experimental Methods for Acoustic Measurements
longitudinal wave with speed vl ¼ c11 =r, and if one in Solids
has =  w ¼ 0, then the second equation becomes a Acoustic measurements with the pulse method are
ffiffiffiffiffiffiffiffiffiffiffiffiequation for transverse waves with speed fairly straightforward; emphasis is on careful bond-
vt ¼ c44 =r. ing of transducers to samples and the use of suitable
Although the pulse ultrasound method has been high-frequency pulse electronics. The RUS method
used extensively in the past, it has a number of di- is less well known, and can be briefly described as
sadvantages, including problems with transducer ring- follows.
ing, beam diffraction, and side-wall scattering, and the In a general RUS measurement, the natural fre-
inconvenience that the sample must be recut, repol- quencies of a sample with stress-free boundary con-
ished, and reattached to a transducer if one wants ditions are determined by measuring the resonance
more than the three elastic constants accessible with frequencies of the sample when held (lightly, with no
one measurement. The second approach to determi- bonding agents, at two positions on the sample sur-
ning elastic constants avoids all of the disadvantages. face) between two transducers. One transducer acts
The second approach is the one described earlier as as a drive to excite vibrations in the sample at a
RUS, which involves the use of a computer to nu- tunable frequency, and the second measures the
merically solve the elastic constants given a set of amplitude (and possibly the phase) of the response
measured natural frequencies for a solid with a given of the sample; as the frequency of the drive is swept,
shape and boundary conditions (usually stress-free a sequence of resonance peaks may be recorded. The
conditions). The computer processing involves sol- positions of the peaks will determine the natural
ving a ‘‘forward problem’’ (finding the natural fre- frequencies fn (and hence the elastic constants), and
quencies in terms of the elastic constants) first and the quality factors (Q’s, given by fn divided by the
then inverting. Unlike the conventional pulse ultra- full width of a peak at its half-power points) will
sound approach, the forward problem does not provide information about the dissipation of elastic
provide a simple relationship between the modes of energy.
vibration and the elastic constants; the displacements RUS may also be used to measure the properties of
in the various modes involve all of the elastic con- thin films on a substrate, to determine the effects of
stants in a complicated manner, and a numerical induced strain from lattice mismatch, etc. In this
computation is required to sort it all out. case, the natural frequencies of the substrate alone
The forward problem may be posed as the min- are measured, then the same sample is again meas-
imization of a Lagrangian L given by ured with the film in place. From the shifts in the
ZZZ natural frequencies, the properties of the film may be
L¼ ðro2 ci ci  cijkl eij ekl Þ dV ½7 determined.
A simple apparatus for making RUS measure-
ments is illustrated in Figure 2. In the illustration, a
The minimization is accomplished numerically with rectangular parallelepiped sample is supported by
a Rayleigh–Ritz method, and the results yield dis- transducers at diametrically opposite corners. Cor-
crete resonance frequencies, fn ¼ 2pon , given the ners are used for contact because they provide elas-
elastic constants, cijkl. For the RUS technique, what is tically weak coupling to the transducers, greatly
needed is the inverse. In most cases, there will be reducing loading, and because the corners are al-
more measured frequencies than independent elastic ways elastically active (i.e., they are never nodes),
constants; so what is required is to find a set of in- and thus can be used to couple to all of the normal
dependent elastic constants which best fits the meas- modes of vibration.
ured frequencies, usually in a least squares sense.
Furthermore, when there are more measured fre-
quencies than independent elastic constants, then
Acoustics in Fluids
other parameters may be varied in order to best fit
the measured frequencies. Such parameters may in- The thermo-hydrodynamic state of a fluid may be
clude the shape and dimensions of the sample (al- specified with five fields, which may be taken as the
though one known length is necessary), and the mass density r(r, t), the pressure p(r, t), and the mean
orientation of the crystallographic axes relative to flow velocity u(r, t). The five equations needed to
the faces of the sample. In any case, it is not neces- determine the five fields are conservation of mass,
sary that crystallographic axes be oriented with re- Newton’s law for the motion of the center of mass of
spect to faces of a sample, although computations are a fluid element (three components), and conservation
greatly simplified if they are oriented. of energy for motion about the center of mass. These
Acoustics: Physical Principles and Applications to Condensed Matter Physics 5

Electrical leads
to transducer


active area

transducer mount

Figure 2 An apparatus for measuring elastic constants and acoustic attenuation with RUS.

equations are respectively: fixed through its viscosity, and the inviscid superfluid
@r component is still free to flow. The four sound modes
þ =  ðruÞ ¼ 0 ½8 are (1) a pressure wave with the two components
moving together (referred to as ‘‘first sound’’), (2) a
  temperature wave with the two components moving
r þ u  =u ¼  =p þ Zr2 u in counterflow (‘‘second sound’’), (3) a pressure wave
  with the normal fluid clamped (‘‘fourth sound’’), and
þ Z þ x =ð=  uÞ ½9 (4) a temperature wave with the normal fluid
3 clamped (‘‘fifth sound’’). An illustration of these
   2 modes is presented in Figure 3. ‘‘Third sound’’ is a
@S 2Z 1 surface wave which propagates on a thin film of su-
T þ u  =S ¼ eij  ð=  uÞ
@t r 3 perfluid; it is not a fundamental sound mode because
x 1 its restoring force is not intrinsic to the superfluid,
þ ð=  uÞ2 þ =  ðk=TÞ ½10
r r but rather is determined by the substrate on which
the superfluid film is formed. With both mechanical
where S is entropy, eij ¼ ð@ui =@xj þ @uj =@xi Þ=2, Z and and thermal properties represented in the sound
x are the shear and bulk viscosity, and k is the thermal modes, the sound speeds, measured as functions of
conductivity. The transport terms (involving Z, x, and temperature and pressure, may be used to determine
k) give rise to dispersion and dissipation. If these all of the thermodynamics of the superfluid.
terms are dropped, then the last equation simply gives
S ¼ constant, and the first two (for small displace-
ments from equilibrium) are easily combined to give a
Applications of Acoustics in Condensed
wave equation with a sound speed given by ð@p=@rÞS.
Matter Physics
Superfluids are modeled as having a superfluid
component (in a macroscopic quantum ground state) In areas of condensed matter physics which involve
and a normal-fluid component (a gas of excitations the development of exotic materials, it is often of
above the ground state). With two fluid components, great value to use acoustics simply to measure and
nine fields are required to specify the state of the tabulate sound speeds or elastic constants. This is the
system; in addition to the ones for a classical fluid, case for such materials as alloys, composites, porous
one also has the mass density and the mean flow materials, sintered materials, polymers (plastics,
velocity for the superfluid component alone. Now epoxies, elastomers, etc.), cements, piezoelectrics,
the linearized equations admit two sound speeds viscoelastic and non-Newtonian fluids, and fluid
for the unconstrained fluid, and two more sound mixtures. In some cases, it is of significant benefit
speeds when the normal fluid component is held to use acoustics to monitor systems which evolve in
6 Acoustics: Physical Principles and Applications to Condensed Matter Physics

First Second elastic strain energy, may be expressed as a fourth-

n n order equation in some order parameter C:
s s F ¼ 12c0 e2 þ 12a1 ðT  Tc ÞC2 þ 14a2 C4 þ 12a3 eC2 ½13

Fourth Fifth where c0 and e are some nominal elastic constant and
n n strain (ignoring the tensor nature), T c is the critical
temperature, and a1, a2, and a3 are constants. The
s s
effective elastic constant is ð@ 2 F=@e2 Þ. Minimizing
Figure 3 An illustration of the basic sound modes in superfluid the free energy with respect to the order parameter
helium. The ‘‘n’’ and ‘‘s’’ lines indicate the motion of the normal
gives the result that the effective elastic constant is c0
fluid and superfluid components, respectively. The dots indicate
that the normal fluid has been clamped by its viscosity. The when T4T c, and is (c0  a23/2a2) when ToT c. Thus
dashed lines indicate motion under ‘‘pressure release’’ conditions. an experimentally measured jump in an elastic con-
stant at a second-order phase transition gives access
time, as during the curing of epoxy, the hardening of to the parameters in a Landau expansion.
cement, the progress of chemical reactions, etc. Other An extensive application of RUS has been in geo-
special condensed matter systems which benefit from physics, where the measurement of the thermody-
basic acoustic measurements are quantum solids, su- namic properties and anharmonic effects of materials
perfluids, quasicrystals, granular media, foams, at high temperatures (exceeding twice the Debye
rocks, etc. High amplitude acoustic fields are used temperature) are a high priority. Elastic data can
as driving mechanisms in sonochemistry, cavitation check theoretical models and their extension to
studies (including cavitation in quantum fluids), high temperature and pressure, where some asymp-
acoustic levitation (facilitating noncontact sample totic behavior may be convenient for other geophys-
manipulation), and sonoluminescence, in which a ical calculations and for extrapolations to even
collapsing bubble creates such extreme conditions higher temperatures. Anharmonic effects are evident
that light may be emitted. Acoustic emission, where in the Gruneisen relation and in the departure of
signals are generated by the system itself, is impor- heat capacity from the Law of Dulong and Petit,
tant in studies of cracks, fracture, earthquakes, etc. clat ¼ 3ðn=rÞkB .
A particular application of acoustic measurements
Acoustic Dissipation
in solid state physics is the determination of the con-
tribution of lattice acoustic modes to the specific Acoustic dissipation may be discussed with two basic
heat, clat. This is given by pictures. In one picture, energy in an ordered form in
   3 Z YD =T 4 x the acoustic field is lost to some disordered form, so
n T x e dx that the amplitude of a sound wave decreases with
clat ¼9 kB ½11
r YD 0 ðex  1Þ2 distance x as e  ax, where a is the attenuation coef-
ficient. From the functional form e  ax, the attenu-
where n is the number of lattice sites per unit volume, ation coefficient a may be considered as an imaginary
kB is Boltzmann’s constant, and YD is the Debye tem- part of a complex wave vector, k̃ ¼ k þ ia. Acoustic
perature. It is the Debye temperature which may be dissipation in this picture is typically modeled with a
expressed in terms of acoustic parameters, as follows: coupling term in a Hamiltonian between phonons
and some other system into which energy is lost
" Z #1=3 (electrons, magnetic spins, etc.).
_ 1 X dO 1
YD ¼ ½12 In the second picture, one notes that changes in
kB 18p2 n s 4p vs ðk̂Þ3 stress (or strain) are not immediately followed by
changes in strain (or stress), and there is a time lag,
Here the sum in s is over the three different acoustic with a characteristic relaxation time, between energy
modes in a particular direction k̂, which have sound being stored in kinetic and potential forms. If there
speeds vs ðk̂Þ. For isotropic solids, the average of 1=v3s were no time lag, then acoustic variations would os-
is ð1=v3l þ 2=v3t Þ. cillate back and forth along a single path (typically an
As mentioned in the introduction, an important isentrope) in the stress–strain plane. If there is a time
application of acoustics is in the study of phase tran- lag, then the path opens up into a loop, and acoustic
sitions. Acoustic studies are typically used in con- energy would be dissipated as ‘‘lost work’’ equal to the
junction with a model of the transition, for example, nonzero area of the loop. The time lag may be rep-
a Landau expansion in the case of second-order resented with a time dependence f ðtÞ ¼ ð1  et=t Þ,
transitions. In this case the free energy F, including where t is the characteristic relaxation time from a
Acoustics: Physical Principles and Applications to Condensed Matter Physics 7

sudden change. The time lag is incorporated in the and Soft Modes; Lattice Dynamics: Vibrational Modes;
equations of motion with additional time derivatives; Mechanical Properties: Anelasticity; Mechanical Proper-
however, the equations may be analyzed by Fourier ties: Elastic Behavior; Specific Heat; Thin Films, Mechan-
transforming in time, and the result involves the ical Behavior of.
Fourier transform of f(t), proportional to f̂ðoÞ given by
PACS: 43.20.  f; 43.25.  x; 43.35.  c; 43.58. þ z;
1 ot
f̂ðoÞ ¼ þi ½14 62.65. þ k; 62.20.Dc; 62.80. þ f; 74.25.Ld
1 þ ðotÞ 1 þ ðotÞ2

The real part gives rise to dispersion in sound

propagation, and the imaginary part results in atten- Further Reading
uation; the effective attenuation coefficient is
Ahlers G (1980) Critical phenomena at low temperature. Reviews
a ¼ amax 2ot=ð1 þ o2 t2 Þ, where the attenuation peak of Modern Physics 52: 489–503.
amax occurs when ot ¼ 1. American Association of Physics Teachers. Superconductivity, Se-
For fluids, the ‘‘time lag’’ derivatives are already in lected Reprints, published by the American Physical Society.
place in eqns [8]–[10]. If the terms containing the Ashcroft NW and Mermin ND (1976) Solid State Physics. Toron-
to: Thomas Learning.
transport coefficients Z, x, and k are not dropped,
Banys J, Macutkevic J, Samulionis A, and Vysochanskii Y (2004)
and the equations are solved to first order in these Dielectric and ultrasonic investigation of phase transitions in
coefficients, then solutions will involve a complex CuInP2S6 crystals. Phase Transitions 77: 345–358.
wave vector k̃ ¼ k þ ia, with the attenuation coeffi- Bhatia AB (1967) Ultrasonic Absorption. New York: Dover.
cient a given by Goto T and Anderson OL (1988) Apparatus for measuring elastic
constants of single crystals by a resonance technique up to
   1825 K. Review of Scientific Instruments 59: 1405–1409.
o2 4 k
a¼ Zþx þ ðg  1Þ ½15 Ichitsubo T, Ogi H, Nishimura S, Seto T, and Inui H (2004)
2rv3 3 gcv Elastic stiffness and ultrasonic attenuation of superconductor
MgB2 at low temperatures. Physical Review B 66: 052514-1–
For solids, forms of acoustic dissipation which are 052514-4.
most readily analyzed with the first picture include Levy M (2000) Handbook of Elastic Properties of Solids, Liquids,
electron–phonon scattering and phonon–phonon and Gases. New York: Academic Press.
scattering. The picture for electron–phonon scatter- Love AEH (1927) Treatise on the Mathematical Theory of Elas-
ticity. Cambridge: Cambridge University Press.
ing is particularly important for studying Bardeen– Maynard JD (1996) Resonant ultrasound spectroscopy. Physics
Cooper–Schrieffer (BCS) superconductors; as more Today 49: 26–31.
electrons pair and enter the superconducting state, Migliori A and Sarrao JL (1997) Resonant Ultrasound Spectros-
the phonon interaction gets used up in the pairing copy. New York: Wiley.
interaction, and part of the electron–phonon scatter- Pandry JD, Dey R, and Bhatt BD (2003) Estimation of molecular
radius of liquids and liquid mixtures from sound velocity. Jour-
ing no longer contributes to acoustic attenuation. nal of Molecular Liquids 111: 67–71.
Thus, a measured drop in acoustic attenuation tracks Putterman SJ (1974) Superfluid Hydrodynamics. Amsterdam:
the number of paired electrons. North-Holland.
Forms of acoustic dissipation which are most read- Schrieffer JR (1964) Theory of Superconductivity. New York: W.A.
ily analyzed with the second picture include time lags Benjamin.
Truell R, Elbaum C, and Chick BB (1969) Ultrasonic Methods in
for energy transfer to magnetic spin systems, electric Solid State Physics. New York: Academic Press.
dipole systems, defect and impurity motion (mechan-
ical diffusion or viscosity), other forms of energy
storage (thermal diffusion), etc. Models based on the Nomenclature
second picture are particularly useful in analyzing
acoustic attenuation measurements in glasses, ‘‘two- B bulk modulus (Pa)
c elastic constant (Pa)
level’’ systems, impurity doped systems, etc.
clat lattice specific heat (J kg  1 K  1)
The list of mechanisms related to acoustic attenu-
cp specific heat at constant pressure
ation just presented is by no means exhaustive. In- (J kg  1 K  1)
deed, the broad range of interacting mechanisms cv specific heat at constant volume
make acoustic attenuation one of the most widespread (J kg  1 K  1)
applications of acoustics in condensed matter physics. F free energy per unit mass (J kg  1)
k wave vector (m  1)
kB Boltzmann’s constant (J K  1)
See also: Crystal Symmetry; Crystal Tensors: Applications; L Lagrangian (J)
Lattice Dynamics: Anharmonic Effects; Lattice Dynamics: n number of lattice sites per unit volume
Aperiodic Crystals; Lattice Dynamics: Structural Instability (m  3)
8 Allotropy and Polymorphism

S entropy per unit mass (J kg  1 K  1) e strain (dimensionless)

t time (s) Z shear viscosity (N s m  2)
T temperature (K) YD Debye temperature (K)
u fluid element velocity (m s  1) x bulk viscosity (N s m  2)
v sound speed (m s  1) k thermal conductivity (W m  1 K  1)
V volume (m3) r mass density (kg m  3)
x spatial coordinate (m) s stress (Pa)
a attenuation coefficient (m  1) t relaxation time (s)
b thermal expansion coefficient (K  1) c displacement (m)
g specific heat ratio (dimensionless) o angular frequency (s  1)

Allotropy and Polymorphism

D R Gaskell, Purdue University, West Lafayette, IN, molar Gibbs free energy which accompany phase
USA transformations or chemical reactions can be ob-
& 2005, Elsevier Ltd. All Rights Reserved. tained from the corresponding changes in H and
S as

The thermodynamic origins of allotropy and poly- DG ¼ DH  TDS

morphism are presented. Allotropy and polymorph-
ism are, respectively, the ability of elements and of The molar enthalpy of an element, relative to that of
chemical compounds to have stable existences in the state in which the atoms are at infinite distances
more than one crystal form. The allotropies of iron, from one another, is the thermal energy transferred
carbon, and sulfur, and the polymorphisms of silica to a thermostat when the atoms come together and
and zirconia are discussed. occupy the sites in a regularly arrayed crystal lattice.
The relative enthalpy is then a measure of the
bonding energy in the crystal.
A chemical element exhibits allotropy when it can The Allotropy of lron
have a stable existence in more than one crystal form.
Polymorphism is the same phenomenon exhibited by Iron has two allotropes: a face-centered cubic (f.c.c.)
a chemical compound. When a range of possible crystal form and a body-centered cubic (b.c.c.) crys-
states of existence is available to an element or com- tal form. In the former, the unit cell has atoms lo-
pound, the stable state is that which has the lowest cated at each of the eight corners, each one of which
molar Gibbs free energy at the constant values contributes one-eighth of an atom to the unit cell.
of pressure and temperature of interest. The molar Atoms located at the centers of each of the six faces
Gibbs free energy, G, in turn, is determined by the of the cell, each contribute one-half of an atom to the
molar enthalpy, H, the molar entropy, S, and the unit cell, to give a total of four atoms per unit cell.
temperature (in kelvins), T, as Alternatively, the f.c.c. crystal structure can be con-
sidered to consist of planes of close-packed atoms
G ¼ H  TS stacked in the sequence ABCABC. The b.c.c. unit cell
has atoms located at each of the eight corners of the
unit cell and one atom located at the center of the
Low values of G are obtained with low values of
cell, giving two atoms per unit cell.
H and high values of S. As only changes in enthalpy
The variations, with temperature at a constant
can be measured by the transfer of thermal energy
pressure of 1 atm, of the molar Gibbs free energies of
between a thermodynamic system and a thermosta-
the f.c.c. and b.c.c. allotropes of iron are shown
ting reservoir, the enthalpy H does not have a definite
schematically in Figure 1a. In Figure 1a
value and thus, also, G does not. In contrast, S,
which is a measure of the thermal and configura-  
tional disorder in a thermodynamic system, does @G
¼ S
have a definite value. Consequently, changes in the @T P
Allotropy and Polymorphism 9

The shapes of the curves in Figure 1a are deter-

mined by the constant pressure molar heat capacities
b.c.c. Fe and the molar entropy of the two crystal forms. In
the temperature range 1100–1700 K, the value of cp
G for b.c.c. Fe is 7.9 J K–1, larger than that for f.c.c. Fe
f.c.c. Fe and the highly ordered f.c.c. crystal structure has a
lower molar entropy than that of the b.c.c. structure.
Thus, in Figure 1, since the rate of decrease of G with
increasing temperature for the b.c.c. Fe is greater
than that for the f.c.c. Fe and the second derivative of
1100 1200 1300 1400 1500 1600 1700
the line for b.c.c. Fe is greater than that of the f.c.c.
(a) Temperature (K) line, the lines intersect twice at 1187 and 1664 K.
−4 Thus, the b.c.c. form is stable at temperatures lower
than 1187 K and at temperatures from 1664 K to the
melting temperature of 1809 K. The f.c.c. form is
stable in the range of temperature 1187–1664 K. The
log p 0 (atm)

−7 low-temperature b.c.c. form is referred to as a-Fe, the

b.c.c. Fe
−8 f.c.c. form is referred to as g-Fe and the high-tem-
−9 f.c.c. Fe perature b.c.c. form is referred to as d-Fe. The
− 10
original assignment of b-Fe to the low-temperature
b.c.c. form, between the Curie temperature of 1033
− 11
and 1187 K, is no longer used.
− 12 Figure 1b shows the variations, with temperature,
1100 1200 1300 1400 1500 1600 1700
(b) Temperature (K)
of the saturated vapor pressures exerted by b.c.c. Fe
and f.c.c. Fe in the range of temperature 1100–
3 ×10 4 1700 K. The saturated vapor pressures are related to
the difference between the Gibbs free energies of the
b.c.c. Fe
two crystal forms by
H ( J mol −1)

2 × 10 4

f.c.c. Fe p0g
DGða-gÞ ¼ RT ln
1 × 104 p0a

0 A negative value of DGða-gÞ makes the p0g value

1100 1200 1300 1400 1500 1600 1700 lower than the p0a value.
(c) Temperature (K) As has been stated, the constant pressure molar
Figure 1 (a) The variations, with temperature, of the molar heat capacity of b.c.c. Fe is 7.9 J K–1 larger than that
Gibbs free energies of b.c.c. Fe and f.c.c. Fe (schematic). (b) The of f.c.c. Fe. Consequently, from eqn [1], the rate of
variations, with temperature, of the saturated vapor pressures of increase of the molar enthalpy of b.c.c. Fe is greater
b.c.c. Fe and f.c.c. Fe. (c) The variations, with temperature, of the than that of f.c.c. Fe. The variations, with temper-
relative molar enthalpies of b.c.c. Fe and f.c.c. Fe.
ature, of the relative molar enthalpies of the two
crystal forms are shown in Figure 1c, in which the
reference state is chosen as b.c.c. Fe at 1100 K. Inc-
  reasing the temperature from 1100 to 1187 K causes
@2G cp
¼ the molar enthalpy to increase along the b.c.c. Fe
@T 2 P T line. At 1187 K, the b.c.c. form transforms to the
f.c.c. form with the required increase in enthalpy
where cp is the constant pressure molar heat capacity (the latent heat of the transformation). Further hea-
of the phase. This quantity is formally defined as ting causes the enthalpy to increase along the f.c.c.
  Fe line (which intersects with the metastable b.c.c.
@H Fe line at 1397 K) and, at 1664 K, the f.c.c. Fe trans-
cp ¼ ½1
@T P forms back to the b.c.c. Fe, again with the required
increase in enthalpy.
that is, it is the ratio of the heat absorbed by a mole The combination of the allotropy of iron and
of substance, at constant pressure, to the consequent Henry Bessemer’s patent no. 356, dated 12 February
increase in temperature. 1856 ‘‘On the Manufacture of Malleable Iron and
10 Allotropy and Polymorphism

Steel without Fuel,’’ gave rise to the Industrial plain carbon steels to be manipulated by a choice of
Revolution. Carbon, which is the alloying element carbon content and heat treatment. Thus, steels can
in plain carbon steel, occupies interstitial sites in the be produced, which, at one extreme, are soft enough
f.c.c. Fe and b.c.c. Fe lattices. The solution of carbon to be plastically deformed to the shape of a paper
in f.c.c. Fe produces a phase called austenite, and its clip, or, at the other extreme, hard and tough enough
solution in b.c.c. Fe produces a low-temperature to be used as a bearing material.
phase called a-ferrite and a high-temperature phase Nickel, which has the f.c.c. crystal structure,
called d-ferrite. As the interstitial sites in f.c.c. Fe are forms substitutional solid solutions with iron and
larger than those in b.c.c. Fe, the solubility of carbon hence stabilizes the f.c.c. structure. Under equilibri-
in austenite is larger than in a-ferrite and d-ferrite. um conditions, a g-Fe,Ni containing 53 wt.% Ni
Thus, carbon is an ‘‘austenite stabilizer.’’ Figure 2, undergoes a eutectoid decomposition to a-Fe and
which is the phase diagram for the system Fe-Fe3C, FeNi3 at 3451C. However, as nucleation and growth
shows the extent of the austenite phase field which of a-Fe from g-Fe,Ni requires significant diffusion by
terminates at the eutectoid point (0.78 wt.% C, migration of vacant lattice sites, the presence of a
7231C). When the temperature of austenite of this few percent Ni in solid solution in the g-Fe,Ni
composition is decreased to a value lower than produces a metastable g-Fe,Ni phase at room tem-
7231C, the austenite undergoes a eutectoid decom- perature. In contrast, chromium, which has the
position to produce a structure consisting of alter- b.c.c. structure, also forms substitutional solid solu-
nating layers of ferrite and cementite (the metastable tions with iron and thus stabilizes the b.c.c. struc-
iron carbide, Fe3C). This structure is called pearlite ture. This causes the formation of a ‘‘g-loop’’ shown
and the fineness of the structure is determined by the in Figure 3. At the minimum temperature of the
rate of cooling of the eutectoid austenite through the g-loop (7 wt.% Cr, 8311C), the g-Fe and a-Fe,Cr
eutectoid temperature. The coarseness of the micro- phases have the same composition. In Fe–Cr solid
structure increases with decreasing cooling rate. solutions containing less than 12 wt.% Cr, the
With increasing rate of cooling a limit is reached, product of oxidation is the highly defective spinel
beyond which nucleation of the ferrite and the ce- FeO  Cr2O3. With Cr contents greater than 12 wt.%,
mentite from the austenite is inhibited. The austenite the oxide Cr2O3 is the product of oxidation. Thus, a
transforms to a metastable body-centered tetragonal ‘‘stainless’’ steel can be produced by having sufficient
phase by means of a diffusionless shear mechanism. Ni in solid solution in Fe to give a single-phased f.c.c.
This very hard and brittle structure is called mar- structure at room temperature and by having suffi-
tensite and, in it, the shear stresses which cause the cient Cr in solid solution to produce a protective
brittleness are relieved by tempering at some tem-
perature less than 7231C. The occurrence of allot-
ropy in iron allows the mechanical properties of 1600

1600 1400
1536° 1493°
1400 1391°
Temperature (°C)

Temperature (°C)

( Fe)
1147° 12.7
Austenite 2.03 4.30 1000

1000 914°C
800 7
0.02 0.78
Ferrite + Cementite (Fe3C)
600 600
0 1.0 2.0 3.0 4.0 5.0 0 5 10 15 20
Weight percent carbon Weight percent chromium
Figure 2 The phase diagram for the system Fe–Fe3C. Figure 3 The iron-rich end of the Fe–Cr phase diagram.
Allotropy and Polymorphism 11

coating of Cr2O3 on the surface of the alloy. Such The crystal structure of graphite is determined by
an alloy is type 304 stainless steel (or 18-8 stain- sp2 hybridization in which a 2s electron is promoted
less steel) which is Fe containing 18 wt.% Cr and to the 2pz orbital and the remaining 2s and the 2px
8 wt.% Ni. and 2py orbitals form a trigonal hybrid containing
three energetically equivalent orbitals which lie in a
plane, forming angles of 1201 with their neighboring
The Allotropy of Carbon orbitals. The crystal structure of graphite, which is
Carbon (element number 6), located at the top of shown in Figure 6, consists of sheets of s-bonded
group four in the periodic chart of the elements, has atoms in the xy-plane, arranged in hexagons which
the electron configuration 2s12s22p2, which would are bonded to one another in the z-direction by
indicate that it has a valence of 2. However, hybrid- p-bonds formed by overlap of the pz orbitals. The
ization of the electron orbits of the carbon atom gives lengths of the s-bonds and the p-bonds are, re-
rise to the occurrence of the familiar allotropes, dia- spectively, 14.2 and 34 nm. As the p-bonds are much
mond, and graphite. In sp3 hybridization, one of the weaker than the s-bonds, this property facilitates
2s electrons is promoted to the 2p orbital and the easy shearing of the sheets of hexagons and makes
four electrons in the second shell undergo hybridiza- graphite a good lubricating material.
tion to produce four energetically equivalent elec-
trons. The 272 kJ mol–1 required to promote the
electron from the 2s to the 2p shell is more than
compensated for by the decrease in energy which oc-
curs upon hybridization. The mutual repulsion be-
tween the hybrid orbitals causes them to point
toward the corners of a tetrahedron, as shown in
Figure 4. The unit cell for the diamond cubic crystal
structure, which is shown in Figure 5, can be
regarded as being based on the f.c.c. unit cell con-
taining eight subcubes. Carbon atoms occupy the
upper-back-left, upper-front-right subcubes, the low-
er-front-left and lower-back-right subcubes with the
bond length being 15.4 nm.The rigidity of the bonds
between the atoms makes diamond the hardest ma-
terial in existence (number 10 on the Moh scale of
hardness). It has a high melting temperature, 38001C,
and a large latent heat of melting, 105 kJ mol–1 at a
pressure of 48 kbar. Also, it has a high index of Figure 5 The unit cell for the diamond cubic crystal structure.
refraction, 2.417, making it a desirable gem stone. In
1913, the determination of the structure of diamond
was one of the early successes of X-ray analysis.


Figure 6 The crystal structure of graphite. The lines join atoms

Figure 4 Lines directed to the corners of a tetrahedron. in successive sheets which are aligned vertically.
12 Allotropy and Polymorphism

10 4
Solid II
1000 Liquid

10 3
10 2

Pressure (atm)

10 10

Pressure (atm)
1 Graphite

10 −1
10 −2
0 1000 2000 3000 4000 5000 10 −3

Temperature (K)

10 −4

Figure 7 The phase diagram for carbon.

The phase diagram for carbon, presented as Figure 10 −5 Vapor

7, shows that a third allotrope, solid II, exists at

pressures in excess of 1000 atm, and that diamond is 10 −6
metastable at low pressures. Diamonds are formed 80 100 120 140 160
Temperature (°C)
under conditions of high pressure experienced in clay
pipes in the earth’s crust. It exists in the metastable Figure 8 The phase diagram for sulfur.
state at room temperature and pressure because of
the extensive rearrangement of atoms required for
the transformation from the diamond structure produced. Amorphous or ‘‘plastic’’ sulfur can be ob-
to the graphite structure. At lower pressures, the tained by fast cooling of the crystalline form and this
melting temperature of graphite increases with inc- amorphous phase also has a helical structure with
reasing pressure, and at higher pressure it decreases eight atoms per spiral. Sulfur is soluble in carbon di-
with increasing temperature. Thus, at lower pres- sulfide and is insoluble in water.
sures, the molar volume of graphite is lower than The complexity of rhombic sulfur allows easy
that of liquid carbon and, at higher pressures, the supercooling of the monoclinic form and has been
reverse is the case. At the ‘‘nose’’ of the melting used as a means of providing experimental evi-
curve, the molar volumes of graphite and liquid have dence to substantiate Nernst’s heat theorem, also
the same value. The densities of diamond and grap- known as the third law of thermodynamics. This
hite at 298 K and 1 atm pressure are, respectively, states that, at 0 K, the entropy of any homogeneous
3.515 and 2.2 g cm–3. substance, which is in complete internal equilibrium,
can be taken as being zero. Summation of the in-
crease in molar entropy of rhombic sulfur when
heated from 0 K to the temperature of transforma-
The Allotropy of Sulfur
tion ð38:86 J K1 Þ, the molar entropy of transfor-
The phase diagram for sulfur, presented as Figure 8, mation of rhombic to monoclinic sulfur at this
shows that sulfur can exist in a rhombic crystal struc- temperature ð1:09 J K1 Þ, and the change in the
ture and in a monoclinic crystal structure. The unit molar entropy of monoclinic sulfur when cooled
cell of the rhombic form contains 128 atoms, existing from the transformation temperature to 0 K ð37:8
as 16 puckered rings of eight atoms which form S8 J K1 Þ gives the molar transformation of rhombic to
molecules. Sulfur melts to form a translucent liquid, monoclinic sulfur at 0 K as 0.15 J K–1. This is less
which, when heated above 2001C, transforms to a than the experimental error involved and, thus, is
red, highly viscous form produced by the breaking taken as an experimental verification of Nernst’s heat
of the rings and the entanglement of the chains theorem.
Allotropy and Polymorphism 13

The Polymorphism of Silica L Coes, Jr. in 1953) and at pressures higher than
105 atm, stishovite is the polymorph. Stishovite dif-
Silicon, occurring below carbon in group 4 in the
fers from the other polymorphs, in that the silicon is
periodic chart of the elements, undergoes sp3 hy-
in octahedral coordination with oxygen. This poly-
bridization of the 3s and 3p electrons to produce
morph was discovered in meteorites found in Russia
energetically equivalent electron orbitals, which
by Stishov. It is believed that the high force exerted
point to the corners of a tetrahedron. In the crystal
on the meteorite by contact with the surface of the
structure of silica (SiO2), the sp3 orbitals of earth caused the tetrahedral coordination in silica to
neighboring silicon atoms form bonds with oxygen
transform to an octahedral coordination.
atoms, such that each silicon atom is tetrahedrally
The differences in standard free energies of for-
coordinated by four oxygen atoms and each oxygen
mation, DðDG Þ, among the various polymorphs of
atom is bonded to two silicons. The basic building
silica, using cristobalite as the reference state, are
block in the silica structure is thus a tetrahedron, at
shown in Figure 11. The relatively small differences
the center of which is a silicon atom with oxygen
in the standard free energies among the polymorphs
atoms located at the four corners. The polymorphism
arise because the enthalpies of formation (the –Si–O–
of silica arises from the number of ways in which Si– bond energies) of the polymorphs and the con-
tetrahedra can be arranged to fill space and one such
figurational entropies of packing SiO4 tetrahedra
arrangement is shown in Figure 9. The phase
together are similar.
diagram for silica, presented as Figure 10, shows
Figure 11 shows another representation of the po-
that, at pressures less than 104 atm, the polymorphs
lymorphism of silica. The horizontal arrows represent
are low quartz, high quartz, tridymite, and cristoba-
reconstructive transformations, which require bond
lite. At pressures between 104 and 105 atm, silica
breaking and complete rearrangement of the SiO4
exists as the polymorph coesite (first described by
tetrahedra. Thus, although the differences in the
standard free energies of formation of the poly-
morphs, shown in Figure 11, are small, the high
energies of activation for these transformations are
such that the high-temperature polymorphs can be
undercooled easily. Used extensively in laboratory
and medical applications, silica glass is obtained by
supercooling liquid silica to form a metastable amor-
phous phase. The vertical arrows in Figure 12 rep-
resent displacive transformations, which do not
involve the breaking of bonds and, consequently, oc-
cur at relatively high rates. The sequence of transfor-
mations during gradual heating of b-quartz depends
on the purity of the quartz. When high-purity quartz

10 5

Pressure (atm)

10 3
High quartz Cristobalite

10 2

500 1000 2000
Temperature (°C)

Figure 9 The arrangement of SiO4 tetrahedra. Figure 10 The phase diagram for silica.
14 Allotropy and Polymorphism


500 163°C
867°C 1470°C 1723°C
∆(∆G °) (J)

−500 573°C



0 200 400 600 800 1000 1200 1400 1600 1800
Temperature (°C)

Figure 11 The differences in standard free energies of formation, DðDG  Þ, among the various polymorphs of silica, using cristobalite
as the reference state.

867°C 1470°C 1723°C

-Quartz -Tridymite -Cristobalite

573°C 163°C 200-270°C

-Tridymite -Cristobalite



Figure 12 Phase transformations in silica.

(o10–2 wt.% impurities) is heated, the b-quartz phase and places the temperature of equilibrium be-
transforms rapidly into a-quartz at 5731C. The a- tween quartz and cristobalite at B10251C. The other
quartz is stable to B10251C, at which temperature it suggests that a sharp polymorphic transformation
transforms into a-cristobalite. However, if the quartz exists between stable tridymite (designated ‘‘tridy-
contains impurities in solid solution, a-quartz trans- mite-S’’) and stable cristobalite at 14701C and that
forms to a-tridymite at B8701C, which, on further quartz and stable tridymite coexist at B8701C. The
heating, transforms into a-cristobalite at 14701C. latter opinion is in accordance with the ‘‘classical’’
The observation of the influence of impurities picture shown as Figure 10.
and the fact that most of the laboratory studies of The precious mineral opal, which has the for-
phase equilibria in silica were conducted under hy- mula SiO2  nH2O, is a form of silica that is wholly
drothermal conditions cast doubt on the validity of amorphous.
the phase relations shown in Figure 10. Evidence,
which suggests that tridymite is not a stable phase in
pure silica, but owes its existence to the presence of
The Polymorphism of Zirconia
impurity ions in the structure, has been provided. Zirconia, ZrO2, has three polymorphs at atmospheric
This would make tridymite a polytype (same struc- pressure: a high-temperature cubic structure which
ture, different composition) rather than a poly- exists from 23701C to the melting temperature of
morph. Two opinions have been expressed on this 26801C, a low-temperature monoclinic structure
topic. One suggests the elimination of the tridymite which exists at temperatures lower than 11741C,
Allotropy and Polymorphism 15

and an intermediate tetragonal structure which is sta- coordinated by seven oxygen ions. This structure is
ble between 11741C and 23701C. The high-temper- shown in Figure 14. The densities and specific
ature cubic form has the fluorite, CaF2, structure volumes of the three polymorphs are listed in Table 1.
shown in Figure 13. In this structure, the Zr4 þ cat- The 4.6% decrease in density accompanying the
ions occupy the corner and face-centered positions in tetragonal to monoclinic transformation cannot be
the f.c.c. lattice and the O2 anions occur at the cen- sustained and the solid exfoliates by means of a
ters of the eight subcells described in Figure 5. The martensite shearing mechanism. The temperature at
unit cell, thus, contains four Zr4þ and eight O2 ions. which this transformation occurs on cooling decreas-
The intermediate tetragonal form is a slightly de- es with decreasing particle size. However, the high-
formed fluorite structure, and the low-temperature temperature cubic form can be stabilized to room
monoclinic form, which occurs naturally as the min- temperature by the substitution, for Zr4 þ , of appro-
eral baddeleyite, has a structure in which Zr ion is priately sized cations of valence less than 4. The
cations Ca2 þ and Y3 þ are used to produce lime-
stabilized zirconia (LSZ) and yttria-stabilized zir-
conia (YSZ). LSZ is considerably less expensive than
YSZ. Several versions of the phase diagram for the
system ZrO2–CaO have been proposed and two of
them are shown in Figures 15 and 16.
Figure 15 shows the existence of the three poly-
morphs of ZrO2 and contains the line compound
CaO  4ZrO2, which undergoes incongruent decom-
position at 13101C. Figure 15 also shows that the

Table 1 The densities and specific volumes of the polymorphs

of zirconia

Polymorph r ðkg m3 Þ V ðm3 kg1 Þ

Cubic 6090 0.001 63

Tetragonal 6100 0.001 64
Figure 13 The unit cell for the zirconia cubic (fluorite) crystal Monoclinic 5830 0.001 72
structure. The solid and white circles represent Zr and O,



2000 Cubic
Temperature (°C)


500 CaO. 4ZrO2

CaO . ZrO2
0 10 20 30 40 50
Figure 14 The crystal structure of monoclinic zirconia (bad- ZrO2
CaO (mol.%)
deleyite). The solid and white circles represent Zr and O,
respectively. Figure 15 One version of the phase diagram for ZrO2–CaO.
16 Allotropy and Polymorphism

Fraction of vacant oxygen sites

0.05 0.10 0.15



Tetragonal ZrO2
Temperature (°C)
1500 Cubic


Monoclinic ZrO2


0 10 20 30 40 50
ZrO2 CaO (mol.%) CaZrO3

Figure 16 Another version of the phase diagram for the system ZrO2–CaO.

substitution of Ca2 þ ions for Zr4 þ in cubic zirconia 3000

stabilizes the cubic structure only to 11401C, at which Liquid
temperature it undergoes a eutectoid decomposition
to tetragonal zirconia and CaO  4ZrO2. It is known 2500
that the substitution of Ca for Zr stabilizes the cubic
structure to room temperature, but it is unlikely that
supercooled cubic zirconia will maintain a metasta- Cubic
ble existence at temperatures near 7001C, at which it
Temperature (°C)

is used as an oxygen sensing device. On the other

hand, Figure 16 shows the existence of lime-stabi- 1500
lized-cubic zirconia at room temperature, but does Tet
not show the existence of cubic zirconia in the pure
state. Figure 17 shows the phase diagram for the 1000
system ZrO2–YO1.5, in which it is seen that the sub-
stitution of Zr3þ ions for Zr4þ ions stabilizes the cu-
bic phase at room temperature. m
Stabilized cubic zirconia is a practical ceramic ma-
terial used as the solid-state electrolyte in a cell used
for measuring the partial pressure (or thermodynam-
ic activity) of oxygen in a gaseous or liquid medium. 0 5 10 15 20
Electroneutrality requires that the substitution of a ZrO2 YO1.5 (mol. %)
Ca2þ ion for a Zr4þ ion in cubic zirconia be accom-
Figure 17 The phase diagram for the system ZrO2–YO1.5.
panied by the formation of a vacant site of the
oxygen anion sublattice. Hence, x moles of CaO þ
(l  x) moles of ZrO2 contain ðx þ 2  2xÞ ¼ ð2  xÞ
moles of O2– and x moles of vacant sites. Thus, the 12–13%. This high percentage of vacant sites
fraction of vacant sites is x=ð2  x þ xÞ ¼ 0:5x. imparts high diffusivity of oxygen in cubic zirconia
Figure 16 shows that the percent of vacant oxygen at elevated temperatures. The working of a lime-
sites in stabilized cubic zirconia can be as high as stabilized cubic zirconia EMF cell is illustrated in
Allotropy and Polymorphism 17

Stabilized cubic zirconia is also used as an elec-

CaO–ZrO2 electrolyte
trolyte in a hydrogen generator, in which water vapor
Pt Pt is decomposed to produce hydrogen gas at the cath-
cathode O2− anode ode and oxygen gas at the anode. Also, it is used in a
Higher fuel cell in which electric power is produced by the
oxygen oxygen oxidation of CO or H2 at the anode and the reduc-
pressure Vacancies pressure tion of oxygen gas at the cathode.
EMF See also: Alloys: Iron; Alloys: Overview; Ceramic Mate-
rials; Electronic Structure (Theory): Molecules; Irrever-
sible Thermodynamics and Basic Transport Theory in
Solids; Molecular Crystallography; Phase Transformation;
Figure 18 The working of an LSZ solid-state oxygen sensor. Phases and Phase Equilibrium; Thermodynamic Proper-
ties, General.

Figure 18. The half-cell reaction at the cathode is

PACS: 82.60.Fa; 81.30.Bx; 81.30.Dz; 61.50.Ks;
þ 2e -O2 81.05.Je

and the half-cell reaction at the anode is

Further Reading
O2 -2e þ 12O2ðgasÞ
Bragg L and Claringbull GF (1965) The Crystalline State, Vol IV:
Crystal Structures of Minerals. Cornell, NY: Cornell University
This gives the cell reaction as O2 ðP ¼ Phigh ; TÞ¼ Press.
O2 ðP ¼ Plow ; TÞ for which Levin EM, Robbins CR, and McMurdie HF (1964, 1969) Phase
diagrams for ceramists. Amer. Ceramic Society (suppl.). Metals
Plow Park, OH: American Society for Materials.
DG ¼ RT ln Massalski TB (ed.) (1986) Binary Alloy Phase Diagrams, vols. 1
Phigh and 2. Metals Park, OH: ASM.
and as Weast RC and Astle MJ (eds.) (1996) CRC Handbook of Chem-
istry and Physics, 77th edn. Boca Raton, FL: CRC Press.
DG ¼ zFE ¼ RT ln
cp constant pressure molar heat capacity
RT Po2ðlowÞ (J K–1)
E¼ ln E electromotive force (V)
4F po2ðhighÞ F Faraday’s constant ¼ 96 487 C mol–1
G molar Gibbs free energy (J)
where E is the measured EMF of the cell, F is H molar enthalpy (J)
Faraday’s constant, and z is the number of electrons P pressure (atm or Pa)
transferred by the electrochemical cell reaction. In the p0 saturated vapor pressure (atm or Pa)
use of the oxygen sensor where the value of one of the R universal gas constant (8.3144 J K–1
oxygen pressures is known, measurement of the EMF
S molar entropy (J K–1)
of the cell and the temperature allows the unknown T temperature (K)
oxygen pressure to be determined. The oxygen V specific volume (m3 kg–1)
content of liquid steel is measured routinely by im- z number of electrons transferred in an
mersion of a disposable LSZ cell in the steel bath in electrochemical reaction
the converter. r specific density (kg m–3)

Alloys See Alloys: Aluminum; Alloys: Copper; Alloys: Iron; Alloys: Magnesium; Alloys: Overview; Alloys:
18 Alloys: Aluminum

Alloys: Aluminum
E A Starke Jr., University of Virginia, Charlottesville, aluminum has a low stiffness, E, of 70 GPa compared
VA, USA to 211 GPa for iron, and low tensile strength,
& 2005, Elsevier Ltd. All Rights Reserved. 80 MPa, compared to 300 MPa for iron. However,
its specific modulus, that is, modulus divided by
density, is almost equal to that for iron, titanium, and
magnesium. An advantageous chemical property of
aluminum is its reactivity with oxygen, which leads
Aluminum is the third most abundant element in the to the formation of a dense layer of Al2O3 on the
earth’s crust but because it is so reactive with other surface, which shields the base metal from further
elements it is not found in the native state. Hans environmental interactions.
Christian Oersted (1777–1851), a Danish physicist Pure aluminum is not used commercially because
and chemist, was successful in isolating aluminum in of its low strength, but this property can be improved
a pure form in 1835. Sir Humphrey Davy (1778– by alloying elements. A major metallurgical
1829) had previously been unsuccessful at such at- development occurred in 1906 when Alfred Wilm
tempts, but it was Davy who named the element (1869–1937) discovered the process of ‘‘age harden-
‘‘aluminum,’’ the name used in the US. The rest of the ing’’ in aluminum alloys. Wilm was conducting re-
world uses the term ‘‘aluminium.’’ Until 1886 a search directed toward improving the strength of
chemical process was used to produce aluminum, aluminum alloys. He knew that steel could be
which employed crystals of corundum that were strengthened if the right compositions were cooled
chemically converted to aluminum chloride and then fast enough from high temperatures, so following
reduced with metallic sodium to form salt and me- this recipe he heated some alloys of aluminum con-
tallic aluminum. From 1825 to 1886, aluminum was taining 3.5–5.5 wt.% copper, plus less than 1%
primarily used for jewelry and as expensive table- magnesium and manganese, to a high temperature
ware due to the difficulty and cost of extracting it and quenched them in water. To his frustration,
from its ore. During the construction of the many of the alloys he tested were softer after quench-
Washington Monument, the world’s tallest structure ing than before the heat treatment. However, after a
of that period, a material was needed to top off the few days he found that their hardness and tensile
structure and to serve as a lightning rod. A cast properties had increased considerably. One of the
aluminum pyramid was produced by William alloys, designated Duralumin, is still in use today.
Frishmuth in 1884 and mounted on the top. It was Much later, in 1919, Merica, Waltenberg, and Scott
the largest aluminum casting ever produced and was explained the phenomenon, which is due to nano-
the first architectural application of this metal. sized clustering and precipitation of solute atoms
In 1886, Charles Martin Hall (1863–1914) in the from the supersaturated solid solution. However,
US and Paul-Louis Toussaint Héroult (1863–1914) in these nano-microstructural features are too small to
France simultaneously developed an economical be resolved by optical microscopy and were only in-
electrochemical method of producing aluminum, ferred by the X-ray diffraction studies of Guinier and
which ultimately led to its widespread use through- Preston in 1938. Direct proof was not obtained until
out the world. Their invention replaced the chemical the development of transmission electron microscopy
reduction process and lowered the metal’s cost from in 1959. Duralumin is probably the first example of a
$15/pound in 1884 to $0.50/pound in 1890. The nanostructured material developed by humans.
versatility of aluminum has resulted in it replacing Aluminum alloys are classified as heat-treatable or
many older, more established materials, and it is now non-heat-treatable, depending on whether or not
consumed, on a volumetric basis, more than all other they undergo precipitation (age) hardening. There
nonferrous metals combined, including copper, lead, are also two different product forms of aluminum
and zinc. alloys; wrought alloys that have been worked or de-
Aluminum is light, ductile, has good electrical and formed after casting and casting alloys, which are
thermal conductivity, and can be made strong by al- used in the ‘‘as-cast’’ condition. Wrought aluminum
loying. It has a low density of 2.7 g cm  3 compared alloys are normally designated by a four-digit nu-
to that of iron (7.9 g cm  3) due to its low atomic merical system developed by the Aluminum Associ-
mass of 27. The ductility and formability of alumi- ation. The nomenclature has now been accepted by
num is due to the high symmetry and thermodyic most countries and is called the International Alloy
stability of the face-centered cubic (f.c.c.) lattice Designation System (IADS). The system used for
and its high stacking-fault energy. In the pure form, wrought alloys is slightly different from that used for
Alloys: Aluminum 19

cast alloys; however, the first digit designates the al- normally added: those for strength and those to
loy group and is essentially the same for both control the grain structures that precipitate as ‘‘di-
wrought and cast alloys. The alloy group is associ- spersoids.’’ During solidification of the ingot, some of
ated with the major alloy addition and the second the alloying elements and impurities may precipitate
digit indicates modification of the original alloy or out of the aluminum, forming coarse ‘‘constituent’’
impurity limits. The last two digits identify the spe- particles within the ingot. For wrought products,
cific aluminum alloy. Experimental alloys also use additional precipitation of the alloying elements may
this system, but are indicated as experimental by the occur as the ingot is worked down to the final prod-
prefix X. The 1XXX alloys contain a minimum of uct form (such as sheet or plate).
99% aluminum. The major alloying addition in the The ‘‘constituent’’ particles potentially impair the
2XXX series is copper; it is manganese for 3XXX, properties of the final product in several ways. By
silicon for 4XXX, magnesium for 5XXX, magne- tying up alloying elements that are added on purpose
sium and silicon for 6XXX, zinc for 7XXX, and to develop the desired properties, these elements are
8XXX is used for others (tin for casting alloys). A not available to impart strength and other beneficial
first digit of 9 is not used for wrought alloys but is properties to the aluminum alloy. In addition, the
used for other alloy additions for casting alloys. constituent particles are more brittle than the sur-
The designation system is slightly different for rounding aluminum and, by fracturing under stress,
casting alloys although, as mentioned, the first digit form and promote the growth of cracks and hence
still refers to the major alloying element. The second impair beneficial mechanical properties. One typical
and third digits serve to identify a particular com- goal of aluminum alloy processing is to reduce the
position. These three digits are followed by a decimal size and amount of the constituent particles in the
point, which is followed by a zero to indicate a cas- aluminum alloy product.
ting. Often, a letter prefix is used to denote either an Beginning with the as-cast ingot, conventional
impurity level or the presence of a secondary alloying processing includes a homogenization treatment. This
element. These letters are assigned in alphabetical is the stage where dispersoids normally form. For al-
sequence starting with A but omitting I, O, Q, and X. loys used in the as-cast condition, homogenization
X is reserved for experimental alloys. For example, may be the final treatment for non-heat-treatable al-
A201.0 has a higher purity than the original 201.0. loys, but for age-hardenable alloys further heat treat-
The heat-treatment or temper-nomenclature sys- ments will be required. For wrought alloys, hot
tem developed by the Aluminum Association has also working follows the homogenization for ingot break-
been adopted as part of the IADS by most countries. down and shape change to the appropriate product
It is used for all forms of wrought and cast aluminum form. Since aluminum and its alloys have high stack-
alloys with the exception of ingot. The system is ing-fault energy, sufficient dynamic recovery normal-
based on the treatments used to develop the various ly occurs during hot deformation to give rise to a
tempers and takes the form of letters added as suf- stable polygonized substructure, an example of which
fixes to the alloy number. One or more digits fol- is shown in the transmission electron micrograph of
lowing the letter indicate subdivisions of the tempers, Figure 1. The grains elongate in the direction of metal
when they significantly influence the characteristics flow and do not recrystallize. An example of the grain
of the alloy. Alloys supplied in the as-fabricated or structure of a hot-worked aluminum-alloy plate is
annealed condition are designated with the suffixes F shown in Figure 2. The dispersoids, and any constit-
and O, respectively. The letter W designates those uents and primary phases that may be present, are
supplied in the solution heat-treated condition. strung out in the working direction; their spacing in-
Alloys supplied in the strain-hardened condition are creases in the working direction and decreases nor-
designated with the letter H and those in the heat- mal to the working direction. In some cases large
treated condition with the letter T. Digits following constituent phases, if present, are broken up during
H represent the degree of strain hardening and those the working operation. The fine distribution of di-
following T the type of aging treatment. spersoids associated with Cr, Mn, or Zr additions
delays or prevents static recrystallization and aids in
retaining the elongated or ‘‘pancake-shaped’’ grains
Processing of Aluminum Alloys during subsequent processing. After processing for
In fabricating aluminum alloy products, the alumi- shape change, non-heat-treatable alloys may then be
num alloy composition is made by adding the alloy- cold worked and/or annealed to develop the desired
ing elements to molten aluminum, usually in the form strength, and heat-treatable alloys may be solution-
of a concentrated hardener or master alloy, with the ized, quenched, in some cases stretched to remove
requisite purity. Two types of alloying elements are residual stresses developed during quenching or for
20 Alloys: Aluminum

Figure 3 Dark-field TEM micrograph of the matrix precipitation

Figure 1 Subgrain structure in a hot-rolled Al–Cu–Mg–Li–Zr from an Al–2.7Cu–1.8Li–0.6Zn–0.3Mg–0.3Mn–0.12Zr alloy nat-
alloy. urally aged for seven days at room temperature after a rapid
quench from the solution heat-treatment temperature. The matrix
precipitation is dominated by fine d0 (Al2Li) spheres.

Figure 2 Optical micrograph illustrating the pancake grain

structure of a hot-rolled Al–Cu–Mg–Li–Zr alloy.

creating nucleation sites for precipitation, and either

Figure 4 Bright-field TEM micrograph of the matrix precipita-
naturally aged at room temperature or artificially tion from an Al–2.7Cu–1.8Li–0.6Zn–0.3Mg–0.3Mn–0.12Zr alloy
aged at some moderate temperature. An Al–2.7Cu– (T6 temper) artificially aged at 1501C for 6 h after a rapid quench
1.8Li–0.6Zn–0.3Mg–0.3Mn–0.12Zr alloy naturally from the solution heat-treatment temperature. The matrix precip-
aged at room temperature after a rapid quench from itation is dominated by fine y00 /y0 (Al2Cu) plates and d0 spheres
(not shown).
the solution heat-treatment temperature is shown in
Figure 3 and the same alloy artificially aged at 1501C
for 6 h after a rapid quench from the solution heat-
Non-Heat-Treatable Alloys
treatment temperature is shown in Figure 4.
Alloys that are susceptible to corrosion in moist air The 1XXX, 3XXX, 5XXX, and some of the 8XXX
may be clad with pure aluminum, which is not sus- series alloys are non-heat-treatable. These alloys are
ceptible and can offer protection in an aggressive primarily strengthened by elements in solid solution
environment, or they may be anodized. Anodizing is and by deformation structures. One of the first non-
a method for producing a very thick, protective oxide heat-treatable alloys, other than the 1100 alloy that
film on the surface of aluminum. The anodized film only contained 0.95 wt.% Fe and Si as impurities,
normally contains chemical compounds such as chro- was an alloy containing 1.25 wt.% Mn and
mates, which are collected from the anodizing bath 0.12 wt.% Cu, now designated as 3003, that was
and render the film more corrosion resistant than introduced in 1906, the same year that Wilm dis-
films that form naturally in air. covered age hardening. Alloy 3003 can develop yield
Alloys: Aluminum 21

strength up to 400 MPa by special thermomechanical alloys contain elements that decrease in solubility
treatments. Magnesium is added to non-heat-treata- with decreasing temperature, and in concentrations
ble alloys for its solid-solution strengthening effect. It that significantly exceed their equilibrium solid sol-
also enhances work hardening and makes aluminum ubility at room and moderately higher temperatures.
more anodic. In general, solid-solution alloys are One of the transition elements Cr, Mn, or Zr, is
more resistant to corrosion than two-phase alloys. added to age-hardenable wrought alloys to control
Al–Mg alloys have a high resistance to corrosion, the grain structure; however, the grain and deforma-
particularly in seawater and alkaline solutions. tion structure play only a secondary role in the
However, when magnesium exceeds the solid solu- strengthening of this class of materials.
bility in binary alloys, it precipitates at grain bound- A normal heat-treatment cycle after deformation
aries as Al3Mg2, which is anodic to the matrix and processing includes a soak at a high temperature to
promotes intergranular attack. dissolve the soluble constituent particles and any
Aluminum combines readily with transition met- other precipitates that may be soluble (dispersoids
als, Ti, V, Cr, Mn, Fe, Co, Ni, and Zr to form in- containing the elements Cr, Mn, or Zr have already
termetallic phases with little or no solubility in the precipitated at this point). After the aluminum alloy
aluminum matrix. The intermetallics phases increase has been held at the solutionizing temperature for a
the strength by enhancing work hardening during sufficient time, the alloy is rapidly cooled or
working operations, and by refining the grain struc- quenched to prevent precipitation of coarse primary
ture. They increase the work hardening since they are phases and to obtain a solid solution supersaturated
usually incoherent with the matrix, are nondeform- with both solute elements and vacancies. The next
able, and must be looped or bypassed by moving step involves aging at room temperature (natural
dislocations. This increases the dislocation density aging) or at an intermediate temperature (artificial
and blocks the dynamic recovery processes. During aging).
hot working, the subgrain structure that is developed Quenching is accompanied by a change in free
can also increase the strength. The smaller (less than energy, which increases progressively as the differ-
0.6 mm) intermetallic dispersoids aid in the stabiliza- ence between the solutionizing temperature and the
tion of the substructure. The intermetallic particles quenching temperature increases. The volume free
generally do not add a component of the particle energy change is the driving force for precipitation
strengthening because of their low volume fraction, and is associated with the transfer of solute atoms to
large size, and interparticle spacing. a more stable phase. However, when precipitation
It is customary to control both composition and occurs there are other factors, which increase the free
cooling rates in order to prevent large primary phases energy, that is, formation of the interface between the
from forming. For example, commercial Al–Mn al- matrix and the precipitate requires an increase in the
loys most often contain less than 1.25 wt.% Mn, al- surface free energy, and, if there is a volume change
though as much as 1.82 wt.% is soluble in pure or interfacial strains associated with the precipitate,
aluminum. The 1.25 wt.% limit is imposed because there is an increase in elastic strain energy. The
Fe, present in most aluminum alloys as an impurity, change in free energy when a precipitate forms, DG,
decreases the solubility of Mn in aluminum. This in- is the sum of these free energy changes and can be
creases the probability of forming large primary par- expressed mathematically as
ticles of Al6Mn, which can have a disastrous effect
on ductility. DG ¼ VDGV þ Ag þ VDGs ½1
Silicon is a principal addition to most aluminum
casting alloys because it increases the fluidity of the where V is the volume of the new phase; DGV is the
melt. In the solid state, the hard silicon particles are free energy decrease due to creation of a volume, V,
the major contributor to the strength of non-heat- of the precipitates and is therefore negative; A is the
treatable casting alloys. An additional improvement area of the interface between the matrix and the
in strength can be obtained by minor additions of Mg precipitate; g the energy of the new surface formed;
or by trace alloy additions, for example, Na that re- and VDGs is the increase in elastic strain energy per
fines the cast structure. The latter also minimizes unit volume of precipitate.
porosity and increases ductility. The critical increase in free energy, DG , required
for a nucleus to become an equilibrium or metastable
precipitate is known as the activation energy barrier
Heat-Treatable Alloys for nucleation and can be expressed as
The 2XXX, 4XXX, 6XXX, 7XXX, and some 8XXX
series alloys are considered heat-treatable. These DG ¼ ð16pg3 Þ=ð3½DGV  DGs 2 Þ ½2
22 Alloys: Aluminum

The nucleation rate, N, can be expressed as the dislocations in aluminum alloy products ar-
range themselves into a structure of subgrains by
N ¼ C expðQ=kTÞ expðDG =kTÞ ½3 the process known as dynamic recovery. As tem-
perature decreases and strain rate increases, the
where C is the number of nucleating sites per unit stored energy increases. This increase in stored
volume, Q is the activation energy for diffusion, k is the energy is manifested by an increase in number and
Boltzmann constant, and T the absolute temperature. consequent decrease in the size of subgrains and
There are a number of requirements for an effective an increase in the dislocation density in cell walls.
age-hardenable alloy. First, there has to be a suffi- During subsequent heat treatments, the stored
cient volume fraction of the second phase for the de- energy of deformation may decrease by either
sired strength. An examination of aluminum binary static recovery or recrystallization. The final prod-
diagrams suggests that the most attractive alloying ad- uct may be completely unrecrystallized, partially
ditions would be, in decreasing volume fraction of the recrystallized, or completely recrystallized.
second phase, Ag, Mg, Li, Zn, Ga, and Cu. The re- 4. Grain size and shape. These are also controlled by
quired volume fraction will depend on the effectiveness thermomechanical history and may be influenced
of the second phase on impeding dislocation motion, by ingot grain refining practice.
that is, whether it is a ‘‘soft’’ or ‘‘hard’’ phase, and its 5. Crystallographic texture. The deformation proc-
size and spacing. Second, the aging potential must be ess used to produce the product and the thermo-
adequate and the nucleation of the precipitates should mechanical history determines this.
be as close as possible to being homogeneous. This is 6. Intermetallic constituent particle. These form by a
generally accomplished by the addition of a second liquid–solid eutectic reaction during solidification,
alloying element. The addition of the second alloying primarily from impurities, for example, iron and
element may: (1) reduce the solubility of the first el- silicon. Because the low solubility of iron in pure
ement, (2) increase the diffusion rates by trapping aluminum is reduced by alloying elements, con-
vacancies, (3) increase the driving force for nucleation, stituent particles containing iron are insoluble.
or (4) reduce the activation energy against nucleation. The size and size distribution of insoluble constit-
If the strain term in eqn [1] is large, nucleation may be uent particles are controlled by the rate of ingot
aided by defects, for example, dislocations. For this solidification, the chemical composition, and the
type of nucleation, the precipitate distribution depends extent and nature of bulk deformation. Constitu-
on the distribution of the defects. ents generally range in size from 1 to 30 mm.
7. Porosity may be a factor in thick plates and, par-
ticularly, in castings. Individual pores can range
Microstructural Features
up to and beyond 200 mm.
Chemical composition and processing control the
microstructure and thus the physical, mechanical,
Aluminum Alloy Design
and corrosion properties of aluminum alloy prod-
ucts. In general, microstructural features of impor- After Wilm discovered age hardening of the Al–
tance for property control include the following: Cu–Mg–Mn alloy through serendipity, traditional
aluminum alloy development primarily relied on a
1. Coherency, volume fraction, and distribution of trial-and-error philosophy. Most aluminum alloys
strengthening precipitates. These form intention- were developed by simply identifying a given alloying
ally during aging treatments and unintentionally element that led to gains in a certain mechanical
during quenching. The precipitates normally property, typically strength, and continuing to add
range in size from 1 to 10 nm. that element until the property began to degrade. The
2. Size, distribution, and coherency of dispersoid amount of secondary element used would then be the
particles. These form during the ingot preheat or percentage that corresponded with maximum per-
homogenization treatment by precipitation of the formance. This process would then follow in ternary
transition elements chromium, manganese, or zir- and higher-order systems in order to optimize a given
conium. They are present by design to control mechanical property. Often this occurred with little
the grain structure and degree of recrystallization. knowledge of the corresponding microstructural
Dispersoids normally range in size from 10 to components or appropriate thermodynamic unders-
200 nm. tanding, for instance the phase field of operation.
3. Degree of recrystallization. This is determined by Unfortunately, this practice led to the slow insertion
the thermomechanical history. During bulk metal of novel alloys into various applications creating a
deformation processing at elevated temperatures, disparity with the design community.
Alloys: Aluminum 23

Table 1 Property–microstructure relationships in aluminum alloys

Property Desired microstructural feature(s) Function of feature(s)

Strength Fine grain size with a uniform dispersion of Inhibit dislocation motion
small, hard particles
Ductility and toughness Fine structure with clean grain boundaries Encourage plasticity and work hardening,
and no large or shearable precipitates inhibit void formation and growth
Creep resistance Thermally stable particles within the matrix Inhibit grain boundary sliding and coarse
and on the grain boundaries microstructure
Fatigue crack initiation Fine grain size with no shearable particles Prevent strain localization, stress
resistance and no surface defects concentrations, and surface slip steps
Fatigue crack propagation Large grain size with shearable particles and Encourage crack closure, branching,
resistance no anodic phases or hydrogen traps deflection, and slip reversibility
Pitting No anodic phases Prevent preferential dissolution of second-
phase particles
Stress corrosion cracking Hard particles and no anodic phases or Homogenize slip and prevent crack
and hydrogen interconnected hydrogen traps propagation due to anodic dissolution of
embrittlement phases

More recent aluminum-based alloy development has wide variety of applications. The 1XXX alloys are
begun to utilize the knowledge of phase diagrams and strain-hardenable, have high formability, corrosion
the productive interplay between empirical research resistance, and good electrical conductivity. They are
and theoretical science. Phase diagrams represent the often used for electrical applications and as foil and
state of an alloy as a function of temperature, pressure, strip for packaging. The 2XXX alloys are heat-treat-
and alloy concentration and can be used for alloy able, and possess excellent combinations of high
design and process development. There are currently strength and toughness, but are not resistant to at-
software programs available that use thermodynamic mospheric corrosion; so, they are usually painted or
functions for the calculation of phase diagrams. Such clad in such exposures. They are primarily used for
calculations reduce the effort required to determine aircraft and truck body applications. The 3XXX al-
equilibrium conditions in a multicomponent system. loys have high formability, corrosion resistance, and
Materials design can now be viewed as the best ap- can be joined by a variety of methods. They are
plication of available models for the prediction and widely used in cooking utensils, beverage cans,
synthesis of alloys with desired properties. Of course, chemical equipment, roofing and siding, and in heat
all theoretical calculations must be verified experimen- exchangers. The 4XXX alloys are heat-treatable and
tally, but the experiments should be based on sound have good flow characteristics along with medium
theory and not on a trial-and-error approach. strength. They are primarily used in forging applica-
The most general understanding of alloy design tions, for example, aircraft pistons and for welding
must derive itself from the metallurgical paradigm of wires. The 5XXX alloys are strain-hardenable
processing–structure–property relationships. In order and have excellent corrosion resistance, even in
to optimize a given property, one must determine the salt water, toughness, and weldability along with
necessary ideal microstructural components, which moderate strength. They find wide application in
are dictated by the alloy composition and processing highway bridges, storage tanks, marine and auto-
from solidification through final manufacture. Fun- motive applications, and pressure vessels. The 6XXX
damentally, to properly determine the resulting alloys are heat-treatable, have high corrosion resist-
microstructure for a given thermomechanical proc- ance, excellent extrudibility, and moderate strength.
essing routine, the pertinent literature and equilibri- They are primarily used in building and construct-
um phase diagrams must be consulted. A summary of ion, automotive and marine applications, and re-
various empirical structure–property relationships cently in aerospace applications. The 7XXX alloys
found in aluminum alloys is summarized in Table 1. are heat-treatable and can provide the highest
This table highlights the general structural features strengths of all aluminum alloys. They are primari-
desired to optimize a given property. ly used for automotive and aerospace applications.
The 8XXX series is used for alloys that contain
lesser-used alloying elements such as nickel and lith-
Applications of Aluminum Alloys ium. Some are used for electrical conductors and
Aluminum alloys are lightweight, resistant to atmos- the lithium-containing alloys have found application
pheric corrosion, and have good mechanical proper- in the aerospace industry due to lithium’s effect in
ties, which make them a popular material for use in a reducing density.
24 Alloys: Copper

Summary Further Reading

Aluminum alloys possess a number of very attractive Altenpohl DG (1998) Aluminum: Technology, Applications, and
characteristics that, together with their light weight, Environment. Washington, DC: The Aluminum Association.
Aluminum Association (1997) Aluminum Standards and Data.
make them extremely attractive for many applications.
Washington, DC: The Aluminum Association.
Since aluminum comprises 8% of the Earth’s crust by Hatch JE (ed.) (1984) Aluminum: Properties and Physical Metal-
weight, the overall reserves are adequate to cope with lurgy. Metals Park, OH: American Society for Metals.
anticipated demands for the foreseeable future. Cost Hornbogen E (2001) Hundred years of precipitation hardening.
differentials with respect to competing materials will Journal of Light Metals 1: 127–132.
Lorimer GW (1978) Precipitation in aluminum alloys. In: Russell
probably dictate the extent to which aluminum will be
KC and Aaronson HI (eds.) Precipitation Processes in Solids, pp.
used. The cost includes the price of the electrical 87–119. New York: The Metallurgical Society (AIME).
energy needed for the extraction of the metal from its Polmear IJ (1995) Light Alloys, Metallurgy of the Light Metals.
minerals, and this may be greatly reduced in the future London: Arnold.
since aluminum can be recycled for many applications. Sanders RE Jr., Baumann SF, and Stumpf HC (1986) Non-heat-
treatable aluminum alloys. In: Starke EA and Sanders TH (eds.)
In addition, aluminum often pays off when lifecycle
Aluminum Alloys: Their Physical and Mechanical Properties,
costs, instead of acquisition costs, are considered. vol. III, pp. 1441–1484. West Midlands, UK: EMAS.
Starke EA Jr. and Staley JT (1996) Application of modern al-
uminum alloys to aircraft. Progress in Aerospace Science 32:
See also: Deformation Processing; Mechanical Proper- 131–172.
ties: Strengthening Mechanisms in Metals; Phase Trans- Vasudevan AK and Doherty RD (eds.) (1989) Aluminum Alloys –
formation; Phases and Phase Equilibrium; Recovery, Contemporary Research and Applications. London: Academic
Recrystallization, and Grain Growth; Solidification: Mod- Press.
els and Simulations. Zhu AW, Gable BM, Shiflet GJ, and Starke EA Jr. (2002) The
intelligent design of age hardenable wrought aluminum alloys.
Advanced Engineering Materials 4(11): 839–846. Germany:
PACS: 60; 61.66; 62.20 Wiley-VCH Verlag GmbH.

Alloys: Copper
W F Hosford, University of Michigan, Ann Arbor, MI, Australia as shown in Figure 1. Other sources include
USA Zambia, Peru, Canada, Russia, and Mexico. Recycle
& 2005, Elsevier Ltd. All Rights Reserved. scrap accounts for B50% of copper consumed in
the US.
Copper has good strength and ductility as well as
reasonably good corrosion resistance. The unique
Copper Production features of copper are its color and very high elec-
trical and thermal conductivities.
Copper is among the few metals known to ancient
civilizations. Copper, gold, silver, platinum, and me-
teoric iron are the only metals found in their native
state. However B90% of copper comes from sulfide 6
ores. Most copper ores contain 0.5% Cu or less, iron

being the principal metallic impurity. The ores are 5

Millions of metric tons

ground to a fine powder and concentrated (normally

by floatation) to 20–25% Cu. This concentrate is 4
melted as a matte of mixed Cu and Fe sulfides con-
taining up to 60% Cu. The matte is oxidized to 3

remove the iron as an oxide and burn off the sulfur.



The product is called blister copper which contains 2

98.5% Cu. Blister copper is fire refined to a tough-
pitch (99.5% Cu). It is, often, then electrolytically 1
refined to 99.95% þ Cu. Au, Ag, and Pt metals are
recovered from slime. 0
Chile is the single largest producer of copper. Figure 1 World production of copper in 2000. (Data from Chil-
The other major producers are US, Indonesia, and ean Copper Commission.)
Alloys: Copper 25

Table 1 Physical properties of copper 60

Physical property Value

Atomic number 29
Atomic mass 63.54 amu
Cu 69%, 65Cu 31% 40

Stable isotopes


All other isotopes have very
short half-lives (o13 h)
Crystal structure f.c.c.
Lattice parameter (251C) 0.3615 nm


Density (201C) 8930 kg m  3 20


Electrical conductivity 103.06 IACS ¼ 59:77 
106 ohm1 m1
Melting point 1084.9 K
Heat of fusion 205 J kg  1
Volume change on 4.92% 0
solidification Figure 2 Uses of copper. (Data from ASM (2001) Copper and
Boiling pointa 2967, 2995 K Copper Alloys, ASM Speciality Handbook. Materials Park, OH:
Heat of vaporization 4730 J kg  1 ASM.)
Linear thermal expansion 16.7 mm mK  1
coefficient (201C)
Thermal conductivity (01C) 403 W mK  1 Table 2 Electrical and thermal conductivities of several metals
Specific heat (201C) 386 J kg  1 K  1
Young’s modulus (GPa) Metal Electrical Electrical Thermal
polycrystal 128 resistivity conductivity conductivity
/1 1 1S direction 191.7 ðnO mÞ ð% IACSÞ ðW m1 K1 Þ
/1 0 0S direction 99.9
Silver 14.7 108.4 428 (201C)
Shear modulus 46.4 Gpa
Copper 16.73 103.06 398 (271C)
Poisson’s ration 0.34
Gold 23.5 73.4 317.9 (01C)
Self-diffusion coefficient, Db
Aluminum 26.55 64.94 247 (251C)
830–10301C 47  10  4 exp(259 000/RT)
Beryllium 40 43 190 (201C)
m2 s  1
Magnesium 44.5 38 418 (201C)
750–9501C 11  10  4 exp(240 000/RT)
Zinc 59.16 28.27 113 (251C)
m2 s  1
Nickel 68.44 25.2 82.9 (1001C)
where R ¼ 8:314 J mol1 K1 Iron 98 17.59 80 (201C)
Energy of vacanciesc 84 kJ mol  1 Platinum 106 16.3 71.1 (01C)
Liquid/solid surface energyd 17.7 J m  2
High-angle grain-boundary 60 J m  2 Data from Metals Handbook, 9th edn. vol II. Metals Park, OH:
energye ASM.
Solid/gas surface free energyf 170 J m  2
Stacking fault energyg,h 70 J m  2 use, principally as wire. Corrosion resistance, ther-
All data are from (1979) Metals Handbook, 9th edn. vol. II. Metals mal conductivity, formability, and unique color ac-
Park, OH: ASM. except where noted. count for almost the rest of the consumption of
(2004–2005) Handbook of Chemistry and Physics, 85th edn. copper.
Cleveland: CRP.
Jost W (1952) Diffusion in Solids, Liquids, Gases. New York:
Academic Press. Electrical Properties
Reed-Hill R (1973) Physical Metallurgy Principles. Princeton,
NJ: Van Nostrand-Reinhold. The International Association of Classification Soci-
Hollomon JH and Turnbull D (1953) Progress in Metal Physics 4. eties (IACS) standard of conductivity was set up with
Gjostein NA and Rhines FA (1959) Acta Mettallurgica 7: 319.
f what was thought to be the conductivity of pure
Udin H, Schaler AJ and Wulff J (1944) Journal of Metals 1: 1936.
Fullman F (1951) Journal of Applied Physics 22: 448. copper as 100%. Later it was found that copper had
Thoron PR and Hirsch PB (1958) Philosophical Magazine a higher conductivity than what was originally
3: 738. thought, but the standard was not changed. Pure
copper has a conductivity of 104% according to the
The physical properties of copper are listed in IACS standard.
Table 1. Only silver has higher electrical and thermal
conductivities than copper. Table 2 lists the con-
ductivities of various metals. The resistivity increases
by 43% between 01C and 1001C.
Figure 2 shows the principal uses of copper. Its ex- All impurities raise the electrical resistivity of
cellent electrical conductivity accounts for most of its copper. At low concentrations, the increase is
26 Alloys: Copper

0.08 100
P Fe Zr

Increase of resistivity (µΩ m)


Reflectance (%)
0.04 Mg

0.20 Al 20
In Ag 0
0 Cd 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1 Au Wavelength (µm)
Weight percent solute
Figure 4 The color of copper is a result of the dependence of
Figure 3 All impurities increase the resistivity of copper. The reflectivity on wavelength. (Data from ASM (2001) Copper and
increase is proportional to the concentration. (Adapted from ASM Copper Alloys, ASM Speciality Handbook. Materials Park, OH:
(2001) Copper and Copper Alloys. ASM Speciality Handbook. ASM.)
Materials Park, OH: ASM.)

solution. This lowers the conductivity to below that

proportional to the concentration of the impurity. In of oxygen-free copper. Lead is often added to copper
general, the effect of solutes in raising the resistivity and copper alloys to form free machining character-
is greater for large differences between the atomic istics. Lead is virtually insoluble in copper as shown
diameter of the solute and that of copper as shown in by the copper–lead phase diagram (see Figure 7).
Figure 3. Also solutes of high valency (P, Si, and As) Lead particles appear as a separate phase in the grain
have a greater effect on resistivity than those of low boundaries (see Figure 8).
valency such as Ag, Au, and Cd. Zirconium copper (C15000) contains 0.13–0.20%
Zr. It can be heat treated to yield strengths of
400 MPa while retaining a conductivity of 84%
IACS. Copper dispersion strengthened by 0.2–0.7%
Probably the most distinctive property of copper is Al2O3 retains reasonable strength at temperatures up
its color. The reddish hue is a result of how the to 10001C. It finds applications as electrodes for re-
reflectivity varies with wavelength. The dependence sistance welding.
of reflectivity on the wavelength of a polished copper Bismuth is a very detrimental impurity. It com-
surface is shown in Figure 4. pletely wets the grain boundaries and because it is
brittle, its presence renders copper and copper-based
alloys brittle. It must be kept o0.003%.
Commercial Grades of Copper
There are several commercial grades of copper. The
principal use of oxygen-free copper (C10100) with a Copper Alloys
minimum of 99.99% Cu is for wire. Fire-refined
tough-pitch copper (C2500) is fire-refined copper Copper is too soft for structural applications. Its
containing between 0.02% and 0.05% oxygen in the strength is markedly increased by alloying. For dilute
form of Cu2O and B0.5% other elements. The cop- substitutional solutions (see Figure 9), the yield
per–oxygen phase diagram is shown in Figure 5. strength increases in proportion to the solute con-
There is a eutectic at 0.38% oxygen and 10661C. centration. The rate of this increase is proportional
Figure 6 is a typical microstructure of cast tough- to the 4/3 power of a misfit parameter defined as
pitch copper showing primary copper dendrites sur- e ¼ ðda=aÞ=dc, where da/a is the fractional change in
rounded by a copper–copper oxide eutectic. the lattice parameter with concentration c, expressed
Often tellurium (B0.5%) or sulfur (B0.5%) is as an atomic fraction of solutes,
added to copper to promote free machining. Copper
is frequently deoxidized with phosphorus. Deoxida- Dt
¼ CGe4=3 ½1
tion leaves B0.01% residual phosphorus in solid Dc
Alloys: Copper 27

L1 L1 + L2

Temperature (°C)

L1 + Cu2O

 + Cu2O
0 2 4 6 8 10 12
Weight percent oxygen
Figure 5 The copper–oxygen phase diagram. (Adapted from (1973) Metals Handbook, 8th edn., vol. VIII. Materials Park, OH: ASM.)

increases from B0.4 for pure copper to B0.6 or 0.65

at 35% Zn. The greater strain hardening raises the
tensile strength as shown in Figure 12.
Finer grain size also increases strength. The grain-
size dependence of the yield strength follows the
Hall–Petch relation

sY ¼ sO þ Kd1=2 ½3

Scale line Deformation Mechanisms

The slip system in copper, as in all other f.c.c. metals,
Figure 6 The microstructure of cast toughpitch copper. The is {1 1 1}/1 1 0S. The critical shear stress for slip is
scale line is B125 mm. Note the primary copper dendrites and the
B3–4 MPa. The extent of work hardening increases
Cu–Cu2O eutectic. (Reproduced with permission of the Copper
Development Association Inc.) as the temperature is lowered below room temper-
ature as indicated in Figure 13.
Until the early 1950s, it was believed that me-
where G is the shear modulus of copper and C is a chanical twinning did not occur in copper. However,
constant. The misfit, e, is the fractional difference of it is now known that copper can deform by mechan-
atomic diameters of copper and solute. Figure 10 ical twinning on {1 1 1} planes, and in /11c2S di-
shows the misfits for several solutes. On an atomic rections, at low temperatures. The critical shear
basis Sb, Sn, and In are the most potent hardeners. stress for twinning at 4.2 K is B150 MPa. In copper
Although different measurements of the stacking alloys, twinning depends on the stacking fault energy.
fault energy of pure copper have ranges from 40 to The critical resolved-shear stress for twinning is low-
169 mJ m  2, the best estimate is probably ered by solid solutions that reduce the stacking fault
80 mJ m  2. Zinc, tin, and aluminum form solid so- energy (see Figure 14). It is significant that copper is
lutions that lower the stacking fault energy to not embrittled at low temperatures and does not
o5 mJ m  2 at high concentrations as shown in fracture by cleavage, because of which copper finds
Figure 11. The stacking fault energy depends on the use in cryogenic equipment.
ratio of valence electrons to atoms. A lower stacking
fault energy leads to more strain hardening. The
Commercial Alloys
value of n in the power-law equation,
Copper alloys are classified as being either wrought
s ¼ Ken ½2 alloys or casting alloys, depending on their usage.
28 Alloys: Copper

Atomic percentage lead

10 20 30 40 50 60 70 80 90

1900 F
1000 991°,−63
1700 F 36 955°
1600 F

1400 F
Temperature (°C)

1200 F


1000 F

326° 327.502°
Cu 10 20 30 40 50 60 70 80 90 Pb
Weight percentage lead
Figure 7 The copper–lead phase diagram. (Data from Metals Handbook 8th edn. vol. VIII. p. 296. Materials Park, OH: ASM.)

Grain Lead 3
boundary particles
Shear yield strength (MPa)



Au Mn

Ge Ni Si

0 1 2 3
Figure 8 Lead appears as a separate phase in copper alloys.
Atomic percent solute
(Adapted from ASM (2001) Copper and Copper Alloys, ASM
Specialty Handbook. Materials Park, OH: ASM.) Figure 9 Effect of solute concentration on shear yield strength
of dilute copperbase alloys. Note that the strength is proportional
to concentration. (Data from Linde JO, Lindell BO, and Stade CH
(1950) Arkiv für Fysik 2.)
The unified numbering system (UNS) designations
for some of the wrought alloys are listed in Table 3.
Sheets produced by cold rolling and rods or wire
produced by drawing are sold with a temper designa-
tion indicating the amount of cold reduction. The Brasses are alloys of copper and zinc. Figure 15 is the
system is outlined in Table 4. copper–zinc phase diagram. The solubility of zinc in
Alloys: Copper 29

0.1 400

0.05 Sn

Tensile strength (MPa)

0.025 mm
0.015 mm
0.02 0.035 mm
(1/G)d / dc

Mn 300

Ge 0.070 mm

0.002 0.050 mm

0.001 200
0 5 10 15 20 25 30
0.01 0.02 0.05 0.1 0.2 0.5 1
% Zinc
Figure 12 The effect of zinc content on increasing the tensile
Figure 10 The effect of the atomic misfit on the strengthening
strength. Finer grain size has a similar effect. (Adapted from ASM
effect of the solute. (Data from Linde JO and Edwardson S (1954)
(2001) Copper and Copper Alloys, ASM Speciality Handbook.
Arkiv für Fysik, 8.)
Materials Park, OH: ASM.)

60 120
4.2 K
Stacking fault energy (mJ m−2)

100 78 K

Shear stress (MPa)

200 K

60 300 K

0 20
1 1.1 1.2 1.3
Electron/atom ratio
Figure 11 The stacking fault energy of alloys. (’Cu–Zn, mCu– 0 0.2 0.4 0.6 0.8
Al,  Cu–Ge) (Adapted from Venables JA (1963) Deformation Shear strain
twinning in fcc metals. In: Reed-Hill RE (ed.) Deformation
Twinning. Proceedings of the Metallurgical Society Conference, Figure 13 Tensile stress–strain curves for copper crystals at
vol. 25, p. 77. Gordon and Breach. temperatures down to 4.2 K. (Adapted from Blewitt TH, Coltman
RR, and Redman JK (1955) Proceedings of the Conference on
Defects in Crystalline Solids, Bristol, p. 369. London: Physical
the f.c.c. lattice of copper is 435%. At nearly equal Society.)
atomic percentages of copper and zinc, there is a
b.c.c. intermetallic phase b, that has a wide range of
solubility. At temperatures oB4601C, b-brass un- As the lower stacking fault energy increases the
dergoes an ordering reaction to form an ordered exponent n, the uniform elongation also increases in
b.c.c. phase b0 with a B2 (CsCl) structure. Each cop- the power-law approximation of the stress–strain
per atom is surrounded by eight zinc and eight cop- curve. Another effect of the lower stacking fault
per atoms. A eutectoid transformation, b0 -a þ g at energy is that annealing twins are much more fre-
2501C has been reported. However, the transforma- quent. Figure 18 compares the annealed microstruc-
tion is rarely, if ever, encountered. ture of copper with brass containing 30% Zn. The
Zinc provides solid solution hardening to copper. microstructures of brass containing 40% zinc or
Both the yield strength and the tensile strength in- more consist of two phases, a and b. The two-phase
crease with zinc content, as indicated in Figures 16 microstructure of Muntz metal (40% Zn) is shown in
and 17. Figure 19. At 8001C the microstructure is entirely b,
30 Alloys: Copper

but as the alloy is cooled, a precipitates. The effects lists the zinc equivalents of the common alloying
of other alloying elements to brass on the solid elements. The relative values of the zinc equivalents
solubility of zinc can be approximated by ‘‘zinc of the elements can be understood in terms of the
equivalents.’’ A concentration of 1% tin is equivalent number of valence electrons per atom. Silicon and
to 2% zinc, so tin has a zinc equivalent of 2. Table 5
903°C 37.5
900 32.5
Twinning stress (MPa)

56.5 835°C

Temperature (°C)


39 465°C


0 20 40 60 80 300
Stacking fault energy (mJ m−2) 0 10 20 30 40 50 60
Weight percent zinc
Figure 14 The critical shear stress for twinning decreases with
decreasing stacking fault energy. ( , Cu–Zn; , Cu–Al; , Cu–Ge) Figure 15 The copper–zinc phase diagram. (Adapted from
(Adapted from Venables JA, In: Hirth and Rogers (eds.) Defor- ASM (2001) Copper and Copper Alloys, ASM Specialty Hand-
mation Twinning, Reed-Hill: AIME.) book. Materials Park, OH: ASM.)

Table 3 Commercial wrought-copper alloys

Designation Composition Common name Uses

C21000 5% Zn Gilding metal Coinage, jewelry

C22000 10% Zn Commercial bronze Architectural, jewelry
C23000 15% Zn Red brass Architectural, plumbing
C24000 20% Zn Low brass Ornamental, musical
C26000 30% Zn Cartridge brass Lamps, cartridge cases
C27000 35% Zn Yellow brass Architectural, lamps
C28000 40% Zn Muntz metal Architectural, heat exchangers
C36000 35.5% Zn, 2.5% Pb Free cutting brass Screw machine parts
C44300 28% Zn, 1% Sn Admiralty brass Heat exchangers
C51000 5% Sn, 0.2% P Phosphor bronze Hardware
C71500 30% Ni Cupro nickel Heat exchangers
C75200 18% Ni, 17% Zn Nickel silver Hardware, jewelry
C172000 1.8–2.0% Be, 0.2–0.8% Ni þ CO Beryllium copper Springs, instruments

Table 4 Temper designations

Temper designation B&S Gage numbers Rolled sheet Strain Drawn wire Strain
Thickness reduction ð%Þ Diameter reduction ð%Þ
H01 1/4 hard 1 10.9 0.116 10.9 0.232
H02 1/2 hard 2 20.7 0.232 20.7 0.463
H03 3/4 hard 3 29.4 0.347 29.4 0.694
H04 hard 4 37.1 0.463 37.1 0.926
H06 extra hard 6 50.1 0.696 50.1 1.39
H08 spring 8 60.5 0.925 60.5 1.86
H10 extra spring 10 68.6 1.16 68.6 2.32
Alloys: Copper 31




Other Alloys
True bronzes are copper–tin alloys. Tin is much less
soluble in copper than zinc as shown in the copper–
20% Zn
30% Zn
15% Zn 10% Zn tin phase diagram (see Figure 20). The e phase is very
Yield strength (MPa)

sluggish to form and is rarely encountered. Tin is a

400 5% Zn
potent solid solution strengthener. However, alloys
with 48% Sn are too brittle to be formed mechan-
ically. Because bronzes have a much greater differ-
ence of the liquidus and solidus temperatures than
brass, they are much easier to cast. Casting alloys
Flat rolled product often contain up to B20% Sn. These are used ex-
tensively for pipe fittings and bells.
Alloys of copper with aluminum and silicon are
0 0.4 0.8 1.2 called aluminum-bronze and silicon-bronze, even
True strain though they contain no tin. Manganese-bronze is a
brass with only 1% Sn and 0.1% Mn. Aluminum-
Figure 16 The yield strength of brasses increases with the
amount of cold work and the zinc content. The true strain, e ¼ bronzes containing up to 7% Al behave much like
lnðto =tÞ where t is the thickness and to is the thickness before cold brass. The phase diagram (see Figure 21) shows that
rolling. (Adapted from (1979) Metals Handbook, 9th edn., vol. II. a b.c.c. b-phase occurs at B12% Al.
Materials Park, OH: ASM.) There is a eutectoid transformation of b-a þ g2 at
5651C and 11.8% Al. Figure 22 shows the pearlite-
like microstructure of alternating platelets of a and g2
in an alloy of Cu–11.8% Al after slowly cooling from




8001C. With rapid cooling, the eutectoid transforma-

30% Zn tion is suppressed and the alloy transforms by mar-
tensitic shear to a new hexagonal phase b0 . Figure 23
Tensile strength (MPa)

20% Zn
600 15% Zn shows the martensitic b0 needles in the same alloy
after rapid cooling. The martensitic reaction is almost
500 10% Zn instantaneous. It starts when b is quenched below the
5% Zn Ms temperature and is virtually complete when the
400 Pure Cu temperature is at the Mf. The Ms and Mf temperatures
depend on the composition as shown in Figure 24.
The oxidation resistance of aluminum bronzes is
Flat rolled product somewhat superior to brass.
Silicon bronzes are more easily cast than brass and
0 0.4 0.8 1.2 have better resistance to oxidation and acid corro-
True strain sion.
Copper and nickel are completely miscible in both
Figure 17 The tensile strength of brasses increases with the
the liquid and solid state as shown in the copper–
amount of cold work and the zinc content. The true strain, e ¼
lnðto =tÞ where t is the thickness and to is the thickness before cold nickel phase diagram (see Figure 25). The low-tem-
rolling. (Data from (1979) Metals Handbook, 9th edn., vol. II. perature miscibility gap is rarely encountered.
Materials Park, OH: ASM.) Nickel imparts improved corrosion and oxidation
resistance to copper. Cupronickels containing 20–
30% Ni are widely used for heat exchangers.
aluminum have high valencies (4 for silicon and 3 Alloying copper with 20% Ni causes a loss of the
for aluminum) and low atomic masses (28 for silicon copper’s familiar yellow–reddish color. Coinage ac-
and 27 for aluminum), so 1 wt. % causes a large counts for B1% of the consumption of copper. The
increase of the electron-to-atom ratio. Magnesium, five-cent pieces (‘‘nickels’’) in the US and Canada are
like zinc has a valency of 2, but the atomic mass alloys of 75% copper and 25% nickel. The 10-cent
of magnesium is less than half that of zinc (24 for (‘‘dime’’) and 25-cent (‘‘quarter’’) pieces are sand-
magnesium vs. 65 for zinc), so 1 wt.%Mg has wiches of pure copper core clad with 75% copper–
over twice the effect of 1 wt.%Zn. Tin has a valence 25% nickel alloy as shown in Figure 26. The Susan B
of 4, but its atomic mass (119) is roughly twice that Anthony US dollar is a sandwich of pure copper clad
of zinc. with 77Cu–12 Zn–7 Mn–4 Ni. European coins are
32 Alloys: Copper

(a) (b)
Figure 18 Microstructures of (a) pure copper and (b) 70/30 brass (200). Note that the frequency of annealing twins is higher in the
brass. ((a) Adapted from (1972) Metals Handbook, 8th edn., vol. VII. Materials Park, OH: ASM. (b) Adapted from Hosford WF (1993) The
Mechanics of Crystals and Textured Polycrystals. London: Oxford Science Pub. Courtesy A Graf).



Temperature (°C)



0 10 20 30 40
Weight percent tin
Figure 20 The copper–tin phase diagram. b is a b.c.c. phase.
In alloys containing B20% Sn, formation of the e-phase is so
Figure 19 Photomicrograph of a Muntz metal (40% Zn) after sluggish that it is seldom encountered. (Adapted from (1973)
air cooling. The dark phase is b and the light phase is a, which Metals Handbook, 8th edn., vol. VIII. Materials Park, OH: ASM.)
has precipitated during cooling. (Adapted from Brick RM, Gordon
RB, and Phillips A (1965) Structure and Properties of Alloys. New
York: McGraw-Hill.)
Nickel Silvers
Nickel silvers are ternary alloys containing 12–18%
Table 5 Zinc equivalents of several alloying elements in brass Ni and 17–27% Zn. They have good strength and
Element Zn equivalent Element Zn equivalent good corrosion resistance. They are used for springs
and corrosion-resistant hardware. Costume jewelry
Si 10 Al 6
Mg –2 Sn 2
is often made from nickel silver because of its pleas-
Fe 0.9 Mn 0.5 ant silvery color.
Co –0.8 Ni –1.3

Beryllium Copper
either copper-based alloys or combinations of two Copper alloys containing 1.8–2.0% Be and 0.2–
copper-based alloys. Constantan (45% Ni) has a 0.8% Ni þ Co are precipitation-hardenable. The
nearly zero-temperature coefficient of resistivity, alloy C172000 is solution treated at B8001C and
hence is useful in instruments. rapidly cooled. The precipitation is done between
Alloys: Copper 33

1100 Form 
Liquid ′needle grain boundry

7.4 1035°C 9

Temperature (°C)

 + 1

 + 2

 + 2 100 µm
500 Figure 23 Martensitic structure of Cu–11.8% Al after rapid
0 5 10 15 cooling from 8001C. (Adapted from ASM (2001) Copper and
Weight percent aluminum Copper Alloys, ASM Specialty Handbook, p. 48. Materials Park,
Figure 21 The copper–aluminum phase diagram. b is a b.c.c.
phase. Note that eutectoid transformation at 5651C. (Adapted
from (1973) Metals Handbook, 8th edn., vol. VIII. Materials Park,

2 (dark)
Temperature (°C)


Mf Ms


11 12 13 14
Weight percent aluminum

Figure 24 The Ms and Mf temperatures of aluminum bronzes.

(Adapted from Flinn RA (1961) The Non-Ferrous Copper,
Brass, and Bronze Castings, p. 20. Cleveland, OH: Founders

3001C and 4001C. Figure 27 shows the copper-rich

end of the Cu–Be phase diagram. Precipitation hard-
ening produces very high strengths. Even higher
100 µm strengths can be achieved by cold working after the
solution treatment and before precipitation as shown
Figure 22 Pearlite-like structure in Cu–11.8% Al after slow
cooling from 8001C. (Adapted from ASM (2001) Copper and
in Table 6. The principal drawback is that the tox-
Copper Alloys, ASM Specialty Handbook, p. 48. Materials Park, icity of beryllium is an extreme danger in preparing
OH: ASM.) this alloy.
34 Alloys: Copper

Atomic % at Ni
10 20 30 40 50 60 70 80 90

2600 F
1400 L
2500 F

Temperature (°C)

2300 F


2100 F

1100 (Cu, Ni)

500 F 1 2
Curie temperature
1 + 2
Cu 10 20 30 40 50 60 70 80 90 Ni
Weight percentage nickel

Figure 25 The copper–nickel phase diagram shows complete miscibility. (Adapted from (1973) Metals Handbook, 8th edn., vol. VIII.
Metals Park, OH: ASM.)

Figure 26 Cross section of a US dime. (Adapted from Copper and Copper Alloys, ASM Speciality Handbook. Materials Park, OH:

Shape Memory Alloys increases with alloying. The compression texture of

copper, like all f.c.c. metals, has /1 1 0S directions
There are two groups of shape memory alloys: one
parallel to the axis of compression.
group consists of Cu–Zn–Al alloys with 10–30% Zn
Sheet textures are best described by orientation
and 5–10% Al. Cu–Al–Ni alloys with 11–14.5% Al
distribution functions (ODF). The rolling texture
and 3–5% Ni form the other group. The recoverable
of copper may consist of two continuous fibers or
strain for both groups is 4%. The martensitic trans-
tubes in the ODF. The major fiber runs between
formation temperature is o1201C for the Cu–Zn–Al
alloys and o2001C for the Cu–Al–Ni alloys. The {1 1 0}/1 1 2S (B) at F ¼ 45 , f2 ¼ 90 , f1 ¼ 35
through {1 2 3}/6 3 4S (S) at F ¼ 37 , f2 ¼ 63 ,
exact temperatures depend on composition. The
f1 ¼ 59 to {1 1 2}/1 1 1S (C) at F ¼ 35 , f2 ¼ 45 ,
temperature hysteresis is in the range of 15–251C.
f1 ¼ 90 . A second minor fiber runs between {1 1 0}
/0 0 1S (G) at F ¼ 45 , f2 ¼ 90 , f1 ¼ 0 and
{1 1 0}/1 1 2S (B) at F ¼ 45 , f2 ¼ 90 , f1 ¼ 35 .
Deformation Textures
The corresponding ODF is shown in Figure 28
The wire texture of copper consists of 35% /1 0 0S and represented schematically in Figure 29. This
and 65% /1 1 1S. The ratio of /1 0 0S to /1 1 1S texture has been characterized as consisting of
Alloys: Copper 35

1100 1

1000 Liquid Φ
Temperature (°C)

2.7 866°C

 + 1 1


1.55 605°C
600 6
 + 2

0 2 4 6 8
Weight percent beryllium
Figure 27 The copper–beryllium phase diagram. (Adapted
from (1973) Metals Handbook, 8th edn., vol. VIII. Materials Park,
OH: ASM.) 1 3 7

14 18 20

Figure 28 ODF for cold rolled copper. (Adapted from Hump-

Table 6 Strength of beryllium copper sheet (C17200)b hreys FJ and Hatherly M (1996) Recrystallization and Related
Conditiona Yield strength Tensile strength Phenomena. London: Elsevier.)

Annealed 250 465

1/4 H 485 570 0° 90°
1/2 H 555 605 1
H 690 730 Φ 2
Annealed & HT 1060 1260
1/4 H & HT 1125 1290 90°
1/2 H & HT 1180 1325
H & HT 1195 1360
HT indicates age-hardening treatment. (Solution treatments were
done prior to work hardening.)
1/4 H, 1/2 H, and H indicate increasing strengths by work hard- C
ening. III I -fiber
Data from ASM (2001) Copper and Copper Alloys, ASM Spe-
ciality Handbook, pp. 256. Materials Park, OH: ASM.
27% C, 38% S, 8% B, 18% between S and B, 3% G,
and 6% other. G B
The rolling texture of low stacking fault energy -fiber
alloys is different. The orientation {1 1 0}/1 1 2S (B)
Figure 29 Schematic representation of the ODF shown in
at F ¼ 45 , f2 ¼ 90 , f1 ¼ 35 is much stronger and
Figure 28. (Adapted from Humphreys FJ and Hatherly M
the fiber G–B is more prominent. There are two ad- (1996) Recrystallization and Related Phenomena. London:
ditional fibers. Fiber g corresponds to orientations Elsevier.)
having {1 1 1} aligned parallel to the rolling plane. It
extends from {1 1 1}/1 1 2S at F ¼ 55 , f2 ¼ 45 ,
f1 ¼ 0 to F ¼ 55 , f2 ¼ 45 , f1 ¼ 90 . Fiber t fibers. As the amount of deformation increases the a-
consists of orientations with /1 1 0S parallel to the fiber remains strong, the b-fiber increases, and the C
transverse direction. For alloys with a stacking fault and S components decrease.
energy of B40 mJ m  2, there is an additional fiber Textures are often represented by pole figures.
(t). Figure 30 shows the ODF of cold-rolled brass Figures 32a and 32b are the pole figures of copper
and Figure 31 is a schematic representation of the rolled at room temperature.
36 Alloys: Copper

1 texture depends on the stacking fault energy of the

The rolling texture also depends on the rolling
temperature. The pole figure of copper rolled at
 1961C (see Figure 34) is very similar to brass
rolled at room temperature. Figure 35 shows that the
texture transition with temperature is gradual.

The effects of annealing cold-worked metal are usu-
ally divided into three stages: recovery, recrystalliza-
tion, and grain growth. During recovery, the excess
vacancies anneal out and dislocations are rearranged
into lower-energy configurations. There are no major
changes in hardness or strength. Figure 36 shows the
changes of stored energy and conductivity during
recovery and recrystallization of copper.

Figure 30 ODF for cold-rolled brass. (Adapted from Hump- Recrystallization

hreys FJ and Hatherly M (1996) Recrystallization and Related
Phenomena. London: Elsevier.) During recrystallization after cold work, new strain-
free grains are formed. The extent of recrystallization
depends on both temperature and time. It is custom-
0° 1 90° ary to define the recrystallization temperature as the
temperature at which 50% of the grains will be re-
G B crystallized in a half-hour. For most metals of com-
90° mercial purity, the recrystallization temperature is
2 S between 1/3 and 1/2 of the melting point on an ab-
solute scale. However, the recrystallization tempera-

C D ture depends on purity, the extent of cold work


ibe before annealing, and the prior grain size. Figure 37

S -f
-fiber C D shows the kinetics of recrystallization of copper that
is 99.999% pure after cold reduction of 98%.
-fiber The fact that the recrystallization temperature of
1121C in this case is only 28% of the melting tem-
r S
fibe perature is explained by the extreme purity and very
- heavy reduction. The time, t, to achieve 50% recrys-
tallization is given by the Arrhenius equation,
Figure 31 Schematic representation of the ODF in Figure 30. t ¼ t0 exp ½4
(Adapted form Humphreys FJ and Hatherly M (1996) Recrystal-
lization and Related Phenomena. London: Elsevier.) where Q is the activation energy, R is the gas con-
stant, and T is the temperature. Figure 38 is an
Arrhenius plot of the data in Figure 37. The activat-
This texture has been described as consisting of
ion energy is 93 kJ mol  1.
ideal components of B39% {1 2 3}/41c2S, B34%
According to the Johnson–Mehl–Avrami equation,
{1 4 6}/21c1S, B14% {1 1 0}/1c12S and B13%
the fraction, f, of the microstructure recrystallized at
{1 1 2}/11c1S. Pole figures for 70:30 brass are
any temperature should be given by
shown in Figures 33a and 33b.
The relative amounts of these components change f ¼ 1  expðBtn Þ ½5
as the temperature of rolling decreases or as the cop-
per is alloyed with zinc or some other element that where the exponent n is the sum of the contributions
lowers its stacking fault energy. The transition from from nucleation nn and growth ng. If the nucleation
copper-type recrystallization texture to the brass-type rate is constant, nn ¼ 1 and if the growth rate is
Alloys: Copper 37


0.5 1
0.5 1
5 8 0.5 1 2
12 15 6
0.5 2
1 1
− 2 0.5
3 −C.D.
− −C.D.

Near (1 2 3) [4 1 2] Near (1 2 3) [4 1 2]
Near (1 4 6) [2 1 1] Near (1 4 6) [2 1 1]
(1 1 0) [1 1 2] (1 1 0) [1 1 2]
(1 1 2) [1 1 1] (1 1 2) [1 1 1]
(1 1 0) [0 0 1] (1 1 0) [0 0 1]

(a) (b)
Figure 32 The (1 1 1) and (2 0 0) pole figures of electrolytic copper rolled 96.6% (Adapted from Hu H and Goodman SR (1963) Texture
transition in copper. Trans. Met. Soc. AIME 227: 627–639.)

R.D. R.D.

200 350

500 200
400 450
300 550
200 650

(a) (b)
Figure 33 The (1 1 1) and (2 0 0) pole figures 70:30 brass cold rolled 95%. (From Hu H, Sperry PR, and Beck PA (1952) Rolling
textures in face centered cubic metals. Journal of Metals (4)1: 76–81.)

constant ng ¼ 3. Figure 39 is a log–log plot of ‘‘Commercially pure’’ copper recrystallizes at 1801C

ln(1  f) versus t of the data for Figure 34 at 1351C. while zone-refined copper recrystallizes at 801C. Al-
The slope is close to 4, indicating constant nucleation loying elements raise the recrystallization tempera-
and growth rates. ture as shown in Figure 41. The recrystallization
The effect of the amount of cold work on the re- temperature also depends on prior grain size as
crystallization of brass is illustrated in Figure 40. shown in Figure 42. With greater amounts of prior
38 Alloys: Copper


0.5 1
2 3
4 5
9 11

− −C.D.

0.5 1 2 3 4 5 6 7

(1 1 0) [1 1 2]
Near (1 4 6) [2 1 1]
Near (1 2 3) [4 1 2]
(1 1 2) [1 1 1]
(1 1 0) [0 0 1]
Figure 34 (1 1 1) Pole figure of electrolytic copper, rolled 96.6% at  1961C. (Adapted from Hu H and Goodman SR (1963) Texture
transition in copper. Trans. Met. Soc. AIME 227: 627–639.)

50 cold work, the recrystallization temperature is low-

Brass-type Copper-type ered. (see Figure 43).
The grain size resulting from recrystallization de-
40 (1 2 3) [4 1 2] pends on the relative rates of nucleation and growth
of recrystallized grains. Increasing amounts of cold
Texture composition (%)

work before recrystallization decreases the resulting

(1 4 6) [2 1 1]
grain size because it increases the nucleation rate
more than the growth rate. Solutes in solid solution
decrease the recrystallized grain size because they
decrease the growth rate by slowing grain-boundary
motion. Second-phase particles have the same effect
(1 1 0) [1 1 2] but they may increase the nucleation rate. Finally, the
grain size before recrystallization affects the recrys-
10 tallized grain size. With a finer prior grain size, there
(1 1 2) [1 1 1]
are more sites for nucleation, so the resulting grain
(1 1 0) [0 0 1]
size is also smaller.
−200 −160 −120 −80 −40 0 40
Grain Growth
Rolling temperature (°C)

Figure 35 Change texture of cold-rolled copper with tempe-

With continued heating, the larger recrystallized
rature. (Adapted from Hu H and Goodman SR (1963) Trans. grains grow at the expense of other smaller recrys-
AIME 227.) tallized grains, with the result that the grain size
Alloys: Copper 39

Range of
Range of recrystal-

Release of stored energy (relative)

recovery lization

Change in resistivity ∆ , 10−10 Ω m


100 8
Read right

Read left

100 200 300 400 500
Temperature (K)
Figure 36 Energy release and changes of conductivity on annealing of cold worked copper. The changes o3001C are caused by
annealing out of vacancies. (Adapted from Guy AC and Hren JJ (1974) Elements of Physical Metallurgy. Reading, MA: Addison-

Recrystallization (%)


60 (135°C) (113°C) (88°C) (43°C)



1 10 102 103 104 105
Time of heating (min)

Figure 37 Isothermal recrystallization of 99.999% pure copper cold-worked 98%. (Adapted from Guy AG and Hren JJ (1974)
Elements of Physical Metallurgy. Reading, MA: Addison-Wesley.)

100000 Simple theory gives the exponent 1/n as 1/2, but it

is usually much lower. Solutes and second-phase par-
ticles have a large effect.
Time (min)


100 Annealing Textures

10 ODFs for copper and copper with increasing
amounts of zinc are given in Figure 44.
0.0024 0.0026 0.0028 0.003 0.0032 The transition from copper-type recrystallization
1/T (1/K)
texture to the brass-type texture depends on the
stacking fault energy of the alloy. This correlation is
Figure 38 An Arrhenius plot of the time–temperature relation shown in Figure 45. A marked change occurs at 8%
for 50% recrystallization of the copper in Figure 37.
Zn, 3%Al, 2% Ge, or a little over 1% P. The strength
of the cube texture is reduced by light rolling reduc-
tions, large grain sizes, and low recrystallization
increases. The increase of grain size is often charac-
terized by
Figure 46 is the {1 1 1} pole figure for copper cold
1=n rolled 97% at room temperature and recrystallized at
d1=n  d0 ¼ Ct ½6 2001C. It consists almost entirely of cube texture and
40 Alloys: Copper

10 400

Recrystallization temperature (°C)

−In (1−f )

0.1 135°C

0.01 Pure Ni
1 10 100 copper
Time (min) 100
0 0.01 0.02 0.03 0.04 0.05 0.06
Figure 39 A Johnson–Mehl–Avrami plot of the recrystallization Solute content (wt.%)
of the copper in Figure 34 at 1351C. The slope is nearly 4, which
indicates that the nucleation rate is constant and there is a con- Figure 41 The effect of various solutes on the recrystallization
stant rate of growth in three dimensions. temperature of copper. (Adapted from ASM (2001) Copper and
Copper Alloys, ASM Speciality Handbook. Materials Park, OH:
Annealing temperature (°F)
400 800 1200

140 Reduction
by rolling:
Brinell hardness (ksi)

Fraction recrystallized

25% 15 µm 50 µm
100 10%
3% 0.4

80 0.2

10 100 1000
0 100 200 300 400 500 600 700 800 Time (s)
Annealing temperature (°C)
Figure 42 Dependance of recrystallization kinetics on prior
Figure 40 The recrystallization temperature of copper depends grain size. Copper cold-rolled 93% and recrystallized at 2251C.
on the amount of cold work. (Adapted from ASM (2001) Copper (Adapted from Humphreys FJ and Hatherly M (1996) Recrystal-
and Copper Alloys, ASM Speciality Handbook. Materials Park, lization and Related Phenomena. London: Elsevier.)

its twins. Other textural components of copper,

rolled at lower temperatures and recrystallized at Table 7 Texture components of copper, rolled, and recrystal-
4001C, are indicated in Table 7. The recrystallization lized at 4001C
texture of copper rolled at  1961C is very similar to
Component Percentage of texture
the recrystallization texture of cold-rolled brass (see
Figure 47). 80 C 140 C 196 C

(2 1 4)[c5 c2 3] 0 21.4 57.4

(3 1 0)[0 0 1] 0 8.1 21.3
Anisotropy (1 0 0)[0 0 1] 94.5 60.4 12.8
The anisotropy of plastic behavior of sheet material (1 2 2)[c2 c1 2] 5.7 10.1 8.5
is conventionally documented by the Lankford pa- Adapted from Hu H and Goodman SR (1952) Trans. Met. Soc.
rameter or R-value which is defined by the ratio of AIME 277: 637.
Alloys: Copper 41

contractile strains during a tension test,

450 ew
R¼ ½7
Recrystallization temperature (°C)


where w and t denote the width and thickness di-

400 rections in a tension test. Because R changes with the
direction of the tensile axis, it is common to measure
the values at 01, 451, and 901 to the rolling direction
and to define an average cR as

350 ðR0 þ 2R45 þ R90 Þ

cR ¼ ½8

Like all f.c.c. metals, copper and copper alloys have

R-values considerably o1.0. For annealed copper,
0.1 0.2 0.3 0.4 typically, R0 ¼ 0:9, R45 ¼ 0:4, R90 ¼ 0:9, and cR ¼
Total strain 0:6 or 0.7 and for annealed 70:30 brass, typically,
R0 ¼ 0:9, R45 ¼ 0:9, R90 ¼ 0:8, and cR ¼ 0:9. The
Figure 43 The dependence of the recrystallization temperature
on prior strain (J, simple tension, compression þ tension).
cube texture component lowers R45 considerably,
(Adapted from Humphreys FJ and Hatherly M (1996) Recrystal- while components with {1 1 1} parallel to the sheet
lization and Related Phenomena. London: Elsevier.) raise the R-values.


0% 5%
Zn Zn

(a) (b)

6.3% 22%
Zn Zn

(c) (d)
Figure 44 Recrystallization textures of rolled and recrystallized copper and copper alloys; (a) Cu, (b) 5% Zn, (c) 6.3% Zn, and (d) 22%
Zn. (Adapted from Humphreys FJ and Hatherly M (1996) Recrystallization and Related Phenomena. London: Elsevier.)
42 Alloys: Copper

{2 3 6}<3 8 5>

(%) Texture component



20 Zn
Al P

10 Ge Goss

0 2 4 6 8 10 12 14 16 18 20 22 24 %Zn
0 1 2 3 4 5 6 7 8 9 %Al
0 1 2 3 4 5 6 %Ge
0 1 2 3 %P
4 3 2 1 RSFE (×10−3)

Figure 45 The transition from copper to brass texture depends on the stacking fault energy of the alloy. (Cube: Zn ( ), Al ( ), Ge ( ),
P ( ); {2 3 6} /3 8 5S: (Zn ) ( ), Al ( ), Ge ( ); Goss: Ge ( ), P ( ).) (Adapted from Humphreys FJ and Hatherly M (1996)
Recrystallization and Related Phenomena. London: Elsevier.)

RD (see Figure 48) show the combinations of strains that

lead to the formation of localized necks. Minima
occur near plane strain, e2 ¼ 0. The level of e1 at
100 this point depends strongly on the strain-hardening
exponent, n in eqn [2]. Cold-worked sheets tend to
have lower forming limits. The forming limits tend to
100 rise with sheet thickness. A high strain-rate exponent
m, in
s ¼ Cem ½9

is also useful in forming. However, it is found to be

very low (0.002–0.005) for copper and brass at room
The compositions and conditions of these sheets
are listed in Table 8.
The limiting drawing ratio (LDR) for flat bottom
cups depends mainly on the R-value of the sheet
material. Because the R-values of copper and brass
sheets tend to be o1, the values of LDR tend to
Figure 46 The {1 1 1} pole figure of copper cold rolled 97% at fall between 2.1 and 2.2. Formability in sheet form-
room temperature and annealed at 2001C. n show the cube tex-
ture and the m show its twin components (1 2 2)[2 1 2]. (Adapted
ing that involve biaxial stretching depends mainly
from Hu H, Sperry PR, and Beck P (1952) Trans AIME 194.) on the strain-hardening exponent n. Because of this,
copper and especially brass have very good stretch-

Forming Extrusion and Forging

Sheet formability benefits from high strain-hardening Hot forging temperatures for most copper alloys lie
exponents and high R-values. Forming limit diagrams in the range of 600–7501C.
Alloys: Copper 43

RD Powder Processing
Parts of copper and copper alloys are sometimes
fabricated by conventional powder metallurgy that
involves pressing powder to the desired shape and
500 700 sintering to achieve nearly full density. Porous bear-
600 800 300
900 ings and filters are made by stopping the sintering
before complete densification.

In general, alloys with a greater separation of liq-
uidus and solidus are regarded as easier to cast. In a
fixed thermal gradient, the length of dendrite L, arms
is proportional to this temperature difference, DT.
The liquid-to-solid shrinkage can occur interdendri-
tically. This simplifies risering. Lower liquidus tem-
peratures are also beneficial.
The solubility of hydrogen, like other gases, is
Figure 47 The {1 1 1} pole figure of 70:30 brass cold rolled 97%
much greater in liquid copper than solid copper.
at room temperature and annealed at 3401C. n show the ideal
(2 2 5)[3 7 4] orientation and the m show the (1 1 3)[2 1 1] orien- Figure 49 shows the solubility of hydrogen in copper
tation. (Adapted from Hu H, Sperry PR, and Beck P (1952) Trans. and copper–aluminum alloys. Alloying with alumi-
Met. Soc. AIME 194.) num and some other elements decreases the solubility.

80 90
C10200 C15500
Major strain e 1 (%)

Major strain e1 (%)

70 80
60 70

50 60
C11000, lot 1 C19400
40 50
C11000, lot 2 C17200
20 40

−30 −20 −10 0 10 20 30 40 50 60 70 80 −30 −20 −10 0 10 20 30 40 50 60 70 80

(a) Minor strain e2 (%) (b) Minor strain e2 (%)

80 80
C23000 C72500
70 70 C51000
Major strain e 1 (%)

Major strain e1 (%)

60 60
C26000, lot 1
50 50
40 40
C26000, lot 2 C75200
30 30
20 20
C26000, lot 3
10 10

−30 −20 −10 0 10 20 30 40 50 60 70 80 −30 −20 −10 0 10 20 30 40 50 60 70 80

(c) Minor strain e2 (%) (d) Minor strain e2 (%)

Figure 48 Forming limit diagrams for copper and copper alloy sheets.
44 Alloys: Copper

Table 8 Alloys shown in Figure 40 and their conditions

Designation Common name Condition Thickness ðmmÞ Grain size ðmmÞ UTSa ðMPaÞ

C10200 Oxygen free copper annealed 0.66 0.014 234

C11000 lot 1 ETP copper annealed 0.74 0.016 224
C11000 lot 2 ETP copper H4 0.69 268
C15500 Silver copper annealed 0.71 0.009 288

C220000 Commercial bronze annealed 0.69 0.0067 234

C19400 HSM copper annealed 0.69 319
C17200 Beryllium copper annealed 0.25 0.019 491

C2300 Red brass annealed 0.69 0.024 293

C2600 lot 1 Cartridge brass annealed 0.64 0.025 345
C2600 lot 2 Cartridge brass H4 0.69 407
C2600 lot 3 Cartridge brass H8 0.51 531

C72500 Cu–Ni–Sn annealed 0.69 0.023 356

C51000 Phosphor bronze annealed 0.69 0.014 374
C74300 Nickel silver annealed 0.69 0.035 387
C72500 Nickel silver annealed 0.69 0.020 405
C70600 Cu–10% Ni annealed 0.81 0.016 361
UTS – ultimate tensile strength.
Adapted from Copper and Copper Alloys, ASM Specialty Handbook. Materials Park, OH: ASM.


Hydrogen solubility (mm3 kg−1)

Pure Cu Dezincification
Zinc content (%)

3.3% Al

8.1% Al


800 1000 1200
0 100 200 300 400 500
Temperature (°C)
Penetration (µm)
Figure 49 Solubility of hydrogen in copper and Cu–Al alloys.
(Adapted from Flinn R A (1963) Fundamentals of Metal Casting. Figure 50 The effect of zinc content on corrosion of brass
Reading, MA: Addison-Wesley.) when exposed to 0.01 M NH4Cl at 451C for 60 days. (Adapted
from ASM (2001) Copper and Copper Alloys, ASM Specialty
Handbook. Materials Park, OH: ASM.)
As the metal solidifies, dissolved gas is released.
For hydrogen,
H2 O þ 2Cu-2H þ Cu2 O ½10 resistance. A few specific forms of corrosion are given
below. One is pitting corrosion in which the attack is
This is likely to cause gas porosity. The source of
very localized. Pitting attacks cause much more
dissolved hydrogen may be from wet scrap or fur-
damage than the same amount of metal loss uni-
nace gases.
formly spread over the corroded surface. A second
type of corrosion is dezincification of brasses con-
taining 415% Zn. Both copper and zinc atoms go
Many of the applications of copper and copper-based into solution, but copper then precipitates as a porous
alloys depend on their generally good corrosion plug. Figure 50 shows the effect of composition on
Alloys: Copper 45

Time to failure (h)

Cu−10% Zn


280 300 320 340 360
Figure 51 A plug of dezincified metal on a brass pipe. (Adapted
1/T (K−1)
from ASM (2001) Copper and Copper Alloys, ASM Specialty
Handbook. Materials Park, OH: ASM.) Figure 53 Arrhenius plot of time to failure of copper and brass
under constant load in NH4OH in the temperature range of 40–
701C. (Adapted from ASM (2001) Copper and Copper Alloys,
ASM Specialty Handbook, Materials Park, OH: ASM.)

(a) (b)
Figure 52 Examples of stress-corrosion cracking in brass. Both condenser tubes of drawn C12200 copper. (Adapted from ASM
(2001) Copper and Copper Alloys, ASM Specialty Handbook. Materials Park, OH: ASM.)

the tendency to pitting and dezincification. An ex- See also: Deformation Processing; Mechanical Proper-
ample of dezincification is shown in Figure 51. ties: Elastic Behavior; Mechanical Properties: Plastic
Brass containing X15% Zn is susceptible to stress Behavior; Mechanical Properties: Strengthening Mecha-
corrosion or season cracking. It occurs only when the nisms in Metals; Mechanical Properties: Tensile Proper-
ties; Orientation Texture; Powder Processing: Models and
brass is under tensile stresses and specific environm-
Simulations; Solidification: Models and Simulations.
ents. For brass, ammonia is the most common agent.
Figure 52 shows examples of stress-corrosion crac- PACS: 81.05.Bx; 81.20.  n; 81.30.Dz; 81.40.  z;
king of brass. The propensity to stress corrosion in- 81.70.  q
creases with zinc content and temperature as shown in
Figure 53. Removal of residual stresses by stress-relief
Further Reading
anneals is sufficient for many applications. Hydrogen
is not a problem for most copper alloys. However, if Haasen P (1978) Physical Metallurgy. Cambridge: Cambridge
tough-pitch copper containing Cu2O is exposed to University Press.
Hu H and Goodman SR (1952) Trans. AIME 194.
hydrogen at high temperatures, the reaction is
Joseph G (1999) Copper; it Trade, Manufacture, Use and Environ-
mental Status. Materials Park, OH: ASM.
2H þ Cu2 O-H2 O þ 2Cu ½11 Mendenhall JH (ed.) (1980) Understanding Copper Alloys. Olin
Brass, New York: Wiley.
Reed-Hill RE and Abbaschian R (1994) Physical Metallurgy Prin-
Such exposure could, for example, occur during torch ciples, 3rd edn. Boston: PWS-Kent.
welding. The formation of H2O in the form of steam Smithells Metals Reference Book, London; Butterworths, 1983.
causes embrittlement. West EG (1882) Copper and Its Alloys. New York: Wiley.
46 Alloys: Iron

Alloys: Iron
M Ferry, University of New South Wales, Sydney, classed as steels or cast irons, and are by far the most
NSW, Australia widely used iron alloys but it is pertinent to note that
& 2005, Elsevier Ltd. All Rights Reserved. many others are also important. Depending on al-
loying additions and the complexity of the various
processing stages, microstructures (and properties) of
steels and cast irons vary dramatically, and include
one or more of the following phases: ferrite, pearlite,
Introduction bainite (range of structures), austenite, and marten-
Iron alloys are arguably the most important class of site as well as second-phase dispersions such as ce-
engineering materials as a huge number of types can mentite, graphite, intermetallic compounds, and
be produced by controlling the type and quantity of precipitates. A detailed treatment of the structure
alloying elements, casting method, thermal and me- and properties of Fe–C alloys is beyond the scope of
chanical processing, etc. Factors contributing to the this overview but several excellent books are given in
attractiveness of iron as the principal component of the ‘‘Further reading’’ section.
many engineering alloys include: the ability to be There are a range of essentially carbon-free iron
produced relatively inexpensively; it has a high mel- alloys that also generate microstructures with at-
ting temperature (15381C) which allows thermally tractive properties. Examples include Fe–Be and FePt
activated processes over a range of temperatures and that exploit the martensitic reaction to generate
it is allotropic thereby allowing its alloys to undergo shape memory behavior and those alloys that exhibit
several useful phase transformations to generate a important physical properties such as Fe–Ni invar
wide variety of microstructures, and mechanical and alloys with very low thermal expansion coefficients
physical properties. and multicomponent alloys such as (Fe, Co)44Zr7-
Figure 1 presents a subgroup of a much wider Cu1B15 (HITPERM) and Nd2Fe14B with exceptional
range of commercially available iron alloys. The al- magnetic and magnetomechanical properties. The
loys given in this figure contain carbon, are generally ability to generate iron alloys with such a large range

Iron−carbon alloys

C<2.11% C>2.11%

Steels Phases present Cast Irons

Dependent on alloying additions/
thermal and mechanical treatment

Ultralow & low

Ferrite Gray iron
carbon steels
Plain carbon /
Low carbon C as graphite:
Ferrite + Pearlite Ductile iron
steels flakes/nodules

Medium carbon White iron

steels Pearlite (contains Fe3C)

High carbon
Bainite (various) Malleable iron
Tool and die
High alloy steels
(exc.Cr alloys)

Stainless steels
(>12% Cr)
Ferrite + Austenite

Figure 1 Classification scheme for iron–carbon alloys (range is not exhaustive).

Alloys: Iron 47

of microstructures and properties has resulted in the Table 1 Approximate diffusivity of various interstitial and sub-
use of these materials in an enormous number of stitutional elements in iron
applications. Phase Element Activation Frequency Temperature
energy, factor, range ð CÞ
1 2 1
Q ðkJ mol Þ Do ðmm s Þ
The Production of Iron Alloys
a-Iron Fe 240 200 700–750
Iron alloy components have been produced from the C 80 0.62 700–750
liquid state for many centuries with mass production N 76 0.3 700–750
of steel commencing in the late nineteenth century. Co 226 20 700–790
Ni 258 970 700–900
Molten steel was traditionally cast into standing g-Iron Fe 284 49 900–1050
moulds but continuous casting was developed in the C 135 15 900–1050
1950s which enabled the production of huge ton- Co 364 3000 1050–1250
nages of steel. Most of the world’s steel production is Ni 280 77 930–1050
now obtained through this casting route. Following d-Iron Fe 239 190 1400–450
casting, as-cast slabs are thermal and mechanical
processed to the desired final shape. In the last few
decades, however, near-net-shape casting processes vector of the minimum energy configuration in a-iron
have been developed that are capable of producing is a=2/1 1 1S and while dislocations can theoreti-
near-final-shape products directly from liquid with cally dissociate into segments of a/0 0 1S, the stack-
secondary processing such as hot rolling reduced to ing fault energy (SFE) is so high that substantial
an absolute minimum. On a reduced scale, there are dissociation has not been observed. In contrast,
various types of novel iron alloys produced in the g-iron has a relatively low SFE ðo80 mJ m2 Þ and
amorphous, nanocrystalline, and intermetallic form. dislocations readily dissociate by the reaction
These materials are either manufactured by convent- a=2/0 1 1S-a=6 /1 1 2S þ a=6/2 1 1S, which
ional casting or by rapid solidification processing, generates a ribbon of stacking fault. The difference
spray deposition, mechanical attrition, and sintering in both crystal structure and SFE of a-iron and g-iron
or physical and chemical vapor deposition, etc. Iron strongly influences the deformation and annealing
alloys produced by these more exotic techniques are behavior of these phases.
generally restricted to specific applications. Due to the various allotropic transformations in
iron ðd2g2aÞ, many physical properties, such as
specific volume and electrical conductivity, change
Structure and Properties of Iron discontinuously through these critical tempera-
The production of pure iron is not possible, but tures. A notable consequence of the f.c.c. 2 b.c.c.
numerous studies have been carried out on the pro- transformation is the change in magnitude of the self-
perties of high-purity iron. Iron is situated toward diffusion coefficient, D, which follows an Arrhenius-
the end of the first transition series in the periodic type relationship:
table with an electron configuration 1s22s22p63s2
3p63d64s2. Iron lies at the left of the larger atomic D ¼ Do expðQ=RTÞ
species Co, Ni, and Cu which are known to crystal-
lize with a close-packed structure. To the left of where Do is a frequency factor that is virtually in-
iron are the less-filled states of Ti, V, Cr, and Mn dependent of absolute temperature (T) and Q is the
which subsequently crystallize with a body-centered activation energy for self diffusion. Table 1 shows
cubic (b.c.c.) crystal structure. Iron lies between the values of Do and Q for iron and other alloying
these groups of elements and, depending on tempe- elements. It is easily shown that the rate of self-dif-
rature, may exist either as the b.c.c. structure fusion of iron decreases by two orders of magnitude
ð1394 C4Ta o912 CÞ or the face-centered cubic when f.c.c. iron transforms to the more open b.c.c.
(f.c.c.) structure ð912 CoTg o1394 CÞ. The high- structure.
temperature b.c.c. structure is not expected under
normal circumstances but is a result of magnetic
Alloying Elements in Iron
energy effects. The b.c.c. structure exists at standard
temperature and pressure, but extreme pressures will A number of elements are added to iron to produce a
generate a hexagonal close-packed (h.c.p.) structure variety of alloys with a wide range of microstructures
(e-iron). and properties (Figure 1). Alloying elements have a
An important difference between a-iron and g-iron strong influence on the stable temperature range of
is the structure of their dislocations: the Burgers b.c.c. and f.c.c. phases. The Wever classification
48 Alloys: Iron

X = C, N, Cu, etc. X = Ni, Mn, Co, etc. 102



Diffusivity (mm2 s−1)


(a) (b)

X = B, Ta, Nb, Zr, etc. X = Si, Ti, Mo, Cr, etc.




0 1 2 3 4 5 6
(c) (d) 1000/T
Fe−X Alloys
Figure 3 Variation of average diffusivity with temperature for
Figure 2 Classification of iron alloy binary phase diagrams: (a) various substitutional and interstitial elements in a-iron.
and (b) austenite stabilizers; (c) and (d) ferrite stabilizers.

particularly in the less densely packed b.c.c. struc-

(Figure 2) has shown that most alloying elements fall ture, and are considerably more mobile than
into four categories with each generating a particular substitutional elements (Figure 3 and Table 1).
phase diagram. This figure shows that a particular Nevertheless, interstitial elements have low solubi-
alloying element in iron may either expand the g-field lity, particularly in a-iron (ferrite), due to the less
(austenite stabilizer) or contract this field (ferrite sta- favorable geometry of the b.c.c. lattice. The solubi-
bilizer). Elements may also form iron-rich com- lity of interstitial elements decreases markedly
pounds that reduce the extent of the g-field. The with temperature and obeys a similar relation as
overall behavior is a result of the different enthalpies diffusivity:
of solution in both the crystal structures of iron
which leads to either an increase or decrease of the CFeX ðTÞ ¼ A expðB=TÞ
temperature range in which austenite is stable. Al-
loying elements can also be grouped into three major where A and B are constants and X is either an
classes: those that form interstitial and substitutional equilibrium or metastable phase such as graphite,
solid solutions, and those that are almost completely Fe2B, Fe3C, and Fe3N4 in equilibrium with either
immiscible with the crystal lattices of iron. ferrite or austenite. Table 2 gives values of the max-
imum solubility of various interstitial elements in
ferrite and austenite which shows that hydrogen has
Interstitial Alloys
the highest solubility in ferrite which is demonstrated
The elements H, C, N, and O are considerably by a small value of the constant, B (not shown).
smaller than Fe and, therefore, tend to occupy inter- Table 2 also shows that the solubility of C and N in
stitial sites in both the lattices of iron. Figure 2a both forms of ferrite are considerably lower than in
shows how the g-field is stabilized for alloys con- austenite.
taining interstitial elements since the f.c.c. lattice In the Fe–C system, graphite does not usually form
provides sites of lower strain energy than does the in either ferrite or austenite due to its high activation
b.c.c. lattice. These elements have a high rate of energy of nucleation. The cementite phase (Fe3C) or
diffusion in both the crystal structures of iron, other less stable carbides tend to form instead. The
Alloys: Iron 49

Table 2 Solubility of various interstitial elements in iron into hypoeutectoid (Co0.77 wt.%) or hypereutec-
Phase Element Equilibrium Temperature of Maximum toid (C40.77 wt.%) while cast irons are classified as
phase maximum solubility either hypoeutectic (Co4.30 wt.%) or hypereutectic.
solubility ð CÞ ðat:%Þ

a-Iron H H2 905 1  2  10  3 Substitutional Alloys

B Fe2B 915 o0.02
C Fe3C 727 0.095
There are many substitutional elements that are par-
N Fe4N 590 0.4 tially or completely soluble in both the crystal struc-
g-Iron C Fe3C 1148 8.8 tures of iron. Elements with the largest solubility fall
N Fe4N 650 10.3 close to Fe in the periodic table: Cr, Ni, Co, Mn, and
d-Iron C Fe3C 1495 0.09 V. A number of elements, such as Ti, Ca, and K, have
almost no solubility in iron. For alloying elements
with an f.c.c. or h.c.p. structure, the g-field is stabi-
lized (Figures 2a and 2b) whereas the g-field is re-
duced by elements that form Hume–Rothery phases
with Cu, Ag, and Cu (Figure 2c) or when iron is
alloyed with the b.c.c. transition metals V, Ti, Mo,
1538 L W, Cr, etc. (Figure 2d). For those elements that re-
1495 strict the g-field completely, austenite cannot exist
which eliminates the possibility of generating several
1394 important decomposition products (pearlite, bainite,
or martensite). These alloys are non heat-treatable in
Temperature (oC)

the sense that phase transformations cannot be ex-

1154 ploited to control the microstructure and properties.
 To achieve the desired mechanical properties such as
yield strength, these alloys must either be work hard-

912  + Fe3C ened (which can also recrystallize to produce a fine

grain size) or elements are used to promote solid so-
738 lution strengthening. The heat-treatable alloys are far
727 more amenable to strengthening over a wide range
 by controlling the transformation products during
austenite decomposition.
 + Fe3C
Most alloys of iron usually consist of a combina-
tion of interstitial and substitutional elements which
results in various complex interactions involving at-
0 0.09 0.77 2.11 4.30 6.67 om species. For example, the rate of diffusional
Carbon (wt.%) transformation is usually controlled by the more
Figure 4 Schematic diagram of the metastable (constitutional) sluggish substitutional elements (as shown by the
Fe–Fe3C and the stable (equilibrium) Fe–graphite diagram lower diffusivity of these elements, Figure 3) which
(dashed) showing important temperatures and compositions. favors the formation of nonequilibrium microstruc-
tures following austenite decomposition. In addition,
elements can combine to produce enhanced
difference between the equilibrium Fe–graphite and strengthening by the generation of atom complexes
metastable (constitutional) Fe–Fe3C phase diagram is (Mn–C, etc.) in the lattice or from the formation of
shown in Figure 4, which illustrates the subtle dif- alloy carbides ([FeaXbYcy]Cj).
ferences in the phase boundaries. Fe–Fe3C is prob-
ably the most important binary system with the
following array of characteristics: (1) various types
Phase Transformations in Iron Alloys
of invariant phase transformations are possible (pe- The allotropic transformations that occur in pure
ritectic, eutectic, and eutectoid) and (2) the solubility iron can result in a number of important phase trans-
of C in a-iron at 7271C is only 0.022 wt.% whereas formations in its alloys. As indicated in Figure 2,
g-iron can accommodate 0.77 wt.% C at this tem- a given concentration of a ferrite stabilizer can
perature. Carbon levels below B2.11 wt.% are clas- eliminate the austenite phase field, whereas an au-
sified as steels while higher carbon alloys are stenite stabilizer will expand this field to allow the
classified as cast iron. Steels are classified further alloy to undergo several useful transformations. For
50 Alloys: Iron

example, Figure 4 shows that a number of transfor- promote the formation of graphite rather than ce-
mations are possible in the Fe–C system. The follow- mentite (Figure 1). Similar to steels, cast irons can
ing brief discussion is devoted mainly to this system generate a wide range of microstructures and prop-
as carbon is the most important alloying addition to erties depending on alloying additions, cooling
iron. behavior, etc.
During solidification, the peritectic reaction can It is clear that the transformation products that
occur over a certain compositional range. For the form during austenite decomposition in iron-base al-
peritectic composition (0.09 wt.% C), the following loys are expected to be diverse and are influenced by
invariant reaction can occur at 14951C: L-L a large number of processing and material-related
þd-g. In this reaction, liquid þ d-iron transforms variables. A particular transformation is achieved by
isothermally to austenite and has several important the control of the type and quantity of alloying el-
consequences during continuous casting of steels. ements, the cooling rate through the transformation
At higher carbon levels, that is, 0.77 wt.% C, the range, and thermomechanical treatments prior to or
eutectoid reaction occurs at 7271C: g-a þ Fe3 C. For during transformation, etc. In addition to the classic
slow cooling rates, austenite (0.77 wt.% C) decom- phase transformations associated with Fe–C, several
poses via a diffusional transformation to produce additional transformations are known to occur in
pearlite which is a lamellar structure containing suc- iron alloys. The ability of the alloys of iron to un-
cessive layers of ferrite (0.022 wt.% C) and cementite dergo such a diverse range of phase transformations
(6.67 wt.% C). With increasing cooling rate, the for- is a major factor contributing to their immense pop-
mation of diffusional products such as pearlite and ularity as an engineering material.
ferrite is suppressed thereby generating nonequilib-
rium microstructures ranging from a series of bainitic
structures to martensite. Thermomechanical Behavior
At very high cooling rates, diffusional transforma-
An understanding of deformation of ferrite and au-
tion products do not form since atomic diffusion is
stenite is extremely important as these phases un-
necessary for nucleation and growth of these phases.
dergo a significant amount of plastic deformation
This results in the formation of martensite; a term
during thermomechanical processing, the most com-
not restricted to iron-base alloys but used to define
mon method for producing useful final components.
the product of a nondiffusional phase transformation
During deformation, many microstructural changes
both in metals and ceramics. The martensitic trans-
occur: the original grains change shape and an inter-
formation has the following important characteris-
nal substructure forms, texture changes take place,
tics: each atom retains its original neighbors and
precipitation may occur, dynamic recovery (DRV) or
there is no interchange among the atoms; the trans-
dynamic recrystallization (DRX) processes are pos-
formation does not involve individual atomic jumps
sible and the constituent particles may fracture and
characteristic of diffusion-controlled and interface-
redistribute. The most notable material factors that
controlled transformations; the reaction is diffusion-
affect both the deformation microstructure of iron
less which means the martensite product has the
alloys include SFE, crystal structure, initial grain size
same composition as the parent phase. In Fe–C al-
and shape, and the size, shape, and volume fraction
loys, the martensite transformation generates a hard,
of a second phase.
brittle phase whereas, for essentially carbon-free al-
loys, the martensite is ductile.
Cold Deformation
For very low carbon levels (o0.022 wt.%), Fe–C
may undergo a series of precipitation reactions after Cold deformation is restricted to low homologous
cooling through the solvus where supersaturated car- temperatures (To0.5Tm), where Tm is the absolute
bon in ferrite decomposes at low temperatures to melting temperature. During cold deformation of ei-
produce a series of precipitates. This can result in ther ferrite or austenite, there is a marked increase in
considerable hardening that may or may not be use- dislocation density resulting in considerable work
ful depending on the desired application of the alloy. hardening and grains subdivide in a complex manner
For carbon levels greater than B2.11 wt.%, Fe–C to produce a range of features such as a cellular
undergoes the eutectic reaction: L-g þ Fe3 C. At the substructure, microbands, deformation twins, defor-
eutectic composition (4.3 wt.% C), liquid iron trans- mation bands, and larger-scale heterogeneities such
formations take place at 11471C to produce austenite as shear bands. In high SFE b.c.c. a-iron, slip is the
and cementite with subsequent cooling to room tem- principal deformation process and occurs in the
perature promoting additional phase changes. It is close-packed /1 1 1S direction but the slip plane
pertinent to note that alloying additions such as Si may be any of the planes {1 1 0}, {1 1 2}, or {1 2 3}. In
Alloys: Iron 51

contrast, g-iron has a relatively low SFE

(o80 mJ m  2) which results in deformation by slip
on {1 1 1}/0 1 1S systems as well as by twinning.
The different modes of deformation result in differ- DRX
ences in the work-hardening rate of these phases as
well as in the development of deformation textures. -iron

Hot Deformation

Hot deformation is carried out at T40.5Tm and is -iron
distinguished from cold working by the absence or
near absence of strain hardening and lower matrix DRV
dislocation content. Deformation becomes more
homogeneous with increasing temperature with the
frequency of microstructural inhomogeneities such as
microbands, transition bands, and shear bands re-
duced considerably. The driving pressure for soften-
ing by recovery and recrystallization (either dynamic c p
or static) is the stored energy of deformation. The
process of recovery reduces the internal energy of the Strain
metal through mechanisms such as annihilation dis- Figure 5 Schematic diagram of typical flow curves associated
locations and their re-arrangement into low-angle with the hot deformation of ferrite (a-iron) and austenite (g-iron).
grain boundaries.
The relatively low SFE of austenite does not allow
include the mode of deformation, strain, strain rate,
for substantial DRV whereas DRV occurs extensively
and temperature of deformation. In general, both
in ferrite since dislocation climb occurs readily and,
SRX and DRX are initiated by nucleation of strain-
for a given temperature, the rate of diffusion of iron
free grains at microstructural inhomogeneities such
atoms in ferrite is B100x greater than in austenite.
as transition bands, shear bands, and grain bound-
In ferrite, a significant proportion of the stored
aries. Cold-deformation of fine-grained ferrite results
energy is released by DRV which reduces the driving
in nucleation predominantly at prior grain bounda-
force for both dynamic and static recrystallization
ries but a coarse grain size results in substantial shear
(SRX). Since the reorganization of dislocations is
band nucleation, particularly in those iron alloys
more difficult in austenite, DRX can occur readily.
containing high levels of interstitial elements.
These differences in the dynamic restoration proc-
The most desirable nucleation sites in cold-rolled
esses generate two distinctive stress–strain (flow)
a-iron are grain boundaries as these nuclei often have
curves, Figure 5. Deformation of austenite results in
orientations where /1 1 1S is almost parallel to the
an increase in flow stress via work hardening but at a
normal direction (ND) of the sheet. This is significant
critical strain (ec), there is sufficient stored energy to
since a strong /1 1 1S8ND texture in cold-rolled
initiate DRX which tends to eventually decrease the
and annealed steels results in superior sheet form-
flow stress since the rate of work hardening is offset
ability. Austenitic alloys recrystallize in a similar
by the softening caused by recrystallization. In con-
manner to ferritic alloys, that is, nucleation occurs
trast, extensive DRV in ferrite leads to a balance be-
mainly at grain boundaries but other microstructural
tween dislocation generation and annihilation which
heterogeneities characteristic of deformed austenite
rapidly results in a steady-state flow stress. There is
also play an important role. Due to the physical
also insufficient stored energy to dynamically recrys-
processes associated with slip and twinning during
tallize this phase. During hot working, successive
cold rolling, austenitic alloys do not produce a high
cycles of DRX in austenite have the capability of
fraction of favorable /1 1 1S8ND nuclei resulting in
refining the grain size whereas grain refinement in
reduced formability.
ferrite can only be achieved by SRX.

Static Annealing
Recent Developments in Iron Alloys
The principal material factors influencing static an-
Magnetic Alloys
nealing in iron alloys include composition, initial
grain size, initial texture, and second-phase precip- Iron is ferromagnetic below the Curie temperature
itates. Processing variables are also important and and these alloys exhibit a remarkably wide range of
52 Alloys: Iron

Table 3 Some notable ferromagnetic iron alloys showing typical properties

Material Initial relative Hysteresis loss per Saturation induction ðBHÞmax ðkJ m3 Þ
permeability ðme Þ cycle ðJ m3 Þ ðWb m2 Þ

Soft magnetic alloys

Commercial iron ingot 1.5  102 270 2.14
Fe-3Si (oriented silicon iron) 1.5  103 40–140 2.0
Fe-45Ni (PERMALLOY 45) 2.7  103 120 1.60
Fe73.5Si13.5B9Nb3Cu1 (FINEMET) 5  103–2  105 1.0–1.2
Fe88Zr7B4Cu1 (NANOPERM) 1  104–1  105 1.5–1.8
Fe44Co44Zr7B4Cu1 (HITPERM) 2  103–3  104 1.6–2.1

Hard magnetic alloys

Fe-1Mn-0.9C (martensitic) 3.8
Fe-3.5Cr-0.9C-0.3Mn (martensitic) 5.0
Fe-24.5Co-13.5Ni-8Al-2Nb (Alnico-XII) 76.8
Fe60Pt40 120
Nd2Fe14B (oriented) 320–400
Nd2Fe14B (sintered nanophase) 450

magnetic properties. Magnetic materials are broadly 2.4

classified into two main groups with either soft or Fe−Co alloys
HITPERM alloys
hard magnetic characteristics. Soft magnetic alloys
Saturation induction (Wb m−2)

have high permeability, low coercivity, and low hys- 2.0 NANOPERM alloys
teresis loss whereas hard magnetic (or permanent)
Fe−C alloys
magnetic alloys have a sufficiently large resistance to
demagnetizing fields due to their high coercivity, re- Silicon irons
manence, and energy product.
Table 3 gives some properties of a select range of
ferromagnetic alloys which includes some of the
more recently developed alloys that possess either an 1.2
amorphous, or a nanocrystalline phase embedded in
Amorphous alloys
an amorphous matrix. These materials are usually FINEMET alloys
produced as ribbons or fibers by rapid solidification 0.8
techniques and are often followed by a series of heat 102 103 104 105 106
treatments. Soft magnetic alloys based on these Relative permeability (
structures are generally complex and contain ele-
Figure 6 Estimated relationship between permeability and sat-
ments such as Cr, Mo, and Al and varying amounts
uration induction for some soft magnetic iron alloys.
of metalloids (B, Si, C, and P) that help to form the
glassy phase. These alloys have low anisotropy com-
bined with high resistivity (which reduces eddy cur- that specifies its performance characteristic, that is,
rent losses) and have high permeability and low its resistance to demagnetization, is the maximum
power losses. The most widely investigated alloys in energy product (BH)max where B and H are the
this new class of soft magnetic materials are magnetic induction and magnetic field strength, re-
Fe73.5Si13.5B9Nb3Cu1 (FINEMET), Fe88Zr7B4Cu1 spectively. To date, the highest values of (BH)max are
(NANOPERM), and Fe44Co44Zr7B4Cu1 (HIT- obtained in oriented Nd–Fe–B alloys (320–
PERM) which are produced by rapid solidification 400 kJ m  3) and sintered fine-grained versions of
and crystallization annealing to produce a nanocrys- these alloys (B450 kJ m  3).
talline structure. Compared with more conventional
iron alloys, these materials have very high perme-
Shape Memory Alloys
abilities and large inductions (Figure 6).
Permanent magnetic iron-base materials include a Iron alloys can exploit the martensitic reaction to en-
variety of alloys and intermetallic compounds. The able a component to be deformed at one temperature
most widely studied materials in recent years are with subsequent recovery of the original shape upon
neodymium–iron–boron (Nd2Fe14B) and iron–plati- heating. This so-called shape memory effect (SME) is
num alloys. For permanent magnets, a parameter associated with the ability of some martensites to
Alloys: Iron 53

undergo thermal reversion on reheating which, on a following phases: martensite, bainite, acicular ferrite,
macroscopic scale, corresponds to up to 15% strain and polygonal ferrite.
recovery. There are various types of SME including
pseudoelasticity, one- and two-way shape memory
See also: Ferromagnetism; Magnetic Materials and Ap-
behavior and magnetic field-induced shape memory. plications; Powder Processing: Models and Simulations;
Iron-base shape memory alloys are produced either Solidification: Models and Simulations; Alloys: Overview;
by conventional casting and thermomechanical pro- Mechanical Properties: Elastic Behavior; Mechanical
cessing, or mechanical attrition and sintering, or Properties: Plastic Behavior; Recovery, Recrystallization,
rapid solidification processing. While the various and Grain Growth; Mechanical Properties: Strengthening
types of shape memory phenomena occur in alloys Mechanisms in Metals; Diffusionless Transformations;
of the type Fe–Mn–Si–Cr, FeBe, FePt, and Fe–Pd–Pt, Phase Transformation; Phases and Phase Equilibrium.
they are largely under development and more work is
required to explore the potential of shape memory
PACS: 61.43.Dq; 61.44. þ w; 61.50.Ks; 61.66.Dk;
iron alloys. 61.72.Cc; 62.20.  x
Intermetallic Compounds
Iron alloys in the form of B2-type ordered interme-
tallic compounds based on the stoichiometric ratios,
Further Reading
FeCo, FeAl, and Fe3Al, have received considerable Dunne DP (2000) Functional memory metals. Materials Forum 24:
attention in high-temperature structural applications 95–108.
due to their low density, high electrical resistivity, Gladman T (1997) The Physical Metallurgy of Microalloyed
Steels. London: Institute of Materials.
high strength at temperature, and excellent resistance Honeycombe RWK (1985) The Plastic Deformation of Metals.
to oxidation and corrosion in various aggressive London: Edward Arnold.
chemical environments. These alloys are either Honeycombe RWK and Bhadeshia HKDH (1995) Steels – Micro-
thermomechanically processed in a manner similar structure and Properties. London: Edward Arnold.
to metals or produced by more novel processing Hume-Rothery W (1967) The Structure of Alloys of Iron. UK:
Pergamon Press.
routes. While binary iron-base intermetallics are in- Humphreys FJ and Hatherly M (2004) Recrystallization and
herently brittle and have low creep resistance, con- Related Annealing Phenomena, 2nd edn. Oxford: Elsevier
siderable work is being carried out to improve these Science.
critical properties both by alloy design and by care- Jiles DC (2003) Recent advances and future directions in magnetic
materials. Acta Materialia 51: 5907–5939.
fully controlling the various processing stages.
Leslie WC (1981) The Physical Metallurgy of Steels, 1st edn. Lon-
Direct Strip Cast Alloys don: McGraw Hill.
McHenry ME and Laughlin DE (2000) Nano-scale materials
There have been substantial developments in the development for future magnetic applications. Acta Materialia
mass production of iron alloys by direct strip casting 48: 223–238.
Manohar PA, Ferry M, and Hunter A (2000) Direct strip casting of
(DSC) which produces as-cast strip products of a steel: historical perspective and future direction. Mater Forum
thickness less than 2 mm. Iron alloys currently pro- 24: 15–29.
duced by DSC include carbon and alloy steels, stain- Manohar PA, Ferry M, and Chandra T (2001) Recrystallization of
less steel, iron–silicon alloys, and cast iron. This ferrite and austenite. In: Buschow KHJ et al., (eds.) Encyclope-
dia of Materials Science and Technology, vol. 4, pp. 3019–3024.
casting process often produces a microstructure that
Oxford: Elsevier.
is markedly different from the same material pro- Munroe PR (2000) Intermetallic compound development for the
duced by conventional casting and thermomechani- 21st century. Mater Forum 24: 5–18.
cal processing (TMP). This is a result of the high Ray RK, Jonas JJ, and Hook RE (1994) Cold rolling and annealing
solidification rates that tend to produce far-from- textures in low and extra low carbon steels. International Ma-
equilibrium microstructures. By a careful control of terials Review 39: 129–172.
Sinha AK (1989) Ferrous Physical Metallurgy. Boston: Butter-
casting parameters and alloying additions, as-cast worths.
low carbon steel strip can be generated with a final Zackay VF and Aaronson HI (eds.) (1962) Decomposition of Au-
microstructure consisting of one or more of the stenite by Diffusional Processes. USA: TMS-AIME.
54 Alloys: Magnesium

Alloys: Magnesium
H Jones, University of Sheffield, Sheffield, UK especially when Fe, Ni, or Cu are present as precip-
& 2005, Elsevier Ltd. All Rights Reserved. itated impurities (Figure 1).
There are more than 20 elements with maximum
equilibrium solid solubility exceeding 1 at.% in
magnesium. Of these, eight are used in commercial
magnesium alloys (Al, Zn, Mn, Zr, Li, Ag, Y, Nd)
Fundamental Characteristics together with Si, Cu, Ca, and Ce. Al and Zn have
Magnesium has a density of 1.74 mg m  3 at 201C, large effects in reducing the lattice parameters of
melts at 6501C, has a Young’s modulus of 45 GPa, magnesium, and also raise the critical resolved stress
and is hexagonal close-packed in crystal structure. for slip; so potent are solid solution hardeners in
Slip at ambient temperatures occurs mainly on the magnesium. Al is a less potent precipitation hardener
basal plane (0 0 0 1) /1 1 2% 0S with pyramidal and than Zn or Nd, because it precipitates incoherently
prismatic planes becoming more active with increa- (e.g., as Mg17Al12 at grain boundaries) rather than
sing temperature. Deformation of room temperature via Guinier–Preston zones within the grains. The
therefore involves twinning also (initially on small grain size has a specially marked hardening
ð1 0 1% 2Þ/1 0 1% 2S), especially when a compressive effect on magnesium (Figure 2), approximately dou-
stress acts parallel to the basal plane. ble that in aluminum, for example; the small grain
Pure magnesium, in the absence of moisture, is size also improves ductility.
protected by an adherent thin film of MgO. This Magnesium and its alloys are commonly melted in
MgO reacts with any moisture present to form mild steel crucibles because of their low solubility for
Mg(OH)2 which breaks down in the pressure of iron. The melt surface, however, must be protected
acids which then attack the magnesium. Saltwater by a flux or inert atmosphere. Grain refinement of
is a particularly harsh environment for magnesium, the cast microstructure can be achieved by treating

100 d (µm)
10 4 2 1 0.5 0.3 0.2 0.1

Corrosion rate (mg cm−2 day−1)

60 600
y (MPa)



0 1 2 3

d −1/2 (µm−1/2)
0 0.01 0.02 0.03
Iron content (wt.%) Figure 2 Effect of grain size d on yield strength sy of high-
strength magnesium alloys compared with AZ91 alloy ingot ().
Figure 1 Effect of iron content on corrosion of pure magnesium Key: (J) AZ91, (&) Mg–Al–Zn–Re base, (n) Mg–Y–MM, (v)
produced by alternate immersion in 3% NaCl solution. (Repro- Mg–Ca–Al, (r) Mg–Cu–Y alloy rapidly solidified particulate ex-
duced with permission from Hanawalt JD, Nelson CE, and trusions. (Reproduced with permission from Jones H (1995) Key
Peloubet JA (1942) Trans. AIME 147: 273–299.) Engineering Materials 97/98: 1–12; & Trans Tech Publishing Ltd.)
Alloys: Magnesium 55

with CCl6 (when Al is present) or with Zr (when it is AZ 91B
not) prior to casting. The nucleant is expected to be Brittle
Al4C3, and Zr itself, in the two cases.
Most magnesium alloy components are produced
by high-pressure die casting, using hot chamber ma- 6

Al (wt.%)
chines except for the largest castings (up to 10 kg).
They can also be sand-cast if certain precautions are 4
taken, for example, to inhibit reaction with the sand
and moisture. Ingots are produced by Direct Chill 2 Hot
casting, either for remelting or as feedstock for cracking
wrought products.
ASTM alloy designations identify the main alloy- 0 5 10 15 20 25 30 35
ing elements by letters followed by numbers indica- Zn (wt.%)
ting nominal wt.% of each rounded up to a single
Figure 3 Effect of Al and Zn contents (wt.%) on die castability
digit, e.g., AZ91 ¼ Mg–9 wt.%Al–1 wt.%Zn. Letter
of Mg–Al–Zn alloys. (Reproduced from Foerster GS (1976) In:
codes for the main alloying elements are A ¼ Al, Proceedings of the 33rd Annual Meeting of the International
C ¼ Cu, E ¼ rare earths, K ¼ Zr, L ¼ Li, M ¼ Mn, Magnesium Association, Montreal, pp. 35–39.)
Q ¼ Ag, S ¼ Si, W ¼ Y (yttrium), and Z ¼ Zn.

Casting Alloys Alloy A
Alloy B
AZ91 (9.5Al, 0.5Zn, 0.3Mn, wt.%) is the most 160
A380 Alloy C
common of the Mg–Al-based die casting alloys, ac-
Stress (MPa)

hieving ultimate tensile and yield strengths of 215 120

and 120 MPa, respectively, in the fully heat treated
(T6) condition. High-purity versions (maximum AE42
0.004Fe, 0.001Ni, and 0.015Cu and minimum
0.17Mn, wt.%) are available when corrosion resist- AZ91D
ance is paramount. Reduced Al-contents, such as in 40
AM60 (6Al, 0.3Mn, wt.%) give increased ductility
and fracture toughness, when required. Die cast 0
60 100 140 180 220 260
Mg–Al–Zn alloys are susceptible to hot cracking at
T (°C)
intermediate Zn contents as shown in Figure 3. Im-
proved creep resistance results from added Si (AS21 Figure 4 Effect of test temperature T on stress for 0.1% creep
and AS41: 2 or 4Si, 0.4 or 0.3Mn and 1Si, wt.%) or strain in 100 h for cast alloys based on the Mg–Al system com-
pared with aluminum-based casting alloy A380. Alloys A to C
rare earths (AE42: 4.0Al, 2.5RE, 0.3Mn, wt.%) in
are developmental Mg–Zn–RE alloys. (Reproduced from Villa
which the formation of Mg2Si and AlxRE com- Fanjul F, Srimanosaowapak S, Mc Nee KR, Greenwood GW, and
pounds is more effective in pinning grain boundaries Jones H (2003) The effect of Nd substitution for CeMM on creep
than is the relatively soluble Mg17Al12. ZE41 (4.2Zn, performance of M–2.5mm–0.35Zn–0.3Mn (MEZ) alloy. Z. Metall-
1.3RE, 0.7Zr, wt.%) combines precipitation harden- kunde 44: 25.)
ing from Zn and Ce with grain refinement by Zr to
allow service up to 1501C. EZ33 (2.7Zn, 3.2RE,
Wrought and Powder Metallurgy Alloys
0.7Zr, wt.%) and QE22 (2.5Ag, 2.5Nd, 0.7Zr,
wt.%) offer creep resistance or high proof strength Working of magnesium is carried out at 300–5001C
up to 2501C, again with grain refinement by Zr. by extrusion, rolling, or forging. Basal planes and
WE54 and WE43 (5 or 4.1Y, 3.25 heavy rare earths, /1 0 1% 0S directions tend to orient in the direction of
0.5Zr, wt.%) combine high strength at room and working which can result in proof strengths in com-
elevated temperatures with better corrosion resist- pression that are 0.5–0.7 of those in tension. This
ance than other high-temperature magnesium-based difference is attributable to the occurrence of twin-
alloys. The development of a die-castable magne- ning more readily under compression, and is reduced
sium-based alloy that has creep resistance compara- at small grain sizes.
ble with A380 aluminum casting alloy (Al–8.5 AZ31 (3Al, 1Zn, 0.3Mn, wt.%) is the most widely
Si–3.5Cu–0.5Mn, wt.%) for automotive engine used wrought sheet magnesium-based alloy, compa-
applications is the focus of major R þ D activity in rable in proof and tensile strength with cast AZ91,
Europe, Japan, and the USA currently (Figure 4). but with high ductility and being weldable. ZK31
56 Alloys: Magnesium

800 ∧
ε = 5.6 × 104 s−1 0.25 Vf
300 0.16 Vf
Mg87.5Cu5Y7.5 0.05 Vf

Alternating stress, ± (MPa)

AZ9l (FHT)
Tensile yield strength, 0.2 (MPa)


400 100

200 104 105 106 107

WE54-T6(Mg-Y-Nd) (a) Number of cycles, N

Process parameters
RPM (pin on disk) 250 min−1
40 load 7N
0 pin diameter 6mm
300 400 500 600 retarding roller 42Cr MoV
Temperature (K) Wear (weight loss) (10−3 g) no lubrication
Figure 5 Temperature dependence of tensile yield strength of
Mg–5Cu–7.5Y (at.%) and Mg–20Al–10Ca (at.%) rapidly solidified
alloy powder extrusions compared with WE54-T6 conventionally
processed alloy. (Reproduced from Inoue A and Masumoto T 20
(1993) Material Science and Engineering A 173: 1–8, with AZ91 [extrusion] 83 HV lO
permission from Elsevier.) AZ91 + SiC 135 HV lO
AZ91 + TiB2 140 HV lO
10 AZ91 + Ti (C,N) 150 HV lO
(3.0Zn, 0.6Zr, wt.%) is stronger, but at the expense AZ91 + AIN 130 HV lO
AZ91 + Al2O3 • platelets
of weldability. LA141 (14Li, 1.2Al, 0.15 minimum 140 HV IO
Mn, wt.%) has lower strength than either of these
0 20 40 60 80
but has very low density (1.35 mg m  3) because of
the high Li content which confers a duplex hexago- (b) Time (min)
nal close-packed/body-centered cubic microstruc- Figure 6 Effects of increasing volume fraction Vf of Saffil
ture. Limitations are poor corrosion resistance and (a-Al2O3) short fibers and 15 vol.% of various ceramic particulates
susceptibility to overaging in service which results in on (1) fatigue resistance and (2) wear resistance of AZ91 alloy.
((a) Reproduced from Llorca A et al. (1991) Material Science and
excessive creep at relatively low loads. Sheets of all of
Engineering A 135: 247–252, with permission from Elsevier.
these alloys can be hot-formed readily at 230–3501C, (b) Reproduced from Schröder J et al. (1989) Proc. ECCM3,
but cold formability is limited. pp. 221–226, with permission from Elsevier.)
AZ61 (6.5Al, 1Zn, 0.3Mn, wt.%) and ZK61 (6Zn,
0.8Zr, wt.%) are examples of extrusion alloys, the
latter giving a tensile strength exceeding 300 MPa in of these compositions exhibit the phenomenon of
the T5 condition, at the expense of weldability. Tensile ‘‘bulk metallic glass formation,’’ where section thick-
strengths can be raised to 4500 MPa in EA55RS nesses of up to several millimeters can be produced in
(5Al, 5Zn, 5Nd, wt.%) by rapid solidification proc- the glassy state directly by chill casting.
essing, which involves planar flow casting to sheet Finally, an increased Young’s modulus, fatigue,
o0.1 mm thick, followed by pulverization and then and wear resistance have been achieved in metal
consolidation by vacuum hot pressing to provide a matrix composites based on alloys such as AZ91
bulk feedstock for extrusion, forging, or rolling. The with ceramic reinforcements (Figure 6).
very high strengths obtained are mainly attributable to
ultrafine grain size (down to 0.3 mm) stabilized by
finely dispersed rare-earth aluminides. Strengths as
high as 740 MPa have been developed in partially or Figure 7 summarizes the consumption of extracted
completely devitrified initially glassy Mg–Cu–Y or magnesium in the Western world in 1999. This shows
Mg–Al–Ca rapidly solidified alloys (Figure 5). Some that some 43% was used as an alloying addition to
Alloys: Overview 57

Chemical 1% Wrought products 1% wheel cores, lock housings, inlet manifolds, cross car
Nodular cast iron 2% Other 3% beams, and instrument panel substrates. Predictions
Electrochemical 3% are that the annual demand for magnesium alloy die
castings should exceed 200 kt by the end of 2004. A
further growth in general usage of magnesium alloys is
likely to depend on (1) expanding the range of alloys
steel 11% Aluminum with good creep resistance to higher temperatures and
alloying even better corrosion resistance, (2) further developm-
ent of new casting techniques such as thixoforming
and squeeze casting, (3) improving the properties of
Pressure wrought magnesium alloys and their formability, in-
die castings
36% cluding superplastic forming and diffusion bonding,
and (4) continued development of advanced magne-
sium-based materials, and their metal matrix com-
posites, by such techniques as mechanical alloying,
rapid solidification, and other powder technologies.
Figure 7 Breakdown of usage of magnesium in the Western
world in 1999.
See also: Electron Gas (Theory); Mechanical Properties:
Strengthening Mechanisms in Metals; Phase Transfor-
aluminum followed by 36% for magnesium-based mation; Phases and Phase Equilibrium; Recovery, Re-
alloy die castings, and 13% to desulfurize liquid steel crystallization, and Grain Growth; Solidification: Models
and to nodularize graphitic cast irons, with smaller and Simulations.
percentages for electrochemical, chemical, wrought
alloy and other purposes. A comparison with the
corresponding data for 1991 shows an increase in to- PACS: 81.05.Bx; 71.20.Dg
tal shipments from 250 to 380 ktons (kt) (involving a
7% annual growth rate since 1993) due mainly to
an increase in demand for magnesium alloy die cas- Further Reading
tings from 38.6 kt (15%) to 133.4 kt (36%), mostly Avedesian MM and Baker H (eds.) (1999) Magnesium and Magne-
for applications in the automotive industry (driven sium Alloys. Speciality Handbook. Materials Park, OH: ASM
by weight reduction requirements), but also for com- International.
ponents of computers, laptops, cell phones, etc. Typ- Neite G, Kubota K, Higashi K, and Hehmann F (1996) Magne-
sium-Based alloys. In: Matucha KH (ed.) Materials Science and
ical automotive applications include cylinder head Technology, vol. 8, Weinheim: VCH.
covers, transfer case and transmission housings, seat Polmear IJ (1995) Light Alloys, 3rd edn., ch. 5. London: Arnold
frames, brake and clutch pedal brackets, steering (see also Material Science and Technology (1994), 10: 1–16).

Alloys: Overview
M Baricco, Università di Torino, Torino, Italy Crystal Structures in Metals and Alloys
& 2005, Elsevier Ltd. All Rights Reserved. Chemical Bonds and Crystal Structures

Material properties are related to crystal structure,

which depends not only on the chemistry of constit-
uent atoms, but more strongly on the nature of
chemical bonds. In fact, the electronic configurations
Metallic alloys constitute a fundamental class of of different atoms determine different kinds of chem-
engineering materials. Their properties are related to ical bonds. Though the concept of chemical bond has
electronic and crystallographic structures, as well as been developed for molecules, it can also be applied
to microstructures. In this article, the fundamental to solid state chemistry. The types of chemical bonds
aspects of metallic alloys are described. For details on involved in solids may be distinguished as follows:
single classes of metallic alloys, the reader is ad- ionic, covalent, metallic, van der Waals, and hy-
dressed to specific sections. drogen bonds. Metals and alloys are mainly based on
58 Alloys: Overview

the metallic bond, where valence electrons are delo- solid solution of C in f.c.c. Fe is observed in auste-
calized over the whole crystal, even if some contri- nite. Complete solid solubility may be obtained only
bution from covalent and ionic bonds may be in alloys when components have the same crystal
observed in special cases. Alkali metals are purely structure and the size factor is lower than B8%.
metallic bonded, whereas metals close to metalloids, Au–Ni and Ag–Au–Pt are examples of systems
such as zinc, have increasing covalent bond contri- showing complete solid solubility.
butions. Sometimes the distribution of atoms in a solid so-
Geometrical factors also play a crucial role in de- lution deviates from randomness. When similar at-
fining the crystal structure of metals and alloys. Be- oms group themselves preferentially, a clustering
cause of the isotropic nature of the metallic bond, the effect is observed. On the other hand, when an atom
structure of metals may be described in terms of is preferentially surrounded by different atoms, the
hard-sphere packing. In the case of pure metals, non- solid solution is said to show a short-range ordering.
interpenetrating equal spheres are interconnected so By means of rapid solidification techniques, the
that the volume is filled as much as possible. Con- disordered structure of the liquid phase may be re-
sidering a simple layered structure of spheres, tained in the solid, giving a metallic glass. The for-
hexagonal close-packed (h.c.p.) and face-centred cu- mation of amorphous alloys is due to the strong
bic (f.c.c.) crystal structures may be easily built up. undercooling of the liquid, which gives an increase in
On the other hand, considering a slightly lower the viscosity up to the glass transition.
packing factor, a body-centred cubic (b.c.c.) structure
is formed. The crystal structure of metals is often
Intermediate Phases
changed on heating (allotropy), as observed in iron,
which shows a transition from b.c.c. to f.c.c. at When the limit of solid solubility is exceeded on al-
9101C and a further transition from f.c.c. to b.c.c. at loying, a second phase is formed. It may be the pri-
14001C before melting at 15391C. The crystal struc- mary solid solution of the alloying elements, such as
tures of metals follow the hard-sphere packing as in simple eutectic systems, but more often it appears
long as pure geometrical factors are concerned. As as an intermediate phase. When only metallic com-
soon as covalent bonding contributions and elec- ponents are present in the system, the intermediate
tronic effects become important, less dense and more phase is called an intermetallic compound.
complicated structures are observed, such as in Sn. The simplest intermediate phase is due to long-
Similarly, different packing factors are observed in range ordering of components in the solid solution
metals with a similar crystal structure. For the same (ordered phases). In this case, crystallographic posi-
reason, metals showing an h.c.p. structure may have tions are preferentially occupied by a specific element
a different ratio between lattice constants (c/a), so that an ordering parameter may be defined. Dif-
which turns out to be equal to 1.63 and 1.86 for ferent sublattices constituted by single components
Mg and Zn, respectively. may be defined in the structure, as evidenced by dif-
In metallic alloys, the constituent elements may be fraction techniques. The ordering parameter changes
simply mixed in a solid solution or may form an continuously in a limited temperature range, accord-
intermediate phase. In the case of alloys, the differ- ing to the second-order thermodynamic transition.
ence in electronegativity between constituent atoms Ordering of phases is effective in improving mechan-
becomes the main factor in defining the nature of the ical properties and it is fundamental in superalloys.
chemical bond. In the case of a high difference in When the atoms constituting the intermediate
electronegativity, an ionic contribution to the chem- phase show a particular value of the valency elec-
ical bond may be observed. As a consequence, rather tron concentration (e/a), electron phases are formed.
complicated crystal structures may be formed. Cry- For instance, 50% of Zn atoms in a Cu matrix give
stallographic parameters relative to various alloys an electron concentration of 1.50, where a b.c.c.
are available in structural databases. b-phase appears. The complex cubic g-phase and
h.c.p. e-phase are formed for electron concentrations
equal to 1.62 and 1.75, respectively. The formation
Solid Solutions
of electron phases is based on the empirical Hume–
Solid solutions may be distinguished as substitu- Rothery’s observations, later explained in terms of
tional, when the volume of constituent elements is density of states for valency electrons. Electron phas-
similar, and interstitial, when a volume difference es show a metallic behavior and exist over a range of
(size factor) higher than B15% is observed. A typi- compositions, such as in brass and bronze.
cal example of substitutional solid solution of Zn When the intermetallic phase contains a large
in Cu is observed in brass, whereas an interstitial number of vacant lattice sites, a defect phase may be
Alloys: Overview 59

formed in a limited composition range. For instance,

in the ordered NiAl phase, the presence of vacancies
in the Ni sublattice leads to an increase in the Al-

Gibbs' free energy

concentration up to B54%. Vacancy phases play an
important role in superalloys. -Phase
Very common intermetallic compounds are known
-Phase ∆Ga
as Laves phases. They have a general formula AB2
and may have a cubic structure (MgCu2) or G
hexagonal structure (MgZn2 and MgNi2). For the ∆G
Laves phases, the radii of components have the ratio G
rA =rB ¼ 1:225, so that a very high packing density is
achieved. Laves phases have recently received atten- Arrangement of atoms
tion because of high hydrogen absorption perform-
ances, useful in modern batteries. Figure 1 Free energy as a function of the arrangement of at-
oms in a phase transformation.
The interaction of transition metals (M) with small
nonmetallic elements (X) generally leads to the for-
mation of interstitial compounds, such as hydrides, becomes possible if the activation barrier ðDGa Þ is
borides, nitrides, and carbides. In these compounds, overcome and the new phase is formed at a reason-
the metal atoms form the matrix and the alloying able rate.
elements are in interstitial positions. If the ratio be- The kinetics of a phase transformation is strongly
tween atomic radii rX =rM is lower than 0.59, simple related to the mechanism. In fact, when the cry-
compound structures are obtained, where the tran- stallography and/or the chemical composition of the
sition metal forms an f.c.c. or h.c.p. lattice. For ce- two phases are significantly different, a recon-
mentite (Fe3C) rX =rM ¼ 0:61, so that a complex structive transformation occurs and a motion of
orthorhombic structure is formed. Carbides and ni- single atoms is necessary. The whole kinetics of phase
trides are very useful for strengthening in alloyed transformation may be controlled either by the di-
steels. ffusivity or by interface migration. A typical example
When the components of the intermediate phase is given by perlite formation in eutectoidic steels. On
have a strong difference in electronegativity, ionic the contrary, a cooperative movement of neighboring
contributions to the chemical bond lead to the for- atoms occurs in displacive transformations, which
mation of valency compounds, which are essentially are diffusionless and generally very fast, because no
nonmetallic. They have a fixed stoichiometry and are activation barrier is present. The formation of mar-
often formed between transition metals and metal- tensite in steel by quenching represents an example of
loids. Silicides are an example of such a type of such a kind of phase transformation.
phase. The kinetics of phase transformation in metallic
In recent years, rather complex intermediate phas- alloys is often described in terms of time–tempera-
es have been continuously discovered. Frank–Kasper ture–transformation curves (TTT), which are a grap-
phases show a huge number of atoms in the unary hical representation of the time necessary to start and
cell. Quasicrystalline phases show a fivefold symme- to complete isothermally the phase transformation
try, which was considered forbidden according to the in a specific temperature range. In the case of cool-
classical crystallographic rules. ing treatments, continuous-cooling-transformation
curves (CCT) are more appropriate. In order to
draw thermal treatments in steels, databases of TTT
Phase Stability and Transformations in and CCT diagrams are available. The kinetics of
Metallic Alloys phase transformations may be described in terms of
the Avrami equation y ¼ 1  expðktÞm , where y is
Thermodynamics and Kinetics of the transformed fraction, t is the time, k is a tem-
Phase Transformations
perature-dependent rate constant, and m is an em-
Phase stabilities and transformations in metallic al- pirical parameter ranging from 0.5 up to 4, which
loys are ruled by thermodynamic and kinetic factors. depends on the type of transformation.
As schematically shown in Figure 1, in order to
transform a phase ðaÞ into a phase ðbÞ, a gain in free
Free Energy and Phase Diagrams
energy is necessary. This free energy difference
ðDG ¼ Gb  Ga Þ represents the driving force for The free-energy of a phase depends on several fac-
the phase transformation. The phase transformation tors, such as composition, temperature, pressure,
60 Alloys: Overview

strain, surface, and interfacial energy. In order to 1200

define the equilibrium condition for a generic mul-
ticomponent system, the free energy of all phases
must be known as a function of composition, tem-

Temperature (K)
perature, and pressure. The thermodynamic equilib-
rium is reached when the chemical potential of all
elements is the same in all phases (common tangent
rule). The number of equilibrium phases is defined by 800
the well-known Gibbs’ phase rule and the amount
of each phase can be calculated according to the
lever rule. The description of the equilibrium phases 600
is generally reported as composition–temperature (a)
plots, known as phase diagrams.
In the case of a simple binary metallic system, the 1200
equilibrium thermodynamics may be estimated by
means of suitable models. For instance, the regular
solution model considers only solution phases, where

Temperature (K)
the free energy is described by means of an ideal
entropy term and a temperature-independent term
for the enthalpy of mixing. This term considers
the chemical interactions between the constituent 800
elements and it turns out positive in the case of
repulsion and negative in the case of attraction.
When the enthalpy of mixing becomes zero, the so- 600
lution is ideal. On the basis of the regular solution 0.0 0.2 0.4 0.6 0.8 1.0
model, several binary phase diagrams may be calcu- (b) Molar fraction (B)
lated, giving solubility or immiscibility either in the Figure 2 Examples of binary phase diagrams calculated ac-
liquid or in the solid phases. Examples of phase cording to the regular solution model. (a) Ideal solution; (b) In-
diagrams calculated with the regular solution model teraction parameter equal to 104 J mol  1 and 3  104 J mol  1 for
are reported in Figure 2. the liquid and solid phases, respectively.
As an example, a full miscibility for the liquid and
solid phases is shown in Cu–Ni, simple eutectics are B
observed in Cu–Ag and in Al–Si, and a single pe- 0.0
ritectic is observed in Ag–Pt. Intermediate phases are
shown in the phase diagrams as single lines (line 0.2
compounds) or as compounds with limited solubility, 0.8
such as in bronze and brass. The equilibrium condi-

tions for a ternary system may be represented by 0.4

the Gibbs’ triangle, where isothermal sections of the
phase diagrams are reported as a function of


the composition of components. The method for 0.4

plotting compositions in a ternary phase diagram is
shown in Figure 3. For multicomponent systems, 0.8
only sections of the phase diagram are usually re-
ported (pseudo-binary).
Phase diagrams are generally determined experi- A
0.0 0.2 0.4 0.6 0.8 1.0 C
mentally. Physical properties of alloys with different
compositions are followed as a function of temper-
ature, giving an experimental evidence of transition Figure 3 Representation of compositions for ternary alloys.
temperatures and equilibrium points. Calorimetric The vertices represent 100% of A, B, or C. The binary systems
techniques are used for the determination of thermo- are represented in the boundaries of the triangle. Ternary alloys
are reported in the middle of the triangle. The distances of the
physical quantities. Modern computing techniques
points from the boundaries represent the relative proportions of
allow the description of the thermodynamics of mul- the components indicated in the opposite vertex of the triangle.
ticomponent systems of industrial interest. For in- The point indicated in the figure indicates the composition with
stance, by means of the CALPHAD (CALculation of 20% of C, 50% of A, and 30% of B (A5B3C2).
Alloys: Overview 61

PHAse Diagrams) method, the free energy of all

phases is described as a function of temperature
and composition through parameters obtained by
an assessment of experimental data. Databases of
parameters are available for metallic alloys of indus-
trial interest, allowing the calculation of equilibrium
conditions, phase diagrams, and thermodynamic

Microstructure of Metallic Alloys

Properties of metallic alloys are often related to the 6 µm

microstructure, which depends on phase transforma-
tions and mechanical treatments involved in the (a)
processing. Examples of microstructures are shown
in Figure 4.
Reconstructive phase transformations usually pro-
ceed by nucleation and growth mechanisms. In the
case of a single phase, the grain size determines the
grain boundary area, which affects the strength
through the Hall–Petch relationship. Grain size is
ruled by solidification rate, plastic deformation and
recrystallization treatments. Eutectic and eutectoid
transformations give a lamellar microstructure. The
significant composition difference between the parent
and the product phases needs atomic diffusion, so the
10 µm
interlamellar spacing is related to the temperature of
the phase transformation. As an example, the tem-
perature of thermal treatments in steels selects the (b)
final microstructure, ranging from coarse perlite to
fine bainite. Similar fine microstructures may be also
obtained by tempering of martensite, previously ob-
tained by quenching. For light alloys, precipitation
hardening is related to the microstructure. Metastable
phases with interfaces coherent to the matrix may be
produced by suitable aging treatments of supersatu-
rated solid solutions obtained by quenching. Similar
microstructural effects are observed in spinodal de-
compositions, where composition fluctuations occur
up to the gradient of concentration (uphill diffusion).
In recent years, very fine microstructures may be ob-
20 µm
tained by suitable processing, giving nanostructured
alloys. As an example, ball milling of metallic pow-
ders, electrodeposition or severe plastic deformation
of ingots, may give a grain size down to few tenths of Figure 4 Examples of microstructures. (a) Microstructure in
spheroidal cast iron: black zone is the nodular graphite, sur-
nanometers, leading to improved properties with re-
rounding gray areas is ferrite, and the light area is perlite. (b)
spect to coarse-grain materials. Microstructure in eutectic Pb26Sn74: white areas are Pb-rich solid
solution and dark areas are the Sn-rich solid solution. (c) Micro-
structure in as-cast Al87Ni7Ce6: white area is Al11Ce3; gray area
Properties of Metallic Alloys is Al3Ni; dark area is f.c.c. Al. (Courtesy of Rizzi P, University of
Torino, Italy.)
Mechanical Properties

Synthesis and processing of metallic alloys is mainly thermomechanical treatments. Basic mechanical
aimed at mechanical applications. In fact, metallic properties of metallic alloys are generally determined
materials show a very broad range of mechanical by standard tensile tests, leading to the classical
properties, which may be often modified by suitable stress–strain curves shown in Figure 5.
62 Alloys: Overview

1000 and a strengthening in the alloy is observed. For su-

peralloys, a precipitation of the ordered phase may
y be induced in the disordered matrix, so that dislo-
750 cations may be blocked at the antiphase boundaries.
Stress (MPa)

Often the yielding point cannot be identified easily,

500 so the stress corresponding to a plastic deformation
of 0.2% ðs0:2%y Þ is considered. In the case of an-
nealed low-carbon steels, a nonuniform yield phe-
250 nomenon may be observed, with the resulting
production of Luder’s lines at the surface. In this
Slope = E case, an upper-ðsuy Þ and a lower-ðsly Þ yield stress are
0 5 10 15 identified. As long as a uniform plastic deformation
Strain (%) acts in the metallic alloys, a hardening effect is
Figure 5 Example of a stress–strain curve for steel (0.4% C,
observed, according to the general equation s ¼ Ken ,
4% Cr). Basic mechanical parameters are indicated: E, Young where s is the true stress, e is the true strain, K is the
modulus; sy , yield stress, UTS, ultimate tensile stress, sf , fracture strength coefficient, and n is the strain-hardening ex-
stress. ponent. The last parameter measures the rate at
which an alloy becomes strengthened as a result of
plastic deformation. When the plastic deformation
Elastic properties are obtained from the elastic becomes nonuniform, a neck begins to form in the
range of tensile tests and they are mainly related to test specimen and an apparent maximum is observed
the composition of metallic alloys. The Young mod- in the nominal stress–strain curve. Necking phenom-
ulus (E) ranges from B40 GPa from Mg-based alloys ena are very dangerous for metallic alloys aimed at
up to more than 400 GPa for W and Os. The Young high-plastic deformation, such as in deep-drawing.
modulus for Al-based alloys is B70 GPa, for Cu- Certain metallic materials, under special deformation
based and Ti-based alloys is B100 GPa, and for fer- conditions, behave as superplastic alloys, so they re-
rous alloys is B200 GPa. Metallic alloys have a shear sist necking and show uniform elongations up to
modulus (G) B3/8 of E and a Poisson ratio (v) of 1000 times the normal amount.
0.33. After thermomechanical treatments and Fracture in metallic alloys may be ductile or
changes of the microstructure, the elastic properties fragile. In ductile fracture, a significant plastic defor-
remain rather constant, so that they may be con- mation acts before rupture and a characteristic cup–
sidered an intrinsic property of the alloys. The Young cone fracture is observed at the surface of the test
modulus of metallic alloys generally decreases as a specimen. Brittle fracture is more dangerous, because
function of temperature ðdE=dTo0Þ. For special ap- it happens suddenly without prior evident plastic de-
plications, such as precision springs, specific compo- formation. Fracture may be intergranular when it
sitions have been developed showing dE=dTE0 follows the grain boundaries or intragranular when it
(Elinvar). crosses the interior of the grains. Fracture by
The strength and elongation properties may be cleavage is often observed instead of by shear. A
obtained from the plastic range of the stress–strain ductile-to-brittle transition may be observed in a
curve. The yield stress ðsy Þ represents the transition limited temperature range by standard impact tests
from the elastic to the plastic regime. It depends on (Charpy type). For instance, for ferritic steels it oc-
the dislocation mobility, so it can be significantly curs below room temperature and may be crucial for
modified by thermomechanical treatments. The low-temperature applications, so that austenitic steel
strengthening mechanisms in metallic alloys are re- is preferred. The tendency for crack propagation is
lated to the hindering of dislocation movements. This defined by the fracture toughness (Kc). It is obtained
goal may be obtained by alloying, because the pres- from a specific test, where increasing stresses are ap-
ence of alloying atoms may generate a stress field. In plied to a pre-cracked specimen. Depending on the
work-hardening processes, the presence of a high- mode of loading, different fracture toughness pa-
dislocation density leads to an entanglement between rameters are defined (I, II, and III).
crossing dislocations. For precipitation-hardening, Fracture is generally pictured as consisting of two
widely used in light alloys, the presence of precipi- stages: crack formation and growth. Crack forma-
tates with interfaces coherent with the matrix pro- tion may be due to processing or use of the alloy. It
duces efficient stress fields against dislocation generally happens at the surface and it is often re-
movements. With the reduction of the grain size, a lated to repeated stresses in the elastic range, a phe-
dislocation pile-up occurs at the grain boundaries nomenon known as fatigue. Due to this reason, for
Alloys: Overview 63

metallic alloys it is necessary to define an endurance

limit, which is about one half of the tensile strength:

Electrical conductivity (%)

below this value, the alloy can withstand an unlim-
ited number of stress cycles without fracture. Unfor-
tunately, some alloys (e.g., Al-based alloys) do not
show this limit, so that the fatigue strength must be
defined for a given number of cycles.
Hardness of metallic alloys is determined by spe-
cific tests, which are based on the resistance to pen-
etration. A commercial hardness tester forces a small
sphere, pyramid, or cone into the body of the alloy
by means of an applied load. Hardness numbers may
be roughly related to tensile strength. When the ap- Cu Cd Ag Zn Sn Al Sb Mn As Si Co Fe P
plied load is small, the hardness may be determined
Figure 6 Effect of the addition of 0.1 wt.% of various elements
for single phases evidenced in the microstructure of in solid solution on the electrical conductivity of copper.
the alloy. For instance, a resolution down to 200 mm
may be obtained by the microhardness Vickers test.
For applications at high temperature, the creep applications. The primary applications arise in fer-
phenomena may occur in metallic alloys. They are romagnetic alloys, based on Fe, Ni, Co, and some
due to the occurrence of a plastic strain when the rare-earth metals. On the basis of the values of
alloy is stressed in the elastic regime. Creep is related magnetization induced in the material when subject-
to dislocations climbing induced by the high temper- ed to a magnetizing field, represented graphically by
ature. For moderate stresses, grain boundary sliding the hysteresis loop, ferromagnetic metallic alloys are
may even occur, because of atomic diffusion. Creep distinguished in soft and hard magnets.
may be avoided in superalloys, where precipitates of Soft magnetic alloys are characterized by high
ordered phases hinder the dislocation movements. magnetic permeability, low coercive field, and low
core losses. For these materials, a high saturation
magnetization is generally desirable. These materials
Electrical and Magnetic Properties
are suitable for electrical motors, transformers, and
Electrical properties of metallic alloys are closely relays, because of the fast response of magnetization
linked to the electronic structure, and the conduction to the applied magnetic field. In Fe, there is a strong
of electrons depends on partially filled bands. Ac- relationship between permeability and crystallogra-
cording to Mathiessen’s rule, the electrical resistivity phic directions, so that the magnetic flux runs better
of alloys increases linearly with temperature near along the (1 0 0) direction (i.e., the edge of the cubic
room temperature and above, although the behavior unary cell), giving low-energy losses. For this reason,
at very low temperature is more complex. Electrical commercial steel containing 3% Si is processed so
resistivity is strongly enhanced by the presence of that a preferred orientation of grains is obtained
foreign atoms in a solid solution in the metal matrix. (Goss texture). Amorphous alloys show still lower
As an example, the effect of the addition of 0.1 wt.% power losses and they find applications in small-sized
of different atoms in Cu is shown in Figure 6. transformers. Magnetic shields may be obtained
It is clear that the effect is strongly dependent on using materials with very high magnetic perme-
the electronic configuration of the foreign atom. ability, such as Fe–Ni alloys.
Oxygen has a very big effect and, for the production On the contrary, hard magnets have a high co-
of copper wires, oxygen-free material is necessary. ercive field and a high saturation magnetization.
Cold rolling increases resistivity only slightly and it is A parameter representative of the magnetic energy
often used as a means for strengthening alloys for that can be stored per unit of mass is given by the
electrical conductors. Metallic alloys behave as su- maximum external energy, (BH)max, which is calcu-
perconductors only at very low temperatures and lated on the demagnetizing portion of the hysteresis
they have been fully replaced by high-temperature loop. These materials are mainly used as permanent
oxide superconductors for practical applications. magnets, often used to convert electrical energy to
Thermoelectric effects of metallic alloys are widely mechanical motion (electrical engines) or to convert
used for devices such as thermocouples, thermore- mechanical motion to electrical energy (micro-
sistors, and Peltier junctions. phones). Defects able to hinder the movements of
The magnetic properties of metallic alloys are dislocations are usually also able to hinder the
of fundamental interest and have several industrial movement of magnetic domains, increasing the
64 Alloys: Overview

magnetic hardness of the material. So, high carbon See also: Alloys: Aluminum; Alloys: Copper; Alloys: Iron;
and alloyed steels are often used as permanent Alloys: Magnesium; Alloys: Titanium; Diffusionless Trans-
magnets. In recent years, new metallic alloys, char- formations; Intermetallic Compounds, Electronic States
acterized by a strong magnetic anisotropy in the of; Ionic Bonding and Crystals; Mechanical Properties:
Anelasticity; Mechanical Properties: Creep; Mechanical
unary cell, have been developed. These materials
Properties: Elastic Behavior; Mechanical Properties:
contain rare earths as alloying element, such as in
Fatigue; Mechanical Properties: Plastic Behavior; Me-
SmCo5 and in Fe14Nd2B. chanical Properties: Strengthening Mechanisms in Met-
als; Mechanical Properties: Tensile Properties; Metallic
Chemical Properties Bonding and Crystals; Phase Transformation; Phases
Metallic alloys are not chemically stable. In fact, and Phase Equilibrium; Recovery, Recrystallization, and
metallurgical processes devoted to the production of Grain Growth; Thermodynamic Properties, General; Thin
Films, Mechanical Behavior of; van der Waals Bonding
metals from their oxides are often counter-balanced
and Inert Gases.
by corrosion and oxidation, which progressively
transform the metallic alloys in more stable com-
pounds. Corrosion is a wet phenomenon, which is PACS: 61.66.Dk; 61.82.Bg; 71.22. þ i; 71.55.Ak;
possible only in the presence of humidity. The cor- 73.61.At; 74.70.Ad; 75.20.En; 75.47.Np; 78.30.Er;
rosion resistance of metallic alloys is generally relat- 78.66.Bz; 81.05.Bx; 81.30.Bx; 81.40.Cd; 81.40.Ef;
ed to the properties of parent elements, according to 81.40.Rs
the electromotive series. A control of the microstruc-
ture of the alloys may change the mechanism of cor-
rosion, which should not be localized but uniformly Further Reading
distributed on the whole surface. In several cases, the
Ashby MF (1999) Materials Selection in Mechanical Design. Ox-
addition of elements prone to passivity (Cr, Al) is
ford: Butterworth-Heinemann.
used to increase the corrosion resistance of alloys, Christian JW (1975) The Theory of Transformations in Metals and
such as in the case of stainless steel. Corrosion might Alloys. Oxford: Pergamon Press.
be also avoided by deposition of thin layers of Guy A and Hren JJ (1974) Elements of Physical Metallurgy. Rea-
a metal on the surface of the alloy, such as in ding: Addison-Wesley.
Haasen P (1996) Physical Metallurgy. Cambridge: Cambridge
zinc coating and tinning of steel. Oxidation is a
University Press.
high-temperature process, due to the reaction of the Saunders N and Miodownik AP (1998) CALPHAD: A Compre-
metallic alloy with the atmosphere. At the very hensive Guide. New York: Elsevier Science.
beginning of the reaction, a thin layer of the oxide is (1998) Metals Handbook Materials Park: ASM International.
formed at the surface. If the volume of the oxide is at
least as great as the volume of the metal from which
it formed, the oxide is protective and oxidation pro- Nomenclature
ceeds slowly. On the contrary, if the volume of oxide
is less than this amount, the layer is not continuous (BH)max maximum external energy (J m  3)
E Young modulus (GPa)
and it is less effective in preventing the access of
G free energy (J mol  1), shear modulus
oxygen to the alloy surface, so that the reaction rate (GPa)
becomes high. Often the simultaneous formation of k rate constant (s  1)
two or more oxide layers is observed, such as in steel, K strength coefficient (MPa)
where a variable amount of FeO, Fe3O4, and Fe2O3 m Avrami exponent (dimensionless)
layers are formed at different temperatures. n strain-hardening exponent (dimension-
Materials Selection r atomic radius (m)
t time (s)
For specific applications, it is a combination of ma- T temperature (1C)
terial properties (material index) that characterizes UTS ultimate tensile strength (MPa)
the performance. Properties of alloys are available in v Poisson ratio (dimensionless)
databases, so that merit indices, combined with Ash- y transformed fraction (dimensionless)
by’s charts, allow optimization of the material selec- e strain (dimensionless)
tion process. s stress (MPa)
Alloys: Titanium 65

Alloys: Titanium
T R Bieler, R M Trevino, and L Zeng, Crystallographic Structure of Phases in
Michigan State University, East Lansing, Titanium Alloys
& 2005, Elsevier Ltd. All Rights Reserved.
Titanium exists in two allotropic crystal forms, a,
which is a hexagonal structure with a c/a ratio of
B1.59 (slightly squashed compared to the close
packed ratio of 1.63), and b, which has the body-
centered cubic (b.c.c.) crystal structure. In pure
titanium, the a-phase is stable up to the b transus
Introduction temperature (bT), 8831C. The a-phase is transformed
Titanium and titanium alloys offer a unique combi- upon heating above 8831C to the b.c.c. b-phase. The
nation of physical and mechanical properties that idealized orientation relationship between a (thin
makes titanium an excellent choice for applications lines) and b (dashed lines) is illustrated in Figure 1.
that require high strength at temperatures below However, the actual lattice spacings for a and b do
5501C, good stiffness (104–133 GPa), toughness, not match perfectly. Figure 2 shows how the atomic
and corrosion resistance. Another desirable charac- positions of these phases are related to each other
teristic of titanium is its relatively low density of based on a common origin using lattice constants
4.54 g cm  3, which is between iron (7.87 g cm  3) from a Ti–6Al–4 V alloy. With heating, the a-b
and aluminum (2.70 g cm  3). The high specific transformation requires lattice strains of B10% ex-
strength along with good corrosion resistance pansion along ½2 1% 1% 0a to become a ½1 0 0b direc-
makes titanium an enabling material in the aero- tion, B10% contraction along the ½0 1 1% 0a to
space industry (e.g., airframe components, rotors, become ½0 1 1 % b , and B1% contraction along
and compression blades in jet engines), and it is ½0 0 0 1a that becomes ½0 1 1% b . Due to this imper-
important in the growing biomedical engineering fect geometrical relationship, one of the two
field due to its biocompatibility with living tissues. /1 1 1Sb directions in the {1 1 0} plane will be ro-
Titanium alloys have historically been too expensive tated B51 to align itself with one of the /2% 1 1 0Sa
for the auto industry, but titanium alloys have been directions, leading to six possible variants of b from
recently used for exhaust systems, valves, and one a orientation, as shown in Figure 3a. This means
springs (its high-yield strength and moderate mod- that two b variants can be misoriented from another
ulus value gives titanium alloys tremendous elastic by B101. Thus, b variants arising from one parent a
resilience). Titanium aluminide intermetallic alloys crystal can be misoriented from each other by B101,
are likely to become important in the future, due to 501, 601, or 701 rotation about a /0 1 1Sb axis (a
their lower density and superior high-temperature prior /0 0 0 1Sa axis).
Titanium ranks ninth as the most plentiful ele-
<c> [0001] II [011] II
ments, and is the fourth most abundant structural
metal in the Earth’s crust exceeded only by alumi-
num, iron, and magnesium. Despite this abundance, {0001} II {011} II {001}′′
the difficulty in processing titanium makes it ex-
pensive (B$50/kg for Ti, compared to $4/kg for Al
and $1/kg for Fe); yet it is a strategic metal due to  Hexagonal
which its corrosion resistance and specific strength  Body centered cubic
outweigh its high cost. The cost is likely to drop ″ Orthorhombic
in the next decade, as new processing methods that
reduce the energy requirements and allow conti-
nuous processing are scaled up.
The crystallography is considered first, to establish <a>
[2110] II [100] II [100]′′
how phase transformations between low- and high-
[111] II [1210] II [110] ′′
temperature crystal structures occur. From this, the
<b> [010]′′
rationale and the complexity surrounding alloying
[001] [010]
and processing strategies can be appreciated. Finally,
a few problematic issues associated with titanium Figure 1 Idealized orientation relationship between a, b, and a00
alloys are noted. crystal structures.
66 Alloys: Titanium

1.1 B101 about the c-axis. An example of a 901 misori-

Viewed along entation from the same parent b crystal is shown in
0.9 the c -axis Figure 3b.
or beta (011)
plane normal Although there are 12 variants of the a orientation
0.7 possible for each parent b orientation, most prior b
b direction (nm)

grains will only choose a few of the 12 a variants

0.5 during the a–b transformation process. Recent re-
BCC {011} search has shown that the preferred transformation
0.3 Orthorhombic variants are those that have higher populations of
60° dislocations present on {1 1 0}b planes. Therefore, it
0.1 Beta
54.7° Alpha" is possible to control the transformation process by
Alpha controlling the dislocation density in the b-phase.
−0.2 0 0.2 0.4 0.6 0.8 1 The dislocation density depends on dislocation gene-
a direction (nm) ration (plastic deformation) and annihilation (recov-
ery) processes that are sensitive to prior deformation
1.1 history, heating rates, and time at temperature. The
understanding of what controls variant selection is
Viewed along
0.9 the b -axis, just emerging, and it is likely to be exploited prof-
prism itably in the coming decade to allow the manufac-
plane normal,
c direction (nm)

0.7 or beta turing processes to be designed to achieve optimized

(110) normal properties for a given application.
0.5 In addition to a-and b-phases, there are a number
BCC {110} of subtle variations, including a0 and a00 marten-
0.3 sites that may form during quenching, and ordered
Hexagonal phases such as a2- and o-phases. With quenching,
0.1 Hex prism there is no opportunity for diffusional segregation of
alloying elements; so the resulting hexagonal super-
−0.1 Alpha"
−0.2 0 0.2 0.4 0.6 0.8 1 Alpha saturated a0 has lattice constants slightly different
a direction (nm)
from equilibrium a. The a00 -phase has an ortho-
rhombic crystal structure, which is outlined in bold
Figure 2 Atomic lattice positions of the a, b, and a00 crystal
lines in Figure 1. This crystal structure is between
the a and the b crystal lattice (Figure 2). The a2-
phase (Ti3Al) is an ordered version of the hexagonal
lattice with Al atoms spaced regularly on the
hexagonal lattice, which increases the resistance to
dislocation motion. There has been much interest in
ordered orthorhombic intermetallic Ti–Al–Nb alloys
in the past decade. The o-phase is a different
hexagonal ordered-crystal structure that could
~50° ~70° form in prior b-regions after long-term aging or in
a-regions after stress-induced transformation from
a-o. Ordered phases resist dislocation motion,
(a) (b)
which can increase the strength, but the continuous
o-phase is known to reduce the ductility and facili-
Figure 3 (a) Possible b {0 1 1} plane orientations based upon tate crack nucleation.
parent a during a-b transformation, and (b) two possible a
orientations based upon parent b during b-a transformation.
Figure 2 shows that the atomic arrangement of
the a-, a00 -, and b-phases are very similar; but small
differences in atomic position lead to different cry-
With the reverse b-a transformation upon cool- stal structures that have significantly different pro-
ing, a /  2 1 1 0Sa vector can align with one of the perties. Thus, phase boundaries in Ti alloys are
two /1 1 1Sb directions on each of the six {1 1 0} typically coherent or semi-coherent. These phase
planes in the b-phase. Since the angles between boundaries are quite easily observed in polished and
{1 1 0}b planes are either 601 or 901, the c-axes of etched microstructures, often giving the impression
daughter a can point in as many as six directions, and of a fine grain size, but it is important to realize that
each direction has two variants that are rotated the domains of similar orientations are often much
Alloys: Titanium 67

larger than they appear, and these ‘‘invisible’’ meso-

textural features can account for early dwell fatigue
crack nucleation.
Unlike steels, the martensitic transformations in
titanium result in small volume changes and modest
shears, but the paradigm of using martensitic trans-
formations in steels to manipulate microstructures
and properties is also used in titanium alloys, though − − −
(a) {1 0 1 0} < 1 2 1 0> (b) {0 0 0 1} <1 2 1 0>
the details are completely different (e.g., a00 is softer Prism slip has 3 systems Basal slip has 3 systems
than a). Ordered intermetallic alloys based on Ti3Al
and TiAl can also be manipulated with similar
strategies as those used in titanium alloys, but again,
the details differ. The technology of titanium alloys is
highly dependent on the ability to control the phase
transformation from the b- to the a-phase using
strategic alloying and heat treatment history, which
gives titanium alloys flexible and designable proper-
ties similar to that possible with steels. − − − − −−
(c) {1 0 1 1} < 1 2 1 0> (d) {1 0 1 1} < 2 1 1 3>
3 systems 6 systems

Effects of Crystal Orientations on

(1010) (0110) (1100)
Mechanical Properties (1012) (1011)

The hexagonal crystal structure has considerable (2113)

(0110) (1123)
elastic and plastic anisotropy, which are strongly (2111)

affected by processing history. Young’s modulus is (2112)

highest (143 GPa) along the c-axis, and lowest
(104 GPa) in any direction in the basal plane. Tita- − − − − − −−
(e) {1 0 1 2} <1 0 1 1}> (f) {2 1 1 2} <2 1 1 3>
nium deforms with eight slip and twinning systems Tensile twinning Compressive twinning
are illustrated in Figure 4. Each of these systems shear of 0.169 shear of 0.224
require different critical resolved shear stresses to = tan−1 (9.6°) = tan−1 (12.6°)
operate, making the plastic deformation process
much more complicated than cubic metals. Prism (1010) (0110)

slip in /aS directions is the easiest, leading to plas-

tic deformation that leaves the crystal dimension (1123) (1011)

along the c-axis and the c-axis orientation un- (0110)
changed. The next easiest slip system is basal slip (2112)
in /aS directions, which also does not change the (1100)
dimension of the crystal along the c-axis, but it −− − − − −
causes crystal rotation about a prism plane normal (g) {2 1 1 1} <2 1 1 6> (h) {1 0 1 1} <1 0 1 2>
Tensile twinning Compressive twinning
axis. It is possible to have /aS slip on pyramidal shear of 0.628 shear of 0.103
planes as well. = tan−1 (32.1°) = tan−1 (5.9°)
To change the crystal dimension along the c-axis, Figure 4 (a) Prism slip, (b) basal slip in the /aS direction, (c)
twinning or /c þ aS slip on pyramidal planes is re- /aS slip, (d) /c þ aS slip on pyramidal planes, and (e–h) twin-
quired, but the critical resolved stress for /c þ aS ning systems; all types have 6 systems. Only /c þ aS slip or
slip is high, so twinning systems are usually more twinning causes a change in crystal dimension in the c direction.
easily activated. Consequently, Ti crystals are intrin- Arrows illustrate axis of crystal rotation due to slip; only prism slip
does not change the orientation of the c-axis.
sically hard when stressed along the c-axis. Further-
more, the relative strengths of these deformation
systems change with temperature, for example, twin-
ning does not occur at elevated temperatures. Con- respect to the resolved shear stresses, any crystal
sequently, the distribution of crystal orientations can can be either hard or soft. In Figure 5, shades of
have a large impact on both elastic and plastic prop- gray represent the Taylor factor (proportional to the
erties. Even pure titanium behaves like a composite plastic flow stress of the crystal) in a rolled plate
material; depending on the crystal orientation with deformed in uniaxial tension.
68 Alloys: Titanium

ND RD 100 µm

Ti−6AI−4V Taylor Factor 4.5 (black) 11.1 (white)

Figure 5 Rolled plate (rolling direction and plate normal are indicated) SEM and Orientation Imaging Microscopy Taylor factor maps
before (inset) and after annealing above b transus, followed by controlled slow cooling. Taylor factor was computed based upon uniaxial
deformation in the rolling direction with CRSS ratios of 0.7:1:3 for prism:basal:pyramidal /c þ aS slip. (Images obtained at the Air Force
Research Laboratory, WPAFB, OH.)

The Classification of Titanium Alloys potent b-stabilizer:

Titanium alloys are classified according to the phases Alequiv ¼ %Al þ 13ð%SnÞ þ 16ð%ZrÞ þ 10ð%OÞ
present within their microstructure. Alloys consisting
Moequiv ¼ %Mo þ 0:67ð%VÞ þ 2:9ð%FeÞ
of mainly a-phase are called a-alloys, but if small
amounts of b-phase are present, the alloy is classified þ 1:6ð%CrÞ  %Al
as a near-a alloy. Alloys consisting of a mixture of both
Oequiv ¼ %O þ 2ð%NÞ þ 0:67ð%CÞ
a- and b-phases are termed a–b alloys. Finally, titani-
um alloys that have the majority of b-phase at room
temperature are called b-alloys. Figure 6 shows the Commercially Pure Titanium; 99–99.5% Ti
effects that several important alloying elements have
on the phase diagram, indicating that the equilibrium Commercially pure (CP) titanium is considered an
a-alloy, because the a-phase is the only phase present.
between a, a þ b, and b phase fields (a and b transus
Oxygen is the main alloying element that determines
lines) are highly sensitive to alloy composition. Alumi-
the grade and strength of the alloy (Figures 6b and
num is one of the most important alloying elements
7). CP titanium has lower strength than other alloys,
because it is a potent solid–solution strengthener and it
but it is the alloy of choice for applications requiring
reduces the density, so it is found in virtually all
corrosion resistance, and it has better elevated tem-
titanium alloys. Ti3Al (a2) forms with more than
6 wt.% Al, which can strengthen, but also embrittle perature creep resistance, and is less expensive than
the alloy. Molybdenum and vanadium are the two the other titanium alloys. Figure 7 shows how the
strength and hardness increase with interstitial ele-
most popular b-stabilizing additives, which also
ment concentration. Since oxygen is the main alloy-
provide strengthening of the b-phase. Tin and zirco-
ing element in CP titanium, the Oequiv describes the
nium are the only two alloying elements used which
strengthening effect of the interstitial elements O, N,
have neither an a-stabilizing nor b-stabilizing effect
and C. Each 0.1% Oequiv increases the strength of
on the crystal structure (Figure 6a), but both provide
unalloyed titanium by B120.5 MPa.
solid–solution strengthening to the a-phase; also, Sn is
known to improve weldability. All other elements
either stabilize the a- or b-phase, as summarized in Near a and a Titanium Alloys; Alequivo8 and
Table 1.
The aluminum equivalent and molybdenum Alpha titanium alloys are primarily used in the
equivalent equations (in wt.%) are used to describe chemical and processes engineering industry. These
the degree of stability of the a- and b-phases, applications require excellent corrosion resistance
and to define what class an alloy is in. Oxygen and and ductility. Aluminum is the most important al-
nitrogen are potent a-stabilizers, and iron is the most loying element in order to stabilize a-phase, add
Alloys: Titanium 69

2000 Alpha stabilizers Al and O
Neutral elements Zr and Sn
1600 TiO

Temperature (°C)
Temperature (°C)

1200 1200 TiAl
1000 1000

800 800  Ti3Al

600 600

400 400
5, 10, … 30 wt.% Zr 2 5 10 20 30 40 wt.% Al
200 200
5, 10, … 30 wt.% Sn 0.5 2 5 10 15 20 wt.% O
0 0
0 10 20 30 40 0 10 20 30 40 50 60
(a) At.% Zr or Sn (b) At.% Al or O
2000 2000

1800 1800

1600 1600 Beta eutectoid stabilizers

Fe, Mo, and H
1400 1400
Temperature (°C)
Temperature (°C)

Beta stabilizers Nb and V

1200 1200

1000 1000
800 800
600 600
400 400
5, 10, … 30 wt.% Nb 0.1, 0.2, … 0.6 wt.% H
200 200
5, 10, … 30 wt.% V 5, 10, … 30 wt.% Mo (o), Fe (+)
0 0
0 10 20 30 40 0 10 20 30 40 50 60
(c) At.% V or Nb (d) At.% H, Mo, or Fe
Figure 6 Exemplary phase diagrams for elements that have (a) neutral, (b) a stablizing, (c) b stabilizing, and (d) b eutectoid stablizing
effects. (Adapted from Smithells Metals Reference Book, 7th edn., 1992, London: Butterworth. McCullough C, Valencia JJ, Levi CG, and
Mehrabian R (1989) Phase equilibria and solidification in Ti–Al alloys. Acta Metallurgica 37(5): 1321–1336, for update of Ti–Al system.)

strength, and lower the density (Figure 6b, Table 2). can be combined with higher strength due to a small
Ti–5Al–2.5Sn is the most commonly used a alloy, amount of dispersed b-phase. Small amounts of
and like most a alloys it cannot be age-hardened, but molybdenum and vanadium are added to retain some
it does has excellent weldability. b-phase at room temperature. Ti–8Al–1Mo–1V is the
most commonly used near-a alloy, however the high
aluminum content can cause stress corrosion crac-
Near-a Titanium Alloys; 6oAlequivo10 and king (SCC) problems; therefore, most alloys used to-
day are limited to 6 wt.% Al to avoid SCC problems.
Near-a alloys are ideal for high temperature applica- This alloy has good weldability, but a limited degree
tions B500–5501C, where its excellent creep resistance of hardenability due to a small amount of b-phase.
70 Alloys: Titanium

Table 1 Effects of alloying elements

Alloying element Range ðwt:%Þ Effect on structure and properties

Aluminum, Al 2–7 a-Stabilizer, solid solution strengthener, reduces density, improves

oxidation resistance
Carbon, C 0.05–0.1 a-Stabilizer, interstitial element
Chromium, Cr 2–12 b-Eutectoid stabilizer
Cobalt, Co b-Eutectoid stabilizer
Copper, Cu 2–6 b-Eutectoid stabilizer, improves weldability, a and b strengthener
Hydrogen, H 0.008–0.02 b-Eutectoid stabilizer, interstitial element
Iron, Fe b-Eutectoid stabilizer
Manganese, Mn b-Eutectoid stabilizer
Molybdenum, Mo 2–20 b-Isomorphous stabilizer, moderate solid solution strengthener of b
Nickel, Ni b-Eutectoid stabilizer
Niobium, Nb b-Isomorphous stabilizer, known to improve oxidation behavior of Ti
alloys, moderate solid solution strengthener of b phase
Nitrogen, N 0.015–0.07 a-Stabilizer, interstitial element
Oxygen, O 0.1–0.4 a-Stabilizer, interstitial element, strengthens a phase
Silicon, Si 0.05–1 b-Eutectoid stabilizer, Si atoms tend to segregate at dislocations and
thus effectively prevent dislocation climb, improving creep
resistance, also strengthens a phase
Tantilum, Ta b-Isomorphous stabilizer
Tin, Sn 2–6 Neutral stabilizer, improves weldability, solid solution strengthener of
a phase
Vanadium, V 2–20 b-Isomorphous stabilizer, moderate solid solution strengthener of b
Zirconium, Zr 2–8 Neutral stabilizer, solid solution strengthener of a phase,
Zr also tends to homogenize fine silicide precipitates

Various sources, including (1990) Properties and selection: nonferrous alloy and special-purpose materials. ASM Metal Handbook, vol.
2, p. 605. Metals Park, OH: ASM International.

1200 Tensile strength lower than the a and near-a alloys, primarily due to
Yield strength the presence of the b-phase, which has a much higher
for 25 mm bar
% Elongation

1000 diffusivity and more slip systems.

Hardness Hv
800 % Elongation
MPa, or Hv

600 b Titanium Alloys; Alequivo6 and Moequiv ¼ 15–30

(Metastable b), 430 (Stable b)
400 40
The b titanium alloys are heat treatable, to achieve
200 20 the highest strength levels of the five types of titani-
um alloys (Figures 6c and 6d). The high strength
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
arises from precipitation of very fine a-phase during
Wt.% Oxygen an aging heat treatment. The b.c.c. crystal structure
also gives good formability properties prior to heat
Figure 7 Effect of oxygen on strength and elongation in CP
treatment. Beta titanium alloys contain large
titanium. (Adapted from Jaffee RI (1958) The physical metallurgy
of titanium alloys. Progress in Metal Physics 7:109.) amounts of alloying elements such as Mo, Cr, V,
and Fe which are needed to stabilize the b-phase. The
high degree of alloying elements makes b alloys the
densest of the five alloy types (Table 2).
a–b Titanium Alloys; 5oAlequivo10 and
Tables 2–4 provide further details, using normal
font to identify CP, italics to identify near-a and
The a–b alloy Ti–6Al–4V is the most popular of all a alloys, bold italics to identify a–b alloys, and
titanium alloys, representing more than 50% of the bold to identify b alloys. Table 2 is sorted by density
titanium market (Figures 6b and 6c). The a–b alloys to illustrate the influence of alloying. Table 5
can be solution heat-treated, quenched, and aged to describes weldability issues, and Tables 3 and 4
medium-/high-strength levels and have good form- describe heat treating methodologies, which are dis-
ability, but the creep resistance and weldability are cussed next.
Table 2 Effect of alloying on properties (sorted by density)

Titanium alloy – Condition E ðGPaÞ Poisson’s ratio Density ðg cm3 Þ Room temperature properties KðICÞ ðMPa m1=2 Þ V-notch
CP, near a and a, ðJÞ
a  b, near b, metastable b, or b Y.S. T.S. % Elong.

Ti–8Al–1Mo–1V Duplex annealed 120 0.32 4.37 951 1000 15 82 32

Ti–6Al–4V Annealed 110–140 0.342 4.43 924 993 14 33–110 19
Solution þ age 4.43 1103 1172 10 33–110
Ti–5Al–2.5Sn Annealed 115 0.31 4.48 807 862 16 96 26
Ti–6Al–2Nb–1Ta–1Mo As rolled 2.5 cm plate 113.8 0.31 4.48 758 855 13 31
Ti–3Al–2.5V Annealed 4.48 586 689 20
Ti–7Al–4Mo Solution þ age 4.48 1034 1103 16 18
99.5%Ti Grade 1 Annealed 102.7 0.34 4.51 241 331 30
99.2%Ti Grade 2 Annealed 102.7 0.34 4.51 354 434 28 43
Ti–0.3Mo–0.8Ni (Ti Code 12) 4.51
Ti–6Al–2Sn–4Zr–2Mo Duplex annealed 4.54 896 979 15
Annealed 110–117 4.54 1000 1069 14 30–70 18
Ti–6Al–6V–2Sn (Cu þ Fe) Solution þ age 4.54 1172 1276 10 30–70
Ti–6Al–2Sn–2Zr–2Mo–2Cr–0.25Si Solution þ age 110–120 0.327 4.57 1138 1276 11 65–110
Ti–6Al–2Sn–4Zr–6Mo Solution þ age 114 4.65 1172 1269 10 30–60
Ti–5Al–2Sn–4Mo–2Zr–4Cr (Ti–17) 112 4.65 1050 1100–1250 8–15 30–80
Ti–10V–2Fe–3Al 110 4.65 1000–1200 1000–1400 6–16 30–100
Ti–15V–3Al–3Cr–3Sn 80–100 4.78 800–1000 800–1100 10–20 40–100
Ti–13V–11Cr–3Al Solution þ age 0.304 4.82 1172 1220 8 77 11
Ti–3Al–8V–6Cr–4Zr–4Mo (b C) Solution þ age 4.82 1379 1448 7 50–90 10
Annealed 86–115 4.82 834 883 15 50–90
Ti–8Mo–8V–2Fe–3Al Solution þ age 4.85 1241 1310 8
15Mo–3Al–0.3Fe–2.8Nb–0.2Si (b-21S) 0.33 4.94
Ti–15Mo–5Zr–3Al 5.01
Ti–11.5Mo–6Zr–4.5Sn (b III) Solution þ age 83–103 5.06 1317 1386 11 50–100

Y.S. ¼ Yield strength; T.S. ¼ Tensile strength.

Sources: (1990) Properties and selection: Nonferrous alloys and special-purpose materials. ASM Metals Handbook, vol. 2, p. 621. Metals Park, OH: ASM International; (1993) ASM Metals
Reference Book, 3rd edn. Metals Park, OH: ASM International.
Table 3 Heat treating of titanium alloys

Titanium alloy – T(b) Stress–relief Annealing Solution treating Aging

CP, near a and a 1C715
a þ b, near b, metastable b, b 1C Time ðhÞ 1C Time ðhÞ Cooling 1C Time ðhÞ Cooling 1C Time ðhÞ

99.5%Ti Grade 1 910 480–595 1/4–4 650–760 1/10–2 Air N/A N/A N/A N/A N/A
99.2%Ti Grade 2 913 480–595 1/4–4 650–760 1/10–2 Air N/A N/A N/A N/A N/A

a or near-a alloys
Ti–5Al–2.5Sn 1050 540–650 1/4–4 720–845 1/6–4 Air N/A N/A N/A N/A N/A
Ti–8Al–1Mo–1V 1040 595–705 1/4–4 790 1–8 Air or furnace 980–1010 1 Oil or water 565–595 y
Ti–6Al–2Sn–4Zr–2Mo 995 595–705 1/4–4 900 1/2–1 Air 955–980 1 Air 595 8
Ti–6Al–2Nb–1Ta–1Mo 1015 595–650 1/4–2 790–900 1–4 Air
Ti–0.3Mo–0.8Ni (Ti code 12) 880 480–595 1/4–4

a–b alloys
Ti–6Al–4V 1000 (720) 480–650 1–4 705–790 1–4 Air or furnace 955–970 1 Water 480–595 4–8
Ti–6Al–6V–2Sn (Cu þ Fe) 945 480–650 1–4 705–815 3/4–4 Air or furnace 885–910 1 Water 480–595 4–8
Ti–3Al–2.5V 935 540–650 1/2–2 650–760 1/2–2 Air
Ti–6Al–2Sn–4Zr–6Mo 940 595–705 1/4–4 N/A N/A N/A 845–890 1 Air 580–605 4–8
Ti–5Al–2Sn–4Mo–2Zr–4Cr (Ti–17) 900 480–650 1–4 N/A N/A N/A 845–870 1 Air 580–605 4–8
Ti–7Al–4Mo 1000 480–705 1–8 705–790 1–8 Air
Ti–6Al–2Sn–2Zr–2Mo–2Cr–0.25Si 970 480–650 1–4 705–815 1–2 Air 870–925 1 Water 900–1100 4–8

b or near-b alloys
Ti–13V–11Cr–3Al 720 705–730 1/12–1/4 705–790 1/6–1 Air or water 775–800 1/4–1 Air or water 425–480 4–100
Ti–11.5Mo–6Zr–4.5Sn (b III) 760 720–730 1/12–1/4 690–760 1/6–2 Air or water 690–790 1/8–1 Air or water 480–595 8–32
Ti–10V–2Fe–3Al 805 675–705 1/2–2 N/A N/A N/A 760–780 1 Water 495–525 8
Ti–15V–3Al–3Cr–3Sn 760 790–815 1/12–1/4 790–815 1/12–1/4 Air or water 790–815 1/4 Air 510–595 8–24
Ti–8Mo–8V–2Fe–3Al 775
Ti–15Mo–5Zr–3Al 785
15Mo–3Al–0.3Fe–2.8Nb–0.2Si (b-21S) 800
Ti–3Al–8V–6Cr–4Zr–4Mo (b C) 795 705–760 1/6–1/2 790–815 1/4–1 Air or water 815–925 1 Water 455–540 8–24

Adapted from (1991) Heat treating. ASM Metals Handbook, vol. 4, pp. 914–917. Metals Park, OH: ASM International.
Alloys: Titanium 73

Table 4 Heat treating of a–b alloys

Heat treatment Heat treatment cycle Microstructure


Duplex anneal Solution treat at 50–751C below T(b), air cool and age for 2–8 h at 540–6751C Primary a, plus Widmanstätten
a–b regions
Solution treat and Solution treat at B401C below T(b), water quench(a) and age for 2–8 h at Primary a, plus tempered a0 or
age 535–6751C a b–a mixture
Beta anneal Solution treat at B151C above T(b), air cool and stabilize at 650–7601C Widmanstätten a–b colony
Beta quench Solution treat at B151C above T(b), water quench and temper at 650–7601C Tempered a0
for 2 h
Recrystallization 9251C for 4 h, cool at 501C h  1 to 7601C, air cool Equiaxed a with b at grain-
anneal boundary triple points
Mill anneal a–b Hot work plus anneal at 7051C for 30 min to several hours and air cool Incompletely recrystallized a
with a small volume fraction
of small b particles

In more heavily b-stabilized alloys such as Ti–6Al–2Sn–4Zr–6Mo or Ti–6Al–6V–2Sn, solution treatment is followed by air cooling.
Subsequent aging causes precipitation of a-phase to form an a–b mixture.
Adapted from (1991) Heat treating. ASM Metals Handbook, vol. 4, p. 914. Metals Park, OH: ASM International.

Table 5 Welding of titanium alloys

Alloy type Condition Description

Unalloyed titanium Annealed Welding of cold-worked alloys anneals the HAZ and negates any
strength produced by cold working
Alpha titanium Annealed Ti–5Al–2.5Sn, Ti–6Al–2Sn–4Zr–2Mo, Ti–5Al–5Sn–2Zr–2Mo,
alloys Ti–6Al–2Nb–1Ta–1Mo, and Ti–8Al–1Mo–1V are always welded
in annealed condition
Alpha–beta titanium Annealed or Low weld ductility of most a–b alloys is caused by phase
alloys solution-treated transformation in the weld zone or HAZ
and partially aged
Metastable beta Annealed or solution In as-welded condition, welds are low in strength but ductile; to
titanium alloys heat treated obtain full strength the alloys are welded in annealed condition, the
weld is cold worked by shot peening or planishing, and the
weldament is then solution treated and aged
Weld type Thickness range Description
Gas–tungsten arc Base metal up to
weld (GTAW) 2.5 mm, for
thicker base metal
filler metal is
Gas–metal arc weld More than 3 mm to Applied using pulse current or spray mode and is less costly than
(GMAW) greater than GTAW
13 mm
Plasma arc welding Plate up to 13 mm, Faster than GTAW and can be used on thicker sections
(PAW) filler metal may be
Electron-beam Plates 6 mm to more Used in aircraft and aerospace industry for producing high quality
welding (EBW) than 76 mm welds; performed in high vacuum atmosphere therefore low
contamination of weldament
Laser-beam welding Base metal usually
(LBW) cannot exceed
13 mm
Friction welding Useful for joining tube, pipe, or rods
Resistance welding Used to join titanium sheet by spot welds or continuous seam welds,
(RW) also used to weld titanium sheet to dissimilar metals
Fluxes N/A Fluxes cannot be used because they combine with titanium to form
brittleness and may reduce corrosion resistance

Excerpted from (1990) Welding brazing and soldering. ASM Metals Handbook, vol. 6, pp. 783–786. Metals Park, OH: ASM International.
74 Alloys: Titanium

Heat Treating of Titanium Alloys alloy is in its final age-hardened condition, the SR
temperature should be at or below the aging tem-
Titanium and titanium alloys are heat treated for a
perature to prevent strength reduction.
number of reasons: to reduce residual stresses from
the many different fabrication processes (stress re-
lieving), to obtain optimum combinations of ductil-
ity, machinability, and dimensional stability Annealing is similar to SR but usually done at higher
(annealing), to increase strength (solution treating temperatures. The annealing of titanium and titani-
and aging in a–b and b alloys), and finally to opt- um alloys serves primarily to increase fracture
imize specific properties such as fracture toughness, toughness, room temperature ductility, dimensional
fatigue strength, and high-temperature creep resist- stability, and high-temperature creep resistance. This
ance. In general, a heat treatment is used to rearrange is often the final heat treatment for a and a–b alloys,
phase–volume fractions, phase boundaries in strate- as it yields a good balance of properties. Care must
gic ways. A quench from high temperature is com- be taken when using b alloys in the annealed con-
monly used to provide a thermodynamic driving dition in service temperatures up to 4001C, which
force for nanoscale rearrangement at lower aging could result in embrittlement, and service tempera-
temperatures. More recently, benefits of rapid hea- tures in the range of 400–6001C could age-harden
ting into the b-phase field to obtain desirable micro- the b alloy, resulting in a strength increase at the
structures have been described. Standard practice is expense of ductility. Alpha and a–b alloys are typ-
identified in Tables 3 and 4, and the preferred phase ically annealed in the temperature range of 700–
for alloying elements is in Table 1. However, alloy 9001C, while b alloys are normally annealed in the
element local concentration is highly dependent on temperature range of 690–8151C.
the history of the phase volume fraction and loca- There are four commonly used annealing treat-
tions of phase boundaries in space and hence, on all ments, mill annealing (MA), duplex annealing (DA),
prior heat treating and working history. For this rea- recrystallization annealing (RA), and b annealing
son, the same heat treatment on materials with dif- (BA). MA is a general-purpose heat treatment given
ferent prior history can cause considerable variability to all mill products, but it is not a full anneal and the
in resulting properties. material may retain effects of prior working proc-
Unalloyed a titanium alloys can be stress relieved esses. DA is used for a and a–b alloys; it involves an
and annealed, but cannot be heat treated to increase initial anneal high in the a/b-phase field to generate
their strength. When the heat treatment of titanium significant fractions of both a and b with small grain/
alloys involves heating to temperatures near the b phase sizes near 10 mm, followed by an MA to
transus, it is important that Tb is known for each provide thermal stability in the prior b regions. DA
specific alloy. Table 3 illustrates how the b transus improves creep resistance and fracture toughness
temperature (bT) depends on alloy composition. (resistance to crack nucleation) of the material.
An RA is done at a temperature high enough in
Stress Relieving a/b-phase field to ensure recrystallization. It is then
slowly cooled to form a high-volume fraction of
Stress relieving (SR) is used to remove residual stress-
equiaxed a with islands of retained b at triple points,
es from prior fabrication steps, or even prior heat
and some interfacial b at a/a boundaries. RA
treatments. Unbalanced residual stresses can result in
provides high-damage tolerance properties (fracture
part distortion, and cause problems in those alloys
toughness, crack growth resistance) that are crucial
susceptible to hydrogen embrittlement. SR can result
for fracture critical applications. Finally, BA is done
in aging of all alloy types, a can be aged by a2 pre-
at temperatures above the bT, leading to a large grain
cipitates, and if metastable b is present, a precipita-
size followed by a subsequent MA (Figure 5 shows
tion that provides strengthening can occur. For a and
how a BA converted a small grain size DA micro-
a–b alloys, the SR temperature will be in the range of
structure into large prior b grains that were trans-
480–8151C, and if these alloys were b-annealed,
formed into several a variants during slow cooling).
more rearrangement of phase boundaries is needed
For a and a–b alloys, BA maximizes the damage tol-
and the SR temperature should be increased by
erance properties, since resistance to crack propagat-
B551C. When stress-relieving b alloys, care must be
ion is best when there are fewer grain boundaries.
taken to avoid interfering with the final age-harden-
ing treatment. If a b alloy has not been heat treated,
Solution Treating
the SR temperature should be substantially below the
bT to prevent preferred a precipitation in grain Solution treating (ST) is used to transform a desired
boundaries (which embrittles the material). If the b amount of a- to b-phase in near a and a–b alloys, by
Alloys: Titanium 75

heating the material to a temperature near bT and Ti–5Al–2.5Sn and Ti–2.5Cu, for weldability, Ti–
then cooling strategically to produce a higher ratio of 6Al–2Sn–4Zr–2Mo and Ti–6Al–5Zr–0.5Mo–0.2Si,
b-phase in a desired morphology. Beta alloys are for creep resistance.
normally solution treated above the bT (700–8151C),
while a and a–b alloys are normally solution treated Welding of Titanium Alloys
slightly below the bT (850–10001C) (Table 3). For
most b alloys, the object of ST is to retain 100% Table 5 provides relevant details for welding titani-
b-phase upon quenching, so that upon subsequent um alloys. Unalloyed titanium and a alloys are weld-
aging, decomposition of the metastable b-phase oc- able. These alloys are typically welded in the
curs to generate second-phase precipitates that annealed condition, and have good weldability be-
provide high strength. cause they are insensitive to heat treatments. Alpha–
beta alloys and weakly b-stabilized alloys are also
Quenching weldable, but should be welded in the annealed or
solution-treated conditions prior to aging heat treat-
The rate of cooling from the ST temperature strongly ments. Most b alloys can be welded in annealed or
affects the strength of the material. Slow cooling solution-treated condition, but subsequent aging can
rates allow diffusion to occur, resulting in decompo- cause the weld to become brittle. Most strongly
sition of the b-phase and may prevent effective b-stabilized alloys are embrittled by welding, since it
strengthening during the aging process. Also the is not possible to quench a weld (without quenching,
quench delay, which is the time from which the ma- second phases precipitate preferentially on grain
terial is removed from the furnace and subsequently boundaries).
placed into the quenching media, can affect proper-
ties. Longer quench delays can lower strength; this
reduced strength is ascribed to the formation of Challenging and Problematic Issues with
coarse acicular a prior to immersion in quenchant. Titanium Alloys
Alpha–beta alloys are typically quenched in water, a
Although titanium alloys have many desirable char-
5% brine solution, or a caustic soda solution because
acteristics, they have problematic areas as well.
the rate is sufficient to retain the b-phase obtained by
These include their affinity to oxygen, hydrogen
ST. For highly b-stabilized alloys and small sections,
pickup, susceptibility to certain types of chemical
an air or fan cooling may be adequate.
attack, and damage generation arising from the elas-
tic and plastic anisotropy of the hexagonal crystal
Aging is the final heat treatment step needed in order Heat treating can cause contamination due to ox-
to achieve high-strength levels. Aging causes the de- idation that causes an oxygen-rich brittle layer on the
composition of the supersaturated b-phase retained metal surface called ‘‘a case.’’ Unless surface hard-
upon quenching, through a-precipitation within the ening is needed for wear resistance, this a case must
b-matrix. Martensites also decompose in a manner be mechanically or chemically removed before the
that sharpens interfaces or boundaries that provide part can be put into service. Another concern is em-
strengthening. The aging temperature controls the brittlement due to very low levels of hydrogen, nit-
size and volume fraction of the a-precipitates, and rogen, and carbon absorption at virtually every
determines the final strength of the material. Aging at fabrication stage from casting to heat treating. Due
or near the annealing temperature results in over- to its great affinity to oxygen, alumina (Al2O3) can-
aging, which is desirable if good toughness and di- not be used to machine titanium alloys, so WC tools
mensional stability are needed with only a modest (typically C-2 grades) are used for turning and face
increase in strength. For a and a–b alloys, aging is milling, while high-carbide, high-speed steels are ap-
usually done in the temperature range of 480–6051C, propriate for drilling, tapping, and end milling.
and b alloys are typically aged in the temperature Though CP titanium is highly corrosion resistant,
range of 425–5951C. SCC is more likely with increasing alloy content, and
Suggested heat treatments for various titanium in environments with red fuming nitric acid, N2O4,
alloys are listed in Tables 3 and 4. It is also import- HF, methonal, HCl, and seawater. The susceptibility
ant to note that not all heat treatments are applic- depends greatly on alloy elements, crystal orientation
able to all titanium alloys, because some alloys are and misorientations at grain boundaries, and the de-
designed for specific purposes, for example, Ti–5Al– tails of microstructural evolution.
2Sn–2Zr–4Mo–4Cr (Ti-17) and Ti–6Al–2Sn–4Zr– Due to its hexagonal crystal structure, slip in
6Mo are designed for strength in heavy sections, the /aS direction is much more facile than other
76 Alloys: Titanium

directions, so heterogeneous deformation and shear Bondarchuk VI, Ivasishin OM, Moiseyeva IV, Okrainets PM,
banding is common. Combinations of working and and Pishchak VK (2001) Effect of rapid heat treatment on
heat treatment can generate highly preferred crystal the high-temperature deformation of refractory titanium
alloys. Metal Physics and Advanced Technologies (UK) 19(5):
orientations, so between the coherent phase bound- 743–753.
aries with transformation shears, and the anisotropy Boyer RR and Lutjering G (1996) Heat treatment of titanium
of the thermal expansion coefficient (which is 20% alloys: overview. In: Weiss I, Srinivasan R, Bania PJ, Eylon
higher in the /cS than /aS directions) substantial D, and Semiatin SL (eds.) Titanium Alloy Processing, pp.
residual stresses are common even after a stress relief 349–367. Warrendale, PA: The Minerals, Metals and Materials
anneal. Though titanium alloys can be processed for Boyer R, Collings EW, and Welch G (eds.) (1994) Materials Prop-
excellent high-temperature creep resistance, room- erties Handbook: Titanium Alloys. Materials Park, OH: ASM
temperature creep occurs in some microstructures Publications.
and alloys, and hence dimensional stability can be a Donachie MJ Jr. (2000) Titanium: A Technical Guide, 2nd edn.
problem. To fully optimize material property design Metals Park, OH: ASM Publications.
Exner HE, Muller C, and Schmidt H (2004) Modification of ti-
for particular applications, control of texture and tanium alloys for medical applications. Zeitschrift fur Metal-
microstructure with innovative processing strategies lkunde 95(7): 650–662.
requires predictive understanding of the rules for Froes (Sam) FH, Imam MA, and Fray D (eds.) (2004) Cost-
variant selection during the a-b-a transforma- Affordable Titanium. Warrendale, PA: The Minerals, Metals
tions, which are not yet well established. and Materials Society.
See also: Dislocations; Mechanical Properties: Plastic http://www.matls.com
Behavior; Periodicity and Lattices; Recovery, Recrystalli- Leyens C and Peters M (2003) Titanium and Titanium Alloys;
zation, and Grain Growth. Fundamentals and Applications. Weinheim: Wiley-VCH Verlag
GmbH & Co. KGaA.
Semiatin SL, Seetharaman V, and Weiss I (1996) Hot working of
PACS: 61.66.  f; 62.20.  x; 64.70.  p; 81.05.Bx; titanium alloys – an overview. In: Weiss I, Srinivasan R, Bania
81.40.  z; 81.40.Cd; 81.40.Ef PJ, Eylon D, and Semiatin SL (eds.) Titanium Alloy Processing,
pp. 3–74. Warrendale, PA: The Minerals, Metals and Materials
Further Reading Smith WF (1993) Structure and Properties of Engineering Alloys,
2nd edn., pp. 433–484. New York: McGraw-Hill.
(1991) Heat treating. ASM Handbook, vol. 4. Metals Park, OH: Veeck S, Lee D, and Tom T (2002) Titanium investment castings.
ASM Publications. Advanced Materials & Processes 160(1): 59–62.

Amorphous See Alloys: Overview; Composites: Overview; Disorder and Localization Theory; Disordered
Solids and Glasses, Electronic Structure of; Porous Silicon; Waves in Random Media.

Anelasticity See Mechanical Properties: Anelasticity; Mechanical Properties: Elastic Behavior.

Aperiodic Crystals See Lattice Dynamics: Aperiodic Crystals.

Ballistic Transport
M Dragoman, National Institute for Research and not affect phase relaxation. Lph is of the same order
Development in Microtechnology (IMT), Bucharest, of magnitude as Lfp in high-mobility semiconductors.
Romania Experimental conditions for ballistic transport
D Dragoman, University of Bucharest, Bucharest, observation have become achievable only in the last
decades due to impressive developments in semi-
& 2005, Elsevier Ltd. All Rights Reserved. conductor technology, especially in advanced epitax-
ial growth and lithography techniques. It has thus
become possible to fabricate band-engineered semi-
conductor heterostructures called quantum wells,
Introduction quantum wires, and quantum dots that confine, re-
The transport of charged carriers depends greatly spectively, the electron motion in one, two, and three
on the scale of the sample. At the macroscopic scale, dimensions. The electron density in such structures
in bulk semiconductors with dimensions larger than can be controlled through electric potentials applied
1 mm, conduction electrons move randomly due to on surface gates. Coherent propagation of electron
the high scattering rates with impurities and phonons. waves, electron tunneling, and Coulomb blockade
An applied electric field induces a diffusive electron are examples of specific phenomena associated with
transport described by the stochastic Boltzmann ballistic transport. These phenomena have found
equation. At the microscopic scale of atoms and mol- hosts of applications in the design of new devices
ecules, which is of the order of Å ¼ 1010 m, the such as resonant tunneling devices and the single-
particles evolve according to the laws of atomic and electron transistor, the ballistic transport regime
molecular physics. At the intermediate, mesoscopic being indispensable for high-frequency electronic
scale, distinct features in the particle transport ap- and optoelectronic devices. Not only electrons but
pear, which include the emergence of the ballistic also holes and even quasiparticles, such as phonons,
transport regime characterized by the coherent, wave- experience ballistic transport regime under the same
like behavior of charged particles over length scales conditions as those defined above, but with the cor-
much larger than the dimension of molecules and responding mean free path and phase relaxation
much smaller than macroscopic dimensions. Ballistic length parameters.
transport is defined as the collisionless transport
regime, valid when the dimensions of the sample are
Energy Subbands and Density of
smaller than the mean free path Lfp and the phase
States in Ballistic Transport
relaxation length Lph. The mean free path is a meas-
ure of the distance between successive electron col- Ballistic transport occurs when, at most, only a
lisions with impurities or phonons that destroy the few elastic scattering processes take place, such that
initial electron momentum. Lfp is about tens of nano- electron motion is characterized in this regime by a
meters in polycrystalline metallic films and about tens constant energy E, and can be described by a time-
of micrometers in high-mobility semiconductors at independent Schrödinger equation for the envelope
low temperatures (To4 K); the latter materials are electron wave function C, which varies slowly over
thus more likely to exhibit ballistic transport. The the dimensions of the unit cell:
phase relaxation length is the distance after which
the coherence (the phase memory) of electrons is lost. _2 a
 fm r½mb rðma CÞg þ VC ¼ EC
Phase memory is lost in time-reversal breaking proc- 2
esses such as electron–electron collisions, dynamic
scattering processes, or scattering by impurities with where a and b are material-related parameters for
internal degrees of freedom (spin) that change in the which 2a þ b ¼ 1, m is the electron effective mass
process; any static spin-independent scattering does that incorporates the effect of the periodic lattice
78 Ballistic Transport

potential on electron motion, and V is the potential occupy in the kx  ky plane an area pk2 ¼ p
energy. This scalar equation is valid when coupling ðk2x þ k2y Þ, is Ntot ðkÞ ¼ k2 ðLx Ly Þ=2p (or Ntot ðEÞ ¼
between different electron bands is negligible. Bal- mðLx Ly ÞðE  Es;p Þ=p_2 Þ. The density of states per
listic transport has been mainly evidenced in AlGaAs unit area S ¼ Lx Ly , and per unit energy in the pth
semiconductor compounds for which a ¼ 0 and subband is thus
b ¼ 1. The potential energy V includes the conduc-
tion band discontinuities in heterojunctions (which 1 dNtot ðEÞ m
is the dominant component in the low doping and r2D;p ðEÞ ¼ ¼ 2 WðE  Es;p Þ
S dE p_
the low free carriers case), the electrostatic potential
due to ionized donors and acceptors (which is deter-
mined from a self-consistent solution of the coupled with W the unit Pstep function, the total density of
Schrödinger and Poisson equations), and the self- states r2D ðEÞ ¼ p r2D;p represented in Figure 1a
consistent Hartree and exchange potentials due to being discontinuous, unlike in bulk semiconductors
free carriers. where no size constriction is imposed upon the elec-
The solution of the Schrödinger equation requires tron motion.
proper boundary conditions. The constraints on elec- The equilibrium electron density per unit area is
tron motion present in any ballistic structure impose given by
the appearance of discrete energy levels along the
direction of the constraint and a resulting disconti- Z N X
nuity in the density of states. For example, in a n¼ r2D;p ðEÞf ðEÞdE
quantum well the conduction electrons are free to 0 p
move along the x and y directions, and are confined m X
¼ kB T ln½1 þ eðEF Es;p =kB TÞ 
by potential barriers along z in a region of width Lz. p_2 p
For abrupt infinite-height barriers, thepelectron
pffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi wave
function Cðx; y; zÞ ¼ 2=Lz sinðkz zÞ 1=Lx Ly exp
ðikx xÞexpðiky yÞ, obtained from the Schrödinger
equation for V ¼ 0, with Lx and Ly the sample di-
mensions along x and y, must satisfy the boundary
conditions Cðx; y; 0Þ ¼ Cðx; y; Lz Þ ¼ 0. These induce EF
a discrete spectrum kz ¼ pp=Lz for the momentum
component along z, with p an integer, and an energy
dispersion relation of the form x

_2 pp 2 _2 2 Es,1 Es,2 Es,3 E
Eðkx ; ky ; kz Þ ¼ Ec þ þ ðkx þ k2y Þ
2m Lz 2m
_2 2
¼ Es;p þ ðk þ k2y Þ
2m x
where Ec denotes the bottom of the conduction band
and Es,p is the cut-off energy of the subband or y
transverse mode labeled by p. The energy spacing
between discrete p levels increases as Lz decreases, (b) E
Es,11 Es,12 Es,1.3
that is, as electrons become more confined. Although
periodic boundary conditions imposed on x and y 0D
also require the quantization of kx, ky according to
kx ¼ qð2p=Lx Þ, ky ¼ rð2p=Ly Þ, with q and r integer z
values, the energy spectrum remains quasicontinuous
in the kx  ky plane, since Lx ; Ly cLz . The quantizat- y
ion of the wave vector components along x and y x
mainly influences the density of states by assigning to
an individual state an area ð2p=Lx Þ  ð2p=Ly Þ in the (c) Es,111 Es,112 Es,113 E
kx  ky plane. Then, the total number of spin- Figure 1 Density of states in (a) quantum wells, (b) quantum
degenerate states with a wave number less than k wires, and (c) quantum dots. The Fermi energy position in
(or with an energy less than E, for E4Es,p), which (a) corresponds to 2DEG systems.
Ballistic Transport 79

the energy levels being occupied by electrons according Two-Dimensional Electron Gas
to the Fermi–Dirac distribution function
If, as in Figure 1a, the position of the Fermi level in a
1 quantum well is between the first and the second
f ðEÞ ¼ energy subband, the electron density per unit area,
1 þ exp½ðE  EF Þ=kB T
n ¼ ðm=p_2 ÞðEF  Es;1 Þ, is related
pffiffiffiffiffiffiffiffi to the Fermi wave
At low temperatures or in the degenerate limit, number kF through kF ¼ 2pn. The Fermi wave
when kB T{EF , the Fermi–Dirac distribution function number is determined by the kinetic energy of elec-
becomes proportional to WðEF  EÞ, and all electron trons: Ekin ¼ EF  Es;1 ¼ _2 k2F =2m. The electrons
subbands are filled up to the Fermi energy and are form, in this case, a two-dimensional electron gas
empty above it. In this case, all electrons that partic- (2DEG), which has a metallic behavior since EF
ipate in transport have energies close to EF, unlike in is inside the conduction band.
the diffusive transport regime in which electrons have Two-dimensional electron gas systems form, for
a widespread energy distribution. The number of sub- example, at the interface between intrinsic GaAs and
bands M(E) occupied at low temperatures by electrons n-doped AlGaAs layers in modulation-doped GaAs/
with energy E is obtained by counting the number of AlGaAs heterojunctions, the spatial separation of
transverse modes with cutoff energies smaller than E. dopant atoms in AlGaAs from the free electrons that
In a similar manner, if the electron motion is con- form an inversion layer on the GaAs side of the in-
strained by infinite-height potentials along two di- terface assuring extremely low scattering rates. The
rections, y and z, the boundary conditions imposed electron density in this 2DEG can be changed by
p theffi electron wave pfunction
ffiffiffiffiffiffiffiffiffiffiffi Cðx; y; zÞ ¼ 2= modifying the electron kinetic energy via negative
Ly Lz sinðky Ly Þsinðkz Lz Þ 1=Lx expðikx xÞ require electrostatic voltages applied on Schottky surface
that ky ¼ pp=Ly , kz ¼ qp=Lz , with p, q integers. In gates situated in close proximity to the 2DEG; the
such a one-dimensional ballistic quantum wire, the bias depletes the 2DEG underneath the gate, and
electrons are free to move only along x, the energy laterally from the geometric edge of the gate until no
dispersion relation being free electrons are left, a case in which a barrier for
    electrons is created. The coherence of the electron
_2 pp 2 _2 qp 2 _2 k2x wave function in ballistic structures together with the
Eðkx ; ky ; kz Þ ¼ Ec þ þ þ
2m Ly 2m Lz 2m possibility of manipulating the electron wave number
_2 k2x allows the implementation of electron counterparts
¼ Es;pq þ of both geometrical optical systems and interference
devices. Figures 2a–2d show an electron prism, a
The total number of states with a wave vector
refractive lens, an electron beam splitter, and an
component along x less than kx, which occupy an
interferometer, respectively. As in the classical elect-
‘‘area’’ in the kx plane equal to 2kx, is now
romagnetic theory, the refraction at an interface be-
Ntot ðkx Þ ¼ 2  2kx =ð2p=Lx Þ, where the first factor
tween two 2DEG systems with electron densities n1
2 accounts for spin degeneracy. The density of
and n2 occurs according to Snell’s law for ballistic
states in the kx plane per unit length, r1D ðkx Þ ¼
electrons: sin y1 =sin y2 ¼ ðn2 =n1 Þ1=2 , with yi the
x dNtot =dkx , is constant and equal to 2/p, half
angle in region i with the normal to the interface.
of this value representing the contribution of states
The electron wave function is partly reflected and
for which kx is positive, and half representing the
partly transmitted at the interface. Narrow constric-
contribution of states with negative kx values.
tions in the 2DEG created by the black depletion
r1D ðkx Þ corresponds to a total energy-dependent P gates in Figure 2 act as electron sources with a wide
density of states r1D ðEc Þ ¼ ð1=pÞð2m=_2 Þ1=2 p;q
angular spectrum, the gray electrodes in Figure 2 re-
ðE  Es;pq Þ1=2 , which is represented in Figure 1b.
presenting partially depleting gates that refract elec-
In quantum dots the electron motion is confined
tron waves.
along all directions, and the energy spectrum
The analogy between electron propagation in
    ballistic conductors and electromagnetic wave propa-
_2 pp 2 _2 qp 2
Eðkx ; ky ; kz Þ ¼ Ec þ þ gation is based on the formal analogy between the
2m Lx 2m Ly
time-independent Schrödinger equation for the elec-
2  2
_ rp tron wave function and the Helmholtz equation
þ ¼ Es;pqr
2m Lz
r 2 F þ k2 F ¼ 0
is discrete, as in atoms or molecules. Quantum dots,
with density of states r0D pdðE  Es;pqr Þ (see Figure satisfied by the electromagnetic field component F
1c), can thus be considered as artificial atoms. (electric field, magnetic field, or vector potential).
80 Ballistic Transport

Prism C

E E Lens C


(a) (b)

E1 D1
E2 D2


(c) (d)
Figure 2 Schematic representation of (a) an electron prism, (b) a refractive electron lens, (c) a beam splitter, and (d) an interferometer.
The electrons are emitted from the narrow orifice E and are collected by the collector C. Dashed lines represent electron trajectories.

The vectorial character of the latter implies that the to the current of electrons in each occupied subband
eventual quantitative analogies between electron and being additive if the ballistic conductor has a con-
light parameters (energy, effective mass, potential stant cross section; in this case, there is no scattering
energy and frequency, electric permittivity, magnetic of electrons from one subband (transverse mode) to
permeability, respectively) depend on which elect- another. The net current flow in one subband due to
romagnetic field component is compared to the scalar the extra density of electrons in the left contact,
electron wave function. Different sets of analogous dn ¼ ðdn=dEÞeV, is I ¼ evdn, where v ¼ _1 ðdE=dkÞ
parameters between the electron wave function and is the electron velocity along the current flow direc-
the electromagnetic field are derived if, besides the tion. Then I ¼ ðe2 =_ÞVðdn=dEÞðdE=dkÞ, or I ¼
evident similarity between the equation satisfied by ð2e2 =hÞV for a one-dimensional conductor for which
maC and F if k is replaced by ½2mðE  VÞ1=2 =_, ad- dn ¼ Ntot ðkÞðdk=dEÞdE, with Ntot ðkÞ ¼ 1=p for elec-
ditional different boundary conditions for different trons flowing in the direction of the applied bias. The
field components F are considered. total current can be expressed as I ¼ ð2e2 =hÞMV, if
the number of transverse modes M(E) is constant
over the energy range EFr oEoEFl , the ballistic con-
Resistance of a Ballistic Conductor
ductor having a conductance Gc ¼ I=V ¼ 2e2 M=h
A net electron transfer across a ballistic conductor or, equivalently, a resistance
sandwiched between two contacts that act as electron
reservoirs is possible, when an external bias V is ap- h 12:9 kO
Rc ¼ 1=Gc ¼ D
plied between the contacts. The bias drives the elec- 2e2 M M
tron system in the ballistic conductor away from
equilibrium such that no common Fermi energy exists; In the collisionless propagating regime, the resist-
instead, one can define a spatially varying local quasi- ance can only be caused by the mismatch between
Fermi level that takes the values EFl and EFr in the left the finite number of transverse modes that are al-
and right contacts, respectively. The net current can lowed to propagate in the ballistic conductor and the
then be easily calculated if the contacts are reflection- infinite number of transverse modes carried by con-
less, that is, if the electrons can enter them from the tacts. The resistance is called contact resistance be-
conductor without suffering reflections. If EFl4EFr , cause this mismatch appears at the conductor/
such that for a small applied bias EFl  EFr ¼ eV, only contact interface. Rc decreases with the number of
electrons that flow from filled states on the left to transverse modes (energy subbands) in the ballistic
empty states on the right contribute to the net current. conductor.
At zero temperature, there is current flow only in the Unlike the conductance G ¼ sS=L in bulk materi-
electron energy range EFroEoEFl, the contribution als, where s is the conductivity of the material, L the
Ballistic Transport 81

Gc(2e 2/h)

V Metallic gates 4


Contact Contact V
−V1 −V2 −V3 −V4

(a) (b)
Figure 3 (a) Constriction with variable width in a 2DEG and (b) the conductance dependence on the gate voltage at zero temperature
(solid line) and finite temperatures (dotted line).

length of the sample, and S its cross section, the E Fl E Fr

conductance of ballistic structures does not depend
on the length of the conductor. (Note that an unjus- Lead 1 Lead 2
tified extension of the domain of validity of the con-
ductance formula in bulk materials would imply that
in ballistic conductors, G becomes extremely large as
L decreases.) In ballistic conductors, Gc depends Conductor
however on S (the width of the one-dimensional
Contact Contact
conductor), since the number of transverse modes
occupied by electrons propagating with the Fermi Figure 4 Model of a ballistic conductor with transmission prob-
wave number can be estimated from M ¼ Int½kF S=p, ability T between the contacts.
where Int denotes the integer value. This dependence
can be evidenced by measuring Gc of a ballistic con- transmitted to lead 2, the conductance measured be-
ductor delimited from a 2DEG by a pair of metallic tween the contacts at zero-temperature being given
gates (Figure 3a), the conductor width depending by the Landauer formula
on the negative voltage  VM applied on the gates. A
step-like increase of the conductance with 2e2/h, as 2e2
shown in Figure 3b, is observed at low temperatures h
and/or for confined electrons with large spacings
between the energy levels each time M increases by This expression can be viewed as a mesoscopic
one. This discontinuity is otherwise ‘‘smoothed’’ by a version of the Einstein relation s ¼ e2 rD, if the con-
thermal motion. ductivity s is replaced by G, the density of states r is
replaced by M, and the diffusion constant D is re-
placed by T. The total resistance R ¼ h=ð2e2 MTÞ
Landauer Formula
measured between the contacts is the sum between
The formula derived above for the conductance of the contact resistance Rc ¼ h=ð2e2 MÞ and the resist-
ballistic structures assumed implicitly that electrons ance of the ‘‘scatterer’’ element with transmission T,
injected by the left contact are transmitted with unit Rs ¼ hð1  TÞ=ð2e2 MTÞ, that would be measured
probability to the right contact. This is not always between the two leads.
the case. In particular, a partial (nonunity) transmis- This identification allows the calculation of the
sion of electrons from one contact to the other occurs resistance of multiple scatterers with transmission
if the ballistic conductor is composed of several parts probabilities Ti connected in series as Rs ¼ i Rsi ,
that differ in their width or potential energy. The with Rsi ¼ hð1  Ti Þ=ð2e2 MTi Þ, or the calculationPof
conductance, in this case, can be easily calculated if the total transmission probability as ð1  TÞ=T ¼ i
the conductor with a transmission probability T is ð1  Ti Þ=Ti . This addition law, which can alter-
connected to two large reflectionless contacts by bal- natively be obtained by adding the contribution of
listic leads each of which have M transverse modes. successive partial transmitted waves, is an expression
In the model presented in Figure 4, T is the average of the coherent nature of the electron wave function
probability that an electron injected in lead 1 is in ballistic conductors.
82 Ballistic Transport

At finite temperatures, when the Fermi–Dirac dis- M and the transmission probability per mode Tpq
tribution is no longer step-like, electron transport at the Fermi energy. The requirement that the net
through multiple energy channels takes place in the current
P flow
P is zero at equilibrium implies that
energy range EFr  DEoEoEFl þ DE, where DE is q Gqp ¼ q Gpq .
of the order of a few kBT. Since both the number of At finite temperatures, the net current flow
transverse modes M and the transmission probability through the pth terminal is given by
T generally depend on E, the net current flow be-
tween the left and right contacts in the absence of Z X
inelastic scattering is Ip ¼ ½T% qp ðEÞfp ðEÞ  T% pq ðEÞfq ðEÞdE
h q
I¼ MðEÞTðEÞ½fl ðEÞ  fr ðEÞdE
Transmission Probability Calculation
where fl(E) and fr(E) are the Fermi–Dirac quasi-dis-
Ballistic electron transport is determined by the
tribution functions in the left and right contacts.
transmission probability between the leads that con-
nect the conductor to several contacts. There are
Büttiker Formula many methods to calculate the transmission proba-
bility: the scattering matrix formalism, the Green’s
The expression for the net current flow or for the function approach, the Kubo formalism, or the
conductance of a ballistic structure can be generali- transfer Hamiltonian formalism. The most wide-
zed to account for the existence of several contacts or spread for its ease of application is the matrix for-
terminals. This is the case, for example, of four-ter- malism which, in its simplest case, relies on
minal measurements (see Figure 5), where two addi- approximating the spatially varying potential pro-
tional terminals are used to measure the voltage drop file, electron effective mass, and/or dimensions of the
along a conductor besides the two terminals through conductor as a succession of regions in which this
which the current flows. Modeling the additional variation is step-like.
floating or zero-external current terminals as scatter- The application of the matrix method to ballistic
ers characterized by transmission probabilities, the conductors with a constant cross section is simplified
current through the pth terminal at zero-temperature by the absence of electron scattering from one
is determined by the Büttiker formula transverse mode to another. Then, for a one-dimen-
sional conductor composed of several regions with
2e2 X % X
Ip ¼ ½T qp Vp  T% pq Vq  ¼ ½Gqp Vp  Gpq Vq  different but constant potential and electron effective
h q q mass, the solution of the Schrödinger equation in the
ith region, Ci ðxÞ ¼ Ai expðiki xÞ þ Bi expðiki xÞ, can
where Gpq ¼ ð2e2 =hÞT% pq is the conductance associ- be expressed as a superposition of forward- and
ated to the electron transfer from terminal q with a backward-propagating p waves in the xffi direction with
quasi-Fermi energy level EFq ¼ eVq to terminal p wave number ki ¼ _1 2mi ðE  Vi Þ (see Figure 6).
with a quasi-Fermi energy level EFp ¼ eVp . T% pq is, in Inside each layer the wave function and its x-
general, the product between the number of modes derivative must be continuous, whereas boundary
conditions impose the continuity of the wave func-
tion and rC  x=m # aþ1 at the interface between two

E F2 EF3

Ai Ai +1

Bi Bi +1

Vi , mi , ki Vi +1, m i +1, k i +1

x = xi

Figure 5 Four-terminal configuration: the current flows be- Figure 6 Electron wave function forward- and backward-
tween the first and the fourth terminal, the second and third being propagating components at an interface between layers i and
used to measure the voltage drop along the conductor. i þ 1 with different characteristics.
Ballistic Transport 83

layers, where x# is the unit vector (the versor) along Electron Tunneling
the x direction.
The matrix formulation is valid for either real or
The matrix that expresses the electron wave func-
imaginary values of the wave numbers ki. An
tion form at different x values depends on which
imaginary wave number in at least one region i im-
parameters are related. For example, when a ¼ 0, a
plies that the electron energy is smaller than the po-
transfer matrix at the interface between the layers i
tential energy value in this region, so that electrons
and i þ 1 situated at x ¼ xi can be defined as
are forbidden to enter the ith region from a classical
! !
Ai expðiki xi Þ 1 ð1 þ viþ1 =vi Þ ð1  viþ1 =vi Þ point of view. Such a region acts as a barrier for
¼ electron propagation. The propagation of ballistic
Bi expðiki xi Þ 2 ð1  viþ1 =vi Þ ð1 þ viþ1 =vi Þ
! electrons in regions with imaginary wave numbers
Aiþ1 expðikiþ1 xi Þ can be paralleled with an evanescent propagation of
 electromagnetic waves. As for evanescent elect-
Biþ1 expðikiþ1 xi Þ
romagnetic waves, the electron wave function decays
exponentially inside the region with imaginary ki and
where vi ¼ _ki =mi is the electron velocity in the ith
therefore, the transmission probability through this
layer, whereas at free propagation across the ith lay-
region vanishes unless its dimension along the
er, between planes x ¼ xi1 and x ¼ xi the transfer
propagation direction is smaller than the distance
matrix is diagonal, with elements exp½iki ðxi  xi1 Þ
over which the electron wave function decays to zero.
and exp½iki ðxi  xi1 Þ. The total transfer matrix
Thus, the transmission probability can still be signi-
with elements Mpq, p; q ¼ 1; 2 is obtained by multi-
ficant for thin barriers; electrons can pass through
plying the corresponding matrices at each interface
classically forbidden regions, a phenomenon known
and across each layer, and the transmission proba-
as electron tunneling. Tunneling can also occur
bility across the whole structure, from the first to
through a succession of barriers for electrons sepa-
the last, Nth layer is T ¼ vN jAN j2 =ðv1 jA1 j2 Þ ¼ vN =
rated by quantum wells, that is, by regions with real
ðv1 jM11 j2 Þ:
wave numbers. In ballistic structures, quantum tun-
The matrix that relates the outgoing field ampli-
neling is a coherent phenomenon and thus generates,
tudes Bi expðiki xi Þ, Aiþ1 expðikiþ1 xi Þ to the inco-
in this case, constructive or destructive interference
ming field amplitudes Ai expðiki xi Þ, Biþ1 exp
between partially reflected and transmitted waves.
ðikiþ1 xi Þ on either side of an interface is called the
Then, resonances, which correspond to high-trans-
scattering matrix. For free propagation across the ith
mission probability values, can appear similarly to the
layer the scattering matrix is antidiagonal, with el-
occurrence of high intensity values in light diffraction
ements exp½iki ðxi  xi1 Þ and exp½iki ðxi  xi1 Þ.
due to interference between coherent light beams.
In terms of the scattering matrix elements Spq, p; q ¼
1; 2 of a succession of N regions, the total transmis-
sion probability can be expressed as T ¼ vN jS21 j2 =v1 .
In general, the elements of the scattering matrix are
Coulomb Blockade
S11 ¼ r, S12 ¼ t0 , S21 ¼ t, S22 ¼ r0 , with r, t, r0, t0 the
reflection and transmission amplitudes of the struc- In quantum dots, more than in any other confined
ture from left to right and right to left, respectively. structure, a Coulomb interaction between tightly
For a ballistic conductor with a variable cross sec- confined electrons leads to a high sensitivity of elec-
tion, there is scattering between transverse modes, tronic states on the number of particles in the dot and
which is not caused by collisions but by the to the appearance of specific phenomena such as the
transverse mode matching imposed by the electron Coulomb blockade, which reflect the discrete nature
wave function continuity. The difference from the of the electron charge. A Coulomb blockade has been
former case is that not only do the electron wave observed in semiconductor quantum dots and small
numbers ki in the adjacent regions differ, but so do metallic clusters coupled to metallic leads through
the number of transverse modes Mi. More precisely, tunneling barriers. It consists in the opening of a gap
the electron wave function propagating along x has in the energy spectrum at the Fermi energy when an
PMithe ith layer the expression Ci ðx; y; zÞ ¼ electron is transferred at low temperatures from the
k¼1 ½Aik expðiki xÞ þ Bik expðiki xÞfik ðy; zÞ, where reservoir into the dot or cluster (generically, island).
fik ðy; zÞ is the transverse part of the kth mode in the This gap, caused by charge rearrangement and
ith layer. The transmission probability has a more observed when the associated potential change is
complicated expression, but is also obtained by greater than the thermal energy kBT, leads to the in-
imposing the boundary conditions at each interface hibition of further electron tunneling until this
between different layers. charging energy is compensated by an applied bias.
84 Ballistic Transport

A quantitative description of the Coulomb block- threshold. Tunneling cannot occur below this thresh-
ade is simpler in metallic clusters, where size old, and thus no current flows through the island
quantization effects are absent because the ballistic because the Coulomb gap e2/C that opens symmetric
transport conditions are not met; in semiconductor around the Fermi level at V ¼ 0 and C1 ¼ C2 ¼ C=2
quantum dots, the Coulomb blockade coexists with prevents tunneling, no states being available for elec-
the quantization of energy levels. In either case, the trons to tunnel into from the leads and no empty
Coulomb blockade can be understood in terms of a states existing for electrons in the island to tunnel
macroscopic capacitance associated with the system, out. This Coulomb blockade is a consequence of the
if the electrons that tunnel through either junction additional Coulomb energy e2/2C that must be ex-
relax in the dot. Modeling the thin tunnel junctions pended by an electron to tunnel in or out of the is-
as a parallel combination of a tunneling resistance Ri land, the Coulomb gap resembling the energy gap in
and a capacitance Ci, i ¼ 1; 2, as in Figure 7, a semiconductors.
voltage source between the leads performs the work If a bias is applied now such that V4e/C, an elec-
Ws ¼ eVðC1 n2 þ C2 n1 Þ=C to transfer n2 charges tron can tunnel in the island from one lead, the Fermi
through junction 2, and n1 charges through junction energy in the island raises by e2/C, and further tun-
1. Here, V is the applied voltage and C ¼ C1 þ C2 is neling is prohibited by the appearance of a new
the capacitance of the island. The electrostatic energy energy gap unless the voltage increases to V43e/2C
stored in the capacitors is Es ¼ ðn1 eÞ2 =2C1 þ or the extra electron in the island tunnels out through
ðn2 eÞ2 =2C2 ¼ ½C1 C2 V 2 þ ðneÞ2 =2C, with n ¼ n1  the other junction, a case in which the island reverses
n2 the excess electrons accumulated on the island. to the n ¼ 0 state. These correlated tunneling pro-
Then, the change in the total energy Eðn1 ; n2 Þ ¼ cesses into and out of the island produce a net cur-
Es  Ws when a single particle tunnels through rent flow. The I–V characteristic has a staircase-like
junction 1 is DE7 1 ¼ Eðn1 ; n2 Þ  Eðn1 71; n2 Þ behavior if the capacitances or the tunneling resist-
¼ ðe=CÞ½e=28ðen þ VC2 Þ, whereas the change in ances of the two junctions are very dissimilar. The
energy when a particle tunnels through the second charging effects can be observed if e2 =CckB T and
junction is DE7 2 ¼ Eðn1 ; n2 Þ Eðn1 ; n2 71Þ ¼ ðe=CÞ R1;2 ch=e2 .
½e=27ðen  VC1 Þ. The system evolves at zero tem-
perature only when DE7 1;2 40, a condition which can See also: Effective Masses; Meso- and Nanostruc-
only be satisfied for n ¼ 0 if the voltage exceeds a tures; Nanostructures, Electronic Structure of; Tunneling

Island PACS: 73.23.  b; 73.23.Ad; 73.40.Gk; 85.35.Be

Lead Lead

Further Reading
C1 C2
Datta S (1997) Electronic Transport in Mesoscopic Systems.
Cambridge: Cambridge University Press.
R1 R2 Dragoman D and Dragoman M (1999) Optical analogue struc-
tures to mesoscopic devices. Progress in Quantum Electronics
V 23: 131–188.
(a) Ferry DK and Goodnick SM (1997) Transport in Nanostructures.
Cambridge: Cambridge University Press.
Island Tunnel

EF 2
e /C
C capacitance
kx, ky, kz wave vector components along x, y, z
kF Fermi wave number
e electron charge
V = 0, C1 = C2 = C/2 E electron energy
(b) Ec bottom of the conduction band
Figure 7 (a) Island connected to external leads through tun- Ekin kinetic energy of electrons
neling junctions and its equivalent model, (b) Opening of a gap in EF Fermi energy
the energy spectrum at the Fermi energy when V ¼ 0, C1 ¼ EFl, EFr quasi-Fermi levels in the left and right
C2 ¼ C=2 due to the additional Coulomb energy that must be contacts
spent by an electron to tunnel in or out of the island. Es electrostatic energy
Bioelectronics 85

Es,p cut-off energy of the subband or trans- Ntot(k), Ntot(E) total number of spin-degenerate states
verse mode labeled by p with a wave number less than k, or with
F electromagnetic field component an energy less than E
f(E) Fermi–Dirac distribution function r, r0 reflection amplitudes from left to right
fl(E), fr(E) Fermi–Dirac distribution functions in and right to left
the left and right contact R resistance
Gc conductance of a ballistic conductor Rc contact resistance
Gpq ballistic conductance associated to the Spq scattering matrix elements
electron transfer from terminal q with a t, t0 transmission amplitudes from left to
quasi-Fermi energy level EFq to terminal right and right to left
p with a quasi-Fermi energy level EFp T temperature; also electron transmission
I current probability
KB Boltzmann constant v electron velocity
Lfp mean free path V potential energy; also applied voltage
Lph phase relaxation length Ws work
Lx, Ly, Lz dimensions of the free space propagat- r2D,p density of states per unit area S ¼ Lx Ly
ion region along x, y, z r1D(kx) density of states in the kx plane per unit
m electron effective mass length
M(E) number of subbands occupied by elec- r0D density of states in quantum dots
trons with energy E W unit step function
Mpq transfer matrix elements y angle of propagation
n electron density per unit area C(x,y,z) envelope electron wave function

E Katz, The Hebrew University of Jerusalem, upon an electron transfer between the enzyme active
Jerusalem, Israel center and the respective substrate molecules. For
& 2005, Elsevier Ltd. All Rights Reserved. example, the enzyme glucose oxidase (GOx), which
catalyzes oxidation of glucose, includes the active
center (also called cofactor) flavin adenine dinucleo-
tide (FAD) that accepts electrons from the substrate
Introduction glucose and donates them to the natural acceptor
O2. If the enzyme is immobilized on a conductive
Bioelectronics is a rapidly progressing interdisci-
support and an efficient electron transfer between
plinary research field that aims to integrate bio-
them is provided, the electron transfer from the en-
molecules and electronic elements, for example,
zyme active center could be directed to the con-
electrodes, field effect transistors, or piezoelectric
ductive support instead of the natural acceptor, thus
crystals, into functional systems. Surface engineering
producing an electronic signal. However, most of the
of biomaterials, such as enzymes, antigen–antibod-
redox enzymes deny direct electrical contact with
ies, or deoxyribonucleic acids (DNA), on the elec-
electrodes. The electron-transfer theory developed by
tronic supports controls the electrical properties
R Marcus implies that the rate constant of the elec-
of the biomaterial–transducer interface and enables
tron transfer between an electron acceptor and a
the electronic transduction of biocatalyzed transfor-
donor pair is given as follows, eqn [1]:
mations and biorecognition events on the transduc-
ers. Bioelectronic sensing devices, biofuel elements,
exp½bðd  d0 Þ  expðDG0 þ lÞ2
and biotemplated circuitries have been developed ket p ½1
recently. 4RTl

In eqn [1], d and d0 are the actual distance and the

Bioelectronic Systems Based on van der Waals distance, respectively, that separate
Biocatalytic Redox Enzymes the donor–acceptor pair; DG0 and l correspond to
the free-energy change and the reorganization energy
Natural redox enzymes catalyzing oxidative or re- accompanying the electron-transfer process, res-
ductive transformations of organic molecules operate pectively; and b is the electronic coupling constant.
86 Bioelectronics

Thus, theory predicts that the electron transfer rate reveal electrical contact with the conductive support,
constant between the donor and acceptor sites ex- the experimental electron transfer turnover rates are
ponentially decreases upon increase of the distance far lower than those observed for the enzymes with
separating the couple. As the enzyme active center the natural donor or acceptor substrates. This was
and the conductive support may be considered as a attributed to the random, nonoptimized, position-
donor–acceptor pair, the embodiment of the active ing of the relay units relative to the redox centers
centers in the protein matrices spatially insulates of the respective enzymes, that themselves lack a
the active centers and blocks the electrical contact structural orientation in the integrated enzyme–
between the redox enzymes and conductive supports. electrode system. Thus, for optimal electrical com-
The understanding of the fundamental limitations for munication between the redox centers of the enzymes
the electrical contacting of the redox enzymes with and the electrodes, it is essential to develop means
electrodes enables the development of synthetic to align all enzyme molecules on the electrode in a
methods to electrically wire redox enzymes and elec- configuration so that the redox centers are as close
tronic transducers. Realizing that the lack of electri- as possible to the electrode surface, and to place an
cal contact between redox enzymes and the appropriate electron relay unit between the redox
electrodes originates from the distance separating center and the electrode support. The architecture
the donor–acceptor couple, one may suggest that of electrically contacted enzymes in an aligned
the shortening of the electron-transfer distances by configuration was accomplished by a surface-recon-
intermediate electron relay units may facilitate the stitution method. The surface-confined cofactor
electron transport between the active centers and unit (FAD, (3)) was incorporated into the GOx en-
the electrode (Figure 1). zyme molecule instead of the natural cofactor, thus
The electrical contacting of redox enzymes that allowing the enzyme alignment on the electrode
defy direct electrical communication with electrodes surface in a form of a stable complex. An electron-
can be established by using charge carriers as inter- transfer mediator (e.g., pyrroloquinolino quinone
mediates between the enzyme active center and (PQQ), (4)) located between the FAD-cofactor
the electrode. These artificial electron donor or and the electrode surface provides efficient electron
acceptor molecules (in the case of reductive or transfer from the enzyme FAD active center to the
oxidative enzymes, respectively), usually referred electrode (Figure 3a). Bioelectrocatalyzed oxidation
to as electron-transfer mediators, can be accepted of glucose to gluconic acid by the surface-reconsti-
by many redox enzymes in place of their natural tuted GOx enzyme results in the generation of
oxidants or reductants. The redox enzymes were an electronic signal in the electrode. Similarly,
incorporated into polymer matrices consisting of nicotine adenine dinucleotide (NAD þ )-dependent
electron-transfer mediator units (e.g., Os-complexes enzymes, for example, lactate dehydrogenase (LDH),
(1)) (Figure 2a), or the redox mediator molecules were associated with a surface-confined NAD þ -
(e.g., ferrocene derivatives (2)) were covalently cofactor (5) linked to the electrode surface via a
tethered to the enzyme molecules (Figure 2b) to catalyst molecule (PQQ) providing its fast electro-
provide electron transport between the active centers chemical oxidation (Figure 3b). Since the NAD þ -
of the enzymes and electrodes. Although redox cofactor is weakly bound to the respective enzymes,
enzymes anchored to electrodes by these means the enzyme molecules after formation of the surface-
confined complex were additionally cross-linked
to stabilize the biocatalytic interface. Bioelectro-
Transduced current catalyzed oxidation of lactate to pyruvate by the
Active center
surface-integrated LDH enzyme results in the gene-
ration of an electronic signal in the electrode. There-
Electron-relay fore, an efficient electron transport from the enzyme
e− R active centers (e.g., FAD or NAD þ ) via the inter-
Substrate mediate relay units (e.g., PQQ) to the conductive
support was achieved, thus providing electronic

R+ Enzyme
signals generated by the biocatalytic systems. It

Product should be noted that the relay units are chemically

Long distance for reduced and oxidized upon the electron-mediating
electron transfer process that could introduce some limitations to the
Figure 1 Spatially isolated redox active centers of enzymes rate of the process.
could be electrically contacted on electrode surfaces with the use Application of conductive nanosized objects, such
of electron relay molecules. as metal nanoparticles or carbon nanotubes, for
Bioelectronics 87

Substrate Product

R x y z

e− e−
Enzyme N N N+
R e− Os 2+/3+
(bpy)2 NH2
R Cl
R R e− R 1


Enzyme O

Fe N
(b) 2

Figure 2 Redox enzymes electrically contacted on the electrode by means of: (a) the redox polymer (1) surrounding the enzyme
molecules and (b) the redox mediator (e.g., ferrocene, 2) tethered covalently to the enzyme molecules.

wiring of redox enzymes allowed the effective electron The electrically wired redox enzymes could
transfer between the enzyme active centers and the generate electrical power sufficient for construction
conductive supports using metal-like conductivity in- of biofuel cells. The biofuel cells are primary sources
stead of the chemical redox-mediating process. Gold of electrical energy extracted from organic materials
(Au) nanoparticles or single-walled carbon nanotubes using electrochemical processes catalyzed by bio-
(SWCNTs) were functionalized with the FAD-cofac- logical substances (e.g., redox enzymes). They in-
tor units and assembled on electrodes. Reconstitution clude two enzyme–biocatalytic electrodes: the first
of GOx enzyme molecules on the functionalized inter- electrode (anode) provides oxidation of the organic
faces resulted in the biocatalytic interfaces, where the substrate by an enzyme (e.g., oxidation of glucose
enzyme active centers are electrically wired to the con- by GOx), thus injecting electrons into the external
ductive support via the Au nanoparticles (Figure 4a) electrical circuit, whereas the second electrode
or SWCNTs (Figure 4b). The nanosize of the wires (cathode) is used to withdraw electrons from the
allowed a short distance between them and the FAD circuit and transfer them to a chemical acceptor
cofactor, thus providing effective electron transfer, (e.g., O2). The tailoring of efficient electron transfer
whereas the conductivity of the Au nanoparticles and at the enzyme-modified electrodes could enable
SWCNTs provided the efficient electron transport to specific biocatalytic transformations that compete
the macroconductive support. kinetically with any chemical reaction of the elec-
The electric current provided by the enzyme– trodes or of the biocatalysts with interfering subst-
electrode systems depends on the rate of the bio- rates (e.g., substrate transport from the counter
catalytic reactions and could be proportional to the compartment, oxygen). This would enable the
concentration of the substrate, if the entire process is design of noncompartmentalized biofuel cells, where
limited by the substrate diffusion to the biocatalytic the biocatalytic anode and cathode are immersed in
interface. Thus, the generated electronic signal could the same phase with no separating membrane. In
be used to design amperometric biosensors. For a working example, an anode consisting of GOx
example, the GOx enzyme electrically contacted reconstituted onto a PQQ–FAD monolayer for the
via an Au nanoparticle generates the current pro- biocatalyzed oxidation of glucose was coupled to a
portional to the glucose concentration in a certain cathode composed of an aligned cytochrome c/cyto-
concentration range. chrome oxidase (Cyt c/COx) couple that catalyzes
88 Bioelectronics

e− Glucose N
e− 3
GOx Gluconic H H
acid O O N
4 O− O− H H N N

Lactate N
e− OH OH H H N N
e− LDH O
H NAD+ Pyruvate

Figure 3 Redox enzymes electrically contacted on the electrode by means of: (a) reconstitution of the GOx on the electrode surface
functionalized with the FAD (3)-cofactor and PQQ (4)-mediator; and (b) integration of LDH with the NAD þ (5)-cofactor and the PQQ-

the reduction of O2 to water (Figure 5). Since the the source–drain current (Figure 6). The amplified
reconstituted GOx provides extremely efficient electronic signal is dependent on the rate of the
biocatalyzed oxidation of glucose that is unaffected biocatalytic reaction provided by the enzyme and
by oxygen, the anode can operate in the presence proportional to the enzyme–substrate concentra-
of oxygen. Thus, the biofuel cell uses O2 as an tion, thus allowing potentiometric biosensors. Nano-
oxidizer and glucose as a fuel without the need for sized bio-ISFETs were constructed with a single
compartmentalization. The glucose–O2 biofuel cell SWCNT operating as a gate modified with a few
element, or similar cells, might act as implanted biomolecules (e.g., redox enzymes) providing a signal
self-fueled electrical energy generation devices in further electronically amplified by the source–drain
living organisms providing power for other bioelec- current.
tronic elements.
Redox enzymes could be integrated with a gate
of a field-effect transistor. An ion-selective field-
Bioelectronic Systems Based on Protein
effect transistor (ISFET) operates with a gate open
to a chemical system, which provides electrical
or DNA Biorecognition Systems
signals governing the current between source and Biomolecules could provide very specific recognition
drain electrodes. This signal is a potential change processes resulting in the formation of complexes
generated by chemical means on the gate. The redox between them. For instance, biological immunosys-
enzymes associated with the ISFETs gate could tems protecting organisms from foreign species are
reduce or oxidize co-immobilized electron relay based on affinity complex formation between anti-
units, thus providing the potential change on the bodies (recognition proteins produced by the organ-
gate further amplified by the respective change of ism) and antigens (foreign target molecules).
Bioelectronics 89

Biological genetic information is saved in deoxy- of the affinity complex formation with the respective
ribonucleic acid (DNA) molecules producing complementary biomolecule. The bioaffinity com-
complexes with complementary DNA molecules. plex formation in a thin film near the conductive
Association of a biorecognition element such as or semiconductive surface of the electronic trans-
antigen (or antibody) or a DNA molecule with an ducer yields a chemically modified film on the sur-
electronic transducer allows electronic transduction face that alters its interfacial properties. Blocking
the electrode surface with bulky biomolecules upon
immunorecognition between antigens and antibodies
(Figure 7a), or formation of the charged interface
e− Glucose upon hybridization of negatively charged DNA mol-
ecules (Figure 7b), perturbs the double-charged layer
S S FAD existing at the electrode–electrolyte interface result-
GOx Gluconic ing in the increase of its thickness, ddl, and in the
acid insulation of the electrode surface with respect to
redox labels added to the solution. This results in
(a) the capacitance change and electron-transfer resist-
ance change at the interface, respectively, that could
e− e− Glucose be read out by impedance spectroscopy. Two main
approaches were used to increase these interfacial
changes in order to amplify the electronic signal
GOx Gluconic generated upon the biorecognition events: (1) second-
SWCNT acid ary binding of bulky units (proteins, functionalized
liposomes, or nanoparticles) complementary to the
primarily biorecognition complex formed on the
Figure 4 Redox enzymes electrically contacted on the elec- interface (Figure 8a), and (2) labeling the biorecogni-
trode by means of: (a) reconstitution of the GOx on the electrode tion elements with biocatalytic entities which pro-
surface functionalized with the FAD-cofactor and an Au-nano-
vide a cascade of reactions on the surface, for
particles plugging into the enzyme; and (b) reconstitution of the
GOx on the electrode surface functionalized with the FAD-co- example, resulting in the precipitation of an insolu-
factors covalently linked to conductive SWCNTs assembled on ble material on the sensing surface (Figure 8b). Both
an Au electrode surface. the amplifying processes are triggered by the primary


l cell

A R load

e− e− Glucose O2 e− e−


Gluconic H2O

Figure 5 A noncompartmentalized biofuel cell, which generates electrical energy by the oxidation of glucose and reduction of O2
biocatalyzed by the GOx reconstituted on the anode and Cyt c/COx-assembly on the cathode, respectively.
90 Bioelectronics

recognition event and cannot proceed in the absence biorecognition process and the secondary biocatalytic
of the antigen–antibody or DNA–complementary amplification, respectively, with the electronic trans-
DNA complex at the surface. Combination of high duction of the biochemical signal allows bioelectronic
specificity and sensitivity provided by the primary immunosensing and DNA sensing.
Single-base DNA mismatches analysis, which is
Vgs important for the analysis of genetic disorders, could
exemplify the bioelectronic DNA sensing. The pro-
Substrate cess is exemplified by the identification of the DNA
Reference Product
mutant (6), in which the T-base in the normal gene
(6a) is substituted with the G-base (Figure 9). A
e− Enzyme
thiolated DNA primer (7) complementary to the
Ox /Red SiO2 analyzed gene up to one base before the mutation
site is assembled as the sensing interface on an Au
electrode. The normal gene (6a), as well as the
Electron path
Source Drain mutant (6), associate with the sensing interface to
generate the double-stranded DNA configuration.
Reaction of the resulting interface with biotinylated-
deoxycytosine triphosphate (dCTP) base in the pres-
V ds ence of a polymerase enzyme results in the coupling
Figure 6 Redox-enzyme integrated with a gate of a field-effect
of the biotinylated-tagged-base only to the layer
transistor (Vgs – gate-source voltage; Vds – drain-source voltage; which includes the mutant site. Subsequent associ-
Ox/Red – a redox-relay associated with the redox-enzyme). ation of the avidin-alkaline phosphatase (AlkPh)

dl  dl

+ − + −

+ − + −

+ − Antibody
+ −
+ +
− −
− −
+ +
+ +
− −
+ +

Antigen −
Electrolyte complex

− − + − − − − − − − − +
− −
+ − − − − − − − −
− − − − − − − − − − +
− − Complementary-DNA
+ − − − − − − − − +
− −
− − − − − − − − +
− −
+ +
DNA-primer − − − − − − − −
Double-stranded DNA
Figure 7 Alteration of interfacial properties of the transducer surface upon: (a) binding of bulky biological molecules (e.g., antibody) to
the biorecognition elements (e.g., antigen) on the surface; (b) formation of the charged complexes (e.g., double-stranded DNA) on the
transducer surface.
Bioelectronics 91





Insoluble Substrate
antibody product

Figure 8 Amplification of the electronic read-out signal from the primary biorecognition events achieved by the enhancement of the
interfacial changes upon: (a) secondary biorecognition processes; and (b) biocatalytic reactions yielding insoluble materials on the

N 8

Mutation Avidin
site Biotin-dCTP

A AlkPh
6 B
G C A AlkPh C
Analyte DNA DNA
Polymerase I P
(Klenow fragment)
Br N
N Br
Insoluble H O CI



Figure 9 The detection of a single-base mismatch in DNA amplified by a biocatalytic reaction on the transducer interface and read-out
by impedance spectroscopy.
92 Bioelectronics

biocatalytic conjugate, which stimulates the oxi- which are able to exchange electrons with the
dative hydrolysis of 5-bromo-4-chloro-3-indoyl electrode support, are used as labels upon hybridiza-
phosphate (8) to the insoluble indigo product (9), tion of complementary DNA molecules (Figure 10a),
provides a route for the amplified transduction of the or upon biocatalytic replication of a long DNA
existence of the mutation site in (6a). Faradaic im- sequence (Figure 10b). For example, the amplified
pedance spectroscopy was used as an electronic amperometric transduction of DNA sensing with a
means to transduce the biochemical signal. bioelectronic system was demonstrated by the gene-
Another approach to the bioelectronic sensing of ration of a redox active DNA strand complementary
the recognition events uses incorporation of redox to the analyzed DNA, which activates a secondary
active species into the bioaffinity complex generated bioelectrocatalytic process (Figure 10b). The complex
on a conductive transducer surface. Redox molecules, generated between the 7249-base M13 phage viral

Red / Ox

Analyte DNA
Amplifying DNA-
Red/Ox label

Red / Ox

DNA primer

Au electrode

Viral nucleic acid
M13φ s

Au electrode

Gluconic O O
e− O O O O N

Fc = HO P O P O P O O 12
Fc O− O− O−

Figure 10 Labeling DNA molecules with redox units for amperometric read-out of the DNA reactions on the transducer interface. (a)
Hybridization of complementary DNA molecules and (b) biocatalytic replication of DNA.
Bioelectronics 93

DNA (10) and a 27-base primer (11) associated with carboxylic acid was covalently coupled to N6-(2-
an Au electrode was replicated in the presence of a aminoethyl)-FAD to yield a photoisomerizable nit-
polymerase enzyme and a mixture of nucleotides that rospiropyran-FAD cofactor (13). The native FAD
includes ferrocene-modified deoxyuridine tripho- cofactor was extracted from GOx and the modified
sphate (dUTP) (12). The resulting ferrocene-labeled FAD cofactor (13) was reconstituted into the apo-
DNA replica was then coupled to GOx and, upon glucose oxidase (apo-GOx) (Figure 11a). This recon-
electrochemical oxidation of the ferrocene units, the stituted enzyme includes a photoisomerizable unit
GOx-biocatalyzed oxidation of glucose was acti- directly attached to the redox center of the enzyme,
vated. Because the enzyme exhibits a high turnover and hence the enzyme is predisposed for optimized
rate, the electrical output from this process represents photoswitchable bioelectrocatalytic properties. The
an amplified electrical signal of the primary recognit- photoisomerizable enzyme was assembled on an Au-
ion event between the M13 phage viral DNA and the electrode as outlined in Figure 11b. The bioelectro-
sensing nucleic acid associated with the electrode. catalytic interface was reversibly switched ‘‘off’’ and
Metal nanoparticle labels linked to the bioreco- ‘‘on’’ by light signals resulting in photoisomerization
gnition components (e.g., DNA molecules) were of the molecular unit bound to the FAD cofactor
electrochemically oxidized, thus generating the elec- between spiropyran (13) and merocyanine (14) iso-
tronic signal reporting on the formation of a bioaf- meric states, respectively. The electron transport be-
finity complex on the sensing interface. Catalytic tween the photoisomerizable enzyme cofactor and
features of metal nanoparticles that enable the elect- the electrode surface was achieved with the use of a
roless deposition of metals on the nanoparticle clus- diffusional electron-transfer mediator (e.g., ferrocene
ters allow the enlargement of the particles, thus monocarboxylic acid (15)).
providing further amplification of the read-out signal. The distance between surface-confined redox pro-
teins and the bioelectronic transducer surface could
be controlled by an applied electrical potential. Bi-
asing the electrode surface at potentials more positive
Switchable and Tunable Bioelectronic
than a zero-charge potential results in the positive
charging of the surface, thus repelling positively
Bioelectrocatalytic and biorecognition assemblies on charged protein molecules (e.g., cytochrome c). The
electronic transducer interfaces that can be internally longer distance, which is allowed by a long flexi-
‘‘tuned’’ or turned on and off during their operation ble molecular spacer, results in a slow kinetics of
are highly desirable. The use of such interfaces could the electron transfer between the electrode and the
allow the construction of variable output biofuel protein redox active center. Opposite biasing yield-
cells or variable range biosensors. Various physical ing a negative electrode surface results in a short
input signals such as light, magnetic field, or electri- distance upon electrostatic attraction of the posi-
cal potential applied on the bioelectronic interfaces tively charged protein, thus providing a short elec-
have been used to switch and tune bioelectronic tron-transfer distance and the fast electron exchange
transduction of the generated chemical signals. (Figure 12).
For instance, photoswitchable biomaterials open a The magnetic control of bioelectrocatalysis is a
route to optobioelectronic systems. Photoisomeriz- novel concept of bioelectronics. Figure 13 outlines
able molecules undergo reversible structural changes the method to control by an external magnetic field
when exposed to specific energies of light. Their two the bioelectrocatalytic oxidative functions of a redox
states often differ considerably, for instance, in their enzyme, for example, GOx. Magnetic particles (e.g.,
charge, polarization, or shape. If an enzyme molecule Fe3O4) functionalized with a relay unit, R (e.g., a
is functionalized with photoisomerizable groups, ferrocene unit), are used to electrically contact the
then the local environment of that enzyme will redox enzyme (e.g., GOx) and the electrode. Posi-
depend on the state of those groups. As the confor- tioning of a magnet below the electrode attracts
mation of the enzyme is very sensitive to its the functionalized-magnetic particles to the elec-
environment, the shape and therefore the activity trode. This enables the oxidation of the relay unit by
of the enzyme may be controlled by photonic input. the conductive support, and subsequently activates
To optimize the photoswitchable bioelectrocatalytic the bioelectrocatalyzed oxidation of the substrate
features of redox enzymes, the site-specific function- (e.g., glucose) by the oxidized relay associated with
alization or mutation of the active-site microenv- the magnetic particles. Transfer of the external
ironment is essential. This has been accomplished, magnet to an upper position lifts the functionalized-
for example, by the reconstitution of GOx with a magnetic particles upward, and removes them from
photoisomerizable FAD-cofactor. Nitrospiropyran the electrode support. This prevents the oxidation of
94 Bioelectronics




Native GOx Apo-GOx Reconstituted GOx


Fc Fc
e− CO2H Gluconic CO2H
λ > 475 nm
320 nm
MRH+ <λ< SP
380 nm
+ +
Fc Glucose Fc
CO2H CO2H Glucose


+ OH

13 14 15

Figure 11 (a) Implanting of a photoisomerizable FAD derivative into apo-GOx to yield a photoisomerizable enzyme. (b) Switching ‘‘on’’
and ‘‘off’’ the photoisomerizable enzyme by light signals.

+ − Cytochrome c was accomplished by magnetic attraction of the

Cytochrome c
+ − +
magnetic particles to the electrode support fol-
+ +
+ lowed by the rotation of the magnetic particles on
+ −


+ e− +
the electrode by means of an external rotating
+ e− − magnet. The rotation of the redox-functionalized
+ Slow
− Fast + + magnetic particles turns them into circularly rotating
microelectrodes. As a result, the redox-activated
Distance Distance bioelectrocatalytic process mediated by the function-
(a) (b) al particles is controlled by convection rather than
diffusion of the substrate to the microelectrodes.
Figure 12 Potential-induced changes of the protein-surface
distance resulting in the alteration of the electron transfer kinetics. Accordingly, enhanced amperometric responses of
the particle-mediated bioelectrocatalytic processes
are anticipated, and the resulting currents should be
the relay units and the bioelectrocatalyzed oxidation controlled by the rotation speed of the particles.
of the substrate is switched off. By alternate position- Figure 14 depicts the amplified amperometric anal-
ing of the external magnet below and above the elec- ysis of glucose by the rotation of ferrocene (16)-
trode surface, the bioelectrocatalyzed oxidation of the functionalized magnetic particles on the electrode
substrate is reversibly switched between ‘‘on’’ and support. The rotating magnetic particles are func-
‘‘off’’ states, respectively. The magnetic control of a tionalized with an electron-transfer mediator, and
bioelectrocatalytic reduction process can be similarly the interaction with GOx and glucose is convec-
controlled by the use of relay-functionalized magnetic tion controlled. The bioelectrocatalysis in the pres-
particles that are reduced at the electrode support. ence of rotating magnetic particles can be used to
An important advance in the magnetic control of enhance biosensing events and to yield the amplified
electrocatalytic and bioelectrocatalytic transformations electronic signal.
Bioelectronics 95

Magnet e− Enzyme Substrate

Enzyme Substrate
Fe3O4 R

Product e−

Fe3O4 R

Au Au

Enzyme = GOx R= Fe Substrate = glucose Product = gluconic acid

Figure 13 Magneto-switched bioelectrocatalyzed oxidation of glucose in the presence of electron relay-functionalized magnetic

accepted that different miniaturization methods need

to be developed in order to overcome this barrier.
e Gluconic While the lithographic methods use ‘‘top’’ to ‘‘down’’
Agitation of acid miniaturization of patterns, the alternative approach
solution by of the ‘‘bottom-up’’ construction of objects has been
nanoparticle e−
movement GOx
suggested as a means to overcome the lithographic
limitations. That is, the construction of objects on
molecular or supramolecular templates could gene-
e− rate nanometer-size features. Nanowires are consider-
Fe ed as building blocks for self-assembling logic and
memory circuits in future bionanoelectronic devices.
Rotation Different biomaterials, for example, proteins and
nucleic acids, may act as important building blocks
for functional bionanocircuitry, and eventually may
S N provide bionano-elements for the construction of
nanodevices. Biological templates have been used to
Magnet direct the nucleation, deposition, and assembly of
inorganic micro- and nanostructures. Among the
Rotation different biomaterials, DNA is of specific interest as
a template for the construction of nanocircuitries.
Several arguments support the use of DNA as a
O (CH2)5 N
Fe3O4 Si N N H Fe future building block of nanostructures: (1) Nucleic
O O H H O acids of pre-designed lengths, base-orderings, and
16 shapes can be synthesized, and complex structures
Figure 14 Bioelectrocatalytic oxidation of glucose in the pres- generated by self-assembly methods. (2) Nature
ence of GOx and ferrocene-functionalized magnetic particles en- provides us with an arsenal of biocatalysts that can
hanced upon circular rotation of the particles by means of an manipulate DNA. These enzymes may be considered
external rotating magnet. as tools for shaping the desired DNA and eventually
for the generation of nanocircuitry. For example,
ligase ligates nucleic acid, endonucleases affect the
Biomaterial-Based Nanocircuitry for
specific scission of nucleic acids, telomerase elon-
Bioelectronic Systems
gates single-stranded nucleic acids by telomer units,
The miniaturization of objects by lithographic meth- and polymerase replicates DNA. These biocatalysts
ods reaches its theoretical limits. It is generally represent ‘‘cut’’ and ‘‘paste’’ tools for the formation
96 Bioelectronics


P irradiation



N HN Au =
Au O O O O
O 17 P


50 nm
Figure 15 (a) Assembly of an Au-nanoparticle wire in the polyA/polyT template. (b) AFM image of an Au-nanoparticle wire in the
polyA/polyT template. (Adapted from Patolsky F, Weizmann Y, Lioubashevski O, and Willner I (2002) Au-nanoparticle nanowires based
on DNA and polylysine templates. Angewandte Chemie International Edition 41: 2323–2327, with permission.)

of DNA templates and by the application of the (4) Different proteins bind specifically to certain nu-
replication biocatalyst, the design of future ‘‘facto- cleic acid sequences. This allows the addressable as-
ries’’ of nanowires may be envisaged. (3) The inter- sembly of complex DNA–protein structures. Such
calation of molecular components into DNA and the protein–DNA complexes may either act as address-
binding of cationic species, for example, metal ions able domains other than the DNA for the selective
to the phosphate units of nucleic acids allows the deposition of metals, or alternatively, may act as
assembly of chemically active functional complexes. temporary shielding domains that protect the DNA
Bioelectronics 97

from metal deposition. Such insulated domains may double-stranded DNA template. The continuous ap-
then be used for the deposition of other metals or pearance of the Au-nanoparticle wire is due to the
semiconductors, thus enabling the fabrication of dimensions of the scanning AFM-tip, and in reality,
patterned complex structures. most of the particles are not in intimate contact
With the vision that DNA may act as a template one with another. The possibility to arrange the
for the generation of nanocircuitry, attempts were Au-nanoparticles on the DNA template allows fur-
made to explore the possibility of organizing DNA- ther catalytic enlargement of the particles by an elect-
crosslinked semiconductor nanoparticles and DNA- roless deposition process (e.g., reaction of AuCl–4 with
based metal nanoparticle nanowires on surfaces. NH2OH) to yield continuous conductive nanowires.
Since nanoparticles are loaded on the DNA template Practical applications of the nanowires require
with gaps between them, the conductivity of metallic their electrical contacting with macro- or microelec-
nanoparticle aggregates on a DNA template can be trodes. Toward this goal, a single nanowire produced
enhanced upon the chemical deposition of another on a DNA template bridging two microsize electrodes
metal (e.g., Ag deposition on Au aggregates) filling was constructed. Two microelectrodes facing each
the gaps and forming a continuous conductive nano- other (12–16 mm separation) were functionalized
wire. The binding of the primary metallic clusters to with 12-base oligonucleotides that were then bridged
the template DNA for the subsequent catalytic dep- with a 16 mm long l-DNA (Figure 16). The resulting
osition of wires on the DNA frame may be accom- phosphate units of the DNA bridge were loaded with
plished by several means: (1) The reduction of metal Ag þ ions by ion exchange, which were then reduced
ions linked to the phosphate groups to metallic to Ag metal by hydroquinone. The small Ag
seeds linked to the DNA. (2) The use of metal or aggregates produced along the DNA backbone were
semiconductor nanoparticles functionalized with then used as catalysts for further reductive deposition
intercalator units. Intercalation of the molecular of silver, eventually leading to the formation of an Ag
components into double-stranded DNA leads to the nanowire. This micrometer-sized element had a typ-
association of the nanoparticle to the DNA template. ical width of 100 nm and a granular morphology (as
(3) The synthesis of DNA with functional tethers that determined by AFM).
enable the covalent attachment of the metal or sem- The use of biomaterials as templates for the gene-
iconductor nanoparticles to the DNA. (4) The syn- ration of nanostructures and nanocircuitry in the
thesis of single-stranded DNA that includes constant presence of nanoparticles is in an early phase of
repeat units (e.g., telomers) and the hybridization of development. The viability of the concept has been
metal or semiconductor nanoparticles functionalized proven, and nanowires of controlled shapes and elec-
with short nucleic acids, that are complementary to tronic functions have been generated by the incorpo-
the single-stranded DNA repeat units. ration of nanoparticles on template biomaterials. The
Figure 15a exemplifies the method for assembling biggest challenges are ahead however, and exciting
the Au-nanoparticles on the DNA template using Au- systems are envisaged for the future. Besides the int-
nanoparticles functionalized with an intercalator. eresting nano-architectures that may be generated,
The amino-functionalized psoralen (17) is reacted fundamental problems, such as probing charge trans-
with the Au55-nanocluster (diameter 1.3 nm) that in- port phenomena in such nanoparticle-biomaterial
cludes a single N-hydroxysuccinimide active ester
functionality (18), to yield the psoralen-functionali-
zed Au55-nanoparticle (19). As psoralen acts as a
specific intercalator for A–T base pairs, the func-
tionalized Au55-nanoparticles were reacted with the
pA/pT-ds-DNA (20). Subsequently, the assembly was 12−16 µm
irradiated with UV light to induce the 2p þ 2p cyclo- Ag+
addition reaction between the psoralen units and the
(i) Hydroquinone
thymine base sites of DNA. This latter process fixes OH−
covalently the Au55-nanoparticles to the DNA ma- (ii) Ag+ Ag+ Ag+ Ag+
trix. Figure 15b depicts the AFM image of the re- hydroquinone
H+ Ag+ Ag+ Ag+
sulting nanoparticle wire. A B600–700 nm long
nanoparticle wire is formed. Its width corresponds Conductive
to B3.5–8 nm and it is controlled by the width of the silver wire
DNA template. The height of the wire is B3–4 nm, Figure 16 The construction of a wire bridging two microelec-
consistent with the fact that the Au-nanoparticles trodes by the deposition of nanoparticles on a bridging DNA
intercalate into the DNA on opposite sides of the strand, followed by silver deposition.
98 Bioelectronics

systems, may be highlighted. Nonetheless, the organ- Engineering of; Protein Folding, Evolution and; Red Blood
ization of nanoparticle architectures on biomaterial Cells, Physical Properties of; Rhodopsin and the First
templates involves relatively simple motifs that do Step in Vision; Single-Molecule Methods in Biophysics;
not make full use of the biomaterial’s catalytic func- Solid-State NMR Structural Studies of Proteins.
tions. The replication of nanoparticle-labeled DNA,
and specifically the tailoring of polymerase chain
reaction (PCR) for the formation of nanoparticle
PACS: 81.16. Dn; 81.16. Fg; 82.39.Jn; 82.39.Pj;
82.39.Rt; 82.45.Fk; 82.45.Jn; 82.45.Mp; 82.45.Tv;
wires, represents a novel approach to design biolo-
82.47.Rs; 82.65. þ r; 82.80.Fk; 87.14.Ee; 87.14.Gg;
gical machines for the synthesis of nanoparticle wires 87.16.Xa; 87.68. þ z; 87.80.Tq; 87.90. þ y
and circuitry.

Further Reading
Three facets of bioelectronics – biomaterial-based
Bone SS and Zaba B (1992) Bioelectronics. New York: Wiley.
electronic sensors, biofuel cells, and biomaterial-
Braun E, Eichen Y, Sivan U, and Ben-Yoseph G (1998) Nature 391:
based electronic circuitry – continue to progress rap- 775–778.
idly in research, and some of its applications have Carrara S, Facci P, Sivozhelezov V, Adami M, Erokhin V, et al.
been developed commercially. Other topics, such as (1996) Molecular Bioelectronics. New York: World Scientific.
hybrid systems of neural network and electronic Chakraborty T, Peeters F, and Sivan U (eds.) (2002) Nano-Physics
elements, biomaterial-based computers, and bioma- and Bio-Electronics: A New Odyssey. London: Elsevier.
Christof M, Niemeyer CM, and Mirkin CA (eds.) (2004)
terial-based micromachinery devices represent other Nanobiotechnology: Concepts, Applications and Perspectives.
opportunities in bioelectronics that show promise Weinheim: Wiley-VCH.
for future applications. Despite this outlook, how- Gizeli E and Lowe CR (eds.) (2002) Biomolecular Sensors. Camb-
ever, challenges in bioelectronics remain, including ridge: Taylor and Francis.
the development of implantable fuel cells, biomate- Hoffmann K-H (ed.) (2002) Coupling of Biological and Electronic
Systems. Berlin: Springer.
rial-based micromachines (e.g., prosthetic units), and Katz E, Shipway AN, and Willner I (2002) Mediated electron-
the fabrication of functional electronic circuitry. transfer between redox-enzymes and electrode supports. In:
Recent advances in nanotechnology, the availability Wilson GS, Bard AJ, and Stratmann M (eds.) Encyclopedia of
of quantum-size nanoparticles and nanotubes of Electrochemistry, vol. 9: Bioelectrochemistry, chapter 17, pp.
559–626. Weinheim: Wiley-VCH.
unique electronic and catalytic properties, and
Katz E, Shipway AN, and Willner I (2003) Biofuel cells: functional
nanoscopic tools for manipulating surfaces could design and operation. In: Vielstich W, Gasteiger H, and Lamm A
open nanobioelectronics as a new research field. The (eds.) Handbook of Fuel Cells – Fundamentals, Technology,
electronic detection of biorecognition events of single Applications, vol. 1, part 4, ch. 21, pp. 355–381. Weinheim:
molecules, the optical or electronic read-out of Wiley-VCH.
biological processes by functional nanoparticles, the Nicolini CA (ed.) (1998) Biophysics of Electron Transfer and Mo-
lecular Bioelectronics. New York: Plenum Press.
high-throughput analysis of numerous genes or pro- Patolsky F, Lichtenstein A, Kotler M, and Willner I (2001)
tein functions on dense sensing arrays, the construc- Angewandte Chemie International Edition 40: 2261–2265.
tion of biomaterial-based electronic devices of Patolsky F, Lichtenstein A, and Willner I (2001) Nature
nanodimensions, and the tailoring of ultrasmall Biotechnology 19: 253–257.
self-powered devices for the controlled release of Willner I and Katz E (eds.) (2005) Bioelectronics: From Theory to
Applications. Weinheim: Wiley-VCH.
therapeutic drugs may be possible as well. Willner I, Blonder R, Katz E, Stocker A, and Bückmann AF (1996)
Journal of the American Chemical Society 118: 5310–5311.
Willner I, Katz E, and Willner B (2000) Layered functionalized
See also: Biomedical Materials; Biomembranes; Biomol- electrodes for electrochemical biosensor applications. In: Yang
ecules, Scanning Probe Microscopy of; DNA and RNA, VC and Ngo TT (eds.) Biosensors and Their Applications,
Biophysical Aspects; Electric and Magnetic Fields in chapter 4, pp. 47–98. New York: Kluwer.
Cells and Tissues; Electromagnetic Biological Effects; Willner I, Katz E, and Willner B (2002) Amplified and specific
Electrophysiology; Elementary Excitations in Biological electronic transduction of DNA sensing processes in monolayer
Systems; Fluorescent Biomolecules; Fluorescent Pro- and thin-film assemblies. In: Brajter-Toth A and Chambers JQ
teins; Folding and Sequence Analysis; Genetic Algorit- (eds.) Electroanalytical Methods of Biological Materials, pp.
43–107. New York: Dekker.
hms for Biological Systems; Ionizing Radiation Effects in
Willner I, Willner B, and Katz E (2003) Bioelectronics: develop-
Biological Materials; Ion Transport in Biological Mem- ment of biosensors, biofuel-cells and circuitry. In: Barsanti L,
branes; Membrane and Protein Aggregates; Metallopro- Evangelista V, Gualtieri P, Passarelli V, and Vestri S (eds.)
teins, Electron Transfer in; Metalloproteins, Structural Molecular Electronics: Bio-sensors and Bio-computers, NATO
Determination of; Neuroscience; Photosynthesis, Physics Science Series, II. Mathematics, Physics and Chemistry – vol. 96,
of; Protein Folding and Aggregation; Protein Folding, pp. 311–339. Dordrecht: Kluwer.
Biological Structures 99

Biological Structures
R M Ottenbrite and R Javan, Virginia Biodegradation
Commonwealth University, Richmond, VA, USA
Biodegradation is the chemical breakdown of mate-
& 2005, Elsevier Ltd. All Rights Reserved. rials by the action of living organisms leading to
changes in physical properties. There are other def-
initions of polymer degradation; however, they all
involve a chemical breakdown in an aqueous
Polymers environment caused by hydrolysis (reactions with
water) and/or enzymes. Biodegradable polymers were
Polymers are materials composed of molecules linked first used surgically as sutures, staples, implants, and
to one another in a linear or branched manner as temporary devices. With current developments being
multiple repeat units. Applications of biomedical made in drug-delivery systems, tissue engineering,
polymer structures include: sutures, maxillofacial and scaffolds for regenerative organs, there is a
implants, dental constructs, joint replacements, card- growing need for new biodegradable materials. The
iovascular devices, artificially engineered tissues, and majority of the products presently in use are com-
drug-delivery and gene-transvection systems. posed of poly(lactic acid), poly(glycolic acid), poly
In order to understand biomedical polymer struc- (e-caprolactone), their copolymers and composites.
tures, one should have some knowledge of the phys- In addition to these polymers, several other impor-
ical properties that are essential for their function in tant degradable polymer systems are available and
a biological system. To accomplish specific biorelat- many more are currently being developed (Figure 1).
ed tasks, the materials used must meet many physical
and biological requirements. These include biocom-
Biological Scaffold
patible and nontoxic, degradable or nondegradable,
water soluble or insoluble, hard or soft, rigid or Biological scaffold is an artificial support system to
flexible, hydrophobic or hydrophilic, and bioactive replace, temporarily or permanently, a biological
or bioinactive materials. To help with the reading of structure. Sutures, bone-fixation devices (bone nails,
this article some concepts such as biocompatibility, screws, or plates), and vascular grafts are early ex-
characterization, biodegradation, and applications amples of such support devices. The degradable scaf-
such as biological scaffolds, hydrogels, polymeric fold implant provides temporary mechanical support
drugs, and drug-delivery systems are addressed. until the natural tissue has developed and regained its
strength. As the tissue assumes integrity, the scaffold
material should concomitantly slowly degrade.
Adjusting the rate of degradation to the regenera-
Biocompatibility refers to the ability of a material to tion process of the surrounding tissue presents one of
perform with an appropriate host response in a spe- the major challenges in the design of a temporary
cific application and incorporates blood and tissue scaffold. For temporary scaffolding to function ef-
compatibility. When a material is characterized for a fectively, a gradual stress transfer should occur. As
bioapplication, mechanical properties, such as the natural tissue develops, the degradable implant
stress–strain relationships, surface properties, such should gradually weaken so that regenerated tissue
as wettability, in addition to chemical and biological can resume its function. For example, when a bone is
properties, are thoroughly evaluated. shattered, usually a metal pin is inserted or a metal

Dosage Dosage
Maximum desired level

Minimum effective dose

Time Time
Dose Dose Dose Dose

Figure 1 Ordinary drug delivery (left) and delivery using degradable polymer systems (right).
100 Biological Structures

plate fixed to the bone in order to stabilize the injury. therapies, such as peptide, protein, and DNA type
These supply support and alignment during the heal- drugs, are being developed (Figure 1).
ing process. However, the pin or the plate must Polymeric drugs are drugs that are physiologically
be surgically removed after ossification is complete. active polymers. In nature, these are proteins (en-
While these supports are in place, the bone to be zymes) and polysaccharides (heparin). However, oth-
mended has been stress free. After the pin or plate is er than synthetic mimetics, there are few man-made
removed, the regenerated bone is too weak to sup- polymers that are biologically active. Among these
port normal activity until the bone regenerates suf- are the polyanionic and polycationic polyelectro-
ficiently to assume the normal stress loads. The lytes. Polyanions stimulate the immune system and
major advantage of biodegradable supports is that as have shown anticancer and antiviral activity (HIV
the osteocytes proliferate, the polymer degrades and and hoof and mouth disease). Polycations have an-
is absorbed while stress is gradually transferred from timicrobial activity and are currently being investi-
the scaffold to the bone. Consequently, the bone ad- gated as gene-delivery adjuvants.
justs to the increasing stress as it mends. This signi- Active agents can be attached to a polymer back-
ficantly reduces rehabilitation time and eliminates a bone and are known as drug-conjugates. Polymer–
second surgical process to remove the support. drug conjugates enhance circulation time (due to
increased size) as well as increase solubility which
Tissue Engineering enables transport in the circulatory system. However,
conjugation usually inactivates the drug. Therefore,
Tissue engineering is a new frontier in bioscaffolding
polymeric prodrugs were developed. Prodrugs are
and is defined as ‘‘the application of engineering
conjugate systems that have a drug attached to a
principles to create devices for the restoration, mod-
polymer by a degradable linkage so that the drug
ification, and assembly of functional tissues from
can be released by hydrolysis or enzyme activity.
native or synthetic sources.’’ The goal is to regenerate
Furthermore, polymer conjugate systems can be
tissue, bone, or organs that have been disabled or
modified with solubilizing and targeting groups. This
weakened by disease or injury. The tissue scaffold
system has been successfully used for delivering
has to be nontoxic and biodegradable as well as
anticancer drugs, such as daunarubacin.
possess the right three-dimensional structure to sup-
Implantable drug-delivery systems involve incor-
port cell differentiation. The mechanical properties
porating drugs into a biodegradable polymer matrix.
of the scaffold and the surrounding environment
Using this technology, desirable dosages can be
have to be in harmony during the process. To en-
released over a prolonged period of time at an inter-
hance cell proliferation, slow release of supportive
mittent or constant rate. This achieves effective
bioactive recipients, such as specific enzymes, can be
therapies and prevents under- or overdosing the pa-
included in the scaffold which are then released as
tient. Biodegradable polymers that are most com-
the scaffold degrades.
monly used are poly(glycolic acids), poly(lactic
acids), polyurethanes, and poly(ortho esters) with
Polymeric Drugs and Drug-Delivery Systems
commercial names such as Decapeptyls, Lupron
These systems constitute an important technology Depots, and Sandostatin LARs. Degradation may
that has been developing over the last 30 years. Cur- occur through bulk hydrolysis in which the polymer
rently, polymeric drug delivery is a 35 billion dollar degrades in a fairly uniform manner throughout the
market and will continue to grow 5% per year for matrix. For some degradable polymers, mostly poly-
the next ten years. Many drugs are rapidly cleared anhydrides and poly(ortho esters), degradation and
from the circulatory system through the kidneys or erosion occurs only at the surface of the polymer
taken up by the liver or spleen and, therefore, exhibit which provides a release rate that is proportional to
very little efficacy. Other difficulties that are associ- the surface area of the drug-delivery system.
ated with drug assimilation are due to low solubility, With advancements in pharmaceutical sciences, it
random distribution, and toxicity. Polymers are very has been recognized that constant release is not nec-
versatile and, therefore, offer novel approaches so essarily the optimum method for delivering drugs. As
that systems can be tailored for an intended appli- a result, externally modulated or self-regulating
cation and long-term delivery times (days, weeks, drug-delivery systems, using biomedical polymer
months, and even years). Polymeric drug-delivery structures, are being engineered to respond to
systems are prepared in a variety of sizes and shapes, environmental stimuli, pH, temperature, or pressure.
such as films, fibers, disks, slabs, pellets, rods, and The glucose-sensitive insulin-release system is an
monolithic forms. The development of new polymer example of this application; in this device, high
methodology is critically important as new drug glucose concentrations induce matrix disintegration
Biological Structures 101

with polymer dissociation from the surface, leading This structure provides a water content similar to
to the release of insulin. that of living tissue and was one of the first materials
used for soft contact lenses. This hydrogel is inert to
Hydrogels biodegradation, permeable to metabolites, not ab-
sorbed by the body, endures heat sterilization, and
Hydrogels are water-insoluble, three-dimensional
can be fabricated into different shapes and forms.
networks formed by cross-linking a hydrophilic pol-
However, problems with protein deposits are still an
ymer. These networks absorb large quantities of wa-
issue. Hydrogels are now used for artificial tendons
ter to form a gel. Natural hydrogels, such as acacia,
and muscles, wound-healing bioadhesives, artificial
agar, carrageen, and alginates, have been known for
kidney membranes, artificial skin, and maxillofacial
centuries and are used in food and personal products.
reconstruction materials.
Research to develop synthetic hydrogels started in
Hydrogels have also been made with poly(N-vinyl-
the 1970s and dramatically increased in the 1990s
2-pyrrolidone), poly(methacrylic acid), and poly
due to the recognition of the many potential appli-
(methyl methacrylate-co-maleic anhydride), which
cations that these constructs could have in medicine,
are used in similar biomedical applications. For
agriculture, and industry. This work has led to the
example, polyacrylamides are synthetic long-chain
synthesis of many new hydrogels and a better un-
polymers designed to attract positively or negatively
derstanding of the physicochemical properties in-
charged materials and are widely used in gel elect-
volved. When a hydrogel is in the swollen state, the
rophoresis. In gel extraction processes, an acid is
three-dimensional compartment holding the solvent
added to shrink the hydrogel, and conversely, a base
is called a cell or pore. As the gel dries, it becomes
is added to swell it (Figure 3).
concomitantly smaller in size and finally collapses to
Biodegradable hydrogels have emerged as an im-
the solid state called a sol (thus the term sol–gel). The
portant class of biomaterials for controlled drug
overall chemical structure remains the same during
delivery. The release rate of an excipient is deter-
these processes. The three-dimensional cell structure
mined by drug and polymer properties, such as the
is achieved through cross-linking linear polymer
swellability, degradability, porosity, permeability,
chains either chemically (by covalent bonding) or
and hydrophilicity of the hydrogels.
physically (by hydrogen bonding, ionic, or hydro-
Smart (or intelligent) polymer-structured hydrogels
phobic interactions). Since noncovalent bonds are
are being developed for several biomedical applica-
reversible, physical gels have sol–gel reversibility.
tions. A unique aspect about these hydrogels is that
The extent of swellability depends on the hydrophilic
their swelling ratio (swollen volume divided by dried
nature of the polymer and the cross-linking density.
volume) can be engineered to change abruptly with a
Smaller cells (higher cross-link density) provide more
slight external aberration. Hydrogel materials can un-
stable hydrogels and appear to be more solid-like.
dergo continuous or discontinuous changes in swell-
Methods for preparing hydrogels include irradiative
ing in response to environmental stimuli, such as
cross-linking using electron beams, gamma rays,
changes in pH, temperature, ionic strength, pressure,
X-rays, or UV light, and chemical reactions using
and light. These stimuli–response properties are
di- or multifunctional cross-linking agents (Figure 2).
advantageous in many potential bioapplications. For
Most hydrogels have good mechanical stability,
example, timed control release of insulin has been ac-
good refractive index, and oxygen permeability. Con-
hieved with ‘‘smart’’ pH-sensitive hydrogels. This phe-
sequently, one of the early biomedical applications
nomenon is known as on-demand release of insulin in
of hydrogels was for soft contact lenses. The most
response to increased levels of glucose. Thermosen-
widely used hydrogel for bioapplications is cross-
sitive hydrogels, such as poly(N-isopropylacrylamide)
linked poly(hydroxyethyl methacrylate) (PHEMA).
and poly(N-vinyl caprolactam), have also been engin-
eered for controlled drug delivery, where hydrophobic
interactions respond to small changes in temperature


Figure 2 Network of cross-linking polymer chains in hydrogels
(left) that swell after water absorption (right). Figure 3 Polyacrylamide.
102 Biological Structures

that can vary drug release rates. ‘‘Smart’’ hydrogels are viscosurgery, ophthalmic surgery, and in cosmetic
being used extensively for robotics where responses to applications. Despite its highly attractive rheological
numerous and different stimuli are important features. properties, unmodified HA has a short residence time
Reversible contraction and expansion are essential in and poor mechanical properties. However, HA can
the development of advanced robotics with electrical- be chemically modified to prolong degradation time
ly driven muscle-like actuators. The transformation of and improve mechanical stability in vivo. Recently, it
electrochemical stimuli into mechanical work is being has been formed as a thin film that can be used for
used for such artificial muscle responses. tissue separation. HA-based films or sponges have
Hybrid hydrogel systems are composed of at least been made for implants loaded with therapeutic
two distinct classes of macromolecules. These are of agents for delayed release and prolonged activity.
particular interest because it is possible to combine
and superimpose the component macromolecule Chitosan
properties into the hydrogels. Theoretically, it is pos-
sible to manipulate the DNA sequence and generate Chitosan, poly(b-(1-4)-D-glucosamine), is a water-sol-
unlimited varieties of protein products with optimum uble product obtained by N-deacetylation of chitin
structures and properties. For example, hybrid hy- (Figure 4). Chitin is a naturally occurring polysaccha-
drogels have been prepared, using synthetic polymers ride and second to polysaccharides as the most
and biological macromolecules, as engineered pro- abundant natural polymer. It is mainly isolated
teins and DNA-based biomaterials. from crustaceans, such as crab and shrimp, but is
Microporous and macroporous hydrogels are also found in fungi and the hard shells of insects.
being made with pore sizes in the 10–100 nm and The physical properties of chitosan are related to the
100 nm–10 mm range, respectively. These hydrogels number and the distribution of N-acetyl groups. The
are designed to swell within minutes regardless of the primary amino group at the 2-position forms quater-
size of the matrix. In these cases, the pore size is an nary groups in water that become soluble at pHo8.5.
important property of a hydrogel. Now superporous Chitosan has superior biocompatible and biodeg-
hydrogels, a new type of hydrogel, with supersize radable properties and is used in biomedical appli-
pores are being developed. These pore sizes are larger cations such as wound dressings. It was first used
than 100 mm and can reach millimeter range. Their for burn bandages because it maintains good hydra-
fast swelling kinetics, in addition to super-absorbent tion, allows toxins and metabolites to be transported
properties, allow these superporous hydrogels to be away from the injury, and promotes healing. In
used in special biomedical applications, such as long- wound healing, chitosan facilitates the formation
term oral drug delivery, due to their ability to achieve of granulation tissue with angiogenesis. It induces
high gastric retention times. Superporous hydrogels fibroblasts to release interleukin-8, which is involved
and their composites are also being engineered for in the migration and proliferation of fibroblasts and
artificial pancreas, cornea, skin, and joint cartilage. vascular endothelial cells. However, chitosan’s sur-
They are being used as cell growth substrates in face-induced clot-forming potential has limited its
tissue engineering scaffolding, burn dressings, soft application in blood contact situations. Although
tissue substitutes, and as surgical pads to control chitosan itself has a hemostatic function, its deriva-
bleeding. tives, such as, N-hexanoyl and N-octanoyl chitosan,
have antithrombogenic activity. The antimicrobial
and wound-healing properties along with an excel-
Hyaluronic Acid lent film capability make chitosan suitable for ocular
Hyaluronic acid (HA, Hyaluronans) is a naturally lens bandages.
occurring, biocompatible, and biodegradable linear Although water soluble, chitosan can be formed in
polysaccharide. It is composed of unbranched repea- different shapes such as nanoparticles, microspheres,
ting disaccharide units of glucuronic acid and N-
acetyl glucosamine linked by b-(1-3) and b-(1-4)
glycosidic bonds. HA is present in all soft tissues of
higher organisms but is in very high concentration in O O
the synovial fluid and vitreous humor of the eye. It can
be extracted from various sources of animal tissues, O O O
such as the umbilical cord, skin, and rooster combs.
Hyaluronic acid is used in many medical applica- NH2 NH3+
tions, such as promoting cell mobility and differen-
tiation in wound healing, viscosupplementation and Figure 4 Chitosan [poly(b-(1-4)-D-glucosamine)].
Biological Structures 103

membranes, sponges, rods, beads, and solutions. CH3 O CH3 O CH3 O

Consequently, the porous spongy chitosan matrices O CH CH2 C O CH CH2 C O CH CH2 C
are excellent biomaterials for tissue engineering, n
especially artificial skin. When replacing skin, it is (a)
important that the substitute material completely
adheres, wets, and conforms to the wound surface so C2H5 O C2H5 O C2H5 O
as to prevent small air pockets where bacteria can O CH CH2 C O CH CH2 C O CH CH2 C
proliferate. In addition, antibiotics and growth fac- n

tors can be incorporated into the basic matrix. (b)

The chitosan microspheres were found to be a very Figure 5 (a) Poly(hydroxybutyrate) and (b) poly(hydroxyvale-
favorable form for parenteral controlled drug-deliv- rate).
ery systems requiring low-acting drug delivery to the
systemic circulation, as well as for active or passive and biodegradable materials based on properties
targeting to the treatment sites. Several types of such as low toxicity and degradation to 3-D-hydro-
active reagents, such as DNA, Cyclosporine A, and xybutyric acid, which is a normal constituent in hu-
insulin, have been encapsulated in chitosan nanopar- man blood. In addition, industries now have the
ticles. Chitosan membranes have good mechanical technology to melt-process PHB materials into a
properties and are being evaluated for transdermal variety of shapes and forms. These thermoplastic bio-
drug delivery. polymers can be produced as rigid brittle plastics,
With new biotechnology techniques, peptide ana- flexible materials, or strong elastomeric fibers.
logs that are resistant to enzyme degradation have Pure PHB or PHV are highly crystalline and brittle;
been prepared for oral drug delivery. However, their however, their copolymers and composites provide
hydrophilicity and molecular size present prob- materials that have more desirable properties such as
lems with respect to absorption through tight junc- enhanced rates of degradation and more flexibility.
tions of the intestinal epithelium. Due to chitosan’s Consequently, PHAs have been developed for med-
mucoadhesive characteristics, it has the ability to ical applications, such as sutures, artificial skin, drug
partially open tight junctions and enhance the pene- delivery, and paramedical disposables. Currently, po-
tration of macromolecules across membranes, such lyhydroxybutyrates are being investigated for bone-
as the intestinal barrier. Consequently, chitosan and implants and bone-related treatments. It has been
its derivatives have been found to be very effective found that transected nerves heal and restore their
for oral peptide delivery systems. function after being wrapped in thin PHB films.
Currently, chitosans are being investigated as These unique natural polyesters have not yet
nonviral gene-delivery vectors. However, when load- reached their full biomedical potential due to some
ed with DNA at physiological and alkaline pH, they biocompatibility issues. For example, on contact with
are insoluble. This originally presented a problem, blood, PHA induces platelet adhesion and subsequent
but trimethyl quaternization has increased the solu- thrombi formation. However, new composites are
ble range of this system. being developed, by combining PHB with poly(eth-
ylene glycol) to form ‘‘natural–synthetic hybrid’’
block copolymers, that exhibit enhanced blood-com-
patibility as well as inherent bone-compatibility.
Polyhydroxyalkanoates (PHA)s are natural polyesters
that are produced and stored by a wide variety of
Aliphatic Polyesters
microorganisms. Pseudomonades are the most com-
monly known microorganism that produces and uses Aliphatic polyesters are derived from naturally
PHA as an intracellular energy reserve. When Pseu- occurring compounds, such as lactide (LA), glyco-
domonades are stressed under conditions of excess lide (GL), and e-caprolactone (CL). These represent
carbon substrates and limited essential nutrients, PHA an important class of biocompatible and biodeg-
is formed and stored as nodules that can be up to radable materials. Initially, the main application of
90% of the organism’s dry weight. Polyhydroxybuty- these materials was for reabsorbable sutures. L-lactic
rates (PHB) and polyhydroxyvalerates (PHV) com- and L-glycolic acids, commonly found in animal sys-
prise the majority of the bacterial polyesters produced tems and in low concentrations, are nontoxic and
in nature (Figure 5). These natural polyesters provide readily absorbed by the host. However, pure poly(L-
an important source of highly pure 3-D-hydroxy- lactic acid) and pure poly(L-glycolic acid) are highly
butanoic acid and 3-D-hydroxypentanoic acid, res- crystalline; consequently, they do not have good ma-
pectively. PHB has been developed as biocompatible terial properties, such as flexibility or degradability
104 Biological Structures

(Figure 6). On the other hand, PLLA and PLGA co- Current developments involve applications for
polymers and composites (mixtures) are amorphous bone, cartilage, and vessel engineering and for
and provide desirable properties such as tenacity, mechanically dynamic sites. Recently, unique bio-
flexibility, and facile degradability. In addition, degradable copolymer nanofibrous scaffolds were
PLA/PGA copolymers can be ‘‘tailor-made’’ to meet produced by electrospinning. The diameter of the
the requirements of specific applications. Currently, fibers was approximately 500 nm with an aligned
these copolymers are being widely utilized as tem- topography that mimics the orientation of cells and
porary scaffolds for the regeneration of tissues and as fibrils found in a native artery. This process provides
carriers to deliver bioactive molecules. an almost ideal tissue-engineering scaffold, especially
Poly(e-caprolactone) (PCL) has a low melting point, for blood vessel engineering due to the nanometer-
high solubility, and an exceptional ability to form scale dimension that physically resembles the natural
blends. Biodegradable PCL staples for wound closure material but has the advantage of biodegradability.
are in clinical use in Europe. Studies have shown that
e-caprolactone and its copolymers are tissue-compat- Poly(ortho-esters)
ible and nontoxic. Semicrystalline PCL has good drug
Poly(ortho-esters) are unique polymeric structures
permeability and a very slow degradation rate. The
that undergo degradation more rapidly than the al-
requirements for optimal controlled drug release or
iphatic polyesters. In an aqueous environment, they
mechanical properties for engineered tissue can be
undergo surface erosion and are able to release drugs
achieved through copolymering PCL with other mon-
at a constant rate. However, the degradation is not
omers. When PCL is combined with amorphous PLA/
uniform and the acid that is released catalyzes a lo-
PGA copolymers that have poor drug permeability, a
calized breakdown and causes the formation of local
much higher degradation rate is obtained.
erosion forming large void areas. By varying the
These aliphatic polyester polymers usually degrade
acidity and the amounts of neutralizing excipients
by hydrolysis, and to a lesser extent, by enzymatic
incorporated into the polymer, the erosion rate can
processes in animal systems. Consequently, these
be controlled. Delivery systems are being produced
materials have rapidly gained recognition in the area
for short-term applications such as oral 12–24 h ap-
of tissue engineering. Here, cells and extracellular
plications, intermediate ophthalmic products lasting
matrix components can be implanted into the porous
1–7 days, and subdermal implants that last as long as
scaffolds to make pseudo-anatomical shapes. Al-
one year or more.
though approved by FDA, specific applications, bio-
compatibility, sterilization, and storage issues need
further improvement.
Polyanhydride polymers are synthesized by dehydra-
ting dicarboxylic acid functional monomers to form
CH3 O CH3 O CH3 O corresponding polyanhydride linkages (Figure 7). The
O CH C O CH C O CH C anhydride groups in the polymer chain are very mois-
n ture sensitive and on contact with water rapidly
(a) hydrolyze to form carboxylic acid groups. Because
of the hydrolytic instability of this linkage, these ma-
terials were developed as sutures, drug carriers, bio-
O CH2 C O CH2 C O CH2 C absorbable prostheses, and vascular grafts. In general,
(b) they show some degree of crystallinity and most are
soluble in common organic solvents. The decompo-
O O O sition rate decreases as the aromatic content and hy-
CH2 )5 C O
CH2 )5 C O
CH2 )5 C drophobicity of the polymer backbone is increased.
n Well-defined and predictable degradation rates de-
(c) pend on the polymer composition and the size and
Figure 6 (a) Poly(lactic acid), (b) poly(glycolic acid), and (c) shape of the implant. A unique feature of polyanhy-
poly(caprolactone). drides is their ability to undergo uniform surface

x x x

Figure 7 Polyanhydride.
Biological Structures 105

erosion without additional additives to inhibit pitting Dacrons. Other polymer materials include poly-
and cavity formation. This increases performance urethanes, silicones, and poly(ethylene oxide).
predictability and decreases the presence of extrane-
ous substances; both properties are highly desir- Polycyanoacrylates
able in controlled-release drug delivery, which is the
Polycyanoacrylates are better known as ‘‘super glues’’
most current application of this polymer system. Cur-
as they adhere to many different materials, such as
rent investigations are under way to use polyanhy-
metal, glass, wood, and tissue (Figure 9). Strong
drides for controlled delivery of osteogenic proteins
bonds are formed at room temperature in an aqueous
and insulin. Polyanhydrides and their degradation
environment without the need of adding a catalyst.
products have been shown to be neither mutagenic
The monomers polymerize very rapidly in the pres-
nor cytotoxic and were FDA approved in 1996 for
ence of weak bases such as water (moisture) or amino
local delivery of chemotherapeutics to brain tumors.
compounds. The higher alkyl cyanoacrylates, such as
Polyphosphazenes n-butyl cyanoacrylate, are more hydrophobic and,
therefore, spread more rapidly on surfaces, polymer-
Polyphosphazenes are unique polymers with an in- ize more rapidly and degrade slower than methyl
organic backbone structure consisting of nitrogen– cyanoacrylate. In the body, these materials achieve
phosphorous bonds (Figure 8). By changing the side hemostasis rapidly and bond to wet tissues strongly.
groups on the phosphorous, researchers can easily Consequently, they are useful as bioadhesives.
modify the bulk and surface properties of polyp- However, due to adverse tissue response and pro-
hosphazenes. The side groups can be amino, alkyl, aryl, duction of tumors in laboratory animals, cyanoacry-
alkoxy, aryloxy, inorganic, or organometallic units. lates have not been approved for routine clinical
If both side groups on the repeat unit are identi- applications. The current applications include sur-
cal, the polyphosphazene is homogeneous; if they face wound dressing in dental surgery and ophthal-
are different, then the polyphosphazene is hetero- mic surgical adhesives where monomers are applied
geneous. Due to the facile modification of the poly- directly to the tissue and almost instantaneously po-
phosphazene side chains, interesting new materials lymerize with strong adherence to the tissue. They
have been generated in order to develop inert bio- are also being investigated as adhesives to stop blee-
compatible materials for use in cardiovascular and ding of gastric varices – abnormally dilated and
other biomedical devices as well as biological mem- lengthened vessels. As the cyanoacrylate is injected
branes and coatings. Currently, these polymers are into the varix, it forms a polymer plug that occludes
being investigated as gene and drug-delivery systems. the aperture in the vessel, which is eventually ex-
Several bioactive agents, such as, steroids, local pelled into the lumen of the stomach. Cyanoacrylates
anesthetics, catecholamines, and heparin, have been have also been investigated as potential drug-delivery
attached to the polyphosphazene side groups are matrices. However, they have a tendency to induce
being evaluated as a delivery system. significant inflammatory response at the implant site.

Nondegradable Polymers Polyurethanes

Nondegradable polymers are materials that are not Polyurethanes are unique materials that offer the
degraded in a biological environment. These mate- elasticity of rubber combined with the toughness and
rials are used in devices that are implanted for per- durability of metal (Figure 10). They are more com-
manent use, such as joint prostheses, maxillofacial monly known as wood finishes that are more durable
implants, and heart valves. Many of these materials than varnish. For medical use, they can be manufac-
are made by polymerization of vinyl groups tured as very strong and resilient products with good
(–CHQCH2). These polymer structures include flexible and electrometric properties. Applications
methyl methacrylate, polypropylene, and polyethyl- including artificial heart diaphragms, ventricular
ene and cyanoacrylates. Some are polyesters such as

Cl Cl Cl CH2 CH2 CH2


Cl Cl n Cl OCH3 OCH3 n

Figure 8 Polyphosphazene. Figure 9 Polycyanoacrylate.

106 Biological Structures


Figure 10 Polyurethane.

assist bladders, vascular grafts, mammary prostheses, CH3 CH3 CH3

and pacemaker leads. CH2 C CH2 C CH2 C
Copolymers of urethane with monomers, such C O C O C O
as esters, ethers, and ureas provide a wide range of
mechanical properties that are suitable for tissue OCH3 OCH3 n OCH3
engineering and regenerative medicine. Polyurethanes (a)
have exceptional elasticity and flexibility that are ex-
tremely important qualities for soft tissue engineering CH3 CH3 CH3
and scaffolds. Porous, biodegradable, electrometric C
polyurethane scaffolds combined with the patient’s
own bone marrow have successfully facilitated bone
regeneration. Studies show that polyurethane scaf- OC2H5O OC2H5O n OC2H5O
folds induce crystalline calcium phosphate deposition (b)
similar to that formed in bone structure. When used
Figure 11 (a) Poly(methyl methacrylate) and (b) poly(hydroxy-
in plastic and orthopedic reconstructive surgery,
ethyl methacrylate).
tissue ingrowth occurs with low acute inflamma-
tion. However, after several months these implants
begin to show degradation. Poly(ether urethanes), improve the skin contours and reduce depressions in
however, are resistant to hydrolysis and are much less the skin due to scars, injury, or lines. When applied in
biodegradable. cosmetic surgery, PMMA is used as a suspension of
polymer beads in a vehicle, such as bovine collagen
(Artecolls), hyaluronic acid (MetaCrills), or other
Poly(methyl methacrylate)
biocolloidal suspension.
Poly(methyl methacrylate) (PMMA) is an amor-
phous, transparent, and hydrophobic thermoplastic
polymer that is very hard and stiff but brittle and
notch-sensitive (Figure 11). PMMA, commonly Silicone materials consist of repeating units of
known as ‘‘safety glass,’’ is an ‘‘acrylate’’ with brand inorganic –SiO2– as the backbone structure, with
names such as Plexiglas, Diakon, and Lucite. Based methyl or other functional groups as –Si– side groups
on its strength, hardness, and adhesive qualities it is (Figure 12). These materials are biocompatible, very
used in dentures, dental crowns, and caps. stable, nontoxic, and insoluble in body fluids. For
It has excellent light transmittance and good abra- medical applications, dimethylsiloxane is polymer-
sion and UV resistance, but poor low temperature, ized to form a silicone gel. Cross-linking the dime-
fatigue, and solvent resistance. It is, however, very thylsiloxane polymer chains leads to the formation of
biocompatible and was the original material used for silicone rubber-like elastomers.
hard contact lenses and intraocular lenses. Soft con- High-viscosity silicone rubbers are used as tissue
tact lenses evolved by substituting a –CH2–CH2OH expanders. The more common applications of this
group for the alkyl methyl ester group which was material have been for maxillofacial, breast, chest,
engineered to form a hydrophilic hydrogel lens and and calf augmentation. They are currently being
other biorelated products. evaluated for joint replacement and tendon recon-
Methyl methacrylate (PMMA) is used extensively struction. Although silicone is considered biological-
as a medical adhesive. It is used as bone cement to ly inert, it can elicit a mild foreign body reaction
secure prostheses, such as hip replacements and dental followed by tissue encapsulation. The use of silicone
crowns. In orthopedic surgery, PMMA cement is in- gels became controversial due to the concerns regar-
jected into the collapsed vertebra to reconstruct back ding monomer migration and toxicity as well as (un-
injuries. Although this procedure does not re-expend proven) human adjuvant diseases. These
the collapsed vertebra, it seems to alleviate pain uncertainties lead to a limitation of the use of sili-
by reinforcing and stabilizing the fracture. PMMA cone gel implants by the FDA in 1992. Currently,
is used extensively in maxillofacial augmentation to silicone gel implants are being reevaluated and are
Biological Structures 107


O Si O O Si O O Si O n

CH3 CH3 n CH3 Figure 13 Poly(ethylene glycol).

Figure 12 Silicone.
x CH3 y x
available only under specific guidelines for breast
reconstruction. Figure 14 Pluronics (PEO)x–(PPO)y–(PEO)x.

Poly(ethylene oxide)
composed of triblock PEO–PPO–PEO copolymers of
Poly(ethylene oxide) (PEO) is also known as po-
poly(ethylene oxide) (PEO) and poly(propylene ox-
ly(ethylene glycol) or (PEG) and is one of the few
ide) (PPO). The pluronic PEO block is hydrophilic
polymers that is approved by the FDA for clinical use
and water soluble while the PPO block is hydropho-
(Figure 13). PEG compounds are nontoxic and non-
bic and water insoluble. In an aqueous environment,
immunogenic as well as soluble both in water and
these block copolymers self-assemble into micelles
polar solvents. In the pharmaceutical industry, the
with a hydrophobic PPO center core and a hydro-
use of many drugs has been prevented or limited due
philic PEO outer shell that interfaces with water.
to delivery and solubility problems. Getting a drug to
Since these micelles are amphiphilic, they are able to
target is a difficult task and failure to do so can pro-
accommodate lipophilic molecules in the central hy-
duce serious toxic effects. Currently, the attachment
drophobic core area. Consequently, pluronic micelles
of PEG to various drugs (also known as ‘‘PEGylat-
are effectively used as drug carriers because their as-
ion’’) is widely employed to enhance drug solubility
semblies can act as passive drug containers. These
and efficacy. PEGylation technology consists of
assemblies deliver drugs into subcellular compart-
linking PEG to a bioactive component. The result-
ments by slowly releasing hydrophilic–hydrophobic
ing bioconjugate, when administered intravenously,
encapsulated excipients into physiological fluids.
is stable in the blood with enhanced drug delivery.
Pluronic micelles have many advantageous prop-
The method of attachment may be with a permanent
erties that include a relatively low in vivo toxicity
or a biodegradable linkage depending on the specific
and an appropriate size that restricts renal excretion.
application. The solubility of the bioconjugate can be
They provide the opportunity to deliver drugs from
tailored since it is dependent upon the lipophilicity of
micelles in a spacial and temporal controlled manner
the drug and the length of the PEG. Usually, the
with enhanced intracellular uptake via fluid-phase
molecular weight of the PEG molecules used is be-
endocytosis rather than a passive diffusion. When
tween 200 and 20 000, and their shape can be either
drugs are encapsulated as pluronic micelles, their in-
linear or branched. The in vivo half-life of PEG-
tracellular uptake by normal cells or drug-sensitive
bioconjugates in plasma is prolonged due to the
tumor cells is usually substantially reduced. This
larger molecular size of the PEG-conjugate, which
shielding effect is an advantage as it diminishes drug
diminishes the rate of renal excretion and distribu-
interaction with healthy tissues. Site-specific release
tion in the tissues.
of pluronic-encapsulated drugs can be activated un-
In addition, PEG-bioconjugates can decrease an-
der the influence of ultrasound with an enhanced
tigenicity by masking the immune recognition of an-
penetration and retention effect that provides selec-
tigenic sites. This is known as a ‘‘stealth’’ component
tive accumulation of the encapsulated drugs in solid
of the PEG-bioconjugates since they are not recogni-
tumor cells. The latter factor is important for over-
zed by the immune system as a foreign material.
coming drug resistance, which is a significant prob-
Consequently, PEGulated drugs have become very
lem in cancer chemotherapy. In addition to
important in the design of new pharmaceuticals due
sensitizing tumor anticancer agents and maximizing
to the fact that they enhance bioavailability by inc-
therapeutic outcomes, pluronic block copolymers
reasing drug solubility, as well as increasing residence
also effect gene expression at the transcript level.
time in the body by increasing the size of the con-
The aggregation state of these micelle systems can
jugate drug and so decreasing kidney elimination.
be controlled by choosing the appropriate pluronic
size and PPO/PEO block-length ratio. For example,
pluronic P-105, which is often used in biosystems,
Pluronicss represent an important class of biomed- has a molecular weight of 6500 and PEO/PPO repeat
ical polymers (Figure 14). They are unique materials unit blocks of 37 and 56, respectively.
108 Biomedical Materials

Summary Hong Y, Gao C, and Shen J (2003) Influence of quaternized poly-

urethane membrane surfaces on human endothelial cell attach-
In summary, the application of biomedical polymer ment and growth. 18(3): 191–206.
structures has become one of the most interesting Hoste K, Schacht E, and Rihova B (2002) Macromolecular prod-
and rapidly growing areas in polymer science with an rugs of Mitomycin C. 17(2): 123–139.
Leong K, Domb A, Ron A, and Langer R (1990) Polyanhydrides.
annual economic growth of 5%. Based on current
In: Kroschwitz JI, Menges G, Mark HF, Bikales N, and Over-
research results, conventional commodity polymers, berger CG (eds.) Concise Encyclopedia of Polymer Science and
such as poly(methylmethacrylate), polystyrene, po- Engineering. New York: Wiley-Interscience.
lyethylene, and polypropylene, are being replaced Lichun L, Charles AG, and Antonios GM (1999) In vitro degra-
with novel, high-performance materials. The design dation of thin poly(DL-lactic-co-glycolic acid) films. Journal of
Biomedical Materials Research 46: 236–244.
and synthesis of these exceptional polymer systems
Ottenbrite RM, Huang SJ, and Park K (eds.) (1996) Hydrogels and
are providing new and more effective ways to en- Biodegradable Polymers for Bioapplications. ACS Symposium
hance biocompatibility along with unique and effi- Series 627, Washington DC, pp. 2–10, 68–92, 244–245.
cacious bioapplications. Ottenbrite RM (1998) Frontiers in Biomedical Polymers. Lancas-
ter PA: Technomic Publishing.
Ottenbrite RM and Kim SW (2001) Polymeric Drugs & Drug
See also: Biomedical Materials; Polymer Structures. Delivery Systems. pp. 39–41, 131–155. Lancaster PA: Tech-
nomic Publishing.
PACS: 87.80.Rb; 81.05.Je; 87.64.Je Ottenbrite RM and Kim SW (2001) Polymeric Drugs & Drug
Delivery Systems. Lancaster PA: Technomic Publishing.
Rapoport N (2003) In: Dinh S and Liu P (eds.) Advances in Con-
Further Reading trolled Drug Delivery. ACS Symposium Book Series, pp. 85–
101. Washington DC.
Chiellini E, Sunamoto J, Migliaresi C, Ottenbrite R, and Cohn D Schacht E, Vandorpe, Lemmouchi Y, Dejardin S, and Seymour L
(eds.) (2001) Biomedical Polymers and Polymer Therapeutics. (1998) Degradable polyphosphazenes for biomedical applica-
New York: Plenum Publishers. tions. Journal of Bioactive and Compatible Polymers. In: Ot-
Denkbas E, Ozturk E, Ozdemir N, Kecei K, and Ergun M (2003) tenbrite RM (ed.) Frontiers in Biomedical Polymer Applications,
Journal of Bioactive and Compatible Polymers 18(3): 177–190. pp. 27–42. Lancaster, PA: Technomics Publishing.
Ehrenfreund-Klienman T, Domb AJ, and Golenser J (2003) Poly- Vert M, Feijen J, Albertsson A, Scott G, and Chiellini E (eds.)
saccharide scaffolds prepared by crosslinking of polysaccharides (1992) Biodegradable Polymers and Plastics. Stockholm, Swe-
with chitosan or proteins for cell growth. Journal of Bioactive den: The Royal Society of Chemistry.
and Compatible Polymers 18(5): 323–337. Xuejun X, Sannino A, Ambrosio L, Netti PA, and Nicolais L (2004)
Heller J, Roskos KV, and Duncan R (1993) Use of poly(ortho Journal of Bioactive and Compatible Polymers 19(1): 5–13.
esters) in the controlled release of therapeutic agents. Makro- Zanzig J, Marimuthu B, Werka, and Scholz C (2003) J. Bioact. &
moleculare Chemie, Macromolecular Symposia 70/71: 163–171. Compat. Polym 18(5): 339–354.

Biomedical Materials
J R Jones and L L Hench, Imperial College, London, Bone Defects
Bone is a natural composite of collagen (polymer) and
& 2005, Elsevier Ltd. All Rights Reserved.
bone mineral (ceramic). Collagen is a triple helix of
protein chains, a complex structure that has high ten-
sile and flexural strength and provides a framework
The Need for Biomedical Materials
for the bone. Bone mineral is a crystalline calcium
Improving healthcare and technology are increasing phosphate ceramic (hydroxyapaptite (HA), Ca10
life expectancy but as one ages the body parts cannot (PO4)6(OH)2) that provides stiffness and the high
maintain their function. Tissues such as bone and compressive strength of the bone. The two most
cartilage are needed to support the aging body even important types of bone are cortical and cancellous
though the cells that produce them become less active bone. Cortical bone is a dense structure with high
with age. The heart, kidneys, and liver have to mechanical strength and is also known as compact
operate for much longer than ever before. This entry bone. Cancellous or trabecular bone is an internal
discusses how biomedical materials and biophysics porous supporting structure present in the ends of long
techniques are being used in the development of bones such as the femur or within the confines of the
regenerative medicine procedures. The aim of reg- cortical bone in short bones. The trabecular bone is a
enerative medicine is to regenerate diseased and network of struts (trabeculae) enclosing large voids
damaged tissue to its original state and function. (macropores) with 55–70% interconnected porosity.
Biomedical Materials 109

The mechanism for natural bone generation/ engineering of the replacement tissue in the labora-
regeneration in the body is the secretion of extracel- tory. In a typical tissue engineering application, cells
lular matrix by osteogenic cells (osteoblasts), which would be harvested from the patient (i.e., osteogenic
have developed (differentiated) from stem cells. The cells in the case of bone) and seeded on a synthetic
extracellular matrix is collagen type I, which miner- scaffold that acts as a guide and stimulus for tissue
alizes to form bone mineral, creating a composite of growth in three dimensions creating a tissue engin-
orientated collagen fibrils and apatite. The bone is eering construct or living biocomposite. The bio-
remodeled in response to its local loading environm- composite would then be implanted back into the
ent by the body. Osteoclasts are cells that resorb old patient. Over time, the synthetic scaffold should be
bone and bone that is not required (i.e., not under resorbed into the body as nontoxic degradation
any load), while osteoblasts lay down new bone. products at the same rate that the cells produce their
Osteoporosis is a disease where bone resorption own extracellular matrix.
occurs faster than new bone is produced, causing the In this article, the development of scaffold tech-
trabeculae to become thinner which leads to a re- nology is discussed, focusing primarily on porous
duction in total bone density and strength. The dis- glass scaffolds that bond to bone and the growth of
ease eventually leads to fracture of bones especially bone and cartilage in the laboratory on such scaf-
in the hip, wrist, knee, and spine. At present, when folds. The biophysics techniques that are now being
osteoporotic fracture occurs in knees and hips, joint used to monitor cell behavior as the cells grow are
replacement is often required. Millions of orthopedic also discussed.
prostheses made of bioinert materials have been im-
planted, an example of which is the Charnley total
hip replacement, which is heralded as one of the most Biomedical Materials
successful surgical inventions. Any material that is implanted into the body should
Long-term monitoring of 20 000 Charnley joints be biocompatible, that is, not cytotoxic (not toxic to
has revealed that it has a survivability of 76% after 25 cells). There are three classes of noncytotoxic mate-
years implantation, that is, 24% of hip operations rials; bioinert, bioactive, and bioresorbable. No ma-
required revision surgery. Improved metal alloys, spe- terial is completely inert on implantation, but the
cial polymers, and medical-grade ceramics are the ba- only response to the implantation of bioinert mate-
sis for this success, which has enhanced the quality of rials is encapsulation of the implant by the fibrous
life for millions of patients. Reasons for failure tend to tissue. Examples of bioinert materials are medical-
be aseptic loosening of the femoral stem, where bone grade alumina, stainless steels, and high-density po-
resorption occurred due to a mismatch in the Young’s lyethylene that are used in the total hip replacements.
modulus of the bone and the metal stem. Many mod- Resorbable materials are those that dissolve on
ifications and variations of the Charnley joint have contact with body fluids and the dissolution products
been developed over the years, including coating the can be secreted via the kidneys. The most common
metal stem with a synthetic HA layer that can bond to biomedical resorbable materials are polymers that
the bone mineral in the bone; however, survivability degrade by chain scission such as polyglycolic (PGA)
studies are so far only medium term and have shown and polylactic acids (PLLA) and their co-polymers,
similar results to that of the Charnley prosthesis. which are commonly used as sutures.

A New Direction Bioactive Materials

All present day orthopedic implants lack three of the Bioactive materials stimulate a biological response
most critical characteristics of living tissues: (1) the from the body such as bonding to tissue. Bone is a
ability to self-repair; (2) the ability to maintain a natural composite of bone mineral (HA) and col-
blood supply; and (3) the ability to modify their lagen and other proteins. Therefore, synthetic HA
structure and properties in response to environmen- (with Ca/P ¼ 1.667) and other calcium phosphate
tal factors such as mechanical load. All implants have ceramics, including coral, have gained much atten-
a limited lifespan and as life expectancy is continu- tion as a bone mineral substitute.
ally increasing, it is proposed that a shift in emphasis There are two classes of bioactive materials; class
from ‘‘replacement’’ of tissues to ‘‘regeneration’’ of B bioactive materials bond to hard tissue (bone) and
tissues is required to satisfy this growing need for stimulate bone growth along the surface of the bio-
very long-term orthopedic repair. active material (osteoconduction). Examples of class
One way to restore diseased or damaged tissue to B bioactive materials are synthetic HA and tricalcium
its original state and function would be the successful phosphate ceramics. Synthetic HA has been used in
110 Biomedical Materials

Time 1. Formation of Si−OH (silanol) bonds via cation (Na+ or Ca2+) exchange with
H+ or H3O+ ions from solution;

2. Break-up of the silica network (Si−O−Si bonds) and the continued

formation of Si−OH bonds (silanols) at the glass−solution interface;

3. Condensation and repolymerization of a SiO2- rich layer on the surface,

depleted in alkalis and alkali-earth cations;
2+ 3 −
4. Migration of Ca and PO4 groups to the surface and from solution
forming a CaO−P2O5-rich film on top of the SiO2-rich layer;

5. Crystallization of the amorphous film by incorporation of OH− and CO32

2h anions from solution to form a mixed hydroxyl carbonate apatite (HCA) layer.

Figure 1 Mechanism of formation of the HCA layer on the surface of bioactive glasses in solution.

several clinical applications such as a bone defect Figure 2 shows suggested mechanisms for osteo-
filler and is used to coat the stainless steel or titanium production, where new bone grows on bioactive
alloy shaft of the total hip prosthesis so that it bonds glasses away from the glass–host bone interface. The
to the thigh bone (femur). Tricalcium phosphate sequence follows on from the events in Figure 1 but
(b-TCP, with Ca/P ¼ 1.5) is an osteoconductive concentrates on the effects of the glass on the cells
material that is also resorbable in the body. b-TCP within the bone and bone marrow, however it does
is usually used in conjunction with synthetic HA to not fully explain osteoproduction. Recent findings by
improve the resorbability of HA in applications such Professor Dame Julia Polak’s team at the Tissue
as the filling of bone defects left by cysts, sinus floor Engineering and Regenerative Medicine Centre at
augmentation, and bone cements. Imperial College, London, have shown that the dis-
Class A bioactive materials not only bond to bone solution products of bioactive glasses up-regulate
and are osteoconductive but they are also osteoprod- seven families of genes that regulate osteogenesis and
uctive, that is, they stimulate the growth of new bone the production of growth factors.
on the material away from the bone/implant inter-
face and can bond to soft tissue such as gingival Composites
(gum) and cartilage. Examples of class A bioactive To overcome the low toughness of HA, a polyethyl-
materials are bioactive glasses. A certain composition ene–HA composite was developed by Professor Will-
of melt-derived bioactive glass (46.1% SiO2, 24.4% iam Bonfield and colleagues at the University of
Na2O, 26.9% CaO, and 2.6% P2O5, in mol), called London. It is clinically available under the name
Bioglasss is used in the clinic as a treatment for per- HAPEXTM and has been used as middle ear implants
iodontal disease (Perioglass) and as a bone-filling and orbital floor replacements. The polymer provides
material (Novabones). Bioglasss implants have also toughness strength and the HA increases the com-
been used to replace damaged middle ear bones, pressive strength and provides bioactivity so that
restoring hearing to thousands of patients. implants are not encapsulated by scar tissues. The
The mechanism of bone bonding to bioactive ma- composite is class B bioactive and is easily shaped
terials is thought to be due to the formation of an HA with a scalpel by the surgeon and so is now routinely
layer on the surface of the materials after immersion used to replace the bones of the middle ear.
in body fluid. This layer is similar to the apatite layer
in bone and therefore a strong bond can form. The
layer forms quickest on class A bioactive materials. Scaffolds for Bone Repair
The mechanism of formation of the HA layer on bio-
The Criteria for an Ideal Scaffold
active glasses is shown in Figure 1. Stage 1 involves
release of cations. Stage 2 is the break up of the silica- To be able to regenerate the trabecular bone, a con-
based glass network. These two stages are dissolution struct is required that will mimic the structure and
processes; therefore, these glasses are not only bio- mechanical properties of the trabecular bone and
active but are also resorbable in the body. b-TCP stimulate new bone growth in the shape dictated by
ceramics are also resorbable. the scaffold. Therefore, the scaffold should have a
Biomedical Materials 111

Time 1. Adsorption and desorption of biological growth factors, in the HCA layer
(continues throughout the process) to activate differentiation of stem cells.
10 h
2. Action of macrophages to remove debris from the site allowing cells to
20 h occupy the space.

3. Attachment of stem cells on the bioactive surface.

4. Differentiation of stem cells to form bone growing cells, such as osteoblasts.

100 h 5. Generation of extra cellular matrix by the osteoblasts to form bone.

6. Crystallization of inorganic calcium phosphate matrix to enclose bone

cells in a living composite structure.

Figure 2 Suggested mechanism of cellular interactions between cells and bioactive glasses after implantation, after the formation of
the HCA layer.

structure that acts as a template for tissue growth in the time and financial commitment required for ful-
three dimensions. The template must be an intercon- filling the extensive tests.
nected macroporous network containing pores with
diameters in excess of 100 mm to promote cell migra-
Polymer Scaffolds
tion, tissue ingrowth and vascularization, and nutri-
ent delivery to the center of the regenerating tissue on Resorbable polymeric scaffolds have been developed
implantation. The scaffold material should be able to with porous structures similar to the trabecular bone.
bond to the host bone without the formation of scar There are three reasons why much effort has been
tissue, that is, it should be made from an osteocon- put into using polymers as scaffolds. First, polymers
ductive material. If the bone is to be restored to its are easy to process in the shape of a 3D scaffold with
original state, the scaffold should be resorbable so a pore morphology suitable for a tissue engineering
that eventually there is no trace of the scaffold’s application with techniques such as phase separa-
presence. The degradation rate of the scaffold must tion, gas foaming, freeze drying, combined solvent
be controllable so that it can be tailored to match the leaching and extrusion, and computer-controlled
rate of bone growth. The dissolution products would rapid prototyping techniques. Polymer fibers can be
ideally influence the genes in the bone generating woven together using textile processing techniques.
stem cells to stimulate efficient cell differentiation and Second, polymers can have high tensile properties
proliferation. Importantly, the mechanical properties and high toughness. The mechanical properties of
of the scaffold should match that of the host bone polymers can be controlled very easily by changing
and not decrease too rapidly during degradation so the molecular weight (chain length) of the polymer.
that bone regeneration occurs at load-bearing sites. Third, bioresorbable polymers have been successfully
If the scaffold is to be mass produced so that used as dissolving sutures for many years. Therefore,
surgeons can use it in the clinic, it must be made from these degradable polymers, such as the esters PGA
a processing technique that can produce irregular and PLLA, have passed FDA regulations and can be
shapes to match that of the defect in the bone of the implanted into the body.
patient. It must have the potential to be producible to There are however some problems with using
the required ISO (International Standards Organ- biodegradable polymers as scaffolds for bone regene-
ization) or FDA (Food and Drug Administration) ration. First, while a degradable scaffold is desired,
standards and be easily sterilized. sutures dissolve within 2 weeks of implantation,
The need for implant materials to pass the FDA which is too rapid for bone regeneration applications.
standards may be necessary for the safety of patients, Second, although resorbable polymers can be
but it places a large financial obstacle in the path of made with high tensile strength and toughness and
new scaffold developments. Many researchers use their mechanical properties can be matched with
materials that have already been passed by the FDA collagen, their Young’s modulus is much lower than
so that they are sure their scaffold will be safe for use that of bone, therefore these polymers cannot be used
in the clinic. They are unwilling to risk using mate- in load-bearing sites where they undergo compressive
rials that have not undergone FDA testing because of forces.
112 Biomedical Materials

A third problem is that the mechanical strength of The simplest way to generate porous scaffolds from
polymers decreases rapidly as they degrade. High ceramics such as HA, is to sinter particles, preferably
tensile strength polymers have long chains (high mo- spheres of equal size. The scaffolds can then be
lecular weight), which entangle with each other. As a pressed using cold isostatic pressing. As the sintering
tensile force is applied to the entangled chains, the temperature increases, the pore diameter decreases,
chains begin to unravel until they become straight. and mechanical properties increase as the packing of
The maximum tensile strength of polymers is reached the spheres increases. Mechanical properties can be
when the chains are fully extended. Therefore, for a increased further by hot isostatic pressing (HIPing).
polymer to be tough, it must have a molecular weight High mechanical strengths can be achieved but the
over the entanglement value. The mechanism for pore diameter is not high enough. Porosity can be
degradation of biodegradable polymers is chain scis- increased by adding fillers such as sucrose, gelatine,
sion. The polymers undergo hydrolysis and the and polymethyl methacrylate (PMMA) microbeads to
chains are cut in two. The average molecular weight the powder slurry, which burnout on sintering.
of the polymer is therefore approximately halved However, this technique decreases the compressive
with every chain scission event. The mechanical strength to below that of the trabecular bone.
properties of the polymer are proportionate to the The majority of methods that are used to produce
square root of the molecular weight and therefore polymer foams cannot be applied to ceramic systems.
decrease very rapidly as the polymer degrades. However, a popular method for producing highly
Biodegradable polymers have also been found to porous ceramics is to produce a polyurethane foam
produce an inflammatory response due to the acidic template that can be immersed in ceramic slurries
by-products of the degradation. under vacuum to allow the slurry to penetrate into
the pores of the foam. The organic components are
Bioactive Ceramic Scaffolds burnt out and the ceramic foams sintered (13501C)
Ceramics are crystalline materials and tend to have producing a scaffold with interconnected pore diam-
high compressive strength and Young’s modulus but eters of up to 300 mm.
low toughness, that is, they are brittle materials. The ceramic slurries can also be foamed to obtain
Alumina and synthetic HA are ceramics that are pores in the range of 20 mm up to 1–2 mm. The in-
most commonly used in biomedical applications. corporation of bubbles is achieved by injection of
Alumina is a bioinert ceramic that is very hard and gases through the fluid medium, mechanical agitat-
resistant to wear. It is therefore commonly used as a ion, blowing agents, evaporation of compounds, or
replacement to the ball of the femur in total hip by evolution of gas by in situ chemical reaction. A
replacement. Bioactive ceramics that have been used surfactant is generally used to stabilize bubbles
as bone regeneration materials are synthetic HA and formed in the liquid phase by reducing the surface
b-TCA. Synthetic HA has been used most regularly tension of the gas–liquid interface. Surfactants are
because of its similarity to bone mineral; it has a macromolecules composed of two parts, one hydro-
similar structure and Young’s modulus. b-TCP is phobic and one hydrophilic. Owing to this configura-
similar to bone mineral in that they are both calcium tion, surfactants tend to adsorb onto gas–liquid
phosphate ceramics, however b-TCP is resorbable. interfaces with the hydrophobic part being expelled
In a porous form, HA and b-TCP ceramics can be from the solvent and a hydrophilic part remaining in
colonized by bone tissue. A problem with introducing contact with the liquid. This behavior lowers the
pores into a ceramic is that the compressive strength surface tension of the gas–liquid interfaces, making
of the material decreases dramatically. The strength of the foam films thermodynamically stable, which
the scaffold depends on the thickness and strength of would otherwise collapse in the absence of the
the struts or pore walls. Generally, for the brittle surfactant.
compression of a foam The gel-casting method has been the most success-
ful method used to produce macroporous bioactive
scr pr3=2
r ½1 HA ceramics with interconnected pores of greater
than 100 mm in diameter. Suspensions of HA particles
where scr is the critical strength of the pore walls and and organic monomers are foamed by agitation with
rr is the relative density, expressed as the addition of a surfactant under a nitrogen atmos-
phere. In situ polymerization (cross-linking) of the
rr ¼ rb =rs ½2 monomers, which is initiated by a catalyst, creates a
3D polymeric network (gel). The porous gels are sin-
where rb is the bulk density of the scaffold and rs is tered to provide mechanical strength and to burnout
the skeletal (true) density of the material. the organic solvents. HA foams have been made with
Biomedical Materials 113

compressive strengths in excess of 10 MPa, which is During the foaming process, air is entrapped in the
similar to that of trabecular bone. When the foams sol under vigorous agitation as viscosity increases
were implanted into the tibia of rabbits, bone par- and the silica (–Si–O–Si–) network forms. A surfact-
tially filled the pores after 8 weeks and there was no ant is added to stabilize the bubbles at short times.
inflammatory response. The compressive strength of As the porous foam becomes a gel, the bubbles are
the scaffolds that were colonized by bone trebled. permanently stabilized. The gel is then subjected to
Hydrated silicon (SiOH4) has been found to be a controlled thermal processes of aging (601C) to
major contributor to the mineralization of bone and strengthen it, drying (1301C), and thermal stabiliza-
gene activation, which has lead to the substitution tion/sintering to remove organic species from the
of silica (SiO2) for calcia (CaO) into synthetic HA. surface of the material (500–8001C). Bioactive glass
In vivo results showed that bone ingrowth in silica- foam scaffolds can contain macropores up to 600 mm
substituted HA granules was significantly greater in diameter, connected by pore windows with modal
than that into phase pure HA granules. diameters in excess of 100 mm and compressive
The disadvantages of an HA scaffold over a bio- strengths up to 2.5 MPa. Figure 3 shows a scanning
active glass scaffold with similar morphology are electron microscopy (SEM) micrograph of a pore
that HA resorbs only very slowly, the dissolution network of a typical bioactive glass foam. In vitro
products do not stimulate the genes in the osteogenic cell studies using primary human osteoblasts have
cells and HA is only osteoconductive as it generates shown the foams stimulate formation and mineral-
bone at a slower rate than bioactive glasses. HA is ization of bone nodules within 2 weeks of culture
still a bone replacement material rather than a (Figure 4). Figure 4 shows osteoblasts that have at-
regenerative material. tached to the concave surface of the pore. The cells
have proliferated, coating the pore in an organic lay-
er, and have released extracellular matrix, which has
Bioactive Glass Scaffolds
mineralized to form bone mineral. In vivo studies
Theoretically, gel-casting could be applied to melt- have shown that foam scaffolds implanted on rabbit
derived bioactive glass powders. However, bioactive crania stimulated new bone growth at a similar rate
glasses undergo surface reactions on contact with to that of the melt-derived bioactive glass powder
solutions to produce an HCA surface layer and it is available commercially.
desirable to have control over the reaction before a
scaffold is ready for clinical use.
Sacrificial porogens and foaming agents have also Spectroscopy and Biophotonics
been used to create melt-derived bioactive glass
(Bioglasss) scaffolds. Large pores with diameters in Vibrational spectroscopy is one of the most common
the region of 200–300 mm were created but the total techniques to study the surface reactions of bioacti-
porosity was just 21% and there were large distances ve materials. It is nondestructive, rapid, and can be
between pores, so this process failed to mimic the used to analyze small areas of a sample surface. The
interconnectivity of the trabecular bone. two most commonly used vibrational spectroscopy
The most successful method for the production of
bioactive glass scaffolds of similar structure to
trabecular bone mineral is the foaming of sol–gel
derived bioactive glasses. The sol–gel process in-
volves polymer reactions of the glass precursors in a
solution (sol). The sol is a solution of silica species
that forms a gel by cross-linking together into a silica
network. Sol–gel derived bioactive glasses tend to be
more bioactive and resorb quicker than melt-derived
glasses of similar compositions. The compositions of
gel-glasses can also contain fewer components while
maintaining bioactivity. This is due to them having a
textural nano-sized porosity that is inherent to the
sol–gel process, which provides a specific surface
area of 150–600 m2 g–1, which is two orders of
magnitude higher than melt-derived glasses. Along
this surface there are many silanol groups that act as Figure 3 Scanning electron micrograph of a bioactive glass
nucleation sites for HCA layer formation (Figure 1). foam scaffold.
114 Biomedical Materials

P− O
− O−Si
Si− stretch
P− O
C− O Powder

Absorbance (arbitrary values)


Figure 4 Scanning electron micrograph bone cells and miner-
alized bone extracellular matrix in a pore of a bioactive glass
foam scaffold. (Courtesy of J Gough.)

400 600 800 1000 1200

techniques are Fourier transform infrared spectros- Wave number (cm−1)
copy (FTIR) and Raman spectroscopy.
Figure 5 FTIR spectra for 58S glass powder, foam, and mon-
olith after 2 h immersion in SBF at 371C and 175 rpm agitation.
Fourier Transform Infrared Spectroscopy
FTIR spectra have been used to determine the rate of
spectroscopy with optical microscopy (water immer-
formation of the HCA layer on bioactive glasses and
sion objectives) has allowed noninvasive in situ mon-
the technique is used as a quality assurance (QA) test
itoring of cells in cell culture. This technique is called
for bioactivity. FTIR spectra of the glass surface area
micro-Raman spectroscopy. The optical objective
usually taken after a glass sample has been reacted in
provides high spatial resolution and is used to locate
simulated body fluid (SBF) and dried. The crystalline
the cell or part of the cell, such as the cytoplasm or
HCA layer is characterized by the P–O bending vib-
the nucleus, from which a spectrum will be collected.
ration bands at 560 and 604 cm–1 (infrared wave
A laser is then fired at the identified area and the
number). Stretching and bending vibration bands for
scattered photons are collected by a detector.
the amorphous silica network are also present in the
Low scattering efficiency of the cells makes the
spectra. The relative intensity of the bands can be
measurement of Raman shifts difficult. Raman signals
used to monitor growth of the layer as a function of
can be enhanced using UV lasers; however, high-
time. Figure 5 shows FTIR spectra of three forms of
powered lasers can denature the cells and change cell
the sol–gel derived bioactive glasses of the 58S com-
phenotype. Decreasing the laser power decreases the
position (60 mol.% SiO2, 36 mol.% CaO, 4 mol.%
Raman signal, but reduces the risk of changes to the
P2O5) that were soaked in SBF for 2 h. The spectra
cell. A laser wavelength of 785 nm has been found to
show that the HCA layer thickness was thicker for
produce optimal in situ spectra of living cells. A single
the powders and foams with respect to the monolith.
spectrum obtained from the nucleus of a cell can con-
Although FTIR can be used for the characterization
tain information about the protein and DNA content
of dried materials after soaking or after implantation
of the cell. Figure 6 shows a Raman spectrum of a
and removal, in situ measurements are hindered by
cultured lung cell containing vibrational bands corre-
the strong adsorption of water in the infrared region.
sponding to DNA (A, G, T, C adenine, guanine, thy-
mine, cytosine), BK: backbone, RP: ribose-phosphate
Bio-Raman Spectroscopy and Biophotonics
and the proteins phenylalanine (Phe) and tyrosine
Raman spectra are fingerprints of the chemical com- (Tyr). Raman spectroscopy using 785 nm lasers can
position of the material. Raman spectra are obtained be used to monitor the biochemical changes taking
by collecting electromagnetic radiation that has been place in a cell. During cell death, for example, the
scattered by molecules (Raman effect). The spectral degradation of proteins, DNA breakdown, and the
shifts depend on the vibrational frequencies of the formation of lipid vesicles can be detected in real time.
molecules. The low Raman signal and low Raman Cells cultured on bioactive materials can also
scattering of water, and the combination of Raman be monitored by in situ Raman spectroscopy. A
Biomedical Materials 115

1252: Amide III

1449: CH2 def

1339: CH def

1659: Amide I
937: Prot. Skel. C−C
1004: Phe

1126: C−N, C−C str.



853: Tyr
830: Tyr

Counts (a.u.)



T, A, G: 1375
A, G: 1422

G, A: 1578

C, T, BK: 789

BK: 900

A: 729
T: 751


600 800 1000 1200 1400 1600 1800
Wave number (cm
Figure 6 Raman spectrum of a cultured living lung cell. (Courtesy of I Notingher.)

background spectrum for the unreacted material can glass/polymer foam composite would be developed
be subtracted from the spectrum of a cell on the ma- with improved toughness. The properties of a scaf-
terial. The resultant spectrum will then also contain fold must be optimized with respect to the cell re-
vibrational bands corresponding to the formation of sponse to the scaffold. Biophotonics techniques such
any calcium phosphate layer on the surface of the as Raman spectroscopy have the potential to be able
bioactive material as a function of time. to monitor cell activity and intracellular processes of
Raman spectroscopy may hold the key for the in living cells in real time noninvasively.
situ monitoring of bone cells growing on scaffolds.
When acquisition techniques and interpretation of
the spectra are perfected, the micro-Raman spectro- Acknowledgments
meters can be linked to bioreactors and the effect of
flow rates, growth factors, and dissolution products Lloyds Tercentenary Foundation, EPSRC and MRC
on cell behavior can be monitored in real time. (UK), Dr. Julie Gough, Dr. Ioan Notingher.

See also: Bioelectronics; Biomembranes; DNA and RNA,

Summary Biophysical Aspects; Electromagnetic Biological Effects;
Numerous materials have been developed to replace Genetic Algorithms for Biological Systems; Ion Transport
living tissues. They are bioinert (stainless steel, alu- in Biological Membranes; Neuroscience.
mina), bioresorbable (polyglycolic acid), or bioactive
(synthetic HA and bioactive glasses). New biocom- PACS: 87.80.Rb; 81.05.Je; 81.05.Rm; 81.05.Kf;
posites have been made that combine bioinert and 81.05Qk; 81.20.Fw; 82.39.Wj; 87.64.Je
bioactive properties. The latest generation of mate-
rials are used in combination with cells. Many tech-
niques have been employed to produce porous
Further Reading
scaffolds. The scaffold that most closely matches
the criteria for an ideal scaffold and that most closely Brinker CJ and Scherer GW (1990) Sol–Gel Science: The Physics
and Chemistry of the Sol–Gel Process. Boston: Academic Press.
mimics the structure of the trabecular bone is the
Chalmers J and Griffiths PR (2001) Handbook of Vibrational
bioactive glass foam. However, these scaffolds do not Spectroscopy Volume 5: Applications in Life, Pharmaceutical
yet have the mechanical properties to be directly im- and Natural Sciences. New York: Wiley.
planted into load-bearing sites. Ideally, a bioactive Davies JE (2000) Bone Engineering. Toronto: EM2.
116 Biomembranes

Gibson LJ and Ashby MF (1988) Cellular Solids Structure and Kemp P (eds.) Future Strategies for Tissue and Organ Replace-
Properties. Oxford: Pergamon. ment, pp. 3–19. London: Imperial College Press.
Hastings GW and Williams DF (1980) Mechanical Properties of Notingher I, Verrier S, Haque S, Polak JM, and Hench LL (2003)
Biomaterials. New York: Wiley. Spectroscopic study of human lung epithelial cells (A549) in
Hench LL, Jones JR, Lenza RFS, and Vasconcelos WL (2003) Tis- culture: living cells versus dead cells. Biopolymers (Biospectros-
sue engineering. In: Banner N, Polak JM, and Yacoub M (eds.) copy) 72: 230–240.
Lung Transplantation, pp. 367–373. Cambridge: Cambridge Park J and Lakes RS (1992) Biomaterials: An Introduction, 2nd
University Press. edn. New York: Plenum.
Hench LL and Polak JM (2002) Third generation biomaterials. Thomson RC, Yaszemski MJ, and Mikos AG (1997) Polymer
Science 295(5557): 1014–1020. scaffold processing. In: Lanza RP, Langer R, and Chick WL
Hench LL and Wilson J (1993) An Introduction to Bioceramics. (eds.) Principles of Tissue Engineering, pp. 263–271. Austin,
Singapore: World Scientific. TX: Landes.
Jones JR, Sepulveda P, and Hench LL (2002) Bioactive materials Wise DL (2000) The Biomaterials and Bioengineering Handbook.
for tissue engineering scaffolds. In: Polak JM, Hench LL, and New York: Dekker.

F Bordi and C Cametti, Università degli Studi di basic structure based on two-layered sheets of lipid
Roma ‘‘La Sapienza,’’ Rome, Italy molecules held together by a delicate balance be-
& 2005, Elsevier Ltd. All Rights Reserved. tween hydrophobic and hydrophilic interactions
(Figure 1). The first important fact is that this is a
self-assembling structure. The balance between a
Introduction favorable interaction of the hydrophilic polar heads
of the lipids with the aqueous solvent and the
A biomembrane can be considered as a barrier of
unfavorable hydrophobic interactions of their al-
highly selective permeability that allows and regulates
iphatic chains, represents the driving force to the
the traffic of a myriad of different molecular species
spontaneous formation in an organized closed struc-
between the interior of the cell and its surrounding
ture. Due to hydrophobic/hydrophilic interactions,
environment (as, e.g., in a plasmatic membrane), or
the lipid molecules arrange themselves into two-
between different compartments within the same cell
faced leaflets (the double layer) with the hydrophobic
(e.g., in the membrane of the different organelles). Its
basic structure is formed by a bilayer of amphiphilic
molecules, the most common of which are three dif-
ferent classes of lipids (glycerolipids, sterols, and
sphingolipids). On the extracellular side, this bilayer
couples to a glycocalix, a carbohydrate network
composed of oligosaccharides believed to be respon-
sible for cell–cell recognition and adhesion. On the
intracellular side, the bilayer couples to the cytoskel-
eton, that contributes to the mechanical properties of
the overall composite structure.
The main functions exerted by a biomembrane can
be summarized as follows: (1) the membrane is a Cp Rp
selective filter which controls the transport and the
permeation of ions, molecular aggregates, and even Cm Rm
large particles between the extracellular medium and
the cytosol; (2) the membrane is the site for energy Cp Rp
producing processes and for hormone signal trans-
(b) (c)
duction; (3) the membrane acts as receptor for ex-
tracellular signals and mediates the communications Figure 1 A sketch of a biological cell membrane as seen by
between intra- and extra-cellular media; and (4) the a chemist ((a) a lipid bilayer with membrane proteins) and by
a physicist ((b) a hydrophobic layer covered by a hydrophylic
membrane can perform mechanical tasks as in cel-
interface). (c) The passive equivalent circuit of a plasmatic mem-
lular motion or in eso- or endo-cytosis processes. brane. Cp, Rp and Cm, Rm represent the capacitance and resist-
These demanding tasks are fulfilled thanks to a ance per unit surface of the hydrophilic and hydrophobic regions,
very complex, yet in principle, surprisingly simple, respectively.
Biomembranes 117

moiety in the inside and ‘‘polar heads’’ on the out- (e.g., cholesterol), phospholipids (e.g., phosphatidy-
side, representing the interface with the aqueous lcholine (PC), phosphatidylethanolamine (PE), phos-
environment. The presence of lipids of different phatidylserine (PS), sphingomyelin (SPHM)), and
shapes (different volume ratios of the hydrophobic/ sphingolipids (cerebrosides, glycolipids). Lipids
hydrophilic moieties) in the two leaflets stabilizes the with a predominantly functional role are: phospha-
radius of curvature of the double layer. tidylinositol, phosphatidic acid, and many types of
The second important aspect is that, being mainly gangliosides.
stabilized by ‘‘structure-solvent’’ effects, without The lipid composition of different cell types shows
strong ‘‘links’’ between different molecules, the bilay- great differences and within the same cell, the plasma
er is a highly dynamic structure, where lipids (and membrane, the nuclear envelope, mitochondria, and
proteins) can flex, rotate, and diffuse laterally, as in a other membranous organelles and structures have
‘‘two-dimensional’’ fluid. The lipid distribution re- different lipid compositions. Depending on the
sults from a continuous inward and outward movem- cell type, proteins constitute 20–80% of the mem-
ent between the two monolayers, the lipid asymmetry brane mass. Nevertheless, by comparing the average
being maintained by specific mechanisms that coun- molecular weight of lipid molecules (in the range
terbalance the concentration-driven transbilayer 700–1000 Da) and the typical molecular weight of a
permeation (Figure 2). The lipid bilayer has often protein (in the range of the order of tens of thousands
been considered, in the past, as a passive structure Da), lipids are the more abundant components in
and only recently, more attention is being directed terms of molar concentration.
towards its implications in a variety of membrane Membrane proteins are usually divided into two
biological functions. classes, ‘‘intrinsic’’ and ‘‘extrinsic’’ ones. This differ-
Due to its peculiar structure, the overall shape of entiation is mainly based on the greater or lesser dif-
the membrane during the biological functionality can ficulty in separating the fraction of the protein of the
be easily changed, for example, during pseudopodia two classes from the other membrane components by
movement or eso-/endo-cytosis processes. In end- organic solvent extraction. However, intrinsic pro-
ocytosis, in particular, after a region of the mem- teins are generally considered more deeply embedded
brane has surrounded the particle to be introduced into the nonpolar environment (the interior of the
within the cell, it detaches from the rest, forming a lipid double layer), while extrinsic proteins are as-
vacuum that can move within the cytoplasm. The sumed more ‘‘at the surface’’ or even simply adsorbed
whole process is reversed in eso-cytosis, when a at the double-layer interface. However, most intrinsic
substance has to be expelled from the cell. These proteins possess specific sites that are exposed to the
processes well illustrate the ‘‘modularity’’ of the aqueous environment on one, or on both the sides of
membrane structure, that can easily lose or acquire the membrane.
parts to accomplish particular tasks. The thickness of a lipid bilayer can be estimated in
Among the enormous variety of possible lipids, the range of 4–6 nm. However, in estimating the
only a few classes are used to build up a biomem- thickness of a real biological membrane, the fact that
brane. Lipids that exert a predominantly structural proteins and many lipids (glycolipids) bear groups
role can be grouped into three categories: sterols that extend into the aqueous medium, has to be tak-
en into account. This region (glycocalix), at the out-
er surface of the plasmatic membrane, is typically
10–100 Å thick.
Lateral diffusion
Membranes of cells and organelles are sometimes
corrugated and folded or show protrusions of differ-
ent shape and size. For example, the needle-like pro-
trusions of the plasmatic membrane of some cells
Flip-flop (microvilli) typically measure 50–200 nm in diameter
and may extend for several hundreds of nanometers
Waving and even more.

Functional Cell Membrane Domains and

Influence of the Cytoskeleton on the
Organization of the Cell Membrane
Figure 2 Movements of phospholipids in bilayers. The lipid
distribution results from lateral diffusion, flip-flop, waving and ro- One intriguing issue in membrane biophysics is
tation movements between the two leaflets of the membrane. the presence within the plasma membrane of lipid
118 Biomembranes

microdomains, termed rafts, which are believed to Where the fluid-mosaic model fails is in the as-
be essential for its complex activity. Although the sumption of a uniform protein distribution within the
present view of structural organization of a biological double layer. Limiting ourselves to the transport
membrane is still deeply indebted to the fluid-mosaic within the membrane, a great deal of evidence has
model proposed by Singer and Nicholson in 1972, now been accumulated, showing that a variety of ‘in-
consisting of a fluid-lipid bilayer where lipids and tra-membrane’ movement modes exist for membrane
proteins are more or less randomly distributed, re- proteins. In addition to the simple Brownian diffusion
cently a considerable amount of experimental work assumed by the fluid-mosaic model, membrane pro-
suggests that biomembranes are not laterally homo- teins also appear in a stationary phase and experience
geneous, but floating domains with distinct lipid and different diffusion modes: directed diffusion, confined
protein composition. However, it is not yet complete- diffusion (the particle can diffuse freely but within a
ly understood what the mechanism at the molecular restricted area), diffusion in a harmonic potential well
level is, that determines the composition of these do- (this is, e.g., the diffusion mode experienced by a
mains and what their precise functional roles are. protein bound elastically to the cytoskeleton).
Moreover, in addition to the random motion predict- In particular, single-particle tracking techniques,
ed by the fluid mosaic model, a variety of lateral together with a careful theoretical and statistical
transport modes exist for several membrane proteins. analysis of the particle tracks, have shown that the
Microdomains have been detected in certain lipid plasma membrane is compartmentalized with regard
mixtures, for example, PC and cholesterol, that to lateral diffusion of transmembrane proteins
exhibit coexisting liquid phases under definite con- although, in many cases, inter-compartmental hop-
ditions of temperature, lateral pressure, and com- ping diffusion is observed.
position. In this case, the size of membrane domains The relationship of lipid rafts to the zones detected
ranges from several hundred lipid diameters (100– by single-particle tracking, where proteins do not ex-
200 nm) to one micrometer. perience free diffusion (transient confinement zones
The biological membrane system where the exist- (TCZs)) is still controversial. The membrane lipid
ence of lateral domains has been now evidenced with composition and, in particular, the abundance of
certainty, is the plasma membrane of mammalian cholesterol appears to have an effect on TCZ abun-
cells. In living cells, raft domains appear to be very dance. However, experimental evidence, the observed
small and probably heterogeneous. This may explain ‘hopping’ between different zones, in particular, in-
why they have escaped direct visualization by mi- dicates that trapping cannot simply be ascribed to
croscopical techniques. Indirect evidence for small particles encountering a much more viscous region,
rafts was obtained using single-particle tracking of and suggests that some types of barriers are involved.
the thermal position fluctuation, showing that raft- To account for the presence of these barriers, two
associated membrane proteins are stably associated models have been already proposed. These models,
to a small cholesterol-dependent lipid assembly of of course, do not exclude each other and some com-
B50 nm in diameter. binations of them occur. In the first model, which has
In this case, domains are lipid and protein supra- been given the suggestive title of ‘‘membrane-skele-
molecular complexes which include, for example, ton fence model,’’ the membrane skeleton provides
multimers of receptor molecules or receptor and ef- the barrier to the free diffusion of membrane pro-
fector molecules that are thought to be the first teins. The idea is that the space between the mem-
trigger for the cellular reactions that follow ligand brane and the cytoskeleton is too small to allow the
binding. Moreover, there are specialized membrane cytoplasmic portion of any membrane protein to
domains, such as synapses, caveolae, and cell–cell pass. The membrane compartment would hence
and cell–substrate adhesion structures, where specific ensue from a steric hindrance, and would be differ-
proteins and lipids are assembled together to carry ent for proteins that extend differently into the
out specific functions. cytoplasm. The hopping diffusion between adjacent
The domain distribution does not cover the whole compartments should result from the combination of
membrane area, and domains (rafts) are in dynamic two favorable conditions. As a result of the dynamic
equilibrium with more homogeneous and fluid mem- properties of the cytoskeleton, the distance between
brane regions. These domains are thought to form the membrane and the skeleton may fluctuate (the
because they represent the energetically most favor- membrane skeleton probably dissociates and reasso-
able packing of the membrane components. How- ciates from the membrane in a dynamic equilibrium).
ever, these rafts should be considered as being ‘‘soft’’ In this way, the membrane protein molecules that
structures that, despite a high degree of local order, have sufficient kinetic energy and that approach the
do not have a long-range order. compartment boundaries will be able to cross the
Biomembranes 119

barrier. This model has received strong support where es and ss are the membrane permittivity and
from experiments on red blood cells. It has been the membrane conductivity, d the membrane thick-
observed that an impaired cytoskeleton (as it is found ness, and e0 the dielectric constant of free space. For
in erythrocytes of patients affected by different forms a normal cell membrane, different experimental pro-
of hereditary hemolytic anemia) causes an enhance- cedures yield Cs C1 mF cm2 and Gs C1 O1 cm2
ment of the lateral diffusion of ‘‘band 3’’ protein, a that, assuming a membrane thickness dE80 Å allows
typical transmembrane protein of erythrocyte mem- one to evaluate the membrane permittivity and the
brane, while its rotational diffusion remains practi- electrical conductivity to be of the order of es C5 and
cally unaffected. ss C104 O1 cm1 .
In the second model, TCZs would be delimited by Fixed surface charges give rise to an electrostatic
the presence at their boundaries of a high concentra- surface potential that, in turn, results in the forma-
tion of ‘‘obstacles.’’ The protein confinement within a tion of a diffuse electrical double layer. An accumu-
small zone should result from a repeated bouncing of lation of charged species near the membrane surface
the protein between the obstacles, in a sort of ‘‘pin- influences the transmembrane transport mechanisms.
ball effect.’’ The obstacles could be membrane pro- The conformation and function of many molecules is
teins directly linked to the cytoskeletal network. This generally affected by electrostatic fields. As a conse-
mechanism may justify the transient confinement of quence, the membrane charged surface acts as a cat-
proteins that are known not to have parts in the alytic site and, besides the transmembrane transport,
protrusion into the cytoplasm and that, for this rea- the changes in the surface charge can also influence
son, cannot directly interact with the cytoskeleton. recognition and other biochemical reactions that
Also, the directed diffusion of membrane protein take place at its surface.
has been ascribed to two different mechanisms. Using as an obligatory starting point the Poisson–
Analogous to contractile movements that allow cell Boltzmann equation with the exponential space
locomotion, the first mechanism involves the con- charge density expanded to the first order in the
certed action of a number of molecular motors to electrostatic potential, the electrostatic potential F,
move entire regions of the cortical cytoskeleton and, and consequently the space charge density r, in the
with it, patches of cross-linked membrane proteins. simplified case of a 1 : 1 valent electrolyte solution,
Moreover, a protein can in principle undergo di- can be calculated as a function of the distance from
rected diffusion exploiting the skeleton fence mech- the membrane surface (assumed as an infinite plane)
anism again, with elements of the cytoskeleton from the expression (Gouy–Chapman theory)
tracing a sort of corridor underneath the membrane,
within which the protein can freely diffuse. 2kB T 1 þ gekx
FðxÞ ¼ ln ½1
e 1  gekx
Membrane Electrostatics
Here, g is a constant that depends on the electrostatic
Biological membranes carry a great number of polar potential calculated at the surface F0
or ionized groups. The net charge is usually negative,
because most of the membrane proteins and all eF0
charged natural lipids have their isoelectric point be- g ¼ tanh ½2
4kB T
low neutral pH (in human blood, e.g., physiological
pH must be maintained within narrow limits B7.4).
and k is the Debye constant
The charge density is not usually very high, typically
less than 0.05 C m  2 (corresponding to about 1  1=2
elementary charge on a surface of 250 Å2) and much 2rN e2
k¼ ½3
lesser than the charge density in artificial lipid double kB Te0 er
layer, where the surface charge density can reach
values as high as 0.25 C m  2 (1 elementary charge that depends on the ionic force of the solution. kBT is
per 50 Å2). the thermal energy.
From an electrical point of view, a biomembrane In eqn [3], rN is the electrolyte concentration far
can be considered as an extremely thin (B50 Å) hy- away from the surface. The Debye length can be
drophobic isolating interface between two conduct- considered approximately as the thickness of the
ing phases. The passive electrical behavior of a diffuse double layer at the membrane surface. Equa-
membrane can be described as an R–C element net- tion [1] combined with the boundary condition
work, whose specific capacity Cs is given by Cs ¼ dF=dxjx¼0 ¼ s=e0 er gives the following relation,
e0 es =d and the specific conductance Gs by Gs ¼ ss =d, connecting the potential F0 and the surface charge
120 Biomembranes

density s Such a transmembrane potential is found in almost

  all living cells.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi eF0
s ¼ 8rN e0 er kB T sinh ½4 The efficiency of the membrane as a barrier to the
2kB T
passage of charged molecules is mainly due to the high
It must be considered that, in the case of biological energy cost of transferring a charge from the aqueous
membranes, the charge cannot be considered, not environment into the apolar core of the membrane.
even approximately, homogeneously distributed on a Membrane pores or the formation of ion-carrier com-
flat surface. Charges are, in part, borne by the groups plexes can reduce this energy requirement dramati-
attached to the lipid head groups and to the surface cally. The membrane surface electrostatics plays a role
proteins that extend into the solution, forming the in transmembrane transport by affecting the interfa-
glycocalix. The thickness of this zone is often com- cial ion concentration and by increasing the proba-
parable with the electrical double-layer thickness ob- bility of a pore or an ion-carrier complex formation.
tained from eqn [3]. Consequently, the charge should
be considered distributed tri-dimensionally within a
layer at the membrane surface, and not with a bi- Experimental Techniques
dimensional distribution. Besides all the approxima- A large variety of experimental methods for the study
tions of the model, this is another reason for taking of the biomembrane structure and dynamics have
the values calculated in the framework of the Gouy– been developed, the most important of which are
Chapman theory with some warning. Models based nuclear magnetic resonance (NMR), electron spin
on Bogolyubov–Born–Yvon–Green or Ornstein–Zer- resonance (EPR), patch clamp, and Langmuir–
nike equations, together with a suitable closure re- Blodgett technique, single-particle tracking, radio-
lation, give a better description of double-layer wave dielectric spectroscopy, and many more. A brief
behavior and characteristics, but at the cost of a survey on the applicability to the membrane of each
more complicated formalism. In the understanding of them follows.
of the electrokinetic phenomena that take place at
the membrane surface, the Gouy–Chapman theory Single-Particle Tracking
often gives helpful hints as a working model,
provided that the higher-order effects, mainly arising In ‘‘single-particle tracking,’’ the membrane molecular
from ion–ion interactions and from the nonuniform, species whose diffusion has to be investigated is
tri-dimensional charge distribution, are negligible. marked by specifically bound colloidal gold particles
Across many biological membranes exists an elec- or by fluorescent molecules. The concentration of
trostatic transmembrane potential also, DF, that is, a these probes is kept very low, in such a way that only
difference in the electrostatic potential between the few of them appear within the field of an optical mi-
two surfaces of the membrane. croscope, equipped with a digital camera. From a
Different ion-surface affinities lead to the formation statistical study of deviations from a purely diffusive
of a nonuniform layer of ions more or less intimately Brownian motion, information on membrane struc-
associated with the membrane surface. This affinity ture and particle–particle interactions can be obtained.
can be different in the two surfaces of the membrane In a different approach, known as ‘‘fluorescence cor-
and contribute to the transmembrane potential. relation spectroscopy,’’ the analysis of the motion of
Another contribution stems from the ‘‘dipole po- individual molecules is replaced by the analysis of the
tential’’ that arises from the dipole moment of the time-averaged fluctuations of the fluorescence signal,
membrane constituents, in particular from the head- deriving from a very small volume, at the membrane
groups of (zwitterionic/charged) lipids, the carbonyl surface. The requirement of selecting the signal from a
groups of their hydrocarbon chains, and from the volume small enough to make fluctuations observable
interfacially bound water layers. is fulfilled by using a confocal optical microscope.
However, when the electrolyte composition at the In this way, an ‘‘open’’ volume optically selected
two membrane sides is different, the main contribu- B10  15 l, in which molecules can freely go in and
tion to the transmembrane potential comes from the out, can be investigated.
Nernst diffusion potential. To the first approxima-
tion, Nernst potential is proportional to the loga- Patch Clamp
rithm of the ratio of the concentrations on the two Electrical currents flowing through the cell mem-
sides of the membrane, according to brane can be measured by means of a particular
  technique known as ‘‘patch clamp’’ developed in the
kB T c1 early 1970s by B Sakmann and E Neher, who for
DFN ¼ ln ½5
ze c2 their research in this field obtained the Nobel Prize in
Biomembranes 121

physiology in 1991. A microelectrode of suitable size pressure as a function of the area available to each
can be obtained by rapidly pulling a glass capillary molecule at the aqueous interface. The lowering of
with an inner diameter of one millimeter or less in the surface tension at the interface caused by the
size, intensely heated in the middle (a specially presence of the amphiphile is given by P ¼g0  g,
designed ‘‘puller’’ has been devised to this scope). where g0 and g are the surface tension in the presence
The capillary ultimately breaks into two hollow and in the absence of the amphiphile. P is a measure
spikes at the tip, with a diameter of a fraction of a of the surface pressure needed to prevent the film
micrometer. The rim of the hole on the tip is then from spreading. At very low pressure, P obeys a two-
polished and smoothed with a microforge. These dimensional equivalent ideal gas law
‘‘micropipettes’’ are filled with a suitable electro-
lyte. The micropipette is then implanted through the PA ¼ kB T ½6
membrane into the cell. Two metal electrodes, one
within the micropipette and the other in the bathing where A is the area occupied by each molecule and
solution around the cell, connected to a picoampero- kBT the thermal energy.
meter, allow the measurement of the electrical cur- These monolayers undergo a surface potential due
rent flowing through the membrane. In appropriate to the group dipole moments of the film-forming
configurations, the current through the membrane of molecules. By considering the monolayer as a parallel
the whole cell or through ‘‘patches’’ cut from the cell plate condenser, comprising a sheet of uniformly dis-
membrane can be measured. In the last case, it is tributed dipoles, the surface potential DV is given by
possible to measure the current flowing through an
individual protein channel. With this technique, the 4pN0 mn
DV ¼ ½7
protein channels responsible for the selective trans- er A
port of different ions and charged molecules have
been identified and their properties studied thor- where N0/A represents the array of N0 dipoles by
oughly, discovering, for example, that many of them area A and mn the normal component of the dipole
are controlled by the voltage difference across the moment per molecule. The potential of a monolayer-
membrane (voltage-gated channels). Two different free subphase is taken as reference. The measurement
electrical arrangements are usually employed. In the of the surface potential can be carried out by the
‘‘current-clamp’’ technique, the current is kept con- ionizing electrode method or the vibrating plate
stant and the voltage recorded. In ‘‘voltage-clamp,’’ method. In the first case, the ionization of the air
conversely, the voltage is kept constant and the cur- above the film, induced by an a-emitter, allows the
rent is the measured signal. Voltage-clamp results are potential difference between two electrodes, above
more suitable for very low-current intensities and in and below the aqueous subphase, to be measured. In
studies on voltage-gated channels, where the voltage the latter, the vibrating electrode of one of the plates
level is used to control the opening and closing of the of a condenser generates an alternating signal that
channel. can be measured using a high-gain amplifier.

NMR and EPR Spectroscopies

Langmuir–Blodgett Film
A good picture of the characteristics of lipids and
A suitable model of a biological membrane is repre- protein motion in membranes can be derived from
sented by a lipid monolayer, where the nature and both EPR and NMR techniques. In EPR, the study of
the packing of the lipid molecules, the composition the conformational flexibility of the membrane
of the subphase, and the temperature can be chosen involves the use of spin labels containing an un-
without any limitation. The main characteris- paired electron available to observe transition be-
tics of these systems consist in their homogeneity, tween different energy levels induced by the presence
stability, and planar geometry, with lipid molecules of the external magnetic field. Spin labels of different
undergoing a well-defined orientation, providing a structures have been used to probe the behavior of
bi-dimensional system. lipid hydrocarbon chains in the membrane.
Langmuir monolayers are produced by dissolv- NMR spectroscopy can provide information about
ing at the air–water interface, a known amount of the structure and dynamics of membrane proteins at
the amphiphile under investigation in a Langmuir a molecular level and, in particular, the molecular
trough. After volatile solvent evaporation, the mono- arrangement in the bilayer. For example, because of
layer extension is compressed by two movable bar- the anisotropy nature of the bilayer, the order pa-
riers and monitored with the surface-pressure area rameter, related to the quadrupole splitting arising in
isotherms, that is, a plot of the change of the surface a 2H-NMR spectrum, measured at different positions
122 Biomolecules, Scanning Probe Microscopy of

in the alkyl chain, offers a detailed picture of the the extracellular medium and the membrane, a well-
molecular bilayer arrangement. During the past dec- defined dielectric dispersion appears, generally cen-
ade, considerable advances in both NMR and com- tered B1–10 MHz. The analysis of this dispersion,
putational methodologies have taken place, opening both on the basis of specific dielectric models of a
up the possibility of investigating structural and dy- biological cell suspension and on the basis of a
namic properties of systems that may not be suitable phenomenological approach, results in the evaluation
for crystallography. NMR can reveal dynamic infor- of some characteristic parameters, particularly the
mation about regions of macromolecular structure, permittivity and the electrical conductivity of the cell
and conformation or structural changes that occur membrane, which describe the structural and func-
as a result of ligand binding. For example, by the tional behavior of the membrane. This method, that,
P-NMR method, the fraction of the phospholipid from an experimental point of view requires the
molecules at the outer leaflet of the bilayer can be measurement of the electrical impedance of the sam-
determined. Moreover, dynamical parameters such ple under investigation, has been applied to a variety
as rotational and translational diffusion coefficients, of biological cell suspensions and the parameters
the rate of conformational changes, the activity of associated with the cell membrane have been
membrane embedded proteins up to interactions be- determined.
tween membranes and other macromolecules can be
appropriately investigated. See also: Ion Transport in Biological Membranes; Mem-
brane and Protein Aggregates.
Radiowave Dielectric Spectroscopy Technique
PACS: 87.16.Uv; 87.16.Dg; 87.17.  d
Dielectric spectroscopy in the radiowave frequency
range furnishes a valuable method to investigate the
electrical properties of a biological cell membrane, Further Reading
both concerning the structural properties (charge and Cevc G and Marsh D (1987) Phospholipid Bilayers: Physical
dipole moment distribution) described by the permit- Principles and Models. New York: Wiley.
tivity, and the dynamical properties (ionic transport Gaber BP and Easwaran KRK (1992) Biomembrane structure and
across the membrane) described by the electrical con- function, The State of the Art. New York: Adenine Press.
Lasic DD (1993) Liposomes: From Physics to Applications. Am-
ductivity. The method takes advantage from the fact sterdam: Elsevier.
that, due to the different electrical parameters asso- Lipowsky R and Sackmann E (eds.) (1995) Structure and Dynam-
ciated with the intracellular medium (the cytosol), ics of Membranes. Amsterdam: Elsevier.

Biomolecules, Scanning Probe Microscopy of

C Ascoli, CNR-Istituto Processi Chimico Fisici, Pisa, or distances by piezoelectric actuators was experi-
Italy mentally proved in 1971 by measuring the evane-
& 2005, Elsevier Ltd. All Rights Reserved. scent light on the surface of a prism placed in the
condition of total reflection.
Scanning tunneling microscopy (STM), was first
realized in 1982 and, subsequently, many different
Introduction types of scanning probe microscopy (SPM) tech-
The relevance of microscopy for science, and in par- niques followed, which are now used extensively.
ticular for biology, has always been clear. In the fa- Clearly, the widespread development of SPMs was
mous talk of Feymann, given at Caltech for the made possible by the easy use of personal computers.
annual meeting of the American Physical Society The general structure of each SPM is similar: a
in 1959, the requirement of a better microscope for thin local probe, placed close to the sample (at a
biologists was strongly underlined. The possibility of distance of the order of the resolution), scans the
high-resolution probe microscopy was also clear sample and a quantity that depends on an interaction
for the physicists. The use of local probes to illumi- between the probe and the sample is measured. It
nate and/or to detect light in a near-field configura- is then reported in a plot f(x,y) in a color scale
tion was first proposed in 1928 by Synge, who later or with a three-dimensional (3D) perspective. Virtu-
suggested the use of piezoelectric actuators. The ac- ally each kind of interaction between the protrud-
tual possibility of measuring subatomic movements ing tip of the probe and the sample can originate a
Biomolecules, Scanning Probe Microscopy of 123

particular SPM, but the key for high resolution Atomic force microscopy (AFM) was developed
is always the steep dependence of the measured some years later by Binnig, Quate, and Gerber. AFM
quantity on the tip–sample distance. A steep depend- is based on mechanical interactions: a small flexible
ence, with space constant of the order of 1 Å or cantilever, with a tip on the sample side, is carried in
less, as for STM, implies that at short distances the contact with the sample and makes a regular scan-
predominant interaction is between the most protru- ning of the surface. While doing so, the cantilever
ding atoms on the two sides, and then atomic reso- deflection changes according to the surface profile.
lution is possible. The most relevant SPM techniques, The measured deflections (or other equivalent sig-
in particular for the biological systems, are discussed nals) give information on the sample topography and
briefly below. on other local features. An AFM essentially works as
STM is based on the quantum tunnel effect. In old phonographs did, but the difference in scale im-
quantum mechanics, electrons are described by a plies a fundamental difference. For instance, the
wave function. The Schrödinger equation implies meaning of ‘‘contact,’’ which is clear on a macro-
that the wave function is not zero beyond a potential scopic scale, is rather undefined on the atomic or
barrier even if the electron energy is lower than the nanometer scale. This holds for each SPM, and in
barrier itself. Thus, in quantum physics, particles order to get a better understanding of what is possible
have a nonvanishing probability to be found beyond to measure, it is required to consider, in some detail,
a potential barrier, in a region forbidden by classical the interaction between the probe and the sample.
physics: this is known as the tunneling effect. In AFM can easily work in water solution, namely, in
STM, a sharp conducting tip (the probe) and a con- the normal conditions for biological molecules. Under
ducting sample are separated by an insulator (e.g., a appropriate conditions, STM can also work in liquids.
vacuum gap). By the tunneling effect, electrons have Another SPM technique, which necessarily works
some probability of passing from the tip to the con- in ionic solution, is scanning ion conductance micro-
ducting sample. When a difference of potential (typ- scope (SICM). It was initially proposed by Hansma
ically between 1 mV and 1 V) is applied between the and co-workers, and it is based on the idea of using,
tip and the sample, a tunnel current (typically be- as a scanning probe, a pipette-shaped microelectrode
tween 0.1 and 10 nA) runs through the insulator for of the type employed by electrophysiologists for in-
working distances of the order of angstroms. Figure 1 tracellular recording (the radius of the tip is of the
shows schematically how to get images by measuring order of hundreds of nanometers). The current
the tunneling current: piezoactuators move the sam- flowing in such a probe is limited when the electrode
ple in the xy plane and the tunneling current, meas- is in the proximity of a nonconducting (or only par-
ured and stored in a computer, changes according to tially conducting) object, typically the membrane of a
the underlying surface. cell. After some years of latency, this kind of micros-
copy is now applied successfully to living cells, with
very interesting results. Two points have been of cru-
cial importance for the development of SICM. One is
the use of a small modulation of the tip–sample dis-
V tance; the other is the coupling with classical elect-
rophysiological experiments. The use of a position
modulation allows for a better stabilization of the
tip–sample distance (particularly critical at cellular
scales). Combining this technique with other micros-
copies or with electrophysiological recordings, it is
possible to observe the time evolution of submicro-
metric cellular structures and the localization of spe-
cific ion channels on the cell surface.
The scanning near-field optical microscopy
Piezodriver (SNOM; also called NSOM, PSTM, and by some
other names, according to different experimental
configurations), the first one conceptually proposed,
Figure 1 Scheme of STM. The tunneling current It flows from uses various kinds of proximal probes. By using
the tip to the sample because of the applied bias potential V. The proximal probes to collect light, or as sources of local
piezodriver moves the sample forward and backward along the
x direction; at the end of each line, it makes a step in y.
lighting, it is possible to overcome the half-wave-
The measured current It changes according to the tip–sample length limit of the classical optical microscopy (Abbe
distance and gives tunneling current maps of the surface. or Rayleigh criterion). In fact, the Rayleigh criterion
124 Biomolecules, Scanning Probe Microscopy of

is not a stringent theoretical limit. It takes into Forces between Tip and Substrate
account only the propagating components of the
electromagnetic field, which are commonly called Interactions between single atoms obey the Lennard-
‘‘far-field components’’ and are those involved in tra- Jones (LJ) potential. The LJ potential is characterized
ditional optical devices. However, light diffracted, by two terms: the attractive one, due to the van der
refracted, or reflected by small objects contains more Waals forces, acts on a relatively long range, while the
information than its propagating part; for instance, repulsive term becomes most important at short dis-
very close to a narrow slit, the light intensity obviou- tances. For AFM tip–substrate interaction, it is to be
sly mimics the slit profile. The working distance and taken into account that the number of significantly
the probe used are very important for the resolution. interacting atoms varies with the distance. Far from
Frequently, tapered optical fibers with a metal coa- the sample, many atoms on the tip and on the surface
ting opened on the tip and suitably modified AFM are in a significantly equivalent situation and con-
cantilevers are employed. Moreover, apertureless tribute equally to the attractive force, while, as the tip
probes (gold nanoparticles, quantum dots down to approaches the sample, only the most protruding at-
single molecules) are also in use now. Spatial reso- oms on the two surfaces are relevant in the interac-
lutions up to a few tenths of nanometers have been tion. This implies that at a short distance, the main
reached. SNOM is also used experimentally to obtain contribution comes from the repulsive force between
nanolithography; its specificity in biological systems the two nearest atoms on the two sides, while at
is connected with its capability to detect fluorescence greater distances the resulting attractive force is due
of single molecules with spatial resolution. to the integration on many atoms. Therefore, an LJ-
like potential, with increased attractive forces, is ex-
pected to hold and actually is able to account for the
The AFM Microscopy
basic features of experimental force–distance curves.
AFM can be considered as an extension of the sense Figures 2 and 3 outline a graphical method to obtain
of touch, in all its aspects. Through fingers one is able the force–distance curves from an LJ-like potential.
to perceive the relieves on a surface, to evaluate if it is Let z be the distance (controlled by the piezodriver)
smooth or rough, compliant or stiff, and how sticky between the resting position of the tip and the sample,
it is. Accordingly, by an AFM it is possible to meas- d the actual tip–sample distance, and s the deflection
ure topographic relieves, the local elasticity and plas- of the cantilever. Which deflection and distance are
ticity of a body, the friction between the sample and realized for a given z can be determined by assuming
the tip and their adhesion, etc; but all this is now on a that the elastic force of the cantilever (shown in red
nanometer scale and sometimes on an atomic scale. in Figure 2) balances the LJ-like force. Depending
With regard to the atomic resolution available with on the value of the cantilever elastic constant k, the
AFM, it should be noted that most of the older data intersection between f ¼ ks and the LJ force can
show atomic lattices with no well-defined atomic- have more than one solution, as shown in Figure 2,
scale local defects, while STM data clearly show the where two stable solutions and an intermediate un-
presence of defects at the atomic level. A partial an- stable solution occur. This determines two typical
swer to this issue is that a typical radius of curvature discontinuities in the experimental F(z) force–distance
of the tip ranges 5–50 nm, while typical lattice con- curves, as shown in Figure 3. The ‘‘jump to contact’’
stants are of the order of a few angstroms, so that in occurs during the approach when the attractive force
the tip there are a number of atoms that interact si- suddenly pushes the cantilever down. The ‘‘jump
multaneously with the sample surface. This number off contact’’ occurs while retracting the tip from
increases when a strong force is applied, because of the sample. After the ‘‘jump to contact,’’ the tip and
the deformation of the tip. The force acting on the tip the sample move together (for an ideally rigid sample)
is the composition of the forces due to each single and this region of FD curves is called contact
interacting atom. The periodicity of the lattice can be line. The tip–sample adhesion force is measured by
detected, because during the scanning the force due to the ‘‘jump off contact’’ height and, as discussed
each interacting atom on the tip has the lattice pe- above, can be simply due to the unstable equilibrium
riodicity. So the tip vertical displacements reproduce between the LJ force and the elasticity of the can-
the lattice period and not the profile of single atoms. tilever. However, adhesion is due also to different
However, it has been proved that, working at dis- chemical or physical reasons. For instance, the adhe-
tances where repulsive and attractive forces balance, sion in air is very high for the meniscus formed
the interaction of the most protruding atom is en- because of the water adsorbed on the surfaces.
hanced. In this case, true atomic resolution has been Actually, real force–distance curves give much more
achieved by AFM, revealing atomic-scale defects. information than that predicted from the schematic
Biomolecules, Scanning Probe Microscopy of 125


f (d ) − f (s)
−f = ks

d z d,s
d, z > 0
s<0 Z


(a) (b)
z Distance between sample and cantilever resting position
d Tip−sample distance
s Cantilever deflection
Figure 2 (a) Schematic representation of cantilever and sample relative position. The sign of s is assumed negative when the
cantilever is deflected down (attractive force), so z ¼ d  s. (b) Lennard-Jones force (black curve) and cantilever elastic force (red curve
with the vertical axis inverted) are reported as functions of d. The elastic force, given by felastic ¼ ks ¼ kz  kd ; is zero for d equal to z,
that is controlled by the piezodriver. The cantilever deflection is obtained by the balancing of Lennard-Jones and elastic forces, and is
given by the ordinate of the intersection of the two graphs, where fLJ ¼ felastic : Three possible solutions occur in the graph above.
Multiple solutions disappear for the cantilever with k greater of the maximum slope of LJ force.

discussion above. By suitably choosing and preparing with I1,2 the upper-quadrants signal and I3,4 the
the cantilever, tip, and sample, the aspecific adhesion lower-quadrants signal, while
does not mask other interactions, and it becomes
possible to measure forces between molecular part- IT ¼ ðI1 þ I4 Þ  ðI2 þ I3 Þ
ners (specific adhesion) or unfolding of proteins. In
biological systems, the measurement of the force re- gives a measure of the torsion of the cantilever, pro-
quired for unfolding and the force responsible for portional to the friction the tip encounters while
chemical bonds involved in biological functions is scanning the sample.
very stimulating, as will be discussed below in some The signal-to-noise ratio depends basically on the
detail. photodetector shot-noise, whose spectral density is
Nevertheless, AFM was originally considered given by
mainly as a microscopy, namely as a tool to get pffiffiffiffiffiffiffiffiffiffiffiffiffi
images. The different ways to get and to handle AFM iN ¼ 2eIDf
images are discussed next.
where e is the charge of the electron, I is the gene-
AFM Scheme and Operating Modes rated photocurrent proportional to the laser intensi-
ty, Df is the detection band. The important elements
The most-used scheme for detecting the deflection
in determining the sensitivity of this detection meth-
of an AFM cantilever is schematized in Figure 4: a
od are the dimensions of the light spot reflected by
Gaussian laser beam is focused on the back of the
the cantilever, which are due to the diffraction in-
cantilever and the reflected light is collected by a
duced by the cantilever on the beam.
four-quadrant photodetector, so the displacement
In analogy with STM and other SPMs, AFM
of the laser spot gives a signal proportional to the
could operate either at constant height or at constant
cantilever deflection or torsion, and hence to the
force (Figure 5). In the constant-height mode, the
interacting force. In particular, with a properly
piezodriver moves the sample in the xy plane main-
mounted detector, the deflection signal is given by
taining it at a constant z and the cantilever deflec-
tions are acquired. It is possible to employ this
ID ¼ ðI1 þ I2 Þ  ðI3 þ I4 Þ mode only on very flat surfaces, since steep steps can
126 Biomolecules, Scanning Probe Microscopy of

1 2
A F(d )
f1 a
 c′ d, z
b B
f 2′ B′

f3 C

F( z)

C′ z


Figure 3 Graphical method to obtain experimental force–distance curves from the Lennard-Jones force FLJ. In (a), lines 1, 2, and 3
represent the elastic force for three different values (a, b, g) of the height z. The intersection of these lines with FLJ determines the value
of the cantilever deflection s. A single solution is possible for line 1 (intersection a), two solutions for lines 2 and 3 (intersection points b
and b 0 and c and c 0 ). Lines between 2 and 3 (not drawn) have three intersections with FLJ. While the tip approaches the sample, the
solution moves from c0 to b (the corresponding forces can be read on the vertical axis following the dashed lines and range from f30 to f2);
further, approaching the tip and sample, the solution will jump from b to b 0 . During the withdrawal, a bigger jump will occur from c and c 0 .
(b) Force vs. distance curve during the cantilever approach to and retract from the sample, obtained from the graphs in (a). In the insets,
a representation of the cantilever deflection is shown for different positions. (Adapted from Cappella B, Baschieri P, Frediani C, Miccoli
P, and Ascoli C (1997) Force-distance curves by AFM. IEEE Engineering in Medicine and Biology 58–65.)

Photocell constant via a feedback controlling the height z of the