Professional Documents
Culture Documents
This series is intended for readers whose main interest is in physics, or who
need the methods of physics in the study of science and technology. Some of
the books will provide a sound treatment of topics essential in any physics
training, while other, more advanced, volumes will be suitable as preliminary
reading for research in the field covered. New titles will be added from time
to time.
Lovell, M. C.
Physical properties of materials.
Bibliography: p. 292
Includes index.
1. Solids. 2. Materials. I. Avery, A. J., joint
author. II. Vernon, M. W., joint author. III. Title.
QC176.L66 530.4'1 75-35836
Materials Science has now become established as a discipline in its own right as well as
being of increasing importance in the fields of Physics, Chemistry and Engineering. To
the student meeting this subject for the first time the combination of disciplines which
it embraces represents a formidable challenge. He will require to understand the lan-
guage of the physicist and chemist as well as appreciate the practical uses and limita-
tions of solid materials. This book has been written as an introduction to the Physical
Properties of Materials with these thoughts in mind. The mathematical content has
been limited deliberately and emphasis is placed on providing a sound basis using
simplified models. Once these are understood we feel that a mathematical approach
is more readily assimilated and for this purpose supplementary reading is suggested.
While the authors are deeply aware of the pitfalls in attempting such a treatment this
is meant to be an essentially simple book to point the many avenues to be explored.
We anticipate that the book will appeal to first and second year degree students in
a variety of disciplines and may not prove too difficult for those studying appropriate
Higher National Certificate and Diploma courses. Electrical engineers working in the
field of materials applications may well find it useful as a guide to modern thinking
about materials and their properties.
The book begins with an introduction to some basic ideas of modern physics. The
following three chapters discuss the structure of ideal and imperfect solids. They des-
cribe the techniques whereby samples of pure materials may be prepared in single
crystal or amorphous forms and their structure determined. The mechanical properties
of materials are surveyed in Chapter 5 and thermal properties in Chapter 6. Chapter 7
is devoted mainly to semiconducting materials but the usual introduction to semi-
conductor physics which is found in most texts on semiconductor devices has been
omitted. Here much more emphasis has been placed on a survey of the physical
properties of semiconducting materials including recent ideas and applications. The
basic classical ideas of dielectrics are introduced in Chapter 8 followed by a discussion
of piezoelectric, pyroelectric and ferroelectric materials. These materials are proving to
have many varied and interesting possibilities for applications. A similar treatment is
afforded to the magnetic and optical properties of solids in Chapters 9 and 10. New
developments, for example magnetic bubbles and electro-optics are included and the
relevant material properties presented. The final chapter is concerned with Super-
conductivity and attention is focused on materials which are being or may be used for
superconducting devices.
For the materials scientist the aim of scientific theory is to predict new develop-
ments as well as explain experimental results, and a text such as this must point the
way to new possibilities for materials. For this reason some exciting prospects for
new materials are mentioned as well as new ideas which have not yet developed into
vi PREFACE
practical applications. By introducing the reader to the frontiers of the subject at one
or two points his appetite may be whetted for scientific discovery in a rapidly develop-
ing field.
Throughout the book the MKS(SI) system of units has been used. References to
appropriate review articles in the literature are cited so as to give the reader an oppor-
tunity to pursue a given topic. Examples are given at the end of most chapters and
numerical answers are provided. These questions are designed to test and assist the
student's understanding of the text.
We wish to thank Dr A. J. Woodall who gave us the opportunity to undertake the
task of writing this book and to express gratitude to our colleagues who have helped
in many and various ways. Particular thanks are due to Mr C. G. Wilson for his careful
reading of the manuscript and helpful comments, and to Mr D. Carter for expert
assistance with the illustrations. Finally we thank the Dean of the Royal Military
College of Science for permission to publish.
October 1975 M. C. LOVELL
A.J.AVERY
M.W.VERNON
Acknowledgements
We wish to thank the publishers of the following journals and books for permission to
reproduce the items specified:
Physical Review (Figs. 7.10, 7.l4c, 7.17,10.8,10.14,10.24 and some data in Fig. 10.9);
Annalen der Physik (Figs. 7.3, 7.4); Journal of Physics C (Fig. 7.14b); Journal of Non-
Crystalline Solids (Figs. 7.19, 7.20 and some data in Fig. 10.10); Comptes Rendus (Fig.
7 .14d); Proceedings of the Institution of Electrical Engineers (Fig. 9.22); Proceedings of
the International Conference on Magnetism (Nottingham 1964) (Fig. 7.16); McGraw
Hill for Fig. 8.8 and data of Tables 8.1 and 8.3; Academic Press for Fig. 10.22 and
some data in Tables 8.8, 8.9, 10.4; Wiley for some data in Table 10.4.
Acknowledgements are also due to E.M.1. Valve Division, Hayes, Middlesex for per-
mission to reproduce data of Table 10.21, to Chance-Pilkington Glass Co. for the data
on optical glasses in Table 10.1, and Mr. D. A. James for material on the microwave
properties of ferrites.
Contents
Page
Preface v
CHAPTER 1 FUNDAMENTALS
1.1 Introduction 1
1.2 Wave mechanics 1
1.2.1 The hydrogen atom 3
1.3 Many-electron atoms and the Pauli Principle 6
1.3.1 Electron states in multi-electron atoms 7
1.4 Transition elements 9
1.5 Atomic magnetism 10
1.5.1 Diamagnetism 12
1.5.2 Paramagnetism 13
1.5.3 Exchange 13
1.6 Electrons in solids 16
CHAPTER 8 DIELECTRICS
8.1 Introduction 153
8.2 Mechanisms of polarization 156
8.2.1 Induced polarization 156
x CONTENTS
Page
8.2.2 Orientational polarization 157
8.3 The local field 158
8.4 The Clausius-Mosotti relation 161
8.5 Dielectric relaxation 161
8.6 Applications 165
8.7 Piezoelectric, pyroelectric and ferroelectric materials 166
8.8 Piezoelectricity 168
8.8.1 Electromechanical coupling coefficient 170
8.8.2 Ceramic piezoelectrics 172
8.8.3 Piezoelectric materials and their applications 172
8.9 Ferroelectricity 173
8.10 Classification of ferroelectric materials 175
8.11 Barium titanate 176
8.12 Ferroelectric ceramics 177
8.13 Ferroelectric domains 179
8.13.1 Observation of domains 180
8.14 Pyroelectricity 180
8.14.1 Applications of pyroelectric materials 183
CHAPTER 11 SUPERCONDUCTIVITY
11.1 Introduction 256
11.2 Resistanceless and superconducting states 257
11.3 Superconductivity 258
11.4 Penetration depth 260
11.5 The two-fluid model 262
11.6 The intermediate state 262
11.7 Coherence length 263
11.7.1 Coherence length and surface energy 264
11.8 Type II superconductors 265
11.8.1 The mixed state 266
11.8.2 Critical currents 267
11.9 Theory of superconductivity 269
11.9.1 The energy gap in superconductors 270
11.9.2 Electron interactions 271
11.9.3 Zero resistance 273
11.10 Superconducting materials and their applications 274
11.1 0.1 Superconducting magnets 274
xii CONTENTS
Page
11.10.2 A.c. applications 275
11.1 0.3 Other applications 276
11.1 0.4 The future 276
Bibliography 292
Index 297
1 Fundamentals
1.1 Introduction
In a treatment of the structure and properties of materials some basic ideas are needed
concerning the properties of electrons in isolated atoms and in solids. In this chapter a
few of these concepts are introduced briefly and simply. They include the wave
mechanical picture of the atom, the idea of quantum numbers and the magnetic
properties of atoms. Finally the energy band model for electrons in solids is introduced.
In the period between 1900 and 1927 a number of critical experiments involving both
the interaction of light with matter and the behaviour of electrons were successfully
interpreted in a new language, that of quantum mechanics. Ught and electrons are now
regarded as playing a dual role as wave and particle. A beam of light of frequency v
can be thought of as a stream of 'particles' or photons of energy hv where h is a con-
stant. The intensity of the light ig given by the density of photons, or in terms of
the probability of finding a photon at a particular point. These probabilities are
calculated by wave theory on the assumption that the probability is proportional to
the square of the amplitude of the wave. Wave theory becomes a mathematical
technique for calculating photon densities. A wave aspect can also be ascribed to
particles with mass through the de Broglie relation. This suggests that a particle of
mass m and velocity u has an associated wavelength A given by
This equation can be looked upon as a standard recipe for describing the wave aspects
of a particle in terms of the amplitude of a wave function 1/1 of its coordinates in
space and time x,y, z and t. In many cases the time can be separated as a trivial phase
factor leaving the time independent form of the equation appropriate to a standing
wave
81T 2 m
v2 1/1(xyz) + T (E - V)1/I(xyz) = 0, (1.4)
where E is the total energy and V is the potential energy of the particle. The use of
eqn (1.4) can best be illustrated by considering its application to one electron as in
the hydrogen atom. The problem is defined by the forces acting on the electron des-
cribed through the potential V, generally a function of the spatial coordinates x, y
and z. The solution provides a knowledge of the energy E of the electron and the
corresponding wave function 1/1 which is a mathematical function in space describing
the three-dimensional wave pattern. Knowing 1/1, the probability of finding the electron
at a point is proportional to 1/1 2 • Wave mechanics indicates that in the hydrogen atom
or any other atom, it is no longer permissible to think of an electron as describing an
orbit in which its motion can be followed in detail. Only probabilities of finding the
electron are significant and wave mechanics allows us to calculate these probabilities
through a knowledge of 1/1 and hence 1/1 2 • In dealing with wave motion of light or
sound the ideas of an amplitude and its square, the intensity of the wave, are familiar.
A 1/1 pattern in one dimension could be just the simple one shown in Fig. 1.1 as a
~-----------\------------~----------~~----------~x
function of coordinate x. This 1/1 has both a positive and negative part and so if it is
to be interpreted as a probability it makes sense to take the square, since a negative
probability has no meaning. If referred to a single particle, the value of 1/1 2 (x) dx given
by the figure for any range dx is a measure of the probability of finding the particle
in that range. Clearly if we sum 1/1 2 over the range of x from 0 to 00 the result must be
unity if one particle is involved.
For most purposes, it is justifiable to regard an atom as consisting of a stationary
nucleus of charge +Ze surrounded by a cloud of negative charge whose density at any
point is proportional to the probability 1/1 2 of fmding an electron at that point. This
description of an electron cloud still allows the energy of an electron state, a stationary
state, to be computed exactly. Just as a vibrating string can give rise to stationary
vibrations or standing waves only if the wavelengths are related by whole numbers n
FUNDAMENTALS 3
to the length of the string (Fig. 1.2), so the wave equation (eqn (1.4)) for the motion
of an electron around a nucleus can give rise to stationary states only if whole numbers
are characteristics of the solution. The whole numbers are quantum numbers reminiscent
of those in the work of Planck and Bohr which now turn up naturally from the solutions
of the wave equation.
~---------I-----~~Y
x
Fig.1.3 Spherical polar coordinatesr, 11, cPo
4 PHYSICAL PROPERTIES OF MATERIALS
where the standard abbreviation h =h/21T has been introduced. In polar coordinates
the full expression looks formidable.
where R is a function of r only, F of 0 only and <I> of l/J only. Each of these functions
mirrors a motion of the electron in terms of the corresponding coordinate and each
solution throws up its corresponding quantum number n, 1and m,. These quantum
numbers are not independent since although n can be any positive integer, 1can
only have one of the values 0, 1, ... , (n - 1) for each n, and m, can only have one of
the values
m,=-I, -(/- I),' ... ,0, ... , (1- 1),1
for each L Thus, for n = 1, only 1 =0 and m, = 0 are permitted while for n = 2, 1=0 (in
which case m, = 0) or 1= 1 when m, can be either -I, 0 or +1. These rules result from the
general boundary conditions imposed on 1/1 and do not arise from special assumptions for
each particular problem.
The quantization which gives rise to m,
is the simplest case, since there is only one
term in eqn (1.7) which involves l/J. The solution of the resulting differential equation is
z y
---I~--x x
mr= ± 1
m, = 0
Fig. 1.4. Electron cloud patterns for I = 1 state showing how the limits of significant charge
density (ex",2) depend onml.
necessary to postulate the existence of yet another quantum number which was
attributed by Uhlenbeck and Goudsmit in 1925 to the 'spin' of the electron about
its own axis. This fourth quantum number is called the spin quantum number sand
can only have two components ms = ± ! corresponding to angular momentum !Ii. It
is sufficient to think of this as a motion, not in ordinary space, but in a new spin
space, a necessary corollary to some surprising experimental results. It is comforting
that having invented a fourth quantum number the list is now complete. In a hydrogen
atom, then, an electron state is now described by four quantum numbers n, I, ml and
ms and each state has its own characteristic electron cloud pattern 1/1. The electron
clouds of s states are spherically symmetrical since I and ml are zero and so 1/1 does not
depend on 8 and cpo The state oflowest energy, known as the ground state, is the (IS)l
state (superscript 1 indicating one electron in that state). The electron cloud for this
state is a spherically symmetrical ball of negative charge whose density becomes
negligibly small at distances from the nucleus greater than 2 A. When l*-O the F(8)
function multiplies the R(r) function to give complex shapes to the electron cloud. A
few of the simpler examples are given in Fig. 1.5.
For any given value of n, the general effect of increasing I is to increase the mean
distance of the electron from the nucleus. In the n = 3 quantum states, for example,
the electron is, on the average, closest to the nucleus in the 3s state, and farthest away
in the 3d state.
Note that on this simplified wave mechanical model the energy of any atomic
state is determined only by n. This is a consequence of assuming a strict Coulomb
6 PHYSICAL PROPERTIES OF MATERIALS
z z
y y
~--x ----E---x
potential between the nucleus and the electron. When this assumption breaks down
either through the need to consider interactions involving electron spin, because of
the presence of more than one electron in the atom, or because of the presence of an
external magnetic or electric field, the energy levels are split according to the par-
ticular I, m[ and ms values; in general the lowest I values have the lowest energy for a
particular value of n (see below).
The problem of the electronic structures of atoms containing more than one electron
is extremely complicated, and accurate calculations exist only for helium. The com-
plications arise from the fact that each electron moves in the field, not merely of the
nucleus, but of all the other electrons. For atoms of low atomic number, it is some-
times a reasonable approximation to assume that each electron moves in the field of
the nucleus, and of a spherically symmetrical negative charge resulting from the
smeared out motions of the remaining electrons. In this case wave mechanics still
gives solutions in which the electronic states are dermed by the quantum numbers
n, I, m[ and ms. The same is assumed to apply to the heavier elements. It is
customary to call electrons in states for which n = 1,2,3, . . ., the K, L, M, ... ,
electrons.
It might at first be thought that the lowest energy of an atom would result from
placing all the electrons in the lowest or (Is) state but this is not so. The distribution
of electrons is governed by the Pauli Exclusion Principle, according to which not more
than one electron can be in a given energy state as defined by all four quantum num-
bers. As the spin quantum number ms can only equal ± !, the Exclusion Principle is
FUNDAMENTALS 7
sometimes expressed by saying that a given energy state, as defmed by the quantum
numbers n, I and m" can contain not more than two electrons and that when it is fully
understood from the following argument. The three quantum numbers n, I and
completely determine the shape of the wave function and hence also the charge
m,
occupied, these must be of opposite spin. The origin of this principle may be partly
distribution. If two electrons have the same values for these quantum numbers their
charge distributions are thus directly superimposed on one another, or in other words
the electrons are 'in the same place' . Now this is energetically a very unfavourable
situation, since the two electrons must repel one another strongly. The repulsion can
be minimized if the electron spins are opposed; their quantum numbers, ma are then
different. It seems that two electrons with counter-rotating spins do not repel one
another very strongly. On the other hand, if their spins are parallel the repulsion is so
strong that they cannot occupy the same position and therefore they must have a
different set of values for n, I and mi.
A combination of the Exclusion Principle with the restrictions on the values of I,
m, and ma leads to the following conclusions:
(a) The maximum numbers of electrons that can be contained in states character-
ized by principal quantum numbers 1, 2, 3, ... , n are equal to 2, 8, 18, ... , 2n2
respectively.
(b) For a given value ofn, the maximum numbers of electrons that can be con-
tained in s,p, d and/states are 2,6,10 and 14 respectively, i.e. 2(21 + 1).
The Pauli Principle is of importance because, by placing a limit on the number of
electrons in a given state, it leads to the concept of completed groups and subgroups
of electrons, and this explains a periodic repetition of properties with increasing
atomic number. The Exclusion Principle applies to molecules and to other assemblies
of atoms in which the electrons occupy quantized energy states, and in all such cases
no state can contain more than one electron as defined by all the quantum numbers,
including the spin quantum number.
While helium is an inert gas, lithium is a reactive metal which readily emits electrons
when heated in a vacuum. In assigning the value 1= 0 to the third electron we have
made use of another rule; in a multi-electron atom the levels with lowest 1 fill up first.
The influence of 1 on the energy is illustrated schematically in Fig. 1.6.
The electronic structure of the elements can now be built up assuming that the
levels are filled sequentially with rising values of n and I. Table 1.1 shows the number
of electrons in each subshell for each of the first 18 elements.
Electron
energy
t- - ----------------------::---::- : -
---------------------------
-~--{=2-===:~=~
i=1=
L-O
l: 0
i=3
TABLE 1.1
1 H 1
2 He 2
3 Li 2 1
4 Be 2 2
5 B 2 2 1
6 C 2 2 2
7 N 2 2 3
8 0 2 2 4
9 F 2 2 5
10 Ne 2 2 6
11 Na 2 2 6 1
12 Mg 2 2 6 2
13 AI 2 2 6 2 1
14 Si 2 2 6 2 2
15 P 2 2 6 2 3
16 S 2 2 6 2 4
17 a 2 2 6 2 5
18 A 2 2 6 2 6
FUNDAMENTALS 9
Certain interesting features arise in Table 1.1. The elements with atomic numbers
2, 10 and 18 are the rare gases which are not merely stable but are exceedingly inert
chemically. Thus we may equate inert characteristics with completely filled shells or
subshells of electrons as is illustrated by the variation of ionization energy. This is
the first indication, as we move down the list of elements, that similar arrangements
of outer electrons appear periodically and that this periodic variation is reflected in
the chemical properties. In fact the periodicity of chemical properties was noted
before the electronic structure of the elements was known ana as long ago as 1870
the chemist Mendeleev devised a way of tabulating the elements which demonstrated
it very effectively. This is now known as the Periodic Table of the Elements.
To construct the Periodic Table the elements listed in Table 1.1 are placed in
horizontal rows so that the outer electron structure changes stepwise along the row.
A new row is begun whenever a p shell has just been fIlled with electrons. Ignoring for
the moment elements one and two this means that each row finishes with a rare gas,
for example, neon with its full2p shell or argon with its full 3p shell. In this way
elements with identical numbers of electrons in their outermost shells appear directly
beneath one another. Thus lithium, sodium and potassium, each with one electron
outside closed p shells, appear in the first column or Group. More remarkably, these
elements all display very similar behaviour-they all are very reactive, are metals with
a valency of unity, and they all form similar compounds with, for example, fluorine
or chlorine. This last point may be amplified by noting the properties of the chlorides;
they all form transparent, insulating crystals, which are readily cleaved to form regular
and similar shapes. They all dissolve, to a greater or lesser extent, in water and they
all have fairly high melting points (about 700°C).
Corresponding similarities also occur between the elements in the second group;
beryllium, magnesium and calcium are all light metals with a valency of two; they
form very stable oxides which have higher melting points than the alkali halides and
they tend to be reactive although not to the same degree shown by the alkali metals.
This unity of chemical behaviour is common to all groups of the Periodic Table
and not just those discussed above.
So far the discussion has been carefully limited to the first three rows of the Periodic
Table plus potassium and calcium. The reason is that although the periodicity of
properties continues beyond this point it is not exemplified in such a simple fashion.
The arrangement of the electrons for the elements of Z = 19 to 29 are shown in
Table 1.2, and an anomaly is immediately apparent. The rule concerning the order of
filling the various levels has been broken and instead of the outermost electrons of
potassium and calcium entering the 3d shell, they go into the 4s shell. Only when the
4s shell is full do electrons begin to enter the 3d shell-there is one 3d electron in
scandium, two in vanadium, etc. Elements which possess a partially filled 3d shell are
termed transition elements and constitute the first transition series.
The reason for this oddity is that the 4s levels have lower energy than the 3d levels
and so they fill first in keeping with the minimum energy principle. This is illustrated in
Fig. 1.6 where it can be seen that the 4s level is below the 3d level. Further irregularities
occur at Z =24, chromium, and Z = 29, copper, each of which contains only one 4s
electron instead of two. This is due to the fact that the exactly half-f!lled 3d shell and
the filled 3d shell are particularly stable configurations compared to the neighbouring
occupancies of four and nine electrons respectively.
10 PHYSICAL PROPERTIES OF MATERIALS
TABLE 1.2
19 K 3s 2 3p6 Jdo 4s 1
20 Ca 3s 2 3p6 Jdo 4s 2
21 Sc 3s 2 3p6 Jd1 4s 2
22 Ti 3s 2 3p6 3d 2 4s 2
23 V 3s 2 3p6 Jd3 4s 2
24 Cr 3s 2 3p6 3dS 4s 1
2S Mn 3s 2 3p6 Jds 4s 2
26 Fe 3s 2 3p6 Jd6 4s 2
27 Co 3s 2 3p6 Jd7 4s 2
28 Ni 3s 2 3p6 Jd8 4s 2
29 Cu 38 2 3p6 Jd10 4s 1
These transition elements, with their incomplete 3d shells do not fit quite so neatly
into the Periodic Table and so they are left on their own as illustrated in the periodic
chart of Appendix II.
Having dealt with the first series of transition elements, it is not surprising to fmd
another series in the fifth row of the table, and also in the sixth. The latter series, how-
ever, is more complicated owing to the filling of the two inner sub shells (4f and Sd-see
Appendix III) before the 6p shell begins to fill. Here the series of elements from Z =57
to 71, in which the 4f shell is being filled, have almost identical chemical properties
and as a group they are called the rare earth metals or the lanthanide series. The reason
for their chemical similarity is that these atoms differ only in the number of electrons
in a shell which is far removed from the outermost electrons while it is the latter which
determine the chemical behaviour. Their almost indistinguishable qualities place them
all into one pigeon-hole in the Periodic Table.
Another group of elements similar to the rare earths is found in the seventh row of
the Periodic Table, although the majority of these elements do not occur in nature-
because their nuclei are unstable-they have, however, been manufactured artificially.
The elements in this last series are often referred to as the actinide elements because
they have properties similar to actinium, the first of the series.
The orbital and spin magnetic moments of an electron are a consequence of the orbital
and spin angular momenta. It is a relatively easy matter to deduce the magnitude of
the magnetic moments in terms of the appropriate quantum numbers and the results
are
Jlorb =- m,(eh/41rm) (LlO)
and
Jlspin =- gm,(eh/41rm). (Lll)
The factor g is a multiplying factor which is found to be necessary to bring the relation
for spin moment into line with that for orbital moment. It has the value of almost
exactly 2 and is called the Lande splitting factor. Since m,
is an integer, the quantity
FUNDAMENT ALS 11
eh/41Tm can be regarded as a natural unit of magnetic moment, the Bohr magneton,
/lB, and has the value 9.27 x 10-24 amp m2 .
The magnetic moments of single atoms will be considered before the magnetic
properties of a lattice of atoms are discussed. To obtain the total magnetic moment
of an atom we have to add up the magnetic moments of the electrons in it taking into
account the fact that the direction of the moments is given by the sign of ml and ms.
Thus closed shells of electrons having equal numbers of electrons with positive and
negative values of ml and ms have no net magnetic moment since each electron cancels
another. This is illustrated for a p state in Table 1.3, where the dipole moment of
each electron is represented by an arrow. Therefore, the first important rule to note
is that a net magnetic moment comes only from incomplete electron shells.
TABLE 1.3
Electron 1 2 3 4 5 6
ml - 1 - 1 0 0 +1 +1
Moment due to ml t t t t
ms +.!2 1
-1" +.!2 1
-1" +.!2 1
-1"
Moment due to ms t t t t t t
Now the valence electrons together form an incomplete shell so that one might
expect all but the rare gases to exhibit net magnetic moments. However, this is true
only for atoms in isolation, i.e. in the gaseous state, for when solids are considered only
a few elements are found which are permanently capable of retaining a magnetic
moment at room temperature. If all the elements are studied at very low temperatures
many more are found to be magnetic, and if compounds are included yet more
materials are found to be strongly magnetic. However, with one or two exceptions,
all the permanent magnetic materials (including compounds) contain at least one
transition element.
The distinguishing feature of a transition element is that one of the inner electron
shells is incomplete and this suggests immediately that the inner incomplete shell is
the one which gives rise to the atomic magnetic moment while the valence electrons
are of no importance. This characteristic of the valence electrons is just a reflection of
the properties of atomic bonding since atoms bond together in such a way as to form
closed shells of electrons which have no magnetic moment.
Before calculating the magnetic moment of an atom due to an incomplete inner
shell the values of ml and ms for the electrons in the unftlled shells must be considered
because, in the filling of energy levels in multi-electron atoms earlier, no indication was
given as to the sequence of ftlling energy levels in a subshell. The rule governing this
was first deduced from the emission spectra of such elements, and it is known as
Hund's rule after its discoverer. Hund deduced that the different m1levels are filled
first with electrons having the same ms value, that is ms = ~. The electrons arrange
themselves among the levels to give the maximum possible total spin angular momentum
consistent with Pauli's Principle. Thus there must be some form of interaction between
electrons when they have different wave functions which tend to make their spin axes
parallel.
As an example consider the second element in the first transition series, namely
titanium, which has two 3d electrons, with, according to Hund's rule, quantum numbers
ml =-2, ms =! and ml =-1, ms =~. The total orbital magnetic moment is the sum of
12 PHYSICAL PROPERTIES OF MATERIALS
TABLE 1.4
Element K Ca SC Ti V Cr Mn Fe Co Ni Cu
No of 3d
electrons 0 0 2 3 5 5 6 7 8 10
Spin directions - tt ttt ttttt ttttt tttttj, tttttH ttttttH tttttHtH
Magnetic
moment in
Bohr
magnetons 0 0 2 3 5 5 4 3 2 0
The experimental values for the magnetic moment per atom are less than the
theoretical values in Table 1.4 because the broadening and resulting overlap of the
3d and 4s energy bands in the solid state affects the electron distribution between the
various levels.
1.5.1 Diamagnetism
Diamagnetism is a property of all atoms because of the influence of an applied magnetic
field on the motion of electrons in their orbits. It is a very weak effect and in solids is
often masked by other kinds of magnetism.
Consider an electron moving around a proton in an orbit of radius r (Fig. 1.7) with
angular frequency w in the absence of a magnetic field. There is a magnetic moment 11
associated with the motion of the electron
J1 = current x area
(1.14)
= (-we/2rr) x rrr2.
When an external magnetic field H is applied there is a flux change through the orbit
and by Lenz's law an electric field is set up to oppose this change. Note that this
FUNDAMENTALS 13
happens only while the magnetic field is changing from its zero value to the steady
value H The electric field affects the motion of the electron by exerting a force on
it which changes its angular momentum. The change in angular frequency thus effected
Fig. 1.7 Electron orbit around a proton and the associated magnetic moment iJ..
is called the Larmor frequency and has the value eB/2m where B is the flux density
appropriate to Hand m is the electron mass. The corresponding induced magnetic
moment is opposed to the original magnetic moment /l of the electron in its orbit
when H = O. Since this behaviour is a property of electrons in all materials it leads to
a diamagnetic susceptibility which is very small and negative (Chapter 9).
1.5.2 Paramagnetism
As discussed above, each electron in an orbit has an orbital magnetic moment /lorh and
a spin magnetic moment /lspin. In atoms with filled electronic shells there is no net
magnetic moment but where shells are unfilled there is such a moment. In the absence
of a magnetic field the net moments of the atoms usually point in random directions
because of thermal fluctuations, producing no macroscopic magnetization. When a
field is switched on, there is a tendency for the dipoles to align with the field giving
an induced positive dipole moment which is proportional to the field. The component
of each individual electronic moment in the field direction is governed by the rules of
quantum mechanics and a theoretical calculation of the net moment where both orbit
and spin are contributing can be complicated. The induced magnetisation is the
source of paramagnetism and materials whether gases, liquids or solids, show a
paramagnetic susceptibility (Chapter 9) whenever the paramagnetic atoms can
respond individually to an external field without interaction with their neighbours.
1. 5. 3 Exchange
So far the discussion of the magnetic moments of atoms has assumed that each atom
behaves quite independently of its neighbours with no interaction between them.
However, the electron clouds of two neighbouring atoms of a solid can overlap to some
extent on a wave mechanical picture. If, for example, two adjacent transition metal
atoms are considered, the wave functions of the 3d electrons in the unfilled shells have
characteristic lobes which assist overlap (Fig. 1.8). It is important in these cases to
consider what influence the overlap of charge has on the energy of the system of two
adjacent atoms. Suppose there is one 3d electron on atom A and one on atom B (Fig.
1.9). The energy of interaction depends directly on the Coulomb forces acting between
the charge clouds and this is expressed in wave mechanical language as an integral
involving the wave functions of electron 1 and electron 2. But these wave functions
14 PHYSICAL PROPERTIES OF MATERIALS
z z
,.c;....,_ _ _ _ X
.!I.IWIr---X
Electron 1 Electron 2
Overlap
• •
region
Atom A Atom B
also intimately involve the spin of the electron and so the expression for the energy
is found to involve both Coulomb effects and spin inextricably bound together. So to
minimize one involves some restriction on the other. The details of the theory show
that the new term in the energy which appears because of the intermingling of the
wave functions can be written in a simple way as
(1.15)
where Eex is the contribution to the energy from the exchange interaction as it is
called, J represents an integral which involves the wave functions of the electrons and
the potential and in principle is known either through calculation or measurement,
and SA and SB are the spins associated with the electrons in question, but written in
quantum mechanical language as vectors.
As always, the favoured situation is the one with the lowest energy and it turns out
that because there are two ways in which the wave functions can combine there are
two possibilities for lowering the energy by Eex. These are:
(a) If J is positive and the spins are parallel (tt). This is the situation leading to
ferromagnetism because if two neighbouring spins are held parallel by the
exchange interaction, then their neighbours in turn will also be parallel by the
FUNDAMENTALS 15
same kind of pair interaction even though it does not extend beyond nearest
neighbours. In this way a whole line of spins can be parallel even though the
interaction acts only between nearest neighbour pairs (Fig. 1.10).
(b) If J is negative and the spins are antiparallel (t "'). This situation leads to anti-
ferromagnetism (Fig. 1.11). (See also Chapter 9.)
Subsequent calculations have cast doubt on the exact form of the radial dependence
of J. Without quantum mechanics there would be no explanation for ferromagnetism
even though a detailed understanding of the phenomenon has not yet been achieved.
The same idea leads to Hund's rules for spins in unfilled shells of the same atom where
the preference is for parallel alignment as far as the Exclusion Principle allows.
The task of determining the behaviour of electrons in a solid reduces essentially to the
determination of the quantum mechanical states of the solid and the way in which
these states are occupied by electrons. For electrons numbering about 1028 1m3 , all
interacting with one another, the exact solution of this problem is at present impos-
sible to achieve. However, the use of simple models and approximate methods can
help considerably. Here a simple introduction is given to the energy band model of a
crystalline solid. In Appendix I the same model is derived from a different starting
point using the wave picture of the electron and with the aid of some mathematics.
The ideas developed in this appendix are of fundamental importance for a proper
understanding of the behaviour of electrons in crystalline solids.
Energy
~
:
:, ,.....------
~::==~~!!!~_
,,
Outer
electron
levels
,
I
I
Inner
electron
levels
RO Atomic spacing
Consider a small number of widely separated identical atoms which are gradually
brought together to form a large molecule. At infinite separation the atoms are
independent and have electron energy levels appropriate to single atoms. As the atoms
move closer together the electrons and nucleus of each begin to be increasingly
influenced by those of neighbouring atoms. In a hypothetical six-atom molecule, for
example, a single atomic energy level becomes one which contains six components,
one from each atom, and this can be occupied by six electrons. But the interactions
within the molecule slightly modify the level structure so that the single sixfold level
is broadened into six separate levels spread over a range of energy. The disturbance of
the atomic level structure and degree of broadening depends on the atomic spacing.
Figure 1.13 illustrates the situation for a typical inner electron and outer electron
level as a function of atomic spacing. The outer electron levels split at larger separa-
tions because their charge clouds have a greater radial extent. On average the energy
FUNDAMENTALS 17
levels must move down on the energy scale as the atoms are brought together because
the energy of the solid must be less than that of the individual atoms for stability of
the molecule. The spacing Ro represents that for a stable structure in this case.
In a crystal of N atoms, where N -1 026 the same effects take place, each atomic
level splitting into N levels. The average energy decreases with interatomic spacing
until the atoms are at their equilibrium separation. The N associated levels in any
group lie close together and form a 'band of allowed energies'. In general each band is
separated from neighbouring bands by a 'forbidden band of energies'. The energy
range of these bands at the equilibrium separationRo is shown in Fig. 1.14. Note that
the outer or valency electron levels form the broadest highest energy bands.
Figure tI.1S shows the actual situation for sodium metal. Electron charge clouds of
electrons in the deeper atomic levels overlap hardly at all and so lead to very narrow
bands with N levels packed into a very narrow energy range. Usually these electrons
are not of great interest as they do not partake in thermal and electrical phenomena.
Energy ",",,"7"7"7"7"7"7""7"7"7"7"7"7""7"7"7"7"7'7"
Allowed
bands
'~--
Energy
• I •• -- ••••• 3P
.' 3s
,
,
I
~----r----- 2p
Atomic
spacing
Fig. 1.15 Band structure for sodium.
Since the Exclusion Principle applies to the crystal as a whole, only two electrons of
opposite spin can be in any given energy level in the band and the levels are filled from
the lowest state upwards. At the absolute zero of temperature all the electron states
will be m.led to an energy known as the Fermi level. At higher temperatures some
electrons near the Fermi level may be thermally excited into levels just above it. The
uppermost partially filled band is called the Conduction band because the electrons
in it are able to receive energy from an applied electric field and give rise to an electric
current. The highest filled band, at the absolute zero of temperature, is referred to as
18 PHYSICAL PROPERTIES OF MATERIALS
the valence band because the electrons then filling it are valency electrons involved in
bonding. As an example consider lithium in which the outer band will correspond to
the broadened 28 level so that a crystal of N atoms could accommodate 2N electrons
in this band. But each lithium atom contributes only one electron to the band so that
N electrons only fill half of the available states in the outer band lithium is therefore
a good conductor because the electrons at the Fermi level have empty states above
them immediately available for occupation under the influence of an applied electric
field. In some conductors the valence band is full but overlaps a higher empty band
so that electrons are still free to absorb energy and move into higher states.
E"·~'l
Empty
conduction
band
Empty
T
rEmptybandl
, overlaps I
'valence band,
conduction
band 1
Forbidden gap Eg
Forbidden
gap
,.------.,
1
Eg>2eV
Partially
valence orfilled Empty
_____levels Ferml. IL _ _ _ _ _ J'
Full T Full
conduction Occupied level Valence valence valence
band levels band band band
Fig. 1.16 Simplified energy band diagrams for conductors, semiconductors and insulators (inner
electron bands not represented).
When a forbidden energy gap separates a filled valence band from an empty con-
duction band, no electronic activity will be observed unless electrons in the lower
band are supplied with sufficient energy to enable them to surmount the gap. Thus
the width in energy of the gap is critical in determining the electrical or optical
properties of the material. If the gap is a few electron volts wide the material is an
electrical insulator for moderate electric fields and ordinary temperatures. If the gap
is sufficiently small the effect of thermal excitation may transfer a few electrons into
the conduction band creating a few vacant states (holes) in the valence band. Such a
material is a semiconductor, the difference between this and an insulator being simply
the width of the gap. Typically, materials with an energy gap ofless than 2 eV would
be regarded as semiconductors. Silicon is a semiconductor with an energy gap of 1.1 eV
while diamond is an insulator with a 7 eV gap. Fig. 1.16 illustrates the situation in
several cases.
Reference
From experience one knows that solids are usually moderately strong, slightly elastic
structures. Thus the individual atoms must be held together in solids by interatomic
forces or bonds. However, in addition to these attractive forces there must be a
repulsive force because solids are not easily compressed.
The attractive forces between the atoms are basically electrostatic in origin and the
classification of the different types of bonding is strongly dependent on the electronic
structure of the atoms concerned, and hence directly related to the Periodic Table.
There are basically four classes into which these attractive bonds can conveniently be
divided, although the boundaries between them are not always distinct:
The attraction between the atoms brings them closer together until the individual
electron clouds begin to overlap and a strong repulsive force arises to comply with
Pauli's Exclusion Principle. When the attractive force and the repulsive force between
any two atoms are equal, the two atoms should be in a stable situation with a minimum
potential energy.
Consider two atoms in their ground states with an infinite separation, and con-
sequently an interaction potential energy equal to zero. As the atoms are brought
closer together, the potential energy of either atom will be given by
and in the case of a simple type of interatomic potential this can be written as
A B
V=--+- (2.1)
r" ,m'
where A , B, n and m are constant in any specific case. In order to give a minimum
potential energy condition at the equilibrium separation '0, m must be greater than
n so that 3 2 Vj3,2 is positive. The resultant force of interaction, F, is given by
F=- avia" (2.2)
20 PHYSICAL PROPERTIES OF MATERIALS
so that
nA mE
F = - - +-_. (2.3)
rn+ 1 r m + 1
Equations (2.1) and (2.3) can be represented graphically as shown in Fig. 2.1.
At the equilibrium separation the net force is zero, i.e.
F V
I I
Force Energy
I I
I I
I Repulsive energy
\ /"" Repulsive force
\/
\ \
k
\
Resultant force
\
""
;
,"
;
Fig. 2.1 Force and potential energy as a function of distance between two atoms.
The minimum value of the energy Vo , is the bonding energy of the system and is
approximately equal to the heat of dissociation of the molecule. Typical values are
given in Table 2.3.
The four classes of atomic bonding will now be discussed in more detail.
(a)
(b)
Fig.2.2 (a) Individual atoms (denoting the inner closed shells as a solid sphere). (b) Ionic molecule
(force of attraction = -e 2 /4,"EOr02).
Inner
closed
shells
Historically the study of crystal structures began in about 1700 with the study by
metallurgists of the external form or morphology of crystals. Even before the applica-
tion of X-rays to crystallography it was assumed that the geometrical regularity of
many crystals (e.g. the cubic form of sodium chloride) was the result of a correspond-
ing regularity of atomic structure within it. X-ray studies have confirmed this and
enabled a very detailed picture of the arrangement of the atoms in most materials to
be deduced. The technique of X-ray crystallography will be discussed in more detail
in the next chapter.
Before discussing the possible crystal structures the meaning of crystalline order
will be made clear. In a crystal it is possible to choose a small group of atoms or
24 PHYSICAL PROPERTIES OF MATERIALS
b
L~---
STRUCTURE OF SOLIDS 25
As a result of pure symmetry considerations Bravais showed that corresponding
to these there were only fourteen different types of fundamental network, or unit
cell, which could be built up to form an extended three-dimensional lattice, in
which the surroundings of each of the lattice points are identical.
These fourteen different unit cells include the seven so-called primitive unit cells
which have lattice points only at the corners. Translational repetition in space of each
of the unit cells constructs a Bravais lattice. The repeat distances are the unit cell
edges a, band c and are called the lattice parameters. The unit cells of the fourteen
Bravais lattices are illustrated in Fig. 2.4. Although the symmetry of these lattices
varies considerably they all have one element of symmetry in common in that they
all possess a centre of symmetry. This is defined as a point in the interior of the lattice
through which every lattice point can be taken to an equidistant lattice point on the
other side of the lattice.
Turn no)V from these purely geometric networks of lattice points to the actual
crystal structure exhibited by materials. In general each lattice point is associated with
a group of atoms or molecules rather than just one atom, and the many different
geometric arrangements possible mean that there are far more than just fourteen
different types of crystal structure. In many cases the resulting crystal structure does
not show the full symmetry associated with the basic crystal structure. For example,
in cases where complex molecules with asymmetric shapes are associated with the
lattice points the resulting crystal structure may no longer have a centre of symmetry.
[jJ ~ CWfJ]
Cubic: P I
/
I "' /
... 1 __ /~_:::
P C
P
Tetragonal Orthorhombic:
Trigonal
P C
Monoclinic: (Rhombohedral)
Hexagonal Triclinic
Fig. 2.4 The unit cells of 14 Bravais lattices. (The lattice points are represented by dots.)
26 PHYSICAL PROPERTIES OF MATERIALS
D ~---Rr-----r,
8
+"'---- Y
Unit cell
OA = a
A~-------------' OB = b
OC = c
x
Fig. 2.5 Designation of a crystal plane by Miller indices.
Thus the planeA'B C' is denoted by (OA/OA' OB/OB' OC/OC') which in general is
written as (hkl). If OA' = a12, OB' = bl4 and OC' = cl2 then the plane A 'B' C' is the
(1/1 II! liD or the (242) plane.
Similarly a plane (ABC) with intercepts OA, OB and OC is the (OA/OA OB/OB
OC/OC) or (111) plane, and the plane DFBA is the (OA/OA OB/OB OCloo) or (110)
plane, Le. a plane parallel to an axis is taken to intercept that axis at infinity. If the
intercept is negative then the corresponding Miller index is also negative and is denoted
by placing a bar over the numerical value, e.g. 2. The indices of the sides of a unit cell
of a cubic lattice are (100), (100), (010), (010),(001) and (001) and these are equiva-
lent planes of the same form and are collectively represented by {100}. Further examples
of planes in the cubic system are illustrated in Fig. 2.6.
STRUCTURE OF SOLIDS 27
In the hexagonal system an alternative indexing system, which has four numbers in
each set of indices, is often used because it more readily shows the hexagonal symmetry.
The four axes used to define the indices are the vertical axis of the hexagonal prism
and the three axes in the base, the latter being 1200 apart as illustrated in Fig. 2.7.
The indices of a plane are denoted by h, k, i and I where these indices are the
reciprocals of the intercepts on the a, b, d and c axes respectively. Since only three
noncoplanar axes are necessary to specify a plane in space, these four indices cannot
be independent and they must satisfy the additional condition that h + k = - i. Thus
the base plane in a hexagonal unit cell is the (0001) plane and the vertical prism planes
belong to the {lOlO} family.
It should be noted that all these indices represent a set of equidistant parallel
planes since our reference unit cell could be chosen anywhere in the crystal. In X-ray
crystallography (see Chapter 4) interest lies in the interplanar spacing of these planes,
which is defined as the perpendicular distance from the origin to the plane, i.e. a
similar imaginary plane is assumed to pass through the origin. The indices (nh, nk and
nl) where n is an integer, therefore represent a family of planes which lie parallel to
the family (hkl) but with an interplanar spacing lin of that of the latter. Note that as
far as the external morphology of the crystal is concerned the (111) and (222) planes,
for example, are identical and so it is common in this case to reduce the indices to
their lowest form.
The direction of a line in a lattice with respect to the unit cell vectors can be des-
cribed using a similar notation. The indices of a direction are simply the vector com-
28 PHYSICAL PROPERTIES OF MATERIALS
-r---b
[110]
a
Fig. 2.8 Lattice directions.
The role of interatomic bonding forces in the structure of materials will now be con-
sidered. The atoms in a solid state may be regarded for most purposes as rigid spheres
with all the atoms of one element having identical size. An obvious starting point is,
therefore, the consideration of the kinds of structure that can be built up by packing
identical spheres together as closely as possible. This is especially valid for metallic
and van der Waals bonded solids of identical atoms where the bonding is non-
directional.
A single layer of spheres, pushed together, arrange themselves in a hexagonal close
packed network with six nearest neighbours as illustrated in Fig. 2.9. This layer offers
recesses into which additional spheres can rest to form a second layer. It can be seen,
however, that this second layer will occupy all the X sites or all the Y sites ( occupation
of the X sites is illustrated in Fig. 2.9). A third layer is again presented with alter-
natives, either the spheres can be placed directly over those in the first layer or over
the previously unoccupied Y sites. If the spheres in the third layer are positioned
vertically above those in the first layer a hexagonal lattice is formed which is known
as the hexagonal close packed structure (h.c.p.) with a stacking arrangement
ABABAB . ... This is illustrated in Fig. 2.10.
Placing the third layer of spheres in the Y sites necessitates the fourth layer lying
directly over the first layer. Thus the stacking arrangement is now ABCABCABC ... ,
etc. and a face centred cubic structure is built up. The axes of the cube, however, are
neither in, nor perpendicular to, the plane of the close packed layers and actually the
layers are (I 11) planes. Note the face centred cubic lattice illustrated in Fig. 2.11 is
not the simplest building unit but it is taken as the unit cell because it shows the
maximum symmetry.
STRUCTURE OF SOLIDS 29
x X sites
• Y sites
,I
two atoms (000) and
I (2'3 1/3 1/2 ) associated
with each lattice point.
II
II •
.... -,
II I
(Spheres represented
by lattice points)
Each atom has twelve nearest neighbours in both of these close packed structures;
six in the close packed plane and three in each of the adjacent close packed planes
above and below. The number of nearest neighbours is called the coordinatio n number.
30 PHYSICAL PROPERTIES OF MATERIALS
As mentioned earlier these two close packed structures should be favoured by solids
with non-directional forces, i.e. metallic or van der Waals bonding, and indeed many
metals do crystallize in the f.c.c. structure, as do all the solid rare gases except helium which
adopts the h.c.p. structure. Among the metals the h.c.p. structure is exemplified by
cobalt and zinc.
Although these are the only two ways in which identical spheres can be closely
packed together, a very important structure, where the packing is slightly looser, is the
body centred cubic (b.c.c.) structure. In this case each atom has only eight nearest
neighbours and this is indicative of the relatively loose packing. In fact the packing
fraction, or the fraction of the unit cell occupied by the spherical atoms is 0·74 for
both f.c.c. and h.c.p. but only 0·68 for b.c.c.ln spite of this looser packing many metals
crystallize in the b.c.c. form including all the alkali metals, and iron below 910°C. Two
possible reasons for the formation of this structure rather than the close packed
structures are (a) the thermal energy of the atoms (e.g. alkali metals tend to change to
the f.c.c. phase at low temperatures) and (b) partial covalent bonding.
Close
'_--0-- packed
planes
Fig. 2.11 Face centred cubic lattice showing the close packed planes.
Covalent solids do not form close packed structures because the covalent bonds are
very strong and rigid and their directional nature must be maintained. The simplest
covalent structure is that of diamond (and germanium and silicon) which is a fairly
open, or 'empty' structure and far from close packed (packing fraction 0'34). One-
eighth of the unit cell is shown in Fig. 2.12 illustrating the tetrahedral bonding, i.e.
the coordination number is 4. It is a complicated cubic lattice with two atoms, e.g.
situated at (000) and (Hi), associated with each of the cubic lattice points.
The low pressure crystal form of carbon is graphite which differs in many respects
from diamond. It is a layer structure in which the atoms are covalently bonded to
only three nearest neighbours in the layer, and the layers themselves are only weakly
bonded by van der Waals forces (Fig. 2.13). Since there is one electron per atom that
is not employed in the covalent bonding there is an abundant supply of free electrons,
and so graphite is a good electrical conductor whereas diamond is an almost perfect
insulator. However, the free electrons cannot cross easily from layer to layer and so the
resulting conductivity is highly anisotropic (approximately 106 n-
l m- l parallel to the
layers and 40 n- l m- l perpendicular to the layers). For a long time it was believed
that graphite owed its lubricating properties to the weak van der Waals bonds between
adjacent layers. More recently it has been established that lubrication is associated with
adsorbed gas layers such as oxygen or organic molecules.
STRUCTURE OF SOLIDS 31
T--------~
/ 1 / 1
/ I / I
// I I
/ I
f~-~~- I
1 I
1 I
I I
I I
I J
I /
I //
I / //
1/ 1/
~----------v
Fig.2.12 The diamond structure.
Ionic solids cannot form close packed structures because they do not normally consist
of ions of identical size (see Table 2.2), and the charge on the ions requires an alternat-
ing arrangement of anions and cations. The type of crystal structure is mainly deter-
mined by the packing of the larger ions which are normally the negative anions. Two
simple, common ionic crystal structures are the sodium chloride and caesium chloride
32 PHYSICAL PROPERTIES OF MATERIALS
structures illustrated in Fig. 2.14. The main factor determining which structure is
adopted is the ratio of the ionic radii re/ra' since like ions must not touch because of
the large electrostatic repulsion.
o CI- o CI-
• Na+ • Cs+
(a) (b)
The NaCl structure can be regarded as two interpenetrating f.c.c. lattices and each
ion has six nearest neighbours of opposite charge, Le. the coordination number is 6.
It is favoured by ions having an ionic radii ratio re/ra of approximately 0·5. The CsCI
structure, on the other hand is two interpenetrating simple cubic lattices. It has a co-
ordination number of 8 and re/ra is about 0.9.
TABLE 2.2
Ionic radii
The bonding of ionic crystals is the simplest to analyse quantitatively because the
force of attraction between the ions is the Coulombic force - (e2j41Tfor02) where ro
is the equilibrium separation. Using this value for the force in the analysis of section
2.1 the potential energy for a single molecule is given by
(2.5)
However, in an ionic crystalline solid, each ion has a number of neighbours rather than
a single neighbour as in the case of the isolated molecule, and both terms in the above
expression for the energy will be affected by the presence of these other neighbours.
STRUCTURE OF SOLIDS 33
In the case of NaCI reference to Fig. 2.14a shows that a given ion within the crystal
(e.g. the central Na ion) is surrounded by six nearest neighbours of unlike sign at a
distance TO, twelve next nearest neighbours of like sign at a distance ..;2To and eight
second nearest neighbours of unlike sign at a distance 0To, etc. Thus, the total energy
of a single ion will be
(2.6)
where e is another constant representing the net repulsion due to all the overlapping
electron clouds.
The potential energy can be rewritten as
Vo =(Me 2 /47r€oTo) +elTo m, (2.7)
where M is a constant known as the Madelung constant which is simply a geometric
term. The Madelung constant for the NaCI structure is
6 12 8 6 24
.,ff-..;2+0-.[4+...[5-
and although this series converges slowly it can be shown to equal 1·747558 (for the
CsCI structure M = 1·76267).
Equating the attractive and repulsive forces at To (by analogy with eqn (2.4)) leads
to an expression for C, and enables eqn (2.7) to be rewritten as
Vo=---
47r€OTO m
2
Me ( 1 - - . 1) (2.8)
As explained in section 2.1 this is the bonding energy (b.e.) of the ion. Thus if N is
Avogadro's number the bonding energy per mole will be
NMe
b.e./mol = - -
2
- (1- - ,
47r€oTo m
1) (2.9)
2.7 Summary
The previous sections have shown that the type of atomic bonding plays an important
role in determining the possible crystal structures of solids. In addition the type of
bonding considerably influences the physical properties of the resulting material, as
will be apparent in later chapters. The data in Table 2.3 illustrates in a very general
manner, since many are temperature dependent, some of the properties typical of the
various types of bonding, although it must be stressed again that many solids show
more than one type of bonding. Thus, a classification of solids into the four classes
34 PHYSICAL PROPERTIES OF MATERIALS
(i) ionic, (ii) covalent, (iii) metallic and (iv) van der Waals or molecular, is extremely
valuable.
Some of the materials to be met in later chapters, e.g. typical ferrimagnetic and
ferroelectric materials, have much more complicated crystal structures than the simple
examples discussed in this chapter. However, the general principles of packing different
size anions and cations and the directional nature of covalent bonds is still valid.
TABLE 2.3
Gassification of solids
(i) Typical examples NaO, LiF, Diamond, Si, Na,Fe,Cu Ar, Ne, I
Ca02 Ge, SiC (C6 H 12 0 6h
sugar
(ii) Bonding energy 6-10 x 104 2-10 x 104 1-4 X 104 10 3
(J mole- 1)
(iii) Melting point (K) 700-3300 500-3700 230-4150 < 600
(iv) Electrical resistivity 10 12 _10 20 10-10 20 10- 3 -10- 6 10 12 _10 20
(Om)
(v) Thermal conductivity 2-20 4-600 40-400 0·04-4
(Wm-1K-1 )
(vi) Hardness hard but hard malleable
cleavable
The perfect crystal so far considered is built up by repeated translation of the basic
unit cell along the three crystallographic axes. In practice, however, the real crystal is
not quite this regular and there are a variety of imperfections or defects present in all
crystals. These imperfections playa major role in explaining many of the physical
phenomena exhibited by solids; for example electrical conductivity, mechanical
strength and ferromagnetic hysteresis. Since the effects of the imperfections will be
met many times in the subsequent chapters, it is appropriate to deal with them
collectively at this stage.
Lattice imperfections can be broadly classified into four major classes, namely:
(a) lattice vibrations, or phonons,
(b) point defects,
( c) line defects, and
(d) plane defects.
In addition there may be defects in the electronic configuration of individual
atoms, and, although these have a major influence on electrical and optical properties,
they are not strictly lattice imperfections. Hence, further discussion of these will be
left to the relevant chapters.
At all temperatures the atoms are not stationary in their designated lattice positions
as has been assumed so far, but instead they are vibrating about their mean positions
with a set of characteristic frequencies, v. fhere is a wide range of vibrational
STRUCTURE OF SOLIDS 35
frequencies in solids, with values as high as 10 12 Hz at room temperature, and the
amplitude of vibration may be as much as 10 per cent of the interatomic spacing.
According to quantum theory, the energies of these lattice vibrations are quantized
and for a particular frequency v are given by
En = (n + !)hv, (2.10)
where n is an integer and h is Planck's constant. If energy, in the form of heat, for
example, is supplied to a crystal the energy of the lattice vibrations will increase in
integral steps of hv and similarly if the crystal cools the vibrational energy will decrease
in steps of hv. The similarity between this process and the absorption or emission of
light photons is very close and so the name phonon is used for the quanta of vibrational
energy absorbed or emitted by a solid. Even at absolute zero, where n is zero, the atoms
are not stationary but still possess an energy of !hv, which is called the zero point
energy. The total vibrational energy of the crystal will be given by the sum of the
energies of the individual atoms taking into account the complete vibrational frequency
spectrum of the material.
As well as playing a fundamental role in determining the thermal properties such
as the specific heat of solids, interaction between electrons and phonons partly explains
the electrical resistance of materials.
Point defects refer to those irregularities in the crystal structure which occur over a
region of only a few lattice points. All the atoms in a solid possess vibrational energy
and at all temperatures above absolute zero there will be a fmite number of atoms
which have sufficient energy to break the bonds which hold them in their equilibrium
position. Once free from their lattice sites there are two possible situations that can
occur:
(a) the atom may simply occupy a nearby interstitial position leaving a vacancy at
the original lattice site; the associated vacancy and interstitial is termed a
Frenkel defect, or
(b) the atom may migrate to the surface of the crystal or be trapped at a line or
plane defect in the crystal; vacancies of this type are referred to as Schottky
defects.
These two situations are illustrated diagrammatically in Fig. 2.15.
0000000 0000000
OO~~tOO 0000000
o OO&,? 0
0 000 000
0000000 0000000
0000000 0000000
(a) (b)
Fig.2.15 (a) Frenkel defect. (b) Schottky defect.
36 PHYSICAL PROPERTIES OF MATERIALS
a dislocation of this nature is denoted by the symbol 1. The dislocation line is a region
of higher energy than the rest of the crystal lattice because, as can be seen in the dia-
gram, the lattice above the dislocation line is in a state of compression, whereas below
the dislocation line the lattice is in tension. The abnormal interatomic spacing below
the dislocation line provides a ready 'sink' for both interstitial atoms and chemical
impurities.
order of 10 10 lines m -2. Although dislocations are associated with regions of excess
lattice energy, once formed relatively little energy is required to move them; hence
their importance in mechanical behaviour.
(a) (b)
Fig. 2.18 Burgers vector of edge and screw dislocations. (a) Edge dislocation. (b) Screw
dislocation.
parallel to it in the case of a screw dislocation, and in the case of the 'perfect' dis-
locations shown, the Burgers vector is an integral of the atomic spacing. Thus it is
possible to distinguish the two fundamental types of dislocation by the direction of
their Burgers vector relative to the line of the dislocation.
STRUCTURE OF SOLIDS 39
Naturally the grain boundaries which separate the randomly orientated grains in a
polycrystalline specimen cannot be explained in this way because the misorientations
are much too large. In this case the grain boundaries are a few atomic spacings wide
and in the boundary region almost all semblance of regular ordering is absent and
quite abnormal interatomic spacings can be found. This explains the ability of these
boundaries to act as sinks for defects such as lattice interstitials or impurity atoms.
Plane defects of a more regular nature can occur in many crystal structures either
by introduction during growth or by subsequent mechanical deformation. One example
is the twin boundary in which the boundary acts as a mirror in that the two crystal
orientations on either side of the boundary are mirror images of one another (Fig.
2.20). These two orientations are called twins and they are especially common in
metals with b.c.c. or h.c.p. structure.
Finally, another example is the stacking fault which can arise in the close packed
f.c.c. or h.c.p. structures as a result of an accidental fault in the stacking sequence.
40 PHYSICAL PROPERTIES OF MATERIALS
For example, in the f.c.c. structure the stacking sequence of the close packed planes
should be ABCABCABC ... , however a fault may arise leading to a sequence of the
form ABCABABCABC ... so that a thin h.c.p. region is included in the basic fc.c.
structure.
In conclusion it should be stressed that the discussion in this chapter has been con-
cerned exclusively with the structure and imperfections of crystalline material and no
mention has been made of the structure of amorphous materials. Polymers are a very
important class of material in modern technology and their structure has a very
important influence on their properties. Although some simple polymers may
be prepared as small single crystals, most polymers consist of randomly orientated
long chain molecules. Their basic structure is amorphous although crystalline regions
are by no means uncommon. However, because of their unusual structure, which is so
dependent on their preparation and the nature of the molecules themselves, their
preparation and structure will be dealt with collectively in Chapter 3 (see section 3.9).
Many inorganic materials of interest, although usually encountered in crystalline
form, may also be prepared in the amorphous state. The structure of these materials
in considered in section 4.7.
References
Questions
1. Deduce the nature of the chemical bonding in the following crystals from the
position of their constituent atoms in the Periodic Table.
(a) gallium arsenide, GaAs
(b) neon, Ne
(c) potassium, K
(d) nickel oxide, NiO.
2. Show why (a) F face centred tetragonal and (b) C face centred tetragonal are not
included in the list of Bravais lattices.
3. How many atoms are there in the f.c.c. and b.c.c. unit cells? Calculate the packing
efficiency for these two structures.
4. Calculate the theoretical cia ratio for the h.c.p. lattice and compare this value with
actual values for h.c.p. elements.
S. Calculate the limit of the ionic radius ratio for stability in the NaCI structure, and
compare this with the actual radius ratio for as many ionic crystals with the structure
as you can find.
6. Show that, for any cubic lattice, the separation of the planes corresponding to the
Miller indices (hkf) is given by
d =(h
hk1 2 +:2 + [2 y12.
7. Draw diagrams to illustrate the meaning of the following Miller indices: (a) (121);
(b)(212);(c)(II0);(d) [121];(e) [212] and(f) [110].
8. Platinum (atomic mass 195.1) crystallizes in the f.c.c. form and has a density of
21·4 x 103 kgm- 3 • Calculate the length of the edge of the unit cell. [Avogadro's
number = 6'02 x 1026 k mol-I.]
9. The concentration of Schottky defects in an f.c.c. ionic crystal is 1 in 10 10 at a
temperature of 300 K. Estimate the average separation in terms of the lattice spacings
between the defects at 300 K and calculate the value of the concentration to be
expected at 1000 K.
10. A single crystal of copper contains a low-angle tilt boundary on (001) with a tilt
axis parallel to [010]. Calculate the tilt angle if the spacing of the dislocation in the
boundary is 3 x 10- 6 m and their Burgers vector is 0·4 x 10- 9 m. Give a sketch of the
arrangement.
3 Preparation of Materials
3.1 Introduction
It was seen in the last chapter that materials can occur in a variety of forms, namely,
single crystal, polycrystal and amorphous. Cooling from the liquid state usually
results in polycrystalline material and the common engineering metals and alloys cast
from the melt are composed of numerous crystalline grains. The grain size may vary
from micro metres to millimetres and depends on many factors, such as the casting
temperature and the impurity content. Subsequent mechanical working and heat
treatment produces changes in the grain structure.and this in turn alters the mechanical
properties. As yet little or no attempt has been made by the structural engineer to
utilize single crystal material, but in the field of electronics immense progress has
been made through the increased use of single crystals in solid state devices.
Prior to the Second World War the growth and study of single crystals was some-
thing of a novelty and largely restricted to crystals that could be grown easily from
solution, e.g. common salt and copper SUlphate, and crystal growth was regarded more
as an 'art' than a 'science'. However, in the last twenty years or so synthetic single
crystals of various substances have gained much more practical importance, often for
pure scientific investigation but increasingly for technological and industrial applica-
tions. To date, metal and alloy single crystals have not found much use outside
scientific research, but the current technological applications of single crystals range
from the semiconductor field with silicon and gallium arsenide to the laser field with
doped sapphire, yttrium aluminium garnet and zinc tungstate. More recently, several
niobates and tantalates have found application in non-linear optical devices such as
frequency doublers and parametric amplifiers.
Bulk single crystals can be grown by various methods, including growth from the
melt, from solution or by chemical transport. Thin film single crystals can be grown
epitaxially by vapour deposition onto a single crystal substrate. These various tech-
niques will be described in greater detail in the subsequent sections but initially the
mechanism of the growth process itself is briefly discussed [1, 2].
Consider growth occurring on a crystal surface as shown in Fig. 3.1. When an atom
from the surroundings (Le. from solution, melt or vapour) strikes the surface the
collision will decrease the energy of the atom so that it becomes partially bound to
the crystal surface. Thus any further motion of the atom is restricted to the surface.
Isolated nuclei are then formed from groups of such atoms coming together by chance
PREPARATION OF MATERIALS 43
through their thermal motion (Fig. 3.1a). Of the groups shown in Fig. 3.1 b a mobile
'new' atom is most easily built into the crystal structure at a point such as X, since at
this point the atom will be strongly attracted by three neighbouring atoms and the
local surface energy of the surface due to the unsaturated bonds will be decreased.
Another convenient position will be sites such as Y(Fig. 3.1c) and in this way a con-
tinual supply of atoms will lead to completion of the entire crystal plane.
Once this situation arises continuation of the growth process will rely on the forma-
tion of a new stable nucleus (depicted schematically as a thin cylinder in Fig. 3.2) on the
plane surface. In this new nucleus or seed there will be an increase in free energy due to
the atoms at the edge and a decrease of free energy due to the atoms in the interior.
Thus the stability of the seed will depend upon the ratio of volume to surface, and
there will be a critical radius of the nucleus, rc such that if r > rc the nucleus is stable
and growth can continue whereas if r < rc the nucleus is unstable. Since the lateral
growth mechanism discussed above is a very rapid process the rate of formation of
the new nuclei on the completed planes will be the decisive factor in controlling the
overall growth rate.
The probability of formation of stable nuclei in growth from the melt will obviously
depend on the difference between the melt temperature and the normal melting point
of the solid, i.e. the degree of supercooling of the melt. In vapour growth it will depend
on the relative excess of the pressure of the vapour over the vapour pressure of the
44 PHYSICAL PROPERTIES OF MATERIALS
(a)
(b)
Fig.3.3 Screw dislocation growth. (a) Emergent screw dislocation. (b) Formation of growth
spirals.
spirals in the surfaces of crystals grown from solution or vapour provided one of the
earliest pieces of direct evidence for the occurrence of dislocations.
These two theories of crystal growth differ markedly in their predictions for the
magnitude of the growth rate and its dependence on supercooling (t.T). For example,
nucleation controlled growth predicts a variation with t.T of the form exp {-(B/kt.T)}
where B is a constant, whereas growth on screw dislocations predicts a variation pro-
portional to (t.T)2. However, although some experimental data exhibits the correct
PREPARATION OF MATERIALS 45
dependence on t::.T for one or other of these models, there is still a lack of quantitative
agreement on the magnitude of the growth rate. Moreover this is also true for more
detailed models of growth which treat the interface between the growing crystal and
the surroundings in varying degrees of complexity and, as yet, there is no unified
quantitative theory of crystal growth.
Another important criterion in the case of crystal growth from the melt is the tem-
perature variation in the vicinity of the interface between solid and liquid; two possible
temperature distributions are illustrated in Fig. 3.4. If a projection is formed for any
Solid
Liquid
Interface Distance
(a)
Solid
Liquid
Interface Distance
(b)
reason on the interface during crystallization then, for the temperature distribution
shown in (a) the projection passes into a region of higher temperature and melts,
whereas in case (b) the growth of the projection is enhanced until the evolution of the
latent heat produces a temperature distribution similar to (a). In the latter case, there-
fore, a rapid growth of 'needles' or dendrites may occur in specific crystallographic
directions ahead of the main interface. Dendritic growth is generally a nuisance and is
apt to occur when the rate of cooling is excessive.
46 PHYSICAL PROPERTIES OF MATERIALS
This is by far the most common method of growing large single crystals of a wide range
of materials. Usually growth occurs from a high temperature melt but, before the
various experimental techniques are considered in detail, the growth of single crystals
of gallium will be discussed. Gallium is a most unusual metal in that it melts at 29·7°C
and therefore presents no difficulties as regards melting or containment of the melt,
but the growth technique is a simple example of the more complicated methods to
be described later.
Strain-free crystals can be grown very easily in sectioned perspex moulds or perspex
tubes containing a small seed crystal, maintained at 28° C, at one end of the mould
(Fig. 3.5). Clean liquid gallium at 35°C is poured into the mould and this melts the tip
of the seed crystal thus ensuring intimate contact between the seed and the growing
crystal; subsequent growth occurs at a rate of a few cm/hr. The use of sectioned moulds
enables crystals of predetermined shape to be grown, as for example a dumb-bell shaped
crystal for tensile testing as shown in Fig. 3.5; alternatively crystals up to several feet
in length can be grown in perspex tubing. Seed crystals nucleate readily in the super-
cooled liquid and the subsequent position of the seed crystal in the mould enables
crystals of any desired orientation to be grown, so that the anisotropy of the physical
properties can be studied.
This process of melting the charge and subsequent growth onto a seed crystal is
typical of all techniques of growth from the melt, but the high temperature melts
usually encountered produce many additional complications and a variety of different
techniques have been developed. These techniques fall into two general classes depend-
ing upon whether the complete charge is melted and contained in a crucible, as above
in the case of gallium, or whether only part of the charge is molten at anyone time and
this molten charge is not contained in a crucible.
,,~'I---_+-_Molten
charge
_ _ _ Furnace
windings
Direction
of
motion
Fig. 3.6 Stockbarger method.
The crystal-pulling technique overcomes this latter problem, and is the only way of
growing large perfect crystals from the melt, since the solid-liquid interface is no
longer in contact with the crucible. The essential components of a crystal puller are
(i) a heat source to melt the charge; (ii) a crucible to hold the melt; (iii) a seed crystal;
and (iv) a means of producing relative motion between the seed crystal and the melt.
These essential components are illustrated in Fig. 3.7. The seed holder or chuck is
attached to a shaft that can be raised, or lowered, and rotated. If necessary the entire
arrangement can be contained in a silica enclosure so that growth can take place in a
controlled atmosphere, this is essential if the melt is very volatile. Doped crystals can
be grown by incorporating the impurity in the melt and convection currents provide
adequate stirring to distribute the dopant in the melt. Pulling rates vary between 1 mm
hr- 1 to 2 cm hr- 1 and crystals several cm in diameter and more than 10 cm long have
been grown.
The most common form of heating in both these techniques is electrical since this
type of heating is easily controlled. It may be direct heating (1 2 R) in furnace windings
or r.f. induction heating in which either the crucible and/or the charge, or a graphite
susceptor absorbs the r.f. power. The choice of crucible material is of fundamental
importance in both these techniques for it must not only withstand high temperatures
but it must not react with the melt. In addition, to eliminate strain the coefficient of
expansion of the crucible should be less than that of the crystal. Typical crucible
materials are silica (maximum temperature -1200°C), graphite (2600°C), molybdenum
48 PHYSICAL PROPERTIES OF MATERIALS
Shaft---
Grown
- - - - t - - crystal
Melt Crucible
(2400° C) and platinum (1600° C). In general metallic crucibles are used for the growth
of non-metallic crystals and non-metallic crucibles for metallic crystals; the choice of
insulating or conducting crucibles may well determine the type of heating needed.
However satisfactory the choice of crucible material there will always be some strain
and/or risk of impurity being introduced in the crystal and this places some limitation
on these techniques.
Charge
Mesh - -
- - Hopper
Feed tube - - -
: .../ Flame
Gas ---.., 1.--:::::::--- G
- - _.... I - - - - as
- - - Seed crystal
Alumma
pedestal
_ _ Upper
chuck
Seed
crystal
Grown
- - - crystal
o o
Molten _--,-
0 __ o _ _ Induction
zone 0 o coil
o o
_ _ _ Feed
rod
lower _ _
chuck
drawn and the feed rod simultaneously raised. Thus the two chucks must be capable
of separate translational motion and rotation to provide adequate stirring of the melt.
Again after-heaters can be used to prevent excessive thermal strain in the growing
crystal. The use of r J. heating is almost universal for this technique but other heat
sources including arc-image furnaces using radiant heating from a tungsten filament
lamp or carbon arc have been used [5] . Plate 1 shows a molten zone in a sintered rod
of nickel oxide-the heat source in this case was a carbon arc. This latter heating
50 PHYSICAL PROPERTIES OF MATERIALS
liquid
I
I
I
I
I
Solid
I
I
I
I
I
o Co C,
Percentage impurity le%)
as the molten zone passes until the constant region is reached, as illustrated in Fig.
3.11. Of course the final region solidifies with the impurity concentration in the final
zone. This variation of impurity concentration in all but the final region can be rep-
resented by the following equation:
Cs =Co{1-(I -k) exp(-kx/l)}, (3.2)
where I is the length of the molten zone.
Co - - -- - -'-=-:;.;--....---------.-:.---- ---
kCo
Further purification results from repeated passes of the molten zone, as is illustrated
by the curves shown in Fig. 3.12, with subsequent discarding of the final region. In
practice this is carried out by using a multipass technique as illustrated for the case of
germanium contained in a graphite crucible in Fig. 3.13. Using this technique impurities
which can cause large changes in the electrical properties, i.e. elements from Group III
or Group V can be reduced to as little as 10- 8 per cent. In the case of silicon, zone
refining is equally valuable, but the high reactivity of molten silicon places greater
restrictions on the crucible material. Thin-walled quartz can be used but a valuable
52 PHYSICAL PROPERTIES OF MATERIALS
Cs __ -::::::=;::: nn == 21
~o n '" 3
n =4
n =5
~ 10 '
n =6
~c
CIl
(.)
g 10- 2
(.)
CIl
.~
Ri
Qj
a:
Fig.3.12 Curves illustrating the solute distribution for n passes (k =0'25) (after W. G. Pfann).
Molten
zones
o o o
Direction
of travel
•
o 0'-..,/0 o
Ge ingot r.f. heater
coils
Fig. 3.13 Multipass zone rerming of germanium.
alternative is to use the floating zone technique. Zone refining is not only limited to
the purification of semiconductor material but can also be used for metals although
the degree of improvement is much less and is typically only two orders of magnitude.
Despite the successes of the various melt techniques discussed in section 3.3 there are
a wide range of compounds which, either cannot be grown as single crystals at all by
these techniques, or cannot be grown to a sufficiently high degree of perfection. There-
fore considerable attention is being paid to several other growth techniques which
permit growth to occur at temperatures well below the normal melting point. These
are especially valid for many high melting point materials such as ferrites, which are
usually non-stoichiometric when prepared from the melt; i.e. the constituents will not
be in the exact proportions as indicated by the normal chemical formula, but will show
small deviations. They are also applicable to materials which dissociate below their
normal melting point and for materials which undergo phase transitions at elevated
temperatures. Progress with these techniques should improve on the perfection of crystals
grown by other ways and may lead to the production of completely new single crystal
materials with possible technological applications.
PREPARATION OF MATERIALS S3
3.4.1 Fluxed-melt technique
This is a solution growth process which is no different in principle from the growth of
common salt or copper sulphate crystals from aqueous solutions which may well have
been grown or observed by most readers. Relatively few materials of technological
interest are soluble in water. In the fluxed-melt technique, the solvent or flux is
generally a molten salt, such as lead fluoride or barium borate which melt in the range
600-1000°C. Typical materials which have been grown in this way are high melting
point ionic or covalent solids, for example yttrium iron garnet (YIG) from a lead
oxide-lead fluoride flux at 12S0°C, nickel Oxide, NiO, from lead fluoride at 1280°C
and magnesium aluminate spinel crystals (Mg Al2 0 4 ) from lead fluoride at 12S0°C.
The crucible material poses a critical problem in this technique because the high tem-
perature solutions may be highly reactive and platinum although very expensive is
commonly used. Crystal growth can be initiated by slowly decreasing the solution
temperature typically by 1-20°C/hr, or by slowly increasing the degree of super-
saturation leading to growth runs from one to three days. In general many crystallites
are nucleated and the resulting crystals are rather small, e.g. a few mm edge; however,
crystal pulling techniques are currently being applied to flux growth and much larger
crystals should result, although the pulling rates are usually very slow.
Oosely allied to this fluxed melt technique is the hydrothermal growth process [7]
which uses considerably lower melting point fluxes but, because these are relatively
volatile, elevated pressures are required to retain the flux as a solvent. Typical of this
latter process is the growth of quartz from aqueous solutions.
Unfortunately in both of these processes it is impossible at present to eliminate all
traces of solvent inclusion in the crystals and this imposes a rather severe limitation on
these techniques.
As with all non-melt techniques multiple nucleation is a major problem and the
individual crystal size is still quite small, although one possible solution is to periodically
reverse the temperature difference so that preferential evaporation occurs from the
smallest crystallites. Alternatively the material can be deposited onto a single crystal
substrate-see section 3.5. Rare earth compounds and complex niobates and tantalates
have been grown by this technique.
Concentrated
CaCI 2 solution Concentrated
tartaric acid
Crystal
growth
Gel containing Gel containing
tartaric acid weak tartaric acid
Crystal growth
Fig. 3.1S Gel growth.
Once again the individual crystals produced so far are very small, although calcium
tartrate crystals of I cm edge have been grown in this way and further work may lead
to the growth of considerably larger crystals. Examples of other crystals that have been
grown from similar low temperature gels include gold, lead iodide, PhI 2 , calcium tung-
state, CaW0 4 , and zinc tartrate, ZnC 4 ~ 0 6 •
In addition to the demand for bigger and better single crystals of a wide range of
materials the necessity for micro-miniaturized components in this modem space age
has led to a tremendous increase in the technological application of thin films. Thin
films can be deposited onto solid substrates by a variety of techniques including
PREPARATION OF MATERIALS 55
(a) thermal evaporation by resistive or electron bombardment heating, (b) sputtering
by means of a glow discharge or ion beams, (c) chemical deposition, and (d) electrolytic
deposition. Depending upon the structure and temperature of the substrate the
deposited layers may be single crystal, polycrystalline or amorphous and all three are
finding increased applications in technology.
The growth of thin single crystal fllms on single crystal substrates is referred to as
epitaxial growth [10]. The term epitaxy (arrangement on)-was first introduced by
Royer in 1928 to describe the orientated growth of one crystal upon another. The
earliest examples of epitaxy were observed naturally in minerals but subsequently very
many examples of epitaxy have been observed in the laboratory. In almost all of the
observed cases of epitaxy there is a match oflattice symmetry at the substrate-deposit
interface, although the differences in actual lattice spacings may be relatively large
(>20 per cent). During the last decade or so epitaxial growth has been an active field
of interest both for fundamental studies and device applications. In the former case
the main emphasis has been aimed at understanding the nucleation and subsequent
growth of the deposited fllm and in this connection metal fllms deposited on freshly
cleaved surfaces of rocksalt and mica in ultra high vacuum systems have been studied
extensively.
The early work on the growth of epitaxial fllms of silicon and germanium for the
electronics industry was usually confmed to growth on single crystal substrates of the
same material (auto-epitaxy). However, recent developments in integrated circuits
which require individual devices to be electrically isolated from one another have
necessitated the growth of silicon, germanium and gallium arsenide, for example, on
insulating substrates such as sapphire followed by the isolation of neighbouring regions
by photolithographic techniques.
As a typical example of epitaxial growth we will consider the growth of silicon by
the hydrogen reduction of silicon tetrachloride at elevated temperatures, typically
ll50-1250°C, which involves the follOwing reaction:
SiCl4 + 2H2 ~ Si + 4HCl. (3.4)
Dry hydrogen gas is bubbled through the silicon tetrachloride causing it to vaporize,
and the mixture of vapour plus hydrogen is fed into a horizontal silica furnace tube
and over a substrate, which is usually placed on a conducting heat susceptor of
molybdenum or silicon carbide coated graphite, and heated externally by rJ. induction
heating (Fig. 3.16). The substrate temperature is an important parameter because it
tSilica
~~~ Substrate
Controlled /~_ _ _-,oUQl...JQ:L..P.Q..II.Q..IIQ..II~..II..II~/::"-----===~
/c: :::.(.1
gas inlet - - ' /'
-~ --=-'j~
Susceptor to
r.f. induction absorb the r.f. Gas
coil
outlet
Fig. 3.16 Epitaxial growth of silicon.
governs the surface mobility of the atoms once they are adsorbed onto the surface
and they must be suffiCiently mobile to attain a stable configuration. Control of the
dopant level in the inflowing gas stream results in thin fllms of required electrical
56 PHYSICAL PROPERTIES OF MATERIALS
characteristics. Growth rates are typically a few Ilm/min. The quality of the substrate
surface has a direct effect upon the perfection of the epitaxial fllm and generally care-
fully prepared crystal slices from boules grown by the Czochralski or floating zone
techniques are used as substrates. In general the perfection of crystals grown on foreign
substrates is inferior to that obtained in the case of crystals grown on substrates of the
same material (see section 3.6).
In addition to the wide range of applications of thin fllm single crystals there are
many applications for thin polycrystalline fllrns. These can be deposited using the same
variety of techniques used for single crystal deposition, although there is no longer
any need for a single crystal substrate and glass, quartz or ceramic materials are often
satisfactory. Their uses cover a wide field ranging from the use of ferromagnetic fllms
as ~gnetic switches to optical fllms for mirrors and interferometers and to the
chemical deposition of a compound of niobium and tin, Nb 3 Sn, on foil for the windings
of superconducting magnets.
As was seen in section 3.2, dislocations may play an important role in crystal growth
and, in addition, their presence is essential in explaining the mechanical properties of
materials. Therefore dislocations must be inherently introduced into crystals during,
or immediately following, crystal growth. One obvious source of dislocations is by the
propagation of those dislocations already existing in the seed crystal, but there are
several other mechanisms that can lead to the formation of further dislocations.
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0.0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0-4 ... 0 0
0 0 0 0
t0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Fig. 3.17 Formation of a dislocation loop by construction of a vacancy disc.
In addition to the use of polycrystalline metals and alloys cast from the melt the
metallurgist has made an increasing use of non-crystalline material. If a metallic powder
is subjected to a sufficiently high pressure a degree of bonding takes place between the
particles even at room temperature and a coherent mass of material is produced. Heat-
ing, or sintering the compacted mass improves the coherence between the particles by
promoting intergranular grain growth and diffusion.
Originally powder metallurgy techniques were used to replace castings for those
metals which were difficult to melt industrially because of their high melting points;
for example, the sintered carbides of tungsten and titanium used in cutting tools and
58 PHYSICAL PROPERTIES OF MATERIALS
drawing dies. However, they have now been extended beyond the treatment of re-
fractory materials into many other branches of metallurgy. Practical alloys cannot be
produced from the liquid phase unless the metals concerned are mutually soluble in
the liquid state and the powder technique is of particular interest to the metallurgist
since it frees him from the limitations imposed by phase equilibrium; for instance it is
not easy to produce copper-lead bearing alloys by any other techniques. The addition
of non-metallic substances which would be insoluble in the liquid alloy and consequently
would not be incorporated into its structure, can also be achieved with the knowledge
that they will be evenly distributed throughout the material. Thus, graphite and copper
can be compacted for the manufacture of dynamo brushes, whilst small additions of
thorium oxide can be made to tungsten destined for the production of electric-lamp
filaments. The technique can even be extended to include the manufacture of alloys
which can also be produced by more orthodox methods. For example an alnico alloy
produced by powder metallurgy can be magnetically superior to one produced by
casting from the melt.
Closely associated with this technique of powder metallurgy is the production of
ceramics, which is the general name applied to a wide range of non-metallic compounds
such as silica, Si02 , and alumina, Al2 0 3 . Ceramics are usually hard, brittle, amorphous
or glassy solids which are produced by compacting and subsequent sintering. One of
their most important uses is as insulators, both thermal and electrical; ceramic di-
electrics, including glass, are generally stable at elevated temperatures and have
relatively high permittivities. Certain ceramic oxides are magnetic and, although they
are ferrimagnetic with generally lower saturation magnetization and relative permit-
tivity than metallic ferromagnetics, they have higher resistivities than metals or alloys
and hence have smaller energy losses due to Joule heating.
With the tremendous demands placed on materials recently by the technological
advances of supersonic flight and the exploration of space, the search for new materials
is being actively pursued and increasing use is being made of composite materials. The
principal attraction of composite materials is that they are lighter, stiffer and stronger
than more conventional materials. Considerable increase in strength can be achieved by
fibre reinforcement, i.e. embedding very strong fibres in a suitable matrix such as
polyester resin or a ductile metal, e.g. boron fibres embedded in aluminium. The matrix
itself is important because it must protect the fibre surface from damage due to
abrasion, it must separate the individual fibres so that cracks cannot run from one to
another, and finally it must be able to bond to the fibre surfaces so that the stress can
be transferred to the fibres. The strength of these composite materials tends to be
highly directional and usually they are less strong in compression or shear than in
tension.
Some everyday examples of composite materials are reinforced concrete and glass
fibre reinforced plastics which can be used for such diverse applications as kitchen
sinks, motor car bodies and motor tyres. The major attraction of glass fibre rein-
forcement is the ease with which we can obtain long strong glass fibres. These fibres
have immense strength when freshly drawn, comparable in fact with the strongest
steels, but the strength can only be sustained if the surface of the glass fibres is pro-
tected by some suitable matrix such as a plastic.
Although filaments of glass have proved to be strong as stiffeners, there are other
materials which are even stiffer, stronger and sometimes lighter. For example very
fine filaments have been drawn of such substances as tungsten, carbon, silicon carbide,
silica and boron, and the fibres of carbon and boron now in production are more than
twice as stiff as steel although less than a third as dense. Of all the components to
emerge in recent years the one which has attracted most public interest is the carbon
PREPARATION OF MATERIALS 59
fibre reinforced plastic which has been used with success injet engines. The starting
point for the carbon fibres is usually a man-made fibre; acrylics such as poly acrylonitrile
and Courtelle which can be converted by pyrolysis at about 3000°C into pure poly-
crystalline carbon up to six times as stiff as glass fibres. These fibres can then be
bonded with a suitable resin, for example an epoxy or a polyamide.
However, the ultimate in strengthening materials may finally be achieved by rein-
forcing with single crystal whiskers which approach more closely the theoretical
strength than any other form of the material. Considerable attention has been paid to
the manufacture of whiskers of silicon carbide, boron carbide and aluminium oxide
and test specimens of composites reinforced with these whiskers reveal quite remark-
able strengths.
Although the crystalline semiconductor device industry is very highly developed there
is reason to believe that amorphous devices will prove useful in the future. Currently,
considerable research effort is being directed towards the eventual commercial applica-
tion of amorphous semiconductors.
The term amorphous is sometimes used synonymously with glassy or vitreous states.
The latter are usually restricted to those materials obtained by supercooling a melt
while the term amorphous is reserved for fllms produced by vacuum evaporation,
sputtering, electrolytic deposition and related methods. Semiconducting glasses which
have received most attention are the chalcogenide glasses formed from compounds of
sulphur, selenium and tellurium with a variety of other elements such as phosphorus,
arsenic, antimony and bismuth. The transition metal oxide glasses are those in which
a major constituent is a transition metal oxide and of these the vanadium phosphates
012 0S-P2 Os) have been the most thoroughly investigated. In this system glasses with
up to 95 mol per cent of V2Os can be prepared by cooling from the melt. V2Os -Te02
and V20s-BaO have also been prepared. Ternary systems such as V20S-P20S-R20
(or RO) where R is an alkali or alkaline earth metal are also known. Many of the
chalcogenide glasses can be prepared by cooling from the melt although it may be
necessary to melt under vacuum or in an inert atmosphere and rapid cooling may be
required to inhibit crystallization.
For some glasses, for example AS 2Tea, conventional cooling from the melt is not
sufficiently rapid to prevent crystallization and they can only be prepared by evapora-
tion or similar techniques.
The glass-forming region for the ternary system Ge-Se-As is shown shaded in Fig.
3.18. Well inside this region the metal may be prepared by cooling the melt. Near the
boundary, splat cooling [11] may be used while outside it the amorphous phase can
be obtained by evaporation.
Splat cooling is used where conventional cooling from the melt is not sufficiently
rapid to prevent crystallization and with it cooling rates as high as 106 K S-l can be
achieved. In this technique a liquid droplet is driven at high velocity on to the spinning
surface of a cold metal such as copper. The droplet spreads out rapidly into a very thin
film about IJ,Lm thick in good thermal contact with the metal. An alternative is the
hammer and anvil method where two cold metal blocks are clapped together as the
liquid is dropped between them. Germanium and silicon can only be produced in the
amorphous state by deposition from the vapour. Amorphous Si and Ge fllms up to
0·2-0·3 J,Lm thick can readily be deposited by electron beam evaporation of the element
on to a cold substrate (less than about 670 K for Ge and 870 K for Si) [12]. They can
60 PHYSICAL PROPERTIES OF MATERIALS
Ge
Se As
Fig. 3.18 Ternary system Ge-Se-As showing glass-forming region (shaded).
High polymers, i.e. plastic or elastic (rubbers) materials consist of molecules of large
molecular weight ("'10 3 _10 6 ) and a large number of different polymeric materials
can now be synthesized by chemical reactions in which these large molecules are
built up from relatively simple substances oflow molecular weight, i.e. the monomer.
The production of these high polymers from the simple monomers is termed poly-
merization. The simplest type of large molecule consists of a linear chain in which the
repeat units in the molecule are connected together by covalent bonds. However,
branched chains are by no means uncommon (Fig. 3.19), although they are some-
thing of a nuisance in practice because there is little or no control over the regularity
of the branching. The process of polymerization can occur in the solid, liquid or
S'- S - S - S- -S-S-S-S"
3.9.1 Polymerization
H H H H H H H H H
I I I I I I I I I
c=c -c-c- - -_ .. -c-c-c-c-c-
I I I I I I I I I
H H H H H H H H H
Fig. 3.20 Formation of polyethylene. (a) Ethylene molecule with a double bond between the
carbon atoms. (b) Activated ethylene molecule. (c) Polyethylene.
there must be some agency, called an initiator, to initially break the carbon-carbon
double bond. Fortunately this is not as difficult as it may appear for the double bond
is in a highly strained state, since as we saw earlier, covalent bonds tend to separate
spatially from each other as much as possible. Typical initiators are free radicals or
charged ions which can be produced by thermal or photochemical decompOSition.
Once initiated the polymerization process advances rapidly, because each added mono-
mer produces yet another free radical, until the monomer molecules are gradually con-
sumed and then the probability of the chain terminating increases. This termination
may be caused by the joining of two chains or by the interaction between the end
free radical and another molecular species which does not produce another free radical
to continue the reaction.
Starting with a different monomer other types of polymer can be produced in which
the 'backbone' is still a linear array of carbon atoms but the side groups are not just
hydrogen as in the above example of polyethylene. Other common examples of addition
polymers include:
(i) Polypropylene
HHHHHH
I I I I I I
--C-C-C-C-C-C-
hCHI hCHI ACHI
3 3 3
(iii) Polystyrene
nnvY
-91-Y-I-Y-r-
HUHCJ H
Where 6 represents the benzene ring
H
I
H'C~'C/H
I 11
H.....c::::::C/C'H
I
H
The regularity and complexity of these side groups considerably affects the physical
properties of the resulting polymer, because, if the side groups are large as in the case
of polystyrene then neither molecular packing or relative molecular movement will
be very good (see subsection 3.9.2). Polymerization from a mixture of two monomers
is also possible and this leads to a class of polymers known as copolymers.
external forces which would keep the molecule extended in this way, the large free-
dom of rotation about the single bonds in the chain means that there is a high degree
of flexibility and so extensive bending and twisting of the chain results. Thus, the
general configuration of a linear polymer molecule tends to be that of an irregular
coil.
These long chain molecules are generally mixed together in a solid polymer in a
random manner just like a tangled bundle of coiled fibres. Whereas the molecules in
any individual chains are held together by strong covalent bonds, the bonds between
the various molecules are only weak secondary bonds commonly of the van der Waals
type. The basic structure, therefore, will be amorphous. Occasionally there will be
some regions in the material where the linear chains are aligned close enough to one
another so that the weak intermolecular forces are sufficient to keep them parallel.
Over these regions the material is said to be crystalline. These crystalline regions are
usually quite small and in fact one individual chain may form a crystalline region on
its own due to a regular arrangement of the folds. The random length of the individual
chains means that polymers are never completely crystalline and one individual molecule
may pass through several crystallites. These various polymer structures are illustrated
in Fig. 3.22. The crystallinity of a polymer has a considerable influence on its physical
Fig. 3.22 Polymer structures. (a) Amorphous. (b) Partly crystalline (different molecules). (c)
Partly crystalline due to the regular orientation of the coils of a single molecule.
chain molecules are less likely to form crystalline regions because the presence of the
branches prevents close enough approach of the separate molecules for the weak inter-
molecular forces to orientate them. This is also true for those linear molecules which
have excessively large side groups randomly arranged along the chain.
In addition to the polymer structures discussed above, separate linear or branched
chain molecules may be joined together at random points along the chains by primary
covalent bonds. This process is called cross-linking and results in a more rigid structure
as it tends to prevent slipping between the individual chains. A common example of
cross-linking occurs in the vulcanization of rubber in which the raw material is heated
in the presence of 1-2 per cent by weight of sulphur resulting in a random array of
sulphur cross linkages. If the percentage of sulphur is increased to about 30 per cent
of the rigid material, ebonite is produced. Cross-linking is one of the major effects
produced by irradiating polymers with high energy radiation such as'Y rays. In addition
it may be caused accidentally by oxygen or sulphur atoms catalysed by light, leading
to a decrease in the article's elastic properties.
References
4.1 Introduction
Although optical microscopy was an invaluable tool in the early days of crystallography,
the information that it revealed was rather limited. Measurement of the angles between
the faces of well defmed naturally occurring mineral crystals enabled the overall sym-
metry of the crystal structure to be inferred. However, the large wavelength of visible
radiation meant that the resolution obtainable was much too poor to resolve detail on
an atomic scale, and no informaton as to the size of the unit cell and the actual positions
of the atoms could be deduced. Evaluation of this detailed structure had to await the
advent of X-rays during the first few years of this century. X-rays are electromagnetic
waves like visible light but with a much smaller wavelength. This wavelength is typically
10- 10 m and is of the same order of magnitude as the interatomic spacings in crystals,
and so the network of atoms in solids acts like a naturally occurring three-dimensional
diffraction grating. The techniques of X-ray crystallography have enabled the actual
atomic positions in a vast number of simple and complex structures to be determined
and, in addition, X-ray diffraction has provided valuable information on many other
aspects of structure including preferred orientation in polycrystalline sheets or wires,
and crystal imperfections such as grain boundaries and dislocations.
Initially there was no practical alternative to X-rays for investigating crystal struc-
tures, but following the de Broglie hypothesis of 1924 that moving particles possess
wave properties (eqn. (1.1)), both electrons and neutrons were shown to be diffracted
by crystals. Electron and neutron diffraction in no way replace the much more versatile
X-ray diffraction, but under certain circumstances they can provide complementary
information. All three techniques will be discussed in more detail in the subsequent
sections of this chapter.
Before discussing the experimental techniques used in X-ray crystallography the theo-
retical aspects of the diffraction of X-rays by crystals will be considered briefly. When
X-rays strike an electron the varying electric and magnetic fields associated with the
radiation force the electrons to vibrate at the same frequency as the incident X-rays
and the vibrating electrons in turn scatter secondary radiation of the same wavelength
in all directions. All the electrons in an atom contribute to the scattering of X-rays so
that the higher the atomic number of an atom the more effective it is as a scatterer
of X-rays.
66 PHYSICAL PROPERTIES OF MATERIALS
Incident 'Reflected'
b..m~ ~b.. m
9~9 e
Fig. 4.1 Reflection by a single planar array of atoms.
Consider a simple three-dimensional lattice, as shown in Fig. 4.2, with the incident
parallel beam of X-rays of wavelength, A, making an angle 9 with a set of crystal planes
of interplanar spacing d. Assuming that the X-rays are not refracted on entering the
crystal (they are, in fact, refracted to a very slight extent but this only produces changes
of the order of 1 in 104 ) then each plane of atoms alone will tend to reflect a beam
Incident Diffracted
wave wave
tf,~t
e 9 e e e e
:
e
Fig. 4.2 Explanation of Bragg's law.
of X-rays in a direction making an angle 9 with these planes. However, these beams
may, due to their relative path differences, either interfere destructively or construc-
tively. The criterion for the existence of constructive interference and an intense
diffracted beam is that the reflected rays from the individual planes should be in phase
across the entire wavefront. For this to be so, the path difference between two succes-
sive rays must be equal to an integral number of wavelengths, i.e.
d sin 9 + d sin 9 = nA, where n is an integer
or
2d sin 9 =nA. (4.1)
This result is known as Bragg's law and forms the basis of modern X-ray crystallography.
Although it is common to find the Bragg law stated in this form, where n is the
order of reflection, or diffraction, it is not common usage in X-ray crystallography. If
diffraction occurs from the (100) planes of a lattice for example at an angle 9 1 , then
PRAcrICAL DETERMINATION OF STRUCTURE 67
when the path difference between the beams diffracted from adjacent planes is 2A,
instead of A, diffraction will occur at an angle 8 2 and the above equation will be
satisfied with n =2. These reflections are then called respectively the first and second
order reflections of (100). This terminology is a direct transfer from the science of
optical spectroscopy but it has little to recommend it in X-ray crystallography since
the second order reflection from (100) is equivalent to the first order reflection from
the (200) planes (Le. with d halved) and these planes are as uniquely defined as any
others (Chapter 2). Thus in practical X-ray crystallography the above law is of more
use in the form
2d(hkt) sin 8 =A, (4.2)
and each angle 8 is associated with a specific set of lattice planes (hkl). Thus, if A is
known (it can be measured directly using special ruled gratings) and the various values
of sin 8 can be measured, a set of values of d hk1 can be calculated. The interplanar
spacings d of any set of planes depends only on the type and size of the unit cell and the
Miller indices of the plane, for example:
h2 + k2 + [2
for a cubic crystal (4.3)
dlhkt) = a2
and
1 4(h 2 + hk + k 2 ) f
for a hexagonal crystal -2d = 32 +2", (4.4)
(hkt) a c
where a and c are respective lattice parameters. Combining eqns (4.2) and (4.3) it
follows that, in the case of a cubic crystal,
(4.5)
where A =A2 /40 2 =constant. For a cubic crystal the values of sin 2 8 will form a simple
series consisting of the constant A multiplied by a variable integer (h 2 + k 2 + f). The
Miller indices of each reflection, and the lattice constant can then be determined quite
easily. In the case of a hexagonal crystal the sin 2 9 values will follow a more complex
form, since two constants are involved, but it is not too difficult to manipulate these
figures to determine both the indices of the reflection and the unit cell parameters. The
same principle holds, of course, for all seven types of crystal system but as the sym-
metry of the unit cell decreases so the complexity of the problem increases.
The directions of the diffracted beams given by Bragg's law are governed entirely by
the geometry and size of the unit cell, and, therefore, these directions are not affected
by the arrangement of the atoms associated with each lattice point. However, the com-
plexity of the atomic structure within the unit cell does affect the intensities of the
diffracted beams. The diffracted beams are built up by a combination of the scattered
waves from the electrons of all the atoms in the unit cell and this involves two distinct
contributions:
(1) scattering from electrons in the same atom (the atomic scattering factor,{o), and
(2) the summation of this scattering from all the atoms in the unit cell (the structure
scattering factor, F).
If the atoms were merely points, their atomic scattering factors (the ratio of the ampli-
tude scattered by an atom to that scattered by an electron) would be equal to their
atomic numbers. However, atoms have a fmite size which is of the same order of
magnitude as the X-ray wavelengths and this causes phase differences between the waves
68 PHYSICAL PROPERTIES OF MATERIALS
scattered from the various electrons. Therefore, the atomic scattering factor varies with
(sin ())/'A as illustrated in Fig. 4.3. Only at small values of (sin ())/'A does it approach the
atomic number.
fo
o~------------~----------~
(sin 8"!tA
Consider now the total intensity of a diffracted beam that results from the combina-
tion of the waves scattered by all the atoms in the unit cell; e.g. the (100) reflections from
a b.c.c. crystal. When the diffracted beams from successive (100) planes differ in path
length by one wavelength they will constructively interfere, but the atoms at the centres
of the unit cells will give rise to a scattered beam in the same direction which differs in
path length from these two by exactly X/2 and thus interferes destructively. In this
simple example assume that all the atoms are of the same type so that their atomic
scattering factors are equal, and furthermore, the concentrations of the atoms in the
respective planes are equal so that complete cancellation occurs and the (100)
reflections will not be observed.
In general to calculate quantitatively the total amplitude of a given reflection it is
necessary to sum the waves that arise from all the atoms in the unit cell and mathe-
matically this involves adding waves of the same wavelength but different amplitudes
and phases; the square of the resultant amplitude will be the required intensity. If the
positions of the atoms in the unit cell are denoted by x, y and z, where these coordinates
are expressed as fractions of the unit cell parameters, the resultant amplitude for a
given (hkl) reflection can be shown to be given by
n
Fhkl = '2./; exp {2nj(hx; + ky; + lz;) (4.6)
1
or
n n
Fhkl = '2./; cos 2n (hx; + ky; + lzi) +F'ili sin 2n(hx; + kYj+ lzj) (4.7)
1 1
and the intensity of the reflected beam is proportional to F~kl' For the b.c.c. crystal
considered above, having identical atoms at points (000) and (H!) in the unit cell,
eqn. (4.7) reduces to
I ex F~kl ex 12 [{I + cos n(h + k + l)}2 + sin2 n(h + k + 1)] . (4.8)
PRACTICAL DETERMINATION OF STRUCTURE 69
It can be seen from this equation that f = 0 for every reflection having (h + k + l) an
odd number. Thus the (100) reflection is not observed as was seen earlier. This condi-
tion holds for any body centred structure. A similar analysis in the case of a simple
f.c.c. crystal with atoms at positions (000), (BO), (4 O!)and (OB), reveals that the
intensity is zero for all reflections in which the integers (hkl) are partly even and
partly odd. Because of the systematic absence of particular reflections, the observed
series of (hkl) values obtained from the measured values of sin 2 0 enables the basic
structure to be deduced.
For more complex unit cells containing atoms of at least twO different types the
intensities may not completely vanish because of the difference in the atomic scattering
factors. For example in the CsCI structure where a caesium atom is at (000) and a
chlorine atom at (B!) quite a different result is obtained than that for the simple b.c.c.
structure discussed above. In this case eqn (4.7) becomes:
fa. {fcs+fcICOS1T(h+k+l)}2 + {fClsin1T(h+k+l)}2 (4.9)
which reduces to
f 0: (fcs+ fCl)2, when (h + k + l) is even
and
1.0: (fcs- fcD2, when (h + k + l) is odd.
In general the intensities of the various reflections are very susceptible to slight changes
in the coordinates of the atoms in the unit cell and a correlation between the measured
and calculated intensities enables the positions of the atoms in the unit cell to be deter-
mined precisely.
Thus, to determine completely the crystal structure of a material the various diffrac-
tion images must be recorded in such a way that both the angles of diffraction, 0, and
the intensities of the images can be measured. The most usual way of doing this is to
record the diffraction pattern photographically and measure the intensities either
visually or with some form of densitometer. However, for very large complex unit cells,
where there may be as many as 10 000 diffraction images, the information can only be
treated by recording the positions and intensities of the diffraction images automatically
using a scintillation counter or a proportional counter as detector.
In order that Bragg's law can be satisfied and X-ray diffraction can occur the various
values of 0 and Amust be related. It is necessary in practice to vary either the angle
of inclination of the specimen to the beam or the wavelength of the radiation. The
three standard methods of X-ray crystallography allow for this in the following ways:
(a) in the Laue technique a stationary single crystal is irradiated by a range of
X-ray wavelengths;
(b) in the rotating crystal technique a single crystal specimen is rotated in a beam
of monochromatic X-rays; and
(c) in the powder technique a polycrystalline powder specimen is kept stationary
in a beam of monochromatic radiation.
Each of these three techniques will be discussed in more detail after a brief considera-
tion of the nature of the X-rays emitted from a standard high voltage X-ray tube. The
X-rays are not all of the same wavelength. The X-ray spectrum consists of two com-
70 PHYSICAL PROPERTIES OF MATERIALS
ponents: (i) the continuous or white background radiation and (ii) the superimposed
high intensity peaks, as illustrated in Fig. 4.4. The most important peaks are usually
the Ka. and K(J peaks shown in Fig. 4.4. and, as can be seen, the Ka. peak is strictly a
close doublet of which the Ka.l is about twice as intense as the Ka.2 peak.
For the Laue technique the whole range of wavelengths is used but for the other
two techniques a single wavelength is required. This can be satisfactorily achieved by
passing the X-ray beam through a suitable thin metallic filter which preferentially
absorbs the K(J radiation and some of the white background. For example, copper
radiation is commonly used in X-ray crystallography and a sheet of nickel 0·002 cm
thick will absorb about 99 per cent of the CuK(J but only 50 per cent of the CuKa.. If
it is essential to remove all the white radiation the X-ray beam must be reflected from
a given set of lattice planes of a large single crystal set at the correct angle for diffraction
of the Ka. radiation.
White'
X-ray
beam
~-G::=*:::::::>
/
Collimator SinglLcrystal
/
Screen for back-reflection
/
pattern Screen for transmission
pattern-showing
a typical pattern
Fig. 4.5 Schematic representation of the Laue technique.
this axis. This feature makes the Laue technique particularly convenient for checking
the orientation of crystals. However, the method has fundamental disadvantages for
crystal structure determination because the wide range of wavelengths employed leads
to overlapping of diffraction images from different planes of atoms, e.g. Al from (001)
and Al /2 from (002), and the unknown wavelengths producing the various reflections
makes indexing the reflections almost impossible. Furthermore the variation of intensity
with wavelength in the incident beam makes useful intensity measurements impossible.
Photographic
film
Monochromatic Single
crystal
X-ray beam
- - t - Goniometer
~-tt::::=--- Rotating
spindle
axis unless
c cos tPn = n"A, (4.10)
where n is an integer (Fig. 4.7a). The diffracted beams will therefore be along the
surfaces of a family of cones whose vertices are at the crystal, and whose semi-vertical
angles are given by the above equation (Fig. 4.7b).
(a) (b)
Fig. 4.7 Illustration of the formation of the diffraction pattern in the rotating crystal technique.
(a) Diffraction condition. (b) Cones of diffraction.
Of course, the diffracted beams will only occur along those specific directions lying
on the cones for which the correct phase relationship also holds for planes parallel to
the other two coordinate axes. When the ftlm is flattened out after development these
diffraction images will lie on a series of lines called layer lines, as illustrated in Fig. 4.8.
LaVe\'ine
•• . ., .. • \/ .. •• • •• l=2
•• •• •••
,• • • • • • •
\
,•, • • • • • •
I
--...-.--....--..--<>-----..... -...._-- l=O
•• •• •••
,
I
l: -1
I
~.~.~~.~.~~.~.__~.______~,__~.~.__~.~.__~.~. l=-2
All the images on the zero layer line come from planes parallel to the axis of rotation,
i.e. planes with 1=0 in this example, and the other layer lines arise from planes with
1= ± 1, ±2, ... , etc. Diffraction images from planes with the same values of h and k
PRACTICAL DETERMINATION OF STRUCTURE 73
but different values of 1, all lie on one of a series of curves known as row lines which
are transverse to the layer lines and in the particular case when the a and b axes are
perpendicular to c they intersect the zero layer line at right angles.
Measurement of the separation, Sn, of these layer lines provides a rapid, 1 per cent
accurate, measurement of the c parameter of the unit cell, since from the geometry
of the diffraction, if R is the radius of the camera (Fig. 4.7b),
Sn =R cot cf>n (4.11)
and elimination of cf>n between eqns (4.1 0) and (4.11) leads to
c = (nX/Sn)(R2 + S/)1I2. (4.12)
By subsequently orientating the crystal with the a and b axes parallel to the axis of
rotation, the other unit cell parameters may be qUickly determined.
The values of 8 for each of the reflections can be determined fairly easily from their
positions on the film and so theoretically indexing the reflections should be quite
straightforward. In practice, however, complications may arise because the large number
of possible reflections leads to considerable overlapping of images and this causes un-
certainties in both the indices and intensity of the reflections. Therefore, for a com-
plete structure determination this problem of overlapping reflections must be overcome
and in practice this is done in one of two ways; either
(a) the crystal is oscillated through a small angle of about 10° rather than rotated
through 2rr (however, this cuts down the information available on any single
photograph and so necessitates the taking of many exposures); or
(b) some form of moving film technique can be used, as for example in the
Weissenberg technique. In this method the cylindrical film is moved backwards
and forwards parallel to the axis of rotation in exact synchronism with the rota-
tion of the crystal. Furthermore there is a screen with a narrow aperture between
the crystal and the film so that only reflections from a single layer line are
recorded in anyone exposure and, because of the movement of the film, these
images are spread out over the whole of the film.
The Bragg angles 8 of the various reflections can be calculated by measuring the
separation of the pairs of lines since, from the geometry of Fig. 4.9b,
(21T - 48)/21T = OP/21TR (4.13)
and so the reflections can be indexed and the unit cell parameters evaluated. Since the
diffraction pattern is different for all elements and compounds, this powder technique
provides a fairly rapid method of analysis. In addition it is a useful metallurgical
technique and such things as phase changes can be studied by varying the specimen
temperature and noting the associated changes in the diffraction pattern.
Diffraction
. / cone
I
. COII'~ Beom
r:,c~~~nt. stop
Specimen
(a)
A P o 8 -Camera.
/1 I II ~
I
I I I I I radius R
Collimator hole
(c)
TABLE 4.1
Photographic
plate
I
C::;::::====:::;:::J Diffracted
/ beam
Incident
beam
~
Slit
Diffracting
planes
Fig.4.10 Experimental arrangement for X-ray topography.
As was seen in section 4.1 the relationship between the wavelength and momentum
of a neutron is A= h/mu and if the neutrons are to be diffracted by the three-dimensional
gratings occurring in crystals this wavelength must be comparable to the interatomic
spacing, about 10-10 m. The only practical source of neutrons with a suffiCiently large
neutron flux to give measurable reflections is a nuclear reactor in which the neutron
flux is typically 10 18 m- 2 S-1 and even this figure is considerably lower than the photon
PRACTICAL DETERMINATION OF STRUCTURE 77
flux from a normal X-ray tube. For those neutrons which have suffered enough colli-
sions with a graphite moderator to come into thermal equilibrium with the carbon
atoms at a temperature T the most probable energy is given by
!mu 2 = (3/2)kT (4.14)
or
A2 = h2/3mkT, (4.15)
where k is Boltzmann's constant. This gives values of 1·55 x 10- 10 m and 1·33 x 10- 10 m
at oOe and lOOoe respectively, which are just the order of magnitude that is required.
However, not all the neutrons will have this velocity and so there will be a spread of
wavelengths which follows a Maxwellian distribution as shown in Fig. 4.11. A mono-
chromatic beam, which is essential for quantitative diffraction studies, can be ob-
tained by reflecting a collimated beam of neutrons from a given set oflattice planes
of a large single crystal specimen. It is arranged that the wavelength selected, A, is just
below the maximum and the intensity of the second order reflection (A/2) is then only
a few per cent of the first order reflection.
~
"iii
c:
....c:
G)
o
Wavelength xl 0- 10 m
Fig. 4.11 Wavelength distribution associated with thermal neutrons.
Collimator
Crystal -:1--+----.t.~
monochromator
Rotatable
specimen tablee---
To BF3 counter
Fig.4.12 Typical experimental arrangement for neutron diffraction.
However, it does have several important advantages which stem from the origin of the
scattering of the neutrons. In the case of X-ray scattering it is the electrons of the
atoms that are responsible but in the case of neutrons it is the nucleus of the atom
which causes the scattering. Since the nucleus acts as a point source the atomic
scattering factor,/o, does not vary with (sin 8)/'A and in addition/o does not vary in
any regular way with the atomic number of the specimen. This has two important
consequences:
(0 Ught atoms such as hydrogen and carbon are more effective as scatterers of
neutrons than heavy elements. Thus, in the case of light elements in the pre-
sence of heavy ones, as for example in hydrides or carbides of heavy elements,
neutron diffraction enables the positions of the light elements to be fIxed.
X-ray diffraction can only provide information on the arrangement of the
heavier element because the X-ray patterns are almost entirely caused by
scattering from these elements.
(li) Atoms whicll. are close to one another in the Periodic Table, such as iron and
cobalt, are virtually indistinguishable using X-rays, but have signifIcantly dif-
ferent atomic scattering factors for neutrons. Neutron diffraction is then a very
important tool in the investigation of certain alloys.
Furthermore, in the case of those atoms with a permanent magnetic dipole moment
there is an additional scattering contribution due to the interaction between this
magnetic moment with the small magnetic moment of the neutron. In paramagnetic
materials the scattering from the randomly orientated dipoles simply leads to a diffuse
background, but in ordered magnetic structures such as ferromagnetics, antiferro-
magnetics (see Chapter 9) co-operative diffraction occurs and additional information
on the magntiudes and directions of the magnetic dipoles can be obtained.
Electrons, unlike neutrons, can be accelerated by electric fIelds so that their velocity
and hence effective wavelength can be altered at will. For the non-relativistic case,
PRACTICAL DETERMINATION OF STRUCTURE 79
where the electron velocity is much less than the velocity of light, we have
Ve = !mu 2 and "11.= h/mu
so that
h (150)112
"11.= (2 Vem) 112 ,..... V x 10- 10 m. (4.16)
Electron gun
Anodes
150kV
To vacuum
pump
""
Magnetic lens
to improve
focussing
Thin
specimen
Viewing_
port
Fluorescent
screen
I:::===~ or photographic
::::.-- film
Fig.4.13 Electron Diffraction Apparatus.
atoms cannot be resolved but many otherwise 'invisible' features can be observed, for
example large molecules, minute particles, dislocations and other crystal defects, and
the electron microscope has proved of immense value in all branches of science and
especially to metallurgists and biologists.
Studies of the structure of glasses and amorphous materials have used X-ray, electron
and neutron diffraction techniques. The distinction drawn in section 3.8 between the
two types of material based on the method of preparation is not reflected in differences
in microscopic structure and from this point of view they can be considered to be
identical. However, the crystalline state is always the most stable and amorphous
materials exhibit a greater deviation from the stable equilibrium state than glasses
and generally contain a much higher proportion of defects. Both types will here be
called amorphous.
The essential difference between the amorphous and crystalline states is that in the
former the equilibrium positions of the atoms do not exhibit translational symmetry.
The spatial pattern is aperiodic. In a crystal, interatomic distances are fixed but in an
amorphous material there is short range order but no long range order. The interatomic
spacings of nearest neighbours (nn), next nearest neighbours (nnn) and so on now only
occur with a certain probability. Such a structure is described by a radial distribution
function F
F = 41Tr2 per),
where per) is the atomic or electronic density at a distance r from an arbitrary atom as
centre (r = 0). It is assumed that the density is uniform over a sphere of given radius in
PRACTICAL DETERMINATION OF STRUCTURE 81
order to analyse diffraction patterns but this assumption is not strictly justified. The
radial distribution function is used because diffraction techniques sample a volume of
material which is much greater than that in which local order exists and so give informa-
tion as a probability distribution.
T 1 I I I
•:
:
'J
I
;,
c::
o
,
II
-
'';::;
u
, ,
c::
::J
II
,
c::
o
'';::;
::J
:g....
.~
'0
,.
.
.!!!
.,
, ~!
'0
c: 1
I
II
I
I.
,
,
I I
I
II
'l J
~,
Ii
~b
"...
o 1.0 2.0 3.0 o 4.0 5.0
Radial distance r (A)
Fig.4.14 Radial distribution function of silicon. (a) Amorphous. (b) Crystalline.
The contrast between the crystalline and amorphous states is illustrated in Fig. 4.14
which shows the radial distribution function in each case for silicon.
The first peak is almost the same for both, indicating that the crystalline nearest
neighbour interatomic distance 2·35 A is maintained. The area under the peak, if it is
well separated from the rest, represents the number of nns (four) suggesting that the
82 PHYSICAL PROPERTIES OF MATERIALS
local tetrahedral structure is virtually unchanged in the amorphous state. The second
peaks are also coincident at 3'86 A. A broadening of this amorphous peak is interpreted
as distortion of the bond angle by as much as ±20° from the tetrahedral value of
109°28'. The area under the second peak indicates the crystalline number of nnns
(twelve). The striking difference is the absence of an amorphous peak corresponding
to third nearest neighbours.
Various models have been proposed for amorphous structures. The most successful
one for the amorphous forms of germanium and silicon has been a random network
model [7] in which each atom is bonded to satisfy its normal valence without dangling
bonds but with some distortion of bond angles. Neighbouring tetrahedral units are
rotated relative to each other about a common bond. Enlarging this structure leads to
a random network containing five- and six-membered rings. The presence of the latter
tends to suppress the third neighbour peak because the occurrence of atomic separations
at the third neighbour distance is reduced.
Of the chalcogenide glasses, AS 2 S3 is a monoclinic crystal with a layer structure.
Each As atom has as nn three S atoms and each S atom has two As atoms as nn. In
the amorphous phase the nn and nnn coordinates are increased and it seems that the
layers become wavy and connected at some points. Fig. 4.15 (a) represents the crystal-
line arrangements of AS 2 S3 and (b) the amorphous structure on a random network
model.
(a)
(b)
Fig.4.15 Two-dimensional representation of (a) crystalline AS2S3. and (b) amorphous AS2S3
on a random network model.
TABLE 4.2
Classification of amorphous semiconductors by structure with comparison of number of nearest neighbours and nn distance for amorphous and crystalline states_
(a:xb] means that the average number ofnn of atom a of type b is x (from ref (8])_
'"C
Tetrahedral Elements Ge, Si 4 4 (2-47-2_54)(2-37-2-41) 2-43,2·35 ::0
AIVB IV SiC 4 Si: 4C; C: 4Si >
()
o-j
AIIIB V GaSb 4 Ga: 4Sb: Sb: 4Ga 2-65 2-638
AIlIB VI InSe 4 In: 3Se; Se: 3In: In-Se: 2-60 In-Se: 2-50
n
>
In: lin In-In: 3-15 In-In: 3·16 t""
A~IIB¥I In: 3·46 Se; Se: 3·46 In (Ga, In): 4(Se, Te) 2·58 in In2Se3 t:I
(Ga, Inh(Se, Teh 2·57-2-60 in In2Se3 tT:I
In: 0·95 In Se, Te: 2-66(Ga, In) o-j
tT:I
In: (3-91-4·01 Se) ::0
Se: (2·59-2·67 In) is:
AIIBIVd' CdGe(P, Ash (P, As): 2Cd, 2Ge (P, As): 2Cd, 2Ge 2-50 in CdGeP 2 Cd-P: 2-55 in CdGeP 2 ~
Ge-P: 2·33 o-j
A1II B~V N 4 Si3 Si: 4N Si: 4N -1·7 1-7-1·8 0
Z
0
'Tl
Layer Elements C, As, Sb 3-2-3-4; 3; 3 (diamond 4, graphite -I-50, 2-40, 2·87 0·54,1"41),2-51, til
3)3,3 2·87 o-j
A2 VB¥I ::0
(As, Sb, Bi)liS, Scv (As, Sb, Bi) : (2-9-3·6)(S, Se, Te) (As, Sb, Bi) : 3(S, Se, Te) - c:
Teh + (A I, BI CVIl) (S, Se, Te): (2-0-2-4)(As, Sb, Bi) (S, Se, Te): 2(As, Sb, ()
o-j
Bi) c:
::0
tT:I
Chains and Elements Se, Te 2, - 2,2 2-31-2-34, - 2·32, -
rings
Selenium and tellurium form a third group, trigonal in the crystalline state with a
structure based on helical chains. Selenium also occurs in a structure based on eight-
membered rings. The radial distribution function obtained for amorphous elenium
depends on the method of preparation but explanations have been given in terms of
both six- and eight-ring models.
Table 4.2 classifies amorphous semiconductors giving the number and separation
of nn in both crystalline and amorphous states.
In addition to these three fundamental diffraction techniques for studying the struc-
ture of materials there are many other techniques that can be employed and a few of
them are discussed briefly.
silicon but not by the copper solute atoms. Similarly the dislocations in transparent
AgBr crystals can be decorated by causing print-out of the silver. However, in general,
this technique is rather limited in its range of application.
References
Questions
1. Calculate the Bragg angles for Mo-Ka radiation (X = 71·2 X 10- 12 m) diffracted
from the (110) and (200) planes of tungsten, given that tungsten is b.c.c. with a lattice
parameter of 316 x 10- 12 m. Would you expect a diffracted beam from the (100) or
(111) planes?
2. A powder photograph of a certain metal, taken with radiation of wavelength
154 x 10- 12 m, exhibits diffraction lines corresponding to the following Bragg angles:
20°14',29°23',36°50',43°58',50°45',58°44', 66°35', 78°34'. Assuming that the
metal is cubic, determine whether the lattice is f.c.c. or b.c.c. and calculate the unit cell
parameter. By comparing this value with tabulated values for metals try to identify the
metal used.
3. A specimen of powdered aluminium at a temperature of 293 K gave a diffracted
X-ray beam at an angle of 138° to the direction of the incident beam. When the tem-
perature of the specimen was raised to 593 K the angle between the diffracted beam
(from the same set of planes) and the incident beam changed by 2°10'. Calculate the
mean coefficient of expansion of aluminium between 293 and 593 K.
4. Calculate the Bragg angle at which electrons accelerated from rest through a .
potential difference of 80 V will be diffracted from the (111) planes of an Lc.c.
crystal of lattice parameter 350 x 10- 12 m.
S. A beam of thermal neutrons emitted from the opening in a reactor is diffracted by
the (111) planes of a nickel crystal at an angle of 28° 30'. Calculate the effective tem-
perature of the neutrons given that nickel is Lc.c. with a lattice parameter of
352 x 10- 12 m.
6. A transmission Laue photograph is taken of a b.c.c. crystal of lattice parameter
286 x 10- 12 m. Calculate the minimum distance from the centre of the pattern to the
diffraction images if the specimen is 60 mm from the mm and the X-ray tube voltage
is 35 kV.
5 Mechanical Properties
of Materials
5.1 Introduction
The mechanical properties of materials, their strength, rigidity and ductility, are of
vital importance in determining their fabrication and possible practical applications.
Materials exhibit a wide range of mechanical properties ranging, for example, from the
hardness of diamond to the ductility of pure copper and the amazing elastic behaviour
of rubber. Furthermore many materials behave quite differently when stressed in
different ways; for example cast iron, cement and masonry are much stronger in com-
pression than in tension, a fact that was not lost on early architects who avoided using
materials in tension whenever possible, whereas wood is several times stronger in
tension than it is in compression because the cell walls in the wood structure tend to
buckle in compression. In the subsequent sections of this chapter the various mech-
anical properties of materials are discussed and an attempt is made to explain the
behaviour of materials under stress. Initially, however, a brief review is given of some
of the experimental techniques that are used to investigate these mechanical properties.
The most common mechanical test is the tensile test in which a specimen is stretched
by an increasing load and the resulting stress-strain graph produced. The results of
such a tensile test can be used to introduce most of the terminology to be met later
on. Fig. 5.1 shows a typical nominal (or engineering) stress-strain curve for a metal
such as copper. The nominal stress is defined as the load divided by the original cross
section area and the nominal strain as the extension divided by the original length.
Initially as the stress is increased the strain increases uniformly (over the region
OX) and the deformation produced is completely reversible. Over this elastic
region the strain is extremely small and Hooke's law is obeyed, i.e. the stress
is proportional to the strain. Beyond a limiting stress X, called the yield stress, the
strain increases much more rapidly with increasing stress and the deformation is no
longer reversible; this is called the plastic region. In some materials, such as low carbon
steels the yield point is more clearly defined than as shown in Fig. 5.1 and typified
by a sudden drop in the stress, but because the yield point is not always clearly
defined engineers often use the 0·2 per cent proof stress as the safe working stress,
i.e. the stress required to produce 0'2 per cent plastic strain. This proof stress is marked
88 PHYSICAL PROPERTIES OF MATERIALS
by the point Y in Fig. 5.1 and is exaggerated for clarity. At N necking occurs and the
stress at this point is called the ultimate tensile stress or tensile strength and is in fact
the maximum nominal stress. Beyond N the nominal stress automatically decreases
but the specimen continues to elongate until fracture occurs at F. However, because
F
'"1/1
~1/1
coc:
'E
o
z
o \0.2% 10 20 30
of the appreciable change in cross sectional area that occurs during plastic deforma-
tion it is often desirable to talk about true stress and true strain where the true stress
is defined as the ratio of the load to the minimum cross sectional area and the true
strain is defined as the integral of the ratio of an incremental change in length to the
instantaneous length of the sample, i.e.
true stress = 10adjAminimum
and IF
10
A true stress-strain curve is shown in Fig. 5.2 and it can be seen that the true stress is
not a maximum at the initiation of necking but increases continually until fracture
occurs.
----
_.... .... ....
True
1/1
1/1
...~
(/)
Strain
Fig. 5.2 Comparison between nominal and true stress-strain curves.
MECHANICAL PROPERTIES OF MATERIALS 89
A material whose tensile behaviour is typical of that shown in Fig. 5.1 or 5.2 is
termed a ductile material. However, many materials fracture following little or no
plastic deformation and these are termed brittle materials.
Although the tensile test is the most common there are several other mechanical
tests that are commonly employed. Compression testing is quite common for brittle
materials which are usually much stronger in compression than in tension. Necking
does not occur in compression testing and the resulting curves are much closer to the
true stress-strain curve illustrated above. Ductile materials are rarely tested in com-
pression because of the frictional constraint of the apparatus on the ends of the
specimen which gives rise to a complicated stress distribution at the surfaces. Hard-
ness testing measures the resistance of a material to an indentor of a predetermined
shape and an empirical hardness number may be assigned from measurements of the
applied load and the dimension of the indentation. Finally the static tensile properties
of a material are generally better than the dynamic properties which result from the
repeated application of a load. The fatigue test determines the stress which a specimen
can withstand for a given number of cycles of the applied stress. Specimens are
usually tested to failure for different stresses and the number of cycles prior to
failure is noted.
Elastic deformation is reversible and the stress and resulting strain are proportional to
one another and obey Hooke's law; the constant of proportionality between stress and
strain is called a modulus of elasticity. Elastic behaviour is not restricted to tensile
stresses and three elastic moduli are commonly defined:
(i) Young's modulus, E, relates tensile stress 0t to tensile strain e: E =ot/e;
(ii) the shear modulus, G, relates shear stress r to shear strain r: G = r/r; and
(iii) the bulk modulus, K, relates bulk stress 0 to bulk strain ..1V/V(relative change
in volume): K = o/(..1V/V)
In addition, as a specimen elongates in the elastic region its cross sectional area
decreases and another elastic constant, Poisson's ratio J.l., can be defined as the ratio
of the transverse to the axial strain. These elastic moduli are usually anisotropic in the
case of single crystals but, since the mechanical engineer usually works with poly-
crystalline materials in which the individual crystallites are arranged completely
randomly, it can generally be assumed that the mechanical properties of materials
are isotropic. Then quite simple relationships exist between these four elastic
constants:
E= 3K(1- 2J.l.)
and
(5.1)
On the atomic scale a tensile force, for example, will tend to elongate the specimen
and increase the interatomic spacings of the atoms. Thus Young's modulus will be
proportional to aF/ar and, for the small atomic displacements encountered in the
elastic region (---1 per cent), aF/ar, which is the tangent at r = ro of the interatomic
force-interatomic separation graph of Fig. 2.1, will be approximately constant, as
expected theoretically. In addition the modulus of elasticity will be the same for both
tension and compression. The magnitude of aF/ar depends quite markedly on the
90 PHYSICAL PROPERTIES OF MATERIALS
magnitude of the interatomic forces as shown in Fig. 5.3 and so the values 0 f the
elastic constants are directly related to the fundamental interatomic forces. The cor-
relation between the magnitude of Young's modulus and the type of bonding is
evident from the data in Table 5.1, although it should be noted that the value of
Young's modulus is temperature dependent. The somewhat anomalous elastic prop-
erties of elastomers will be left to section 5.8 when the mechanical properties of
polymer materials will be discussed collectively.
F F
\
\
\
/'0
r r
\
aF'
-
ar ", ,
,
\-aF
\ -
ar
(a) (b)
Fig. 5.3 Variation of aF/ar with interatomic bonding. (a) Strong bonding. (b) Weak bonding.
TABLE 5.1
MgO ionic 30 X 10 10
Copper metallic 13 X 10 10
Diamond covalent 54 X 10 10
TABLE 5.2
F F
Lattice bending
and rotation
Slip
directions
However, these slip systems will only be operative if the applied shear stress exceeds
a critical value. Let us consider a cylindrical specimen of cross sectional area,
A, in which the normal to the slip planes makes an angle () and the slip direction an
angle rp with the stress axis as illustrated in Fig. 5.5. The resolved force in the slip
Slip
plane
direction is F cos cp and since the area of the slip plane is A/cos (), the shear stress, T,
is given by
T = (F/A) cos () cos cp = a cos () cos cpo (5.2)
This expression has a maximum value when both () and cp are 45° which explains why
there is a tendency for specimens under tension to fracture along planes at 45° to the
tension axis. Experimentally slip is observed when T reaches a critical value of the
order of 10 5 to 106 Nm- 2 •
Now consider the atomic processes that must be occurring during slip so that a
theoretical estimate can be made of the critical shear stress required to make two
atomic planes slide over one another. For the two atomic planes in their equilibrium
configuration shown schematically in Fig. 5.6a, assume that under the application of
+I
(SCDCDO
d I I
I
I
I
Shear I
stress I
I
T I
Displacement
x of atoms in
upper layer
(b)
Fig. 5.6 Theoretical mechanism of slip.
an applied shear stress the upper layer moves as a whole over the lower plane by a
distance x. When x = al2 the atoms in the upper layer will be directly over those in
the lower plane and be in a position of unstable equilibrium and so the stress will be
zero whenever x =0, a12, etc. If it is assumed that the stress is periodic in nature, as
illustrated in Fig. 5.6
T =Tc sin (21Tx/a), (5.3)
where Tc is the critical or maximum stress. For small strains (x < a) the usual theory
of elasticity should apply so that T = G(xld) and, since sin () ~ () as ()~ 0
T = Tc sin (21Txla) = G(xld) (5.4)
or
Tc = GaI21Td. (5.5)
For most crystals a "" d and T c "" G/21T. The shear modulus of most materials is of the
order of 1010 Nm- 2 and so this theoretical value for the critical shear stress is several
orders of magnitude greater than that observed experimentally. Furthermore, this
theoretical analysis requires an elastic strain of 25 per cent before any permanent dis-
MECHANICAL PROPERTIES OF MATERIALS 93
tortion could occur whereas elastic strains of only 0·5 per cent are typical. Although
the above calculation is by no means rigorous it is obvious that slip does not occur by
one crystal plane simply sliding over another.
This discrepancy between theory and experiment led to the postulation of dis-
locations in the 1930s although at that time there was no direct evidence for their
existence. The movement of both edge and screw dislocations in their slip planes
requires much less stress than calculated above even though the atoms immediately
ahead of either resist the movement to some extent. This is because the dislocation
forces the atoms out of their stable positions. The atoms immediately behind push the
dislocation forward since they can then fall back into their equilibrium positions. The
Burgers vector of an edge dislocation and the dislocation line itself define the slip
plane and edge dislocations normally move in this plane. In the case of a pure screw
dislocation the Burgers vector and dislocation line are parallel and do not define a
unique plane so that this type of dislocation is free to move in any plane containing
the Burgers vector. In both cases the movement of the dislocation produces a per-
manent deformation and a slip step at the surface as illustrated in Fig. 5.7.
Edge dislocation
" ,
line
h-<::~~~~
_ _T
-~
T _..........- \
Screw
dislocation
Slip line
plane
(a) (b)
Fig. 5.7 Formation of a slip step by the movement of dislocations. (a) Edge dislocation. (b)
Screw dislocation.
result in a highly unstable situation; for example slip in an ionic crystal could disturb
the regular anion-cation arrangement demanded by electrostatic consideration. For this
reason the critical shear stress for slip on the theoretically expected planes may be so high
that slip is impossible before the specimen fractures. The effective number of slip systems in
compounds is therefore reduced and compounds tend to be more brittle than elements.
Slip is not the only mechanism that can produce plastic deformation and a shear
stress may cause mechanical twinning (see section 2.12) with a resulting permanent
shear strain. Unlike slip, twinning produces an abrupt reorientation of the lattice and
it can only occur in those directions and on those planes which produce mirror images
of the lattice. Furthermore, because of this mirror image restriction, the shear strain
produced is limited to specific values whereas the shear strain produced by slip can
increase continuously from zero until the specimen fractures. The critical shear stress
for twinning is higher than that for slip and the former does not generally occur unless
slip is restricted in some way. Twinning occurs most commonly at low temperatures
when the critical shear stress for twinning is less temperature dependent than that for
slip, or as a result of high strain rates. It may also occur in h.c.p. metals when the
resolved shear stress on the basal plane is low since this is normally the expected slip
plane. The actual mechanism of mechanical twinning is not fully understood although
it is again believed to be some form of dislocation mechanism.
(a) (b)
Because climb relies on the process of diffusion it is more likely to occur at elevated
temperatures. Cross slip is the movement of a screw dislocation from one slip plane to
another. Because its Burgers vector is parallel to the dislocation line, a screw disloca-
tion can cross slip quite easily on to any plane which contains the Burgers vector (Fig.
5.9b). Cross slip does not involve the diffusion of atoms and so it is not a thermally
activated process.
The discussion, so far, has been restricted to the behaviour of isolated dislocations
ignoring the fact that most crystals contain fairly high dislocation densities. There-
fore, the effect of these other dislocations that are present is considered briefly.
Because of the stress field associated with each dislocation line there will be an inter-
action force between different dislocations and in the simple case of parallel disloca-
tions several conclusions can be deduced quite readily. Parallel screw dislocations of
the same sign will always repel one another whereas parallel screw dislocations of
opposite sign will attract one another and if unimpeded they may ultimately annihilate
one another and leave a perfect lattice. The nature of the force between two parallel
edge dislocations is more complex since it varies with both the sign of the dislocations
96 PHYSICAL PROPERTIES OF MATERIALS
and their separation. However, in general, parallel edge dislocations of the same sign on
the same or nearby slip planes repel each other, but if they are on more widely spaced
slip planes they tend to attract each other and align themselves above one another in
a plane normal to the slip plane to form a low-angle boundary (section 2.12). Parallel
edge dislocations of opposite sign on the same or adjacent slip planes attract one
another and may annihilate each other to leave a perfect lattice or a row of vacancies.
Of course not all dislocations are parallel to each other and if a pair of non-parallel
dislocations of either type are forced to intersect as they glide under the application of
an applied shear stress, jogs are created in one or both of the dislocation lines as
illustrated in Fig. 5.1 0 for the case of two edge dislocations. The presence 0 f these jogs
in the dislocation lines makes them less mobile and so a greater applied stress is required
to move them. This has important consequences in the strain hardening of metals
(subsection 5.6.1).
5.S Fracture
Fracture is the separation of a specimen into two or more parts by an applied stress
and can be classified as either brittle or ductile. Brittle fracture occurs after little or
no plastic deformation, whereas ductile fracture occurs after extensive plastic defor-
mation.
~m
1"----'1
I I
r---2c~
a
Fig. 5.11 Griffith crack model.
where c is half the length of an internal crack or the length of a surface crack and p is
the radius of curvature at the tip, which may approach atomic dimensions in a brittle
material. With such stress concentrations the cohesive strength can be exceeded at the
tip of a crack under a fairly low applied tensile stress. If the crack is to propagate
through the material the release of stored elastic energy must be sufficient to provide
the energy associated with the two surfaces produced. Consider a surface crack of
depth c and unit width, then the strain energy released by this crack can be found by
integrating the strain field (a 2 /2£) per unit volume from infinity to the surface of
the crack and is given by
UE = a2 rrc 2 /2E. (5.8)
The energy consumed in forming the two crack surfaces will be
Us = 2'Yc, (5.9)
where'Y is the surface energy per unit area. Such a crack can only grow spontaneously
if its growth does not increase the overall lattice energy, i.e. if dU/de ~ 0, where
U= -UE + Us, i.e.
-d (rrc
2 2
-- ( ) d
=-(2'Y c) (5.10)
de 2E dc
or
( £)1/2
a = 2'Y-
rrc
(5.11)
This result shows that the stress necessary to cause brittle fracture varies universally
98 PHYSICAL PROPERTIES OF MATERIALS
as the square root of the crack length and so the tensile strength of a completely
brittle material will depend on the length of the largest crack existing in the material.
Although this derivation strictly applies to a thin plate under tensile stress, similar
results can be deduced for other specimen geometries.
Direct evidence that inherent cracks are the cause of brittle fracture has been
observed in investigations with glass, since freshly prepared glass fibres with high sur-
face perfection exhibit very high tensile strengths. This strength is drastically reduced
by handling which causes minute surface cracks, which need be only a few microns
deep, and glass is usually extremely brittle. Sheet glass can be strengthened for such
applications as car windscreens by heat treatments that produce compressive stresses
in the surface which effectively hold the cracks together.
In the case of crystalline materials the Griffith theory has to be modified somewhat
to take account of the small amount of plastic deformation that occurs. Basically,
however, the fracture mechanism is the same although the critical crack depth is
somewhat larger. In single crystals fracture occurs along specific lattice planes which
are called cleavage planes. In polycrystalline materials brittle fracture can either take
place along the cleavage planes or along grain boundaries due to the weakening effect
of impurities which have segregated to the boundaries, leading to a characteristic
granular appearance of the fracture surfaces.
W'
Strain
Fig.5.12 Increase of yield strength by plastic deformation.
TABLE 5.3
requires an initial solution treatment to dissolve the solute atoms and then a quench
to room temperature to produce a supersaturated solid solution. The precipitation is
then initiated by reheating to a temperature below the solid solubility limit and held
for a predetermined 'ageing' period. Close control of this temperature and time is
maintained to produce an optimum distribution of the precipitate. The amount of
ageing is quite critical. Under-ageing causes relatively few particles to precipitate and
over-ageing increases the size of the particles and hence decreases their number so that,
in both cases, the degree of hardening is less than optimum. An example of age harden-
ing is found in the AI-Cu system from which the 'Duralumin' series has been developed.
Dispersion hardening is the technique of achieving a fine distribution of hard particles
by allowing certain gases to diffuse into the solid. A reaction between the gas and the
alloy matrix produces the hard particles. For example steel which is to be nitrided to
increase wear resistance usually contains aluminium in solid solution and it is heated in
an ammonia atmosphere so that nitrogen diffuses into the steel and a fine dispersion of
aluminium nitride is formed.
5.7 Creep
So far it has been assumed that stress and strain occur simultaneously but in the case
of many materials this is not entirely true and under the influence of a constant applied
stress the material may continue to deform indefinitely. This process is called creep
and is readily observable in the sloping lead-covered roofs of old churches where, over
the years, the lead has deformed under its own weight to produce a sheet considerably
thicker at its lower edge.
Since materials tend to work harden during the process of deformation, continued
deformation under constant stress means that there must be some recovery taking
place. The rate of recovery increases with increasing temperature and so the creep rate
is strongly dependent on temperature and creep is an important consideration in high
temperature applications. A characteristic creep curve is illustrated in Fig. 5.13. There
Fracture
Primary Secondary
Time
Fig.5.13 Creep behaviour under constant applied stress.
are four main areas of interest: (i) an initial instantaneous strain which can be partly
elastic and partly plastic; (ii) a primary region of creep of decelerating creep rate; (iii) a
secondary period of constant minimum flow rate; and (iv) a tertiary creep region of
accelerating strain rate, which ultimately ends in fracture.
MECHANICAL PROPERTIES OF MATERIALS 103
In the primary region the creep rate decreases due to the competing effects of work
hardening and recovery until the rates of work hardening and recovery are equal and
the creep rate is then a minimum (secondary region). At low temperatures the creep
rate decreases to virtually zero so that the dimensions of the specimen stabilize and the
applied stress can be safely supported indefinitely. At higher temperatures tertiary creep
can occur due to necking of the specimen or grain boundary sliding and this continues
until the specimen fractures. Precipitation hardened alloys are generally used in high
temperature creep resistant applications because, in addition to acting as barriers to slip
and grain boundary sliding, the fine dispersion of particles apparently restricts the rate
of recovery.
The mechanical behaviour of plastic materials varies considerably from the brittle
behaviour of perspex to the amazing elastic behaviour of rubber. Unlike the majority
of materials, which can only sustain elastic strains of about 1 per cent, rubber is an
example of a class of materials called elastomers, which can be strained elastically by
several hundred per cent. The tensile properties of an elastomer can be explained
qualitatively from the model of the structure of polymer materials as a tangled bundle
of fibres described in subsection 3.9.2. In rubber the individual molecules are coiled
I
I
E I
~ I
> Glass I
I
1_--------1
------; Crystal I
I I
: I
I
Temperature
Fig. 5.14 Variation of volume with temperature for an amorphous material illustrating the
melting point T m and the glass transition temperature T g.
up on themselves and joined to one another by a few sulphur cross linkages which
prevent plastic or viscous flow (see below). Thus an applied tensile stress can be trans-
mitted from molecule to molecule via the cross linkages and straighten out the molecules
producing an immense increase in length. On removal of the applied stress the molecules
revert to their more stable coiled state. This behaviour is not typical of most polymers
because either the configuration of the side groups or the influence of weak inter-
molecular forces tends to prohibit self coiling of the molecules. In general the mechanical
behaviour of polymers varies markedly with temperature.
In the liqUid state a polymer consists of a random collection of wriggling long chain
104 PHYSICAL PROPERTIES OF MATERIALS
molecules and as the temperature is reduced the thermal agitation decreases until, at
the melting point Tm, crystallization may occur with a corresponding change in volume.
However, as already observed crystallization of polymer material is rather difficult and
amorphous polymers show no change in volume at Tm' but remain supercooled liquids
until a lower temperature is reached, where molecular rearrangement ceases and the
liquid vitrifies into an amorphous solid (Fig. 5.14). At this lower temperature there is
a change in the rate of decrease of volume and hence in the thermal expansion
coefficient. This behaviour is not restricted to polymers and is typical of all amorphous
materials, such as glass, and the temperature at which the supercooled liquid finally
vitrifies is called the glass transition temperature, Tg • The resulting mechanical
behaviour at room temperature is governed by the magnitudes of Tm and Tg relative
to room temperature. If Tm and Tg are both below room temperature the polymer is
a liquid, whereas if T m and Tg are both above room temperature the amorphous regions
behave very much like glass. A completely amorphous polymer such as perspex with Tg
about 60°C, tends to be very rigid and brittle at room temperature. If room tempera-
ture lies between Tm and Tg the situation is rather more complicated because the
polymer is either a completely supercooled liquid or a combination of crystalline and
supercooled liquid regions. The presence of the amorphous liquid regions in a partially
crystalline polymer has a marked effect on the mechanical behaviour of the material
since they give a certain amount of flexibility to the material as in the case of poly-
ethylene. The response of the supercooled liquid regions to an applied stress is rather
complicated because, in addition to an instantaneous strain in the individual molecules,
the stress causes the individual molecules to slide past one another as in a viscoelastic
process. A viscous material is of limited technological application but fortunately the
viscous flow can be decreased by increasing the degree of crystallinity, as in high density
polyethylene. This is much more rigid than the normal material and is widely used for
liquid containers. Alternatively a certain amount of cross linking may be introduced to
restrict the relative motion of the molecules. Both T m and Tg are governed by the
strength of the intermolecular forces and can be adjusted to a certain extent by
varying the chemical composition of the polymer side groups. By a suitable choice
of T m' Tg , the degree of crystallinity and cross linking, a polymer material can be
designed for a specific technological application.
In thermosetting polymers the process of curing creates a rigid amorphous solid.
The cross linking prevents relative motion between the molecules and the resulting
material is fairly brittle.
References
Questions
1. A steel wire of length 1 m and cross sectional area O· 5 mm2 is stretched horizontally
between two pillars. A mass of 0·1 kg is suspended from the mid-point of the wire.
Calculate the depression of the mid point. [Young's modulus of steel = 2 x lOll Nm -2.]
MECHANICAL PROPERTIES OF MATERIALS 105
2. (a) Deduce an expression for the bulk modulus in terms of Young's modulus and
the shear modulus.
(b) Show from theoretical considerations that for an isotropic material Poisson's ratio
cannot be greater than 0·5. Check for as many materials as possible that this
theoretical value is not exceeded.
3. Show that, for a homogeneous isotropic material
Ele = 2(11 + 1).
A gold wire, 0·32 mm diameter, elongates by 1 mm when stretched by a force of
103·3 N and twists through 1 rad when equal and opposite torques of 14·5 x 10- 6 Nm
are applied at its ends. Calculate Poisson's ratio for gold.
4. Aluminium is an f.c.c. metal and deforms by slip on {Ill} planes and in (110)
directions. A single crystal is stressed in tension along its [11 0] direction until slip
starts.
(a) On which {Ill} planes and in which <}10) directions will slip occur?
(b) If the critical shear stress for slip is 106 Nm -2 what is the applied tensile stress?
5. A glass plate has a sharp crack of length 1·5 11m in its surface. Calculate what stress
is necessary for the crack to propagate if Young's modulus for glass is 7 x 10 10 Nm- 2 and
the energy of the fracture surface is 0·5 J m- 2 •
6. The relation between the changes in electrical resistance and strain for a strain gauge
is given by the gauge factor (8RIR)j(0/1l). Show that, if the resistivity is assumed to be
independent of strain the gauge factor is given by 1 + 211. Check the validity of this
assumption by finding the gauge factor for several strain gauges and comparing with
this theoretical value.
7. Cleavage occurs on {100} planes of MgO. If a stress of 40 x 106 Nm- 2 is applied
parallel to the [111] axis will cleavage occur, and if so, on which cleavage planes?
Critical fracture stress for cleavage is 12 x 106 Nm- 2 •
6 Thermal Properties
6.1 Introduction
In this chapter the thermal properties of solid materials are considered from an atomic
point of view. It will be important to understand the fundamental thermal concepts
of temperature and heat capacity from such a viewpoint and to integrate these con-
cepts with others that have been met already. Then it is possible to proceed to an
explanation in atomic terms of familiar macroscopic thermal properties such as thermal
conductivity and thermoelectricity. Thermodynamics shows that it is not possible to
talk precisely about the intrinsic heat content of a body but only about its internal
energy, the total kinetic and potential energy of its constituent parts. This internal
energy may be increased or decreased by the transfer of energy to or from the system
in the form of 'heat' or 'work'. The latter mechanism, which invariably involves a force
moving through a distance, will not be of direct concern here.
The internal energy resides in atomic, molecular and electronic motions and these in
turn can be characterized by the occupation of certain quantized states of the system,
the energy levels. Temperature, the characteristic of thermal equilibrium, can then be
described in statistical terms through the occupancy of these energy levels. At tem-
perature T, the occupancy of each level depends exponentially on the energy of the
level measured in units of kT. It is convenient to consider a crystalline solid as a
regular array of bound atoms, molecules or ions with which there mayor may not be
associated large numbers of 'free' electrons. The lattice and any associated free electrons
can be treated separately although this step is not always justified. The discussion will
centre on the thermal properties of the lattice and of the free electrons. The general
ideas of statistics applicable to both will first be considered.
The understanding of temperature and internal energy derives from the classical work
of Maxwell and Boltzmann and it is instructive to begin with a resume of their results.
Assume that in a gas of free particles at temperature T K the only motions of interest
are the translational movements of the particles associated with their velocities. These
motions can be resolved into three mutually perpendicular directions and each particle
is said to have a 'degree of freedom' for the motion in each of the three directions.
Maxwell showed that ~kT of energy can be associated per particle with each degree of
freedom. k is Boltzmann's constant, the ratio of the molar gas constant R to Avogadro's
number. It has the value 1·38 x 10- 23 J K- 1 • Clearly k measures the thermal energy of
a molecule per degree of 'temperature' above absolute zero. Alternatively it can be
THERMAL PROPERTIES 107
considered as the specific heat of the molecule, the amount of energy required to
produce unit temperature rise.
The total energy per particle is then (3/2) kT. For bound particles, however, there is
the potential energy of the binding forces as well as kinetic energy. Maxwell further
showed that the average potential energy is equal to the average kinetic energy. Accord-
ingly at room temperature each bound atom in a solid will possess an average total
energy of 3 kT (0'075 eV). Where this energy, though small, is comparable with energy
level separations, it can playa significant role; in semiconductors for example, by
ionizing the very shallow impurity levels.
The concept of a mean energy implies that some particles have less and some more.
Fig. 6.1 illustrates the distribution of kinetic energies of vibration, E, deduced by
Maxwell for a collection of particles at temperature T. The ordinate rp(E ) represents
the fraction of particles having energy E per unit energy interval, i.e. the actual fraction
1
2kT
li:J
'--I I
9- - I
4kT
Mean:
energy I
1
o kT 2kT Eo 3kT
E
Fig. 6.1 The Maxwell distribution of kinetic energies.
in the interval D.E will be given by r/>(Eo)D.E (Fig. 6.1). This is a statistical distribution
which is somewhat different when the particles whilst still in thermal equilibrium are
being acted on by an external force, as are bound electrons in a magnetic or electric
field. In such a case the thermal contribution to the total energy will result in an
increase in the potential energy. The distribution of this potential energy among the
particles was shown by Boltzmann to follow an exponential law. If No particles have
a given potential energy then the number with amount U in excess of this will be
No exp (-U/kT). This distribution is illustrated in Fig. 6.2.
An alternative statement of the Boltzmann relation is that the probability that a
given particle has potential energy U is proportional to exp( -U/kT). If U is much
greater than kT the probability is very small and vice versa. An example is the distri-
bution of the molecules with height in the atmosphere where the force acting is gravity.
The potential energy of a molecule is mgh where h is the height above the earth's sur-
face. The concentration at height h is therefore proportional to exp(-mgh/kT) as is
the atmospheric pressure at that height.
Again, in the calculation of the relative population of electronic energy states in
atoms, the potential field is an electric one. Not every atom will be in its ground state.
The receipt of thermal energy by way of the atomic vibrations of the lattice can cause
the potential energy to be raised sufficiently to attain the first excited state. Using the
108 PHYSICAL PROPERTIES OF MATERIALS
3kT
U
Fig. 6.2 The Boltzmann distribution (number of particles N with excess potential energy U as a
function of U).
above statistics, usually called Maxwell-Boltzmann (M.B.) when both kinetic and
potential energies are involved, the relative occupation of the ground state (0) (Fig.
6.3) and first excited state (1), in thermal equilibrium will be given by:
number in state I exp(-EdkT)
(6.1)
number in state 0 exp(-Eo/kT)'
henceN1 =No exp{-(El - Eo)/kT}.
, st excited state
.
>
Cl
41
c:
w
Ground state
--------------------------0
Fig. 6.3
The heat capacity of a body is that thermal energy necessary to raise the temperature
by one degree Kelvin and it is usually expressed as a molar quantity. At constant
THERMAL PROPERTIES 109
volume no mechanical work is performed, all the thermal energy is converted into
internal energy U. Hence the heat capacity is given by
(6.3)
In practice measurements of C are made at constant pressure, Cp ' but the two values
can be related.t
At zero temperature the internal energy is a minimum and the heat capacity is zero.
This is a consequence of the third law of thermodynamics. As the temperature is raised
the heat capacity variations indicate the various mechanisms by which energy is taken
up in the solid. A typical measurement is shown in Fig. 6.4 for germanium, whose heat
25 -
I
0
E 20
Cl
~
I 15
~
-
;:;>
0 10
)(
C. 5
V
capacity exhibits a rise from zero to a plateau followed by a very gradual increase
thereafter. Many materials show this kind of behaviour but differ in the rapidity of the
rise. For example, the plateau in lead is reached by 90 K but in silicon not until 600 K.
A direct application of M.B. statistics would attach an average total energy of 3kT
to each atom and hence of 3NkT to each mole of solid. The heat capacity, from eqn
(6.3), is the derivative of this, namely 3Nk (25 x 103 J K -1 kg mol- 1 ) and indeed the
plateau value is close to this. The explanation of the fall off of Cv at low tempera-
tures was given first by Einstein and then in more detail by Debye in his theory of
specific heats. The ideas behind these theories are directly related to the statistical
ideas discussed above.
Without worrying about the details think of a solid as a set of oscillators vibrating
with a characteristic frequency. These will be the entities which take up thermal energy
during heating or emit energy during cooling. Each oscillator will have a spectrum of
possible energy states because theory shows that for a simple harmonic oscillator the
energy E is related to the frequency v of vibration by
E =(n + !)hv, (6.4)
where n is an integer (Fig. 6.5).
The temperature of the collection of oscillators will decide the statistical distribution
of energy states in which these oscillators find themselves. Suppose the temperature is
such that the great majority of oscillators are in the lowest state. If the temperature is
lowered beyond that point the oscillators are incapable of parting with any more energy
so the specific heat suddenly becomes zero. In practice there is always a statistical
t Cp - Cv =90<2 KVT where 0< is the linear expansion coefficient, V is the molar volume and K
is the bulk modulus.
110 PHYSICAL PROPERTIES OF MATERIALS
spread in the occupation of levels so there is a tailing off in the specific heat rather
than a sudden drop at low temperatures. Similar arguments apply during warming up.
A typical frequency of vibration for an atom in a solid is 1013 Hz and the energy ofits
lowest state is thv (0·02 eV). The temperature appropriate to this energy is given by
the classical expression kT leading to a value of T = 232 K. This is the temperature in
this rough calculation at which the oscillator would be in its ground state. Below this
temperature it would not be contributing to the specific heat. Einstein carried out a
statistical calculation assuming that all the oscillators vibrated independently with the
same frequency. His result was that the average total energy E of the system of N
oscillators of frequency Vo is given by
E = 3Nhvo/ {exp(hv/kT) - 1}. (6.5)
Equation (6.5) gives E ~ 3NkT at high temperatures corresponding to a specific heat
of 3R but at low temperatures gives values which follow the curve of Fig. 6.4 fairly
7
"2hv
>-
...
Cl
GI
c:
w 5
"2hv
3
"2hv
1
- - - - - - - - - - - "2 hv
Fig. 6.5 Energy levels of a simple harmonic oscillator of frequency II.
well. In a real solid the atoms cannot be considered to vibrate independently because
of the bonding between them. The motion of one atom must affect its neighbours and
vice versa. The frequencies of vibration are really a property of the solid as a whole.
In a one-dimensional line of atoms various vibrations are possible of different wave-
lengths and therefore different frequencies (Fig. 6.6). Thus, there is really a spectrum
of frequencies for the atomic oscillators and Debye attempted to take this into account
by modifying the Einstein theory and by considering the vibrations to be those of an
elastic medium. The velocity of the waves is the velocity of sound in the solid, since
the waves are sound waves. Debye's statistics were a little more complicated but his
result was in better agreement with experiment than Einstein's. Suppose the maximum
oscillator frequency is vo, then if T~ hVo/k, the theory predicts that the energy distri-
bution for oscillators is Maxwell-Boltzmann like and the specific heat is 3R. But if
T ~ hVo/k the specific heat decreases with temperature as described above, being pro-
portional to T3 in fact. Thus the temperature dependence of specific heat is determined
THERMAL PROPERTIES 111
by the maximum oscillator frequency and this in turn depends on the strength and
type of binding of the atoms, or in Debye's terms on the elastic properties of the
solid. The temperature hVo/k for a solid is known as the 'Debye characteristic tem-
perature e'.
The ideas of Debye lead on to a very important concept in the theory of solids
namely the phonon. The oscillators of Fig. 6.6 each have an energy which is quantized
Evibration = h x frequency . (6.6)
• • • • • • • • • • • • •
• • • • • • • • •
• • • •
• • • • •
• • •
• •
• • •
• • • • • •
• • • •
• • •
• • • •
• • • • •
• • • •
Fig. 6.6 Atomic vibrational modes of a one-dimensional solid.
These electrons are relatively few at ordinary temperatures so that the electronic
heat capacity depends only on the fraction which contribute (Fig. 6.7). The number
increases linearly with absolute temperature.
...... --
,~\
~O'cPo' -Density of
,pfPJ~ g~ states
/ ~ g Probability of
., •• 0 occupancy at
/ .g T=O
/ 10.. kT
/ o.
Probability of
occupancy at
o g~ temperature T
• 0 •
o •
•
EF Energy
Occasionally, quite sudden changes in the specific heat of a substance are observed over
a narrow range of temperature. One situation for which this can happen is when there
is an electronic energy state of the atoms of a solid not far removed in energy from the
ground state by say, e eV (Fig. 6.8).
The excitation of electrons from the ground state will occur with a probability pro-
portional to exp( -e/kT). When T < e/k most atoms will be in the lowest state and
when T ~ e/k the atoms will be equally divided between the lowest and the excited
state. For a range of temperature intermediate between these extremes there will be a
sharp increase in the number of atoms excited and a corresponding change in the
internal energy. A dramatic increase in specific heat will be observed over a small
range of temperature where kT is near e which may typically be 0·1 eV. The behaviour
of the specific heat as the temperature changes is shown in Fig. 6.9.
THERMAL PROPER TIES 113
Excited state
>-
~
CD
C
W
>< )( )( )( )( )( )( ><
Ground state
Fig. 6.8 Ground state conf'Jguration of atoms giving rise to electronic specific heat anomaly for
€~kT.
Cex 1.0
!
+"
0.8
'"
/
I I\. \
I
!
i
"~ ~
«I
CD 0.6
.c
i
11
t.l
t--
;;::
I
.g 0.4
c.
en
~
---r--r--
0.2 J
j
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
kl/,t
Fig.6.9 Typical specific heat anomaly Cex due to electronic excitation.
8" r
600
I
Cl
.>I.
:J.
....«I
CD 400
.r.
t.l
~
t.l
CD
C.
en
200
800
Temperature (K)
Fig. 6.10 Specific heat of nickel (showing a magnetic order-disorder phenomenon).
114 PHYSICAL PROPERTIES OF MATERIALS
A specific heat anomaly also occurs when some kind of rearrangement of atoms or
molecules of a crystal takes place, for example when a phase change occurs in an alloy.
But there are other cases where anomalies appear during a change of order not directly
related to atomic or crystal structure. Examples which will be met in other chapters
are the alignment of electric dipoles in ferroelectric materials, or of magnetic dipoles
in ordered magnetic materials. A more subtle kind of ordering occurs in superconduct-
ing materials where a change in the 'ordering' of electron momenta leads to a specific
heat anomaly. Fig. 6.10 shows a typical specific heat anomaly in a ferromagnet occur-
ring at the Curie temperature where the ferromagnetism (ordered array of magnetic
dipoles) changes to paramagnetism (disordered array) as the temperature increases.
A considerable part of the thermal energy of a solid is in the form of atomic vibrations;
the higher the temperature the more energetic the vibration and consequently, the
greater the amplitude. As a result of the varying amplitude the measured lattice spacing
between atoms is a mean figure and it is necessary to know how it depends upon the
temperature.
Interatomic distance r
'0
>-
Cl
CD
C
III
m
0';:::;
c
....o
III
a..
The lattice constant for a particular solid arises from the balance in forces between
the repulsion of neighbouring ion-cores due to their electron shells overlapping and the
attraction of the particular bonding type (ionic, covalent, metallic). The eqUilibrium
lattice constant is shown in Fig. 6.11 corresponding to the minimum in the depend-
ence of the potential energy upon interatomic distance. The addition of vibrational
energy to an atom can be represented in Fig. 6.11 by the horizontal line A 1 A 2 , the
height above 0 being the maximum kinetic energy of vibration. The extremes A 1 and
A2 correspond to the separations at which the kinetic energy has been transferred
THERMAL PROPERTIES 115
into potential energy and represent the limits of the vibration. For this motion the
mean interatomic distance is midway between A 1 andA 1 , at Ao and for small energies
lies above O.
For higher temperatures and consequently higher vibrational energies (e.g. A ~ A;) the
mean interatomic distance increases since although A ~ is little different to A 1 due to
the strength of the repulsive ion-core interaction, A ~ can differ considerably from A 2 ,
the attractive force being weaker in general. This resulting expansion of the lattice
structure, due to the injection of thermal energy occurs because the potential VCr) is
not symmetrical. Had it been symmetric the mean position of the atom would always
be above 0 and no expansion could occur on the increase of vibrational energy.
Thermal expansion is defmed as the fractional change in either length or volume
per unit temperature rise, e.g. for the linear coefficient
<lL = (1/1) d//dT. (6.8)
The expansion coefficient is largest for those materials with the shallowest potential
curves. These are the relatively weakly bonded materials such as the alkali metals.
Strongly bound insulators, however, such as ionically bound glass, and quartz, or the
covalently bonded materials such as silicon, would be expected to have a low expansion
coefficient. This dependence is shown in Table 6.1 where the coefficients are for linear
expansion.
TABLE 6.1
Linear expansion coeffiCients (K-l)
Potassium K 83 x 10-6
Aluminium Al 23 x 10- 6
Iron Fe 12 xl0- 6
Silicon Si 2·3 x 10- 6
Germanium Ge 5·5 x 10-6
Quartz (fused) Si02 0·5 x 10-6
Crown glass 9 x 10- 6
Polyethylene 180 x 10-6
Natural rubber 670 x 10-6
Bakelite 40 x 10-6
Molecular solids, such as polymers, have very weak attractive forces and consequently
exhibit large thermal expansion, e.g. rubber and polyethylene. It is interesting to note
that rubber contracts when heated but this is due to the temperature dependence of its
elastic modulus. The strength of the bonding in these materials can be increased by
inducing cross-linking between monomers (Section 3.9.2) and this severely reduces the
expansion, e.g. bakelite (Table 6.1).
It has been seen that at a given temperature a material will contain a 'gas' of energetic
free electrons if the material is metallic or semiconducting, plus a concentration of
lattice phonons. In equilibrium these electrons and phonons are moving in random
directions and hence transporting energy, with no net flow in any particular direction.
This energy is eventually lost by interactions either between electrons and phonons,
116 PHYSICAL PROPERTIES OF MATERIALS
electrons and electrons, or phonons and phonons. In all the interactions energy is
merely being transferred, and so an equilibrium is a dynamic situation with energetic
electrons being continually excited and de-excited and phonons being continually
created and destroyed.
Should a material be subjected to a gradient in temperature, the hot and cold regions
will be characterized by different phonon concentrations and energies and different
mean electron energies. As a result there will be a natural diffusion of the more
energetic electrons to the region where their mean energy is lower, and of phonons
from the higher to the lower concentration. In both cases this corresponds to a net
energy transfer from a high to a low temperature region. The mechanism of transfer will
not be one of direct flow from high to low temperature of appropriate electrons and
phonons but will be a dynamic one in which the process is gradual, energy being lost
and gained often en route. Phonons can be regarded as travelling down the sample with
the velocity of sound in the material and with a characteristic mean free path. Unlike
electrons the total number of phonons is not conserved, but phonons can be created
at the hot end to carry more heat and destroyed at the cold end. The energy transfer
increases the mean electron kinetic energy and the total phonon energy of the low
temperature region, i.e. it raises its temperature, and this will continue until the whole
solid achieves a stable equilibrium temperature. Should the temperature gradient be
continuously impressed upon the material, then the energy transported must be trans-
ferred to the coolant at the low temperature end. Heat is then being continuously con-
ducted along the sample.
The ability of a material to conduct heat is given by its thermal conductivity K
which is defined as the quantity of heat flow Q, per unit area A, per unit-time t, per unit
temperature gradient dT/dx, Le.
dQ/dt = -KA(dT/dx) (6.9)
The negative sign indicates that the heat flow is in the direction of decreasing tempera-
ture. Table 6.2 gives some values of K for a range of materials. By simple kinetic theory,
TABLE 6.2
Thermal conductivities (Wm-1 K-l)
Ai, 420
AI 210
Fe 63
(Nichrome) Ni-cr (70-30) 14
Ge 59
NaO 7
Glass O·g
Diamond 630
Polystyrene 0·1
Expanded polystyrene 0·03
Rubber 0·2
Wood 0·2
Concrete 0·9
thermal conductivity K is related to the mean free path A of the carrier by the relation
K=\CVA,
where C is the specific heat and v the velocity. Assuming the formula applies to phonons,
THERMAL PROPERTIES 117
their velocity and mean free path are therefore important. The velocity of a phonon,
regarded as an elastic wave is determined by the elastic constants of the medium
V =(d/p) 1/2 , (6.10)
where d is an elastic constant and p is the density; d will be Young's modulus for a
longitudinal wave and the shear modulus for a transverse wave. Phonons have mean
free paths which may be one-tenth of those of electrons. In a metal an electron may have
a mean free path of 100 lattice spacings. In disordered alloys, electron mean free paths
are much reduced.
The best thermal conductors are metals in which the free electrons are dominant and
very mobile. Non-metallic materials conduct primarily by phonons and in general K is
much smaller for these materials than for metals because of the reduced mean free path
and velocity. However in both semiconductors and metals the electron contribution is
restricted to that proportion near the Fermi level. This is not the case for phonons which
do not obey the Pauli Exclusion Principle. Note also the influence of density on velocity
(eqn (6.10)); light elements tend to have high thermal conductivities.
Some non-metallic materials have a value for K in excess of that for the metal silver.
They are substances which can be manufactured almost to crystalline perfection. The
lack of defects assures large phonon mean free paths and hence a large K. Typical materials
in this class are diamond and sapphire. In contrast, in amorphous materials the random
structure interferes with the movement of phonons so that K is lower than in the
crystalline counterpart.
It has been seen that for most materials the properties of thermal conduction are
determined by the electronic and crystalline state. Some solids, however, are not perfect
lattices, but are porous containing voids (e.g. concrete) or have been deliberately ex-
panded into a very low density form containing voids (e.g. polystyrene). The voids
represent regions of very poor thermal conduction since air is a poor conductor and
so the material as a whole will be characterized by a low value of K. Such materials
represent a most important group of commercial thermally insulating materials.
+
Hot + Cold
+
For materials in which the dominant carrier is the electron, S is negative (e.g. N-type or
intrinsic silicon); where the dominant carrier is the hole, S is positive e.g. P-type silicon).
The manner in which electrons and holes lead to opposite polarity effects is shown in
Fig. 6.13.
T+ dT V- dV T+dT V+dV
Net Net
+0 -0
-" l
."CCD
EOUl
:unf
.&; 11).-
I-Cij"a
-.- j
EII)Ul
~_:::J
G)O:::
.&;.&; . -
I- "a
Net Net
T -0 V T +0 V
S =~ = negative dV =POSITIve
S =liT
(a) (b)
Fig. 6.13 Thermoelectric power. (a) Metal or N-type semiconductor. (b) P-type semiconductor.
Referring to eqn (6.11) and interpreting d Vas energy per unit charge (dV - kT/e),
S - k/e which is a specific heat per unit charge or the amount of 'heat' carried by
each electron. In a metal only electrons near the Fermi level contribute and that
fraction of the whole is kT/E F for a metal at temperature T in which the Fermi level
isEF • S must then be ofthe order (kT/E F) k/e which is typically 10 JlV/degree. In a
semiconductor, because ofthe band gap, most carriers can have energies above the
Fermi level. For a band gap EG the specific heat of excited carriers is -EG/T, i.e.
S - (EG/kT) k/e which is about 103 JlV/degree, much bigger than for metals. Some
values of S are shown in Table 6.3 where this difference is apparent.
TABLE 6.3
AI - 0·47
Fe - 51·3
Si -408
Ge + 303
THERMAL PROPERTIES 119
6. 7.2 Seebeck effect
S is an absolute property of a material, but it can never be determined directly since
in order to measure the thermoelectric potential, a voltmeter and suitable leads are
required. These leads will have the same temperature drop across them as the material
to be measured, and so the measured potential will be that characteristic of the material
reduced by that characteristic of the leads. Thus, it is only possible to measure differ-
ences in thermoelectric power, e.g. for materials 1 and 2 in Fig. 6.14:
(6.12)
Absolute values of S can be obtained indirectly by use of one material of known S, e.g.
S =0 for superconductors.
Material 2
T+I1T
T+3T
Material 1
T+3T
Material 2
T
If the two materials of Fig. 6.14 are identical, then of course there will be no
thermoelectric potential. Should they be dissimilar, however, there will be a potential,
the production of which is known as the Seebeck effect. This is utilized in the common
thermocouple and Table 6.4 shows some values of S 12 for various thermocouple
combinations at room temperature
TABLE 6.4
S12 (pV K .... ) for some common thermo-
couples
Copper-constantan 43
Orromel-alumel 41
Pt-Pt 10% Rh 6·4
Iron-constantan 54
6. 7. 4 Peltier effect
This effect occurs at a junction of two materials when a current flows through it.
Depending on the circumstances, the current flow produces either a heating or cooling
at the junction and this is known as the Peltier effect. The Peltier coefficient n'(B for
a junction AB is dermed as the rate of heat absorption per unit current flow. There
is a simple relation between Sand n, namely S = nIT. When a molecule moves from
the liquid phase into the vapour phase a latent heat is involved. Similarly an electron
in a material A at a junction, in moving to material B can be thought of as being in a
different phase, a latent heat of transfer from A to B being required. By analogy with
the liquid-vapour transition, if there is a 'vaporization' of the electron (electron goes
A -+ B say) heat will be extracted from the junction to provide the corresponding
transfer energy and the junction will cool (Fig. 6.15). If there is a condensation
--+....:......~e
Material A MaterialS
e---+-o--
latent heat given up by
electron, junction heats
Fig. 6.15 Origin of Peltier heating and cooling at a junction.
Major~y
..Minority
Majority.
..Minority
Majority
..
iii
:;::; •
c .Minority
Ql
>
0
...ClQl
.....
Q.
J
C C
0 CD
u
Ql N N P
iii
+ +
• Distance
(a) (b) (c)
Fig.6.16 Peltier effect in a P-N junction by electron transport. (a) Zero bias-thermal eqUilibrium.
(b) Forward bias-cooling. (c) Reverse bias-heating.
however, the potential energy of the electrons crossing is increased at the expense of
their thermal kinetic energy. Only the most energetic are capable of crossing and so
the effect on the N side is that the mean energy and hence the electron temperature is
lowered. To restore the thermal equilibrium, heat must be absorbed from the lattice
and a cooling effect is produced.
Similar arguments apply to holes in semiconductor junctions and also to electrons
in metal-semiconductor junctions. The largest effect occurs in the combination which
gives the largest energy difference between the conduction states on the two sides of
the junction.
The Peltier effect can be used as shown in Fig. 6.17 where the electron current flow
from left to right is in effect transporting the amount of energy f:lE per electron from
metal 1 to metal 2, cooling metall and heating metal 2. Such an arrangement is the
basis of compact, robust and simple thermoelectric heaters or coolers.
Thermoelectric applications [2] involve the use of the Seebeck effect to generate a
voltage from a temperature difference, the 'thermoelectric generator'; and the Peltier
effect to cool, the 'thermoelectric refrigerator'. Materials for these applications are
assessed through a figure of merit Z =S2 ojK, ojK being the ratio of electrical to thermal
conductivities. Hence it is important for a high Z to maximize S2 0 and, since 0 = nep.,
the ratio of mobility to thermal conductivity p.jK. Fig. 6.18 shows the general depend-
ence of S, 0 and K on the charge carrier concentration.
122 PHYSICAL PROPERTIES OF MATERIALS
>-
Cl
"-
CD
c:
CD
....iiic: 1:1£
~
CD
oa. Fermi level
c: ~ - - - ...ferm·
e %COnduc:~~
--
- - 'jel/e/
~~
t)
CD Fermi level
W
~conduc~i~l
Metal 1
cooled
Semiconductor
-~ Metal 2
heated
Fig. 6.17 Use of the Peltier effect for heat transport.
s
"-
CD
~
a.
u
.;:
....u
CD
Qj
I
o
E
Q;
~
I-
log n
I Kelectron
I
I
-----~---------------~---
I I
I I K phonon
I
I
log n
Insulators Semiconductors Metals
Fig.6.l8 Factors influencing the thermoelectric figure of merit, Z, as a function of log (carrier
concentration, n) showing maximum occurring in the semiconductor region.
THERMAL PROPERTIES 123
The maximum of S2 a occurs in the semiconductor region with K slowly varying. In
practice the most effective materials for thermoelectric applications are bismuth
telluride Bi2 Te3 and antimony telluride Sb 2Te3 or an appropriate alloy of both.
Values of S are typically 300 p.V/degree and z,.., 2·5 X 1O- 3/degree. A very compact
thermoelectric refrigerator, perhaps measuring 1 cm x 1 cm x 0·5 cm can pump 5-10 W
of heat with a current of 2 A and as a generator might produce 500 mWat an e.mJ. of
2 V for a temperature difference across the faces of 120°C.
References
Questions
I. Molecules of an ideal gas are distributed above a plane (z =0) in a force field F
given by F= (-3 - 2z) x 10- 21 where z is the distance above the plane. Calculate the
ratio of molecules per unit volume at a height of 6 m to that at 3 m at 300 K.
2. Show that for the Einstein model of a solid (eqn (6.5» the molar specific heat at
constant volume Cy is given by
(eE) 2 efdE/T
C y = 3R T (eeE/T _1)2 '
where eE = hVo/k.
3. Estimate the relative contributions of electrons and lattice to the specific heat (at
constant volume) of sodium at 20 K. The Fermi temperature of sodium is 3·6 x 104 K
and its Debye temperature is 150 K.
4. Obtain the relevant data and calculate the thermoelectric figure of merit for the
following materials: (a) cO'pper, (b) germanium, (c) silicon, (d) indium antimonide,
(e) bismuth telluride, and (f) antimony telluride.
7 Electrical Properties
7.1 Introduction
The most important property characterizing the electrical properties of a solid is the
electrical resistance R, which is more conveniently discussed in terms of a geometry
independent quantity the resistivity p. For a solid of length I and cross section area A
R = pi/A. The units of R are ohms and those of p ohm metre. Electrical conductivity
a = lip is also widely used and the units of a are reciprocal ohm metre, i.e. (ohm
metrer l often written mho metre- l . Ohm's law may then be stated as a relation
between current density J and electric field 8..
J=a&. (7.1)
The conductivity of solids is remarkable because it spans so wide a range of values,
about twenty-three orders of magnitude. The best metallic conductors such as copper
and silver have values of a about 108 n- l m- l and a good electrical insulator, for
example polystyrene, has a conductivity of la-IS n- I m- I . No other quantity varies
to such an extent and it seems a formidable task to explain such a variation.
The ideas of Drude at the beginning of this century on electric conduction in metals
are fundamental and provide a simple view of the behaviour of electrons in all materials.
Drude's work was based on the idea of a metal as a lattice of positive ions through which
electrons moved freely according to the laws of kinetic theory. The electrons move at
random at thermal speeds but under the influence of an applied field gain an additional
drift velocity. Collisions with the lattice are the source of electrical resistance.
In such a collision an individual electron has its drift velocity reduced to zero, and
immediately afterwards its motion is random. In this situation the current density J
from n electrons per unit volume of charge e and drift velocity v is
J=nev. (7.2)
If the average time between collisions is 2T, then during this time the electron is acted
on by a force F = -e8. where 8. is the applied electric field.
dv
m-=-e8. m = mass of electron
dt
e8.
dv=- -dt
m
&t
v = -e - + constant.
m
But at t = 0, v =0 (immediately after each collision) and the integration constant is zero.
ELECTRICAL PROPERTIES 125
... .. e&2r
Hence the velocity Immediately before a collIsIOn = - - -
m
therefore
mean dn·ft veIoClty
· =- -
e&r, (7.3)
m
r is called the rekzxation time and is the time interval in which there is unit probability
of a collision.
From eqn (7.2)
ne 2 &r
J=-- (7.4)
m
so that the conductivity
When pure, metals have a low resistivity ranging from 1·5 x 10-8 Urn to 150 x 10-8 Urn
at room temperature. The variation of resistivity with temperature of a typical metal
eu is shown in Fig. 7.1. It will be observed that the resistivity increases linearly with
temperature over a wide range of temperature.
The essential feature which distinguishes metallic behaviour is that the d.c. electrical
conductivity tends to a constant value as absolute temperature T ~ O. That is, there are
no temperature induced effects necessary for metallic conduction.
About 80 per cent of all known elements are metals and there is some variation in
resistivity with position in the Periodic Table. Table 7.1 gives some typical values.
126 PHYSICAL PROPERTIES OF MATERIALS
10
E
a
GO
I
o
.-
~
~
'>
:;; 5
'iii
CD
II:
0
Temperature (K)
Fig. 7.1 Resistivity of copper as a function of temperature.
TABLE 7.1
Electrical resistivities of metals
Na IA 4'75
K IA 7'19
Ag IB 1'61
Au IB 2-20
Mg IIA 4'30
Ca IIA 3'6
Ba IIA 3"9
Zn lIB 5'92
Cd lIB 7'27
A1 IIIB 2'74
In I1IB 8'75
Sn IVB 11'0
Ph IVB 21'0
As VB 29'0
Sb VB 41'3
Cr transition metal 12'9
Fe transition metal 9'8
Co transition metal 5-8
Ni transition metal 7'0
Mn transition metal 14'4
La rare earth metal 79'0
Ce rare earth metal 81'0
Eu rare earth metal 89'0
Gel rare earth metal 134'0
Ho rare earth metal 77'7
ELECTRICAL PROPERTIES 127
The lowest resistivities (at room temperature) are generally found in Group I and
the highest among the rare earth metals. The main experimental facts relating to the
resistivities of metals which have to be explained are:
For pure metals
(1) The range and variation, particularly the high values for transition metals and
the agreement with Ohm's law.
(2) The temperature dependence. Over a wide range of temperature near room
temperature P ex T but at low temperature P ex Tn where n '" 3 to 5 and there-
fore P -+ 0 as T-+ O.
(3) The pressure dependence. Resistivity decreases as pressure increases and for
the alkali metals eventually passes through a minimum.
(4) The Wiedemann-Franz law, which states that the ratio of thermal conductivity
K to electrical conductivity a is constant, and the modification of Lorenz that
K/aT is constant (Lorenz number).
(5) For impure metals or dilute alloys, Matthiessen's rule, that the resistivity change
due to a small amount of impurity in solid solution is independent of tempera-
ture. This rule is more generally interpeted by the equation
Ptotai = Ppure + Pimpurity
where Ppure is temperature dependent and Pimpurity is not.
Theories of metals have developed from the early work of Drude and later the
Sommerfeld model to the more recent applications of wave mechanics.
The Wiedemann-Franz law may be deduced from an extension of the Drude theory
of section 7.1 using kinetic theory.
For a particle of thermal velocity u and mean free path I
2T = llu.
Note that the kinetic velocity u is not the same as the drift velocity v. The former
arises from the thermal energy per molecule (or electron) whereas the latter is an
average velocity imposed by an electric field and is superimposed vectorially on u.
A typical value of u at room temperature is 4 x 105 m s -1 whereas v is 0 f the
order of 10- 3 m S-1 in a metal in unit electric field. The relaxation time T is typically
10- 14 s.
The thermal conductivity K of a free electron gas is [2]
K = !nklu, where k is Boltzmann's constant
and from eqn (7.5)
a = ne 2 l/2mu.
Since
(7.7)
i.e.
mu = 3kTIu
ne 2 /u
a=--
6kT'
K nklu6kT k2 T
-= =3-
a 2ne 1u
2 e2 •
128 PHYSICAL PROPERTIES OF MATERIALS
The constant of proportionality 3(k/e)2 is known as the Lorenz number and has
the value 2·23 x to- 8wn deg- 2 • It is a good estimate for experimental values of K/oT
for many metals at room temperature.
Equation (7.7) also predicts the temperature dependence of a for, substituting in
eqn (7.5),
a ex: 7 ex: l/u ex: I/VT; P ex:VT.
This prediction is incorrect since experiment shows that P ex: T. The error lies in the
assumption that the temperature dependence of P arises from the temperature depend-
ence of the electron velocities. A wave mechanical treatment shows that the mean
velocity of electrons in a metal is independent of temperature. The temperature
dependence of the resistivity arises from the scattering effect of the thermal vibrations
of the atoms on the electrons (as waves). The mean free path [is then inversely pro-
portional to the mean square of the amplitude of atomic vibrationsA o.
[ex: I/A 0 2
since
A 0 2 ex: mean potential energy ex: kT; P ex: [-I ex: T.
In the Sommerfeld model (Appendix I) a free electron has a velocity appropriate
to the Fermi level, much greater than kT at room temperature. This leads to a much
greater estimate of the mean free path which can now be several hundred interatomic
distances. The more recent developments of band theory have led to more precise but
complicated estimates of the relaxation time. It is necessary not only to take into
account the details of particular band structures but also the effect of lattice vibrations
on the electron scattering. This type of scattering leads to a well established r5 law
for p at low temperatures.
The variation of band structure explains why p varies with position in the Periodic
Table but the transition metals pose a particularly difficult problem because of the
presence of unfIlled inner d shells. The influence of these d bands on the scattering of
s electrons is thOUght to be the cause of the increased resistivity observed for these
metals.
7.2.1 Impurities
The presence of impurity atoms, for example in dilute alloys, introduces additional
scattering effects which are not easy to quantify. If scattering were of the Rutherford
nuclear type then the effect would be proportional to the square of the excess charge
of the impurity atom. The influence of impurities may be separated from normal
scattering by assigning two different relaxation times, 70 for impurity scattering and 7
for normal scattering.
Since
p=--ml
ne 2 7
p=~(l+~)
ne 70 7 2
E
c:
GO
I
o
~
.~ 4
>
'';::;
III
'iii
Q)
II:
Temperature IK)
Fig. 7.2 Resistivity of copper alloys as a function of temperature illustrating Matthiessen's
rule [3,4).
4
Atomic per cent
Fig. 7.3 Resistivity of binary alloys of copper as a function of concentrations of the second
component, illustrating Nordheim's rule [3, 4 J.
130 PHYSICAL PROPERTIES OF MATERIALS
When alloys are no longer dilute,disorder, or defects or both can cause complicated
behaviour. For transition metal alloys too the simple rules no longer apply. Table 7.2
gives the resistivities of typical alloys.
TABLE 7.2
Electrical resistivities of alloys (s.&,nm)
7.2.2 Pressure
At room temperature the general behaviour of p is to decrease initially with increasing
pressure and it may pass through a minimum for example for Na and K. The initial
decrease is due to the effect of pressure in reducing the amplitude of lattice vibrations
(p ex: mean square amplit~de of vibration, section 7.2). The subsequent increase is
probably due to modification of the electron band structure which leads to increased
phonon scattering.
Conduction
band (empty)
1
Electron
energy
Valence
band (full)
where nand p are the electron and hole densities (number per unit volume) fJe and
fJh their respective mobilities, that is velocity in unit electric field; and e the electronic
charge. In an intrinsic semiconductor the concentration of holes is equal to the con-
centration of electrons since each electron excited into the conduction band creates a
hole in the valence band. The number n depends on the temperature, the energy gap,
the density of states and the position of the Fermi level. In an intrinsic semiconductor
the Fermi level lies half way between the top of the valence band and the bottom of
the conduction band. The formula for the electron density is [5]
where m* is the electron effective mass and Eg is the energy gap width.
Clearly the presence of the energy gap requires carriers to be thermally activated
across it as expressed by the exponential term in eqn (7.8). Thus the number of carriers
and hence the conductivity increases rapidly with temperature in contrast to metallic
behaviour. Just as for metals, lattice scattering has an important effect on the motion
of carriers and for this reason the mobility is found to be dependent on temperature
according to the relation
fJ = CT-(3/2) (C = constant). (7.9)
Impurities are introduced by substituting them for atoms of the host intrinsic semi-
conductor. This may be done either by diffusing in foreign atoms, or by introducing
them during crystal growth. If, as is usual, the valency of an impurity differs from the
host, then it may have one more or one less electron to participate in bonding. If one
more, the excess electron is relatively weakly bound to the parent atom, because
polarization effects screen it from the normal binding force of its parent. This effect is
represented by the relative permittivity of the medium. The binding energy of the carrier
is reduced by the factor 1/(€r€o)2 (see Chapter 8).
The ionization energy thus becomes extremely small; a typical value of €r is 12 for
germanium and the ionization energy for an impurity atom of arsenic is 0·01 eV. This
atom is therefore ionized at quite low temperatures and at room temperature most
arsenic atoms present as impurities would be ionized.
The band picture of such an extrinsic semiconductor looks as if extra energy levels
have been created very near the bottom of the conduction band.
132 PHYSICAL PROPERTIES OF MATERIALS
Figure 7.sa illustrates the band structure at T = 0 when the arsenic atoms provide
extra energy levels which are occupied by electrons, the complete entity (the arsenic
atom) being neutral. At room temperature electrons are ionized into the conduction
band, one from each atom, leaving behind a single positively charged arsenic ion in each
case. The electrons in the conduction band are now free to take part in electrical con-
ductivity when a field is applied.
Electronl
Energy
Conduction
band (empty)
-0.01 ------
ev '....Impurity levels
(neutral)
~ Valence band
(a)
Electronl Conduction
energy e- e- e- e- e- e- band
1111!ol
Positively charged
impurity levels
""~>"'7"':7"'7"'l7"'7"'l'7"7"7"?""7 Valence
~band
(b)
Fig.7.S Band structure of impure (N·type) semiconductor. (a) At T= 0 K. (b) At room
temperature.
If the impurity atom bond is deficient in one electron similar arguments show that
an extra level is provided in the band gap, near the top of the valence band. Such levels
ionize by capturing electrons from the valence band creating holes in that band. These
holes can then contribute to electrical conductivity. Semiconductors which have more
electrons than holes are called N-type and those having more holes than electrons are
called P-type. The balance can be adjusted by control of the impurity content of the
host semiconductor. The impurities of Fig. 7.5 are called donors and those of Fig. 7.6
acceptors.
There is a simple rule which describes the statistical balance between electrons and
holes in extrinsic semiconductors at any given temperature:
If ni =electron or hole concentration in an intrinsic semiconductor at a given
temperature,
n =electron concentration in an extrinsic semiconductor and
p =hole concentration in an extrinsic semiconductor
np = ni 2 • (7.10)
The position of the Fermi level is modified by the presence of impurities. Fig. 7.7
illustrates this for both N- and P-type.
The temperature dependence of electrical conductivity of an extrinsic semi-
conductor will be determined primarily by the variation in the number of holes and
ELECTRICAL PROPERTIES 133
Conduction Conduction
band band
Fermi Fermi
level level
Bottom of C.B.
Nimpurity
r----___
levels
N-type impurity
Intrinsic Intrinsic
level level
P-type impurity P-type impurity
- - - - - - - - - - -Pimpurity
Top 0 f V.B. levels
o o
Impurity density Impurity density
(a) (b)
Fig. 7.7 Fermi level as a function of impurity density for (a) low temperature and (b) high
temperature.
electrons and to a small extent by a mobility factor. At low temperatures carriers from
impurities only will contribute, but at higher temperatures when all the impurities are
ionized, intrinsic carriers will increase rapidly in number. As well as the temperature
dependence of the mobility caused by lattice scattering, the scattering effect of im·
purities leads to a characteristic ]'3/2 dependence. Thus lattice scattering predominates
at low temperatures and impurity scattering at high temperatures. Table 7.3 lists im·
purities which may be used in germanium and silicon to create extrinsic semiconductors.
The classic semiconductors germanium and silicon occur in Group IV of the Periodic
TABLE 7.3
Donor and acceptor impurities in Ge and Si
and their ionization energies
P N 0'12 0'045
As N 0'0127 0'05
Sb N 0'010 0'039
B P 0'0104 0'045
AI P 0'0102 0'06
Ga P 0'0108 0'07
In P 0'0112 0'16
134 PHYSICAL PROPERTIES OF MATERIALS
Table. They are covalently bonded materials with energy gaps of 0·78 and 1·2 eV respec-
tively. Carbon which precedes silicon in the table, is a large gap semiconductor when
in the form of diamond, while tin which follows germanium is metallic in one phase
and semiconducting in another.
Consider indium antimonide, InSb. Indium and antimony are in the same row of
the Periodic Table as germanium. Indium is in Group III and antimony in Group V and
so have the same electronic core of inner shells. Indium has three outer electrons and
antimony five. The structure of indium antimonide is an f.c.c. lattice with A atoms at
the corners, and B atoms one quarter way along a diagonal, belonging to an inter-
penetrating f.c.c. lattice. Indium occupies all A sites and antimony all B sites. The
structure is essentially the same as diamond or germanium but there is some ionic
character to it. There are some residual charges on the Sb not quite neutralized by the
average of four electrons round it and there is a corresponding residual negative charge
on the indium. As a result, the charge clouds shrink giving the effect pictured in
Fig. 7.8.
~Sb~
I
~
~
Fig. 7.8 Polarization of electron charge cloud in InSb.
This structure is known as zinc blende (ZnS) or sphalerite. A related structure is that
of wurtzite (ZnS) which can be obtained by rotating alternate (111) layers of zinc
blende through 180 0 to give a structure with hexagonal symmetry. Both zinc bIen de and
wurtzite structures are closely related to the more densely packed sodium chloride
structure.
Various combinations of elements from Group III and Group V form important
semiconducting compounds. Similarly combinations of Group II and Group VI elements
may be formed and these have increased ionic character. In ZnS for example the sulphur
ion core S6+ exerts a strong electrostatic attraction on neighbouring electrons. Other
examples are combinations of Group N and Group VI elements and silicon carbide
is a Group IV-N compound.
All these possibilities provide a wealth of semiconducting compounds with widely
varying properties each capable of being doped with impurities to extend even further
the diversity of the range. Some examples of these materials will now be considered.
mement o S Se Te
Element N P As Sb
TABLE 7.6
Properties 0[[[/- V compounds
Table 7.7 gives some effective masses of electrons and holes which have been mea-
sured but all these figures are subject to variation with the conditions of measurement
and should be regarded as approximate.
TABLE 7.7
Ratio of effective masses of electrons me
and holes mh to free electron mass for
lll- V compounds
me/mo mh/mO
Plate 1 A molten zone between a single crystal seed and a sintered rod of nickel oxide
(T - 2300 K).
PHYSICAL PROPERTIES OF MATERIALS
Plate 2 Epitaxial growth surface of nickel oxide showing <100> cleavage cracks (x 95).
NiO---
-Interface
MgO---
- Cleavage crack
in MgO
Plate 3 Cross section of 100 I'm epitaxial layer ef NiO on MgO showing slip lines in the MgO
near the in terface (x68).
PHYSICAL PROPERTIES OF MATERIALS
Plate 4 An X-ray back reflection photograph (using monochromatic radiation) of rolled nickel
sheet showing preferred orientation.
PHYSICAL PROPERTIES OF MATERIALS
Plate 6 Uniform regions of antiferromagnetic domains in nickel oxide separated by irregular boundaries (x 155).
PHYSICAL PROPERTIES OF MATERIALS
Plate 7 Topographs illustrating the variation with temperature of the misorientation associated
with domain walls in nickel oxide (x35).
PHYSICAL PROPERTIES OF MATERIALS
Light
ttt
----
R.F. in ----
R.F. out
~Quartz
transducers
Pulsed electric
field
Fig. 7.9 Experimental arrangement for the observation of acoustic amplification.
Acoustic waves are developed in the sample by exciting a quartz transducer with radio
frequency waves and are observed after passage through the sample through their effect
on a second transducer. The sample is illuminated to provide plenty of carriers by
excitation and a pulsed drift field is also applied to the sample.
Figure 7.10 shows typical results in which acoustic gain is observed for drift fields
greater than about 7 x 104 V m- I . This effect offers interesting possibilities for future
developments which may include acousto-optical devices, oscillators and amplifiers.
+
N 25
J:
::i!:
It) 20
.... 15
~
..c 10
'0
CI>
> 5 -
co
;:
'0
0
C
:::J
0
-5 --
Ul
'0 -10
c
0-15
:;:;
co
~ -20
!
-< 16
-
c:
......
CD
::I
U
I
I
I
I ~B
Electric field
Fig.7.11 Current density versus electric field for GaAs showing the influence of variation of
electron mobility with electric field.
Suppose this is the case for the electric field region 8.A to ~. Because the sample
has a negative resistance, the situation becomes unstable as far as the electrons in that
region are concerned. Some have high mobilities and some travel slowly. The result
is that the impurities in part of the specimen become uncompensated, the electrons
having moved on too quickly and a dipole region is formed (Fig. 7.12).
eee+++ - e
eee+++ - e
eee+++ - e
eee+++ - e
Slow Fast
electrons electrons
Ga As crystal
Residual
uncom pen sated
positive charge
from impurities
CD
~ Or-~~--~--~--------
IV
.s::.
U
Distance
Fig. 7.12 Formation of a dipole domain in GaAs.
This dipole 'domain' which is a region of relatively high resistance travels down the
specimen towards the anode. Once there it collapses and a new one is formed at the
cathode. During the period of travel of a domain, the specimen current is low, when
the domain has collapsed it is high. So the repeated motion of a domain along the
ELECTRICAL PROPERTIES 139
specimen causes current oscillations. The specimen is very thin about 30 Jim and the
velocity of a domain about 10 5 ms- I giving rise to a signal frequency of 10 9 Hz. The
device is thus a compact microwave source with a particularly economical power supply.
TABLE 7.8
Properties of IV- VI semiconductors
7.3.5 V- VI compounds
These are the sulphides, selenides and tellurides of antimony, bismuth and arsenic
(Table 7.9). Interest in these compounds has been stimulated by the properties of
Bh Te3 as a material for thermoelectric generators and refrigerators.
TABLE 7.9
Properties of V- VI compounds
Nickel oxide is a classic example of a material which according to simple band theory
should be a metal since it would have a partially filled 3d band (Fig. 7.13).
However, pure NiO has a conductivity of about 10- 10 n -1 m -1 at room tempera-
ture varying exponentially with temperature as in a semiconductor. It seems likely that
the band structure of the 3d levels of nickel oxide and similar compounds is not
determined by crystal structure, as in ordinary semiconductors, but by what is known
as correlation. An electron wishing to leave a Nj2+ site must form Ni3+ and move to
another Ni2+ converting it to Ni+. The Coulomb repulsion between the 'mobile' electron
and the two electrons already attached to the Ni 2 + site render this process very unlikely.
There is a reluctance for such electrons to move because of an effective gap in energy
between the two states equivalent to the energy of formation of Ni+ and Ni3+ sites from
two Ni 2+ sites. Moreover, the mobility of the carriers is so low that they have time to
polarize the lattice ions around them. This further inhibits their mobility which may be as
low as 0·5 x 10-4 m 2 V-I s-I.Carriers can be created more easily by doping with lithium
which is readily ionized forming a Li+ ion and a Ni 3 + ion. One lithium atom thus
effectively creates a 'hole' on a Ni3+ which can now move about by accepting an electron
from another Ni 2 +. The ease with which transition elements can be induced to change
their valency has led to the name mixed valence compounds, and the presence of
carriers is often due to non-stoichiometry of the pure compounds, the presence of
nickel vacancies in NiO for example.
Figure 7.14 shows conductivity-temperature curves for nickel oxide, some com-
pounds of vanadium and nickel sulphide.
7.5 Polarons
The idea that an electron can polarize the lattice around it is an important one in dis-
cussing the mechanisms of electrical conductivity of low mobility materials and so this
concept will be considered briefly.
If a slow electron polarizes the ions around it, another electron distance r from it
has a potential energy due to electrostatic interaction given by
VI = -(e 2 /47r€o€sr),
where €s is the 'static' or low frequency relative permittivity of the material. If, however,
Fig.7.14 (a) Electrical conductivity of NiO as a function of I/Temperature. (b) Electrical con-
ductivity of NiS as a function of I/Temperature [17]. (c) Electrical resistivity of V0 2 as a function
of Temperature [18]. (d) Electrical conductivity OfV 2 0 3 as a function of I/Temperature [19].
ELECTRICAL PROPERTIES 141
10
1Ii' -
E
c:: 10- 2
1 0- 8 '-::---='-=---='"-:----=-"=---='"=:-----='"=- ~ __ l_ I I
3.5 2.5 2.3 21 19 17 1.5 1.3
la)
(Temperature)-1Ixl0-3 K-l)
1- E
E 106 c:: 10- 3
c:: ~
.~ 104 :~
...u
.~
<II
·iii
G>
10-4
cr:
'"
"C
c:
0
(J 103 10-5
10- 6
30
Ib) Ic) Temperature 1°C)
108
(d) I 104
E
-c::.~ 10 2
V2~
.~
u
-6c 10- 2
0
(J 10-4
10-80 1 4 8
(Temperature)-l (x1 0- 3 K-l)
142 PHYSICAL PROPERTIES OF MATERIALS
the ions had not been induced to move, the potential energy would be given by
V 2 = -(e 2 /41r€o€oor),
where Eoo is the high frequency relative permittivity. In this case the ions have been
effectively prevented from moving, for example when an applied electric field
oscillates at too fast a rate for the ions to be able to follow.
The difference in these terms represents the effect of motion of the ions and gives
a potential
(7.11)
where 1/ET = 1/Eoo - liEs, which traps the electron in the potential well it has created
for itself. The shape of the polarization well is shown in Fig. 7.15 where the curve
essentially represents eqn (7.11). Over a range rw within the well the equation does
not apply and V T is assumed constant.
ttl
:;::;
c:
....o
Q)
r
0.
Cl
c:
'0.
0.
.::
ttl
The electron can occupy one of a number of quantized states in the well. This
entity (electron + lattice polarization) is known as a polaron. Another way oflooking
at it is to consider the polarization as a particular set oflattice vibrations (optical
modes) induced by the presence of the electron. Thus the trapped electron may be
thought of as being surrounded by a cloud of phonons, a rather more picturesque
description. The behaviour of a polaron depends on a number of factors. One is the size
of the well compared to the interatomic spacing. If the polarization is very localized the
polaron is called small and usually moves by thermal activation. That is, if thermal vibra-
tions produce by chance a fluctuation in the lattice nearly similar to the distortion the
electron is already in, it can tunnel wave mechanically to the new position. But an
activation of the form e- WjkT has been necessary. Conduction occurring by this
process through random diffusion of the electron is called thermally activated hopping.
It is characterized by an increase of mobility with temperature. It has also been shown
that a polaron can move in a band of levels without activation when the radius of the
polarization well is large compared to an atomic spacing, and sometimes when it is not.
However, the motion of polarons is always characterized by a large effective mass and
low mobility. The electrical properties of a number of transition metal oxide insulators
has been interpreted in terms of the motion of polarons [20] .
The transition metal oxides which are semiconductors, may also be grouped in a wider
context with other compounds containing magnetic ions of the transition or rare earth
ELECTRICAL PROPERTIES 143
series as magnetic semiconductors. The magnetic moments of the individual magnetic
ions are arranged in a regular way on the crystal lattice to give ferro, ferri or antiferro-
magnetic arrangements (see Chapter 9). NiO, MnO and CoO, below a particular tempera-
ture, are examples of antiferromagnets, the ferrites (ferrimagnetic spinels, Mlt-OFe~+03)
and garnets (Y 3FeS012 yttrium iron garnet) are ferrirnagnetic, and the chromium
spinel cadmium chromium selenide CdCr2Se4 and europium chalcogenides (EuO, EuS,
EuSe) are ferromagnetic. An interesting group of compounds is the rare earth pnictides,
i.e. compounds of a rare earth (3+ ion) with N, P, As, Sb or Bi. These are III-V com-
pounds with an NaCI structure. Electrically they range from metallic to insulating
materials and their behaviour is not yet explained. Another group of ferromagnetic
compounds possesses a structure based on the mineral perovskite, CaTi0 3 and are
called perovskites, and have the general formula AB0 3 . They are mixed valence
compounds and typical ones are rare earth mixed oxides such as lanthanum manganate,
lanthanum chromate and the alkaline earth mixed oxides calcium, barium and strontium
manganates and titanates. The actual structures are often distorted forms of the cubic
perovskite structure. Those, having the formula AFe03, are called orthoferrites. If these
E
.,a
...
I
o
8=0
.!5.
x=O
~
'>'';::
UI
8=0
·~1.4
a:
8=1.3 Wbm- 2 1.2
x=0.01 PRT=18x10- 2
200 300
Temperature (K) Temperature (K)
Fig.7.16 Temperature dependence of the resistivity of poly crystalline EUI_XGdxSe for various
Gd concentrations x and magnetic flux densities B [23).
TABLE 7.10
NiO antiferromagnetic
Fe203 antiferromagnetic
MOF~03 ferrimagnetic
CdCr2Se4 ferromagnetic
EuO ferromagnetic
EuSe ferromagnetic
LaMn03 ferromagnetic (in distorted form at low temperatures)
SrMn03 antiferromagnetic (at low temperatures)
MnCr2S4 ferrimagnetic
(Note: M =Fe, Zn, Ba, Mg, Ni)
Critical
point P
300
g
...::::JCD Metal
...to
~
Q)
0.
E
~ 100
o
Pressure (kbar)
Fig.7.17 Phase diagram of (V l_xCrxh03 for x = 0.0375 [24].
t t t t t
(a)
t A t t t t
(b)
Fig.7.18 Motion of a localized electron in an array of atoms with one electron per atom. (a)
Ground state. (b) Excited state.
Now the lowest bound state of the hydrogen atom has a radius
(7.13)
Using this formula here
(7.14)
But if 1< aH there will be no bound state for the electron while if I > aH it can be pre-
vented from leaving the parent atom. Thus a transition from the state when all the
electrons are bound to the state when they are all free will occur when
l=aH.
Substituting in eqn (7.14) the condition for the transition is n- 1I3 ~ 4aH, i.e.
n 1l3 aH ~ 0'25. (7.15)
It is predicted by this simple argument that if the atomic spacing is greater than
4 Bohr units of radius, then the preferred state will be insulating. If the atomic spacing
could be varied a transition would be observed from insulator to metal at a value of
4aH . This prediction was made long before any such behaviour was observed. The first
146 PHYSICAL PROPERTIES OF MATERIALS
experimental results which suggested that the transition might take place were ob-
tained with heavily doped germanium at low temperatures. More recent results on the
transition metal compounds have led to further development of the subject of metal-
non-metal transitions [25,26]
......
... -... _______Nc_rv_stalline
......
-E
I
\
8 \
\
2 P
a \
Fig. 7.19 Temperature dependence of the electrical conductivity of InSb. (1) doped 1 per cent
Te (N-type). (2) doped 1 per cent Zn (P-type) in crystalline and amorphous states (29).
fact that such a well known amorphous material as glass is optically transparent sug-
gests the existence of an energy gap in the available states. Band theory suggests that
energy gaps arise from the regularity of crystal atomic planes. How then c~ such a gap
exist in the energy states of an amorphous material? This contradi~tion is a point of
great theoretical interest and it has been demonstrated that a 'disordered' potential is
capable of having a band structure which is rather different from the one that has been
used so far.
The main features of the electrical properties of amorphous materials have been
found by examining the properties of thin amorphous layers of simple materials
such as selenium, germanium and silicon. In this form these materials behave as
semiconductors with a conductivity of the form
U = Uo exp (-f:lE/kT). (7.16)
There were significant differences in some cases between the behaviour of amorphous
and crystalline materials. Fig. 7_19 shows the electrical conductivity for InSb doped
N and P in both states.
ELECTRICAL PROPERTIES 147
The differences in the two types of the same material are obvious. In the amorphous
state the conductivity is much lower and there is very little difference between the
behaviour of Nand P material. The activation energy AE (eqn (7.16)) is 0·55 eV for
the straight line part abo This would correspond to an energy gap about 0·4 eV greater
than that of the crystalline material. This energy gap can be modified by thermal
treatment of the sample. The behaviour of thermoelectric power emphasizes the
difference. The extrinsic nature of the N- or P-type conductivity completely disappears
in the amorphous state (Fig. 7.20) and it seems then that the effect of impurities is
0.2
2
-----
I
:..::
> 0.1 Crystalline
.s
Gi o-
~
o
~ -0.1
';:
1:)
CD -0.2
Qi
o
E
... -0.3
CD
.r::
~
-0.4
-0.5 '--2:':----!4:------:!:6----:!:S:------
(Temperature)-1( xl 0- 3 K- 1)
Fig.7.20 Thermoelectric poweroflnSb (1) N· and (2)P-type in crystalline and amorphous states
(29).
suppressed. In some materials direct measurements of drift mobility have been made and
found to be as low as 10- 7 m2 V-I S-I for electrons and 10- 5 m 2 V-I S-I for holes. Attempts
to explain the differences in the two states have led to a model which, although incomplete,
is successful in explaining some of the main features of the experimental results. It is
interesting because of the number of new ideas it contains. The basic consideration is
the effect of disorder on the normal band structure of a crystalline material. This
can be investigated theoretically by seeking solutions of the Schr6dinger equation
when the potential is disordered. A simplified approach is to consider the motion of
an electron through a series of square potential wells of random depth. The results
show that bands of energy states exist just as for crystalline materials and that these
bands are separated by energy gaps. At the edges of these bands, however, localized
states are present which spread into the forbidden gap. The extent of this spread
depends on the type of material. Where compositional disorder is present as well
as structural disorder, the spread may be so great that overlap with localized states
from an adjacent band edge may occur. Then the crystalline energy gap is either severely
blurred or non existent (Fig. 7.21a and b).
The density of states within a simple band of a crystalline material is modified as
shown in Fig. 7.22. Tails oflocalized states extend the spread in energy beyond the
crystalline band width E B - EA'
Some explanation is necessary for the distinction between the states in the band and
the localized states, as both have rather special meanings here. A localized state is an
148 PHYSICAL PROPERTIES OF MATERIALS
Conduction
band
:-::-=-=:-=-::-:-::-=-::-=-:-::~:-:-::-:~
Valence
-~------------~~~
Valence ~----------~-----~ states
band band
(a) (b)
Fig. 7.21 Localized states in band structure of amorphous materials. (a) Blurring of band edge.
(b) Overlap of localized states in band gap.
UI
Q)
+'
IU
+'
UI
....o
>-
+'
"iii
c
Q)
o
Energy E
electron trap from which an electron can only escape with the help of phonons. This
implies a mobility of the form /J. = /J.o exp (-E/kT) where E is an activation energy and
the motion is called hopping. The band states in this problem are also rather special
because an electron moves in a diffusive manner in them leading to a mobility which
is not activated. If the band gap is blurred by the presence of localized states, how can
a definite activation energy be observed in experimental measurement of electrical con-
ductivity? This difficulty is overcome by assuming a very sharp drop in carrier mobility
at the band edge of extended states, where the transition occurs to motion in localized
states in which the mobility is very small. Fig. 7.23 shows the expected variation in
mobility.
This sudden drop in mobility is said to derive from the existence of a mobility gap,
and it is the activation in mobility across the mobility gap which gives the form a =
ao exp (-AE/kT) because /J. = /J.o exp (-AE/kT) where AE is the mobility gap.
The mobility varies in the following way near the band edges EBand Ee. Within
the band of states well below EB the mean free path will be low compared to the
crystalline case because of disorder, and /J. is typically 10- 2 m2 y-1 S-l. The mobility may
also be modified by trapping. In the region aboveEB' that is in the localized states, the
mobility is very low, about 10- 5 m 2 y-1 S-l . But immediately below EB in the region
ELECTRICAL PROPERTIES 149
t - - - - Mobility gap-
Band Band
Localized
states
EnergyE
Fig. 7.23 Mobility of electrons as a function of energy showing mobility gap (amorphous state).
EaEB, the motion is diffusive and the mobility is an order of magnitude higher than
in localized states in the tail of the band. This region is a kind of transition region from
extended states to localized states and the motion here is a Brownian movement. These
ideas on mobility result from the mathematical treatment of motion in random
potentials [30] and are rather speculative at present.
The insensitivity of some properties of disordered materials to impurities is due
to the fact that local valency requirements can always be satisfied. For example an
arsenic atom in crystalline germanium disturbs the bonding to fit into the lattice, but
in amorphous germanium it can always be expected to fmd attachments for its five
bonds because the situation is so much more flexible (Fig. 7.24).
___ r Ge
Ge
~ Ge Ge
Ge
(a) (b)
Fig.7.24 Bonding to an impurity atom in (a) crystalline and (b) amorphous germanium.
Certain glassy semiconductors have shown interesting electrical effects involving sudden
changes of resistance. These are generally classed as switching properties and typical
materials which show them are the chalcogenide glasses, which are glasses containing
elements of Groups III, IV, V and VI of the Periodic Table. For example Ge, As, S
and Te are combined to form Teo.sAso·2 SiO'l GeO·2 or ASo.4TeO.SGeo.l' When two
150 PHYSICAL PROPERTIES OF MATERIALS
electrical contacts are made to a thin fIlm of such materials, a two terminal device is
formed which may show either of the following current-voltage characteristics
(Fig. 7.25).
\
\
....c ....c \
,,
...... ...l!!
,
CD
::J \
::J
U U
Voltage Voltage
(a) (b)
Fig. 7.25 Current-voltage characteristic of (a) threshold switch and (b) memory switch.
1/1
.!
...~o
~
1/1
c::
Q)
C
Energy
Fig.7.26 Density of states as a function of energy showing the overlap of states in the energy gap
(amorphous state).
Electrons from the tail of the valence band fall into empty conduction band levels
to produce a set of positive and negative charges above and below the centre of the
mobility gap. In this situation the Fermi level EF is effectively pinned at the centre
of the gap. The mechanism of switching is then based on the idea that if space charge
neutrality prevails in the region, and all possible traps are full, the bulk of the material
is capable in prinCiple of carrying a high current even though nominally an insulator.
The details of the actual mechanism of switching is otherwise highly speculative at
present.
References
Questions
8.1 Introduction
Electrical insulators have very few free electrons to take part in normal electrical con-
ductivity and so it is possible to conceive the ideal insulator which would have no free
electrons. Such a material has interesting electrical properties because of the ability of
an electric field to polarize the material to create electric dipoles. A dipole is an entity
in which equal positive and negative charges are separated by a small distance, the
electric dipole moment p being dermed as p = qr in the simple representation of
Fig. 8.1.
+
++++++++
•
d
t
Fig. 8.2 Capacitor with yacuum dielectric.
the plates is directed as shown and has a magnitude V/d V m- 1 arising from the charge
density ± Q on the plates. The relation between Q and the field & is obtained by
introducing a dimensional factor EO, the permittivity of vacuum:
(8.1)
154 PHYSICAL PROPERTIES OF MATERIALS
The value of Eo is 8·854 X 10- 12 Farad/metre. It is convenient to think of Q as a source
of electric flux lines in the space between the plates; the density of these flux lines is
called the electric displacement D.
D=Q=Eo8,· (8.2)
The need for D as well as Q should become clear in due course.
Now imagine that with the battery still connected a dielectric medium is introduced
so as to just ftll the space between the plates (Fig. 8.3). The medium becomes polarized
+ + + + + + + + +
vI I~ + + +
-
+
- +-
- -
+
-
-+ +-
-
1 I~
+
-±
+
-
±
-
+
-
-
+
-
±
-
+
-
±
-
+
-
±
-
+
-
± ;1
Fig. 8.3 Capacitor with a solid dielectric.
by the field 8, and dipoles appear throughout the material lined up in the direction of
the field. All dipole ends of opposite charge inside the material will cancel but there
will be an uncompensated surface charge, negative at the top, and positive at the
bottom. These surface charges will attract and hold corresponding charges of opposite
sign on the plates, because the latter, unlike dipoles, are able to move freely. The field
between the plates must be that appropriate to the voltage applied, irrespective of
whether a dielectric is present or not. Thus the field in the dielectric must still be 8,.
If the effect of some of the original surface charges has been neutralized by being
bound to surface dipole ends, 8, can only be maintained if more charges flow from
the battery to make up for those which have become bound. There is now more charge
density Q' on the plates some of which is tied up and is not contributing to the field 8,
in the dielectric. The amount of charge that is contributing to the field is the same as
before, and
(8.3)
where QB is the bound charge density. Q has been multiplied by a factor Er such that
~=~a ~~
By eqn (8.2) D is now given by
(8.5)
and by eqn (8.3)
(8.6)
D =€o8+ P. (8.7)
Eqn (8.7) makes clear the need for the concept of D, it takes account of both free and
bound charge. These calculations could have been done under conditions of constant
charge instead of constant voltage and by considering the fields with and without the
dielectric, similar equations would be obtained. The ratio DII is called the permittivity
of the dielectric medium
DI8=€r€o =€.
€r is called the relative permittivity or dielectric constant of the medium. For vacuum
€r = 1 and € =eo.
Another way of looking at the above problem is to consider a field 8' in the di-
electric as being appropriate to the total charge Q' on the plates. This field is of course
different from and greater than the applied field 8 since 8' = 1+ P/€o where the
second term is due to the extra charge which flowed on to the plates whilst inserting
the dielectric. Inside the dielectric the field I' is opposed by another field, that due to
the surface charges on the dielectric itself (Fig. 8.5). This is P/€o and is opposed to the
+++++++++
Polarization
field P/Eo
+++++ ++ ++
field due to the plate charges. Thus the field in the dielectric =8' - P/€o = 8 + p/€o
- P/€o = 8 which was the result obtained previously. The presence of the dielectric
thus increases the capacity of the system for charge by the factor €r.
From eqns (8.5) and (8.7)
therefore
P=€o8(€r- I).
The ratio P/eo I is called the electric susceptibility X of the dielectric:
P/e o8=€r -1 =x, (8.8)
where X measures the ratio bound chargelfree charge =P/Q.
By utilizing the concepts developed above it is not necessary to inquire into the
detailed atomic and molecular processes giving rise to polarization in a dielectric. In
fact comparatively limited progress has been made in properly describing dielectric
phenomena on a microscopic scale. However, a number of important features of
physical behaviour can be explained using quite simple models.
156 PHYSICAL PROPERTIES OF MATERIALS
atomic dipole moment p is proportional to the applied field 8. The constant of pro-
portionality is known as the polarizability a, in this case the electronic polarizability
p= a8. (8.9)
In equilibrium, the effect of the field is balanced by the attraction between the dis-
placed charges in the atom leading to the expression for the polarizability a of the
atom
a = 41T€or 3 , (8.10)
where r is the atomic radius. For hydrogen r = 0·53 x 10-10
m and ahydrogen =
1·66 x 10-41 F m 2 , hence for a field of lOS V m-1 Phydrogen = a8. = 10-36 C.m. The
dipole length d = pie = 10- 17 m, a distance small even compared with atomic dimen-
sions.
Induced polarization can also occur in an ionic crystal because the positive ions can
move slightly in the direction of the applied field and the negative ions move in the
opposite direction until new equilibrium positions are achieved. In an electric field 8,
an ionic polarizability aj can be measured from the dipole moment p due to ionic
displacements.
The contribution of ions to the Molar polarization (see eqn (8.27)) of the alkali
halides which are typical ionic solids is given in Table 8.1. Increasing ionic size and
separation reduces the binding between ions of opposite charge and so increases the
polarization, as shown by the figures in parentheses.
TABLE 8.1
Ionic polarization as a fraction of the total polarization for alkali halides,
and (in parentheses) the ionic polarization of each compound relative to
thato/LiF/l]
F Cl Br I
TABLES.2
Molecular dipole moments
CO 0·1
H2 0 1·S7
HO 1·0S
CH 4 o
CH30 1·S7
CH3CHO 2·72
C2 HS OH 1-69
Such materials will not have a net dipole moment in the absence of a field because,
provided the molecules are free to move, thermal vibrations will ensure that the dipole
directions are random and the net moment averages to zero. In gases and liquids
individual dipole moments are usually free to rotate but in solids circumstances may
not allow it, for example if the temperature is too low or the close environment of
other molecules interferes with the motion.
In the presence of an electric field, an electric dipole behaves like a small magnet
in a magnetic field in wanting to line up with the field. This is called orientational
polarization. It is due to the torque applied by the field as shown in Fig. 8.7.
Fig. S.7 Torque on a dipole at an angle to the direction of an applied electric field.
W 40
....>-
">
~ 32
E
a;
c.
Ol 24
>
:;:;
co
Qi
a: 16
0
-90 -60 -30 0 +30
Temperature (0C)
Fig. 8.8 Relative permittivity of nitro methane (at 70 kHz) as a function of temperature [1].
Equation (8.14) is restricted in its applicability since the field acting on a dipole within
a liquid or solid is not the applied field because of the interaction between dipoles.
DIELECTRICS 159
Lorentz derived an expression for the actual field acting at a point in a dielectric known
as the local field 8 loc to distinguish it from the applied field 8 app' In a solid Lorentz
showed that
(8.15)
where P is the polarization of the medium. For most problems, only approximate
calculations of 8 loc are possible.
The need to distinguish between the applied field and the actual field acting on a
dipole within a material can be seen as follows. The field at a point within a dielectric
arises from the applied field and the cumulative effect of all the dipoles in the material
under the influence of that field. All these dipoles are interacting with each other and
with a dipole at the point in question. The exact calculation is thus extremely com-
plicated and cannot be done unless simplifying assumptions are made. In the following
calculation due to Lorentz, the main assumptions are (i) that the effect of most dipoles
at a distance from a given point 0 is given by the macroscopic polarization provided
they are some dipole lengths away from 0, and (li) the interaction of the dipoles near 0
is zero if the material has cubic symmetry about O. Fig. 8.9a shows a thin slab of di-
electric between two plates. The plates of area A have charges Q per unit area when
+QA
.
+ + + + + + + + +
~2at
+ + + +
~
+ + +
~P
+ +
I~ (~)
- -
l~b
·0
- - - -
- - - -QA
(a) (b)
Fig. 8.9 (a) Fields in a dielectric between charged plates. (b) Field in a spherical cavity about O.
the external applied field is 8 app' Let the polarization of the dielectric be P. What is
the field at the point 0 within the dielectric, that is to say the force on unit positive
charge at O?
The field arises from four parallel components
8 loc = 81 + 8 2a + 1l2b + 8 3 . (8.16)
Suppose the point 0 is surrounded by a sphere which is small compared to the size of
the dielectric slab but large compared with molecular sizes. The effect at 0 can be
divided into the effect due to the material outside the sphere treated macroscopically
and the effect due to material inside the sphere treated microscopically, that is with
the detail of dipole interactions being taken into account. Suppose the material inside
the sphere is temporarily removed. The four components of eqn (8.16) are
81 =field due to the charges on the plates;
1l2a =field due to charges on the surface of the dielectric (the depolarizing field);
8 2b = field due to charges on the surface of the spherical cavity (Fig. 8.9b); and
113 =field due to the dipole interactions of the material within the sphere when
replaced.
The effect of the plate charges is given directly by the Gauss theorem
(8.17)
160 PHYSICAL PROPERTIES OF MATERIALS
Since the surface charges on the dielectric are given by the polarization P and their
field is oppositely directed to 8 1
(8.18)
There is always a geometrical factor in this type of calculation. Here the choice of a
thin cylindrical slab of dielectric with lines of force perpendicular to the surface of the
cylinder leads to a factor of unity. For a sphere the factor turns out to be 1/3.
To calculate 8 2b consider the sphere of Fig. 8.9 in more detail (Fig. 8.10a and b).
z z
+ +
8
~--I--X
~---+--X
(a) (b)
Fig. 8.10 Calculation of field It 2b at 0 due to charges on the surface of a spherical cavity
about o.
The field at 0 in the diiection of 8 app can be found by using Coulomb's law for the
effect of the surface charges at O.
The density of induced charge on the surface of the sphere is the normal component
of polarization Pn = P cos 8.
The radial field at 0 due to charges on surface element of area dA =PdA cos f) /41Teor2.
The component of this in the direction of 8 app is PdA (cos 8 /41Teor2) x cos f).
The integration over the surface can best be carried out by using an infinitesimal
element of area obtained by rotating dA about the axis OZ (Fig. 8.1 Ob). An elemental
surface ring is then formed of area 21Tr sin 8r df). The total field 8 2b is then given by
integrating over the range 8 = 0 to 1T.
f
1r
P cos 2 f) :2.
8 2b = :2 21Tr sm 8 d8 = P/3Eo. (8.19)
41TEor
o
This then is the field at 0 due to most of the dipoles in the material expressed through
the macroscopic polarization P, excepting only those within the sphere.
Finally the field 8 3 due to the material within the sphere is often very difficult
to calculate. If the material is assumed to have cubic symmetry the effect of the dipole
interactions averages to zero. For a symmetrical array of dipoles about 0 there is
always a cancellation at 0 of the effect of dipoles at corresponding points on the
opposite sides of the centre point. Thus, if this simplifying assumption is made, then
1t3 =0. (8.20)
The magnitude of the total internal field 8 10c is then the sum of the magnitudes of all
four components:
Q P P
~loc = - - -+-. (8.21)
eo eo 3Eo
DIELECfRICS 161
But
D=Eo&app +P=Q. (8.22)
Substituting eqn (8.22) into (8.21) gives
8 10c = 8'app + (P/ 3E o). (8.23)
This may be written vectorially since the three components are parallel.
The Lorentz equation for IJ loc leads quickly to other standard results. Since
P = Eo(Er - 1)8app
(8.24)
and P = No: Bloc where 0: is the total polarizability and N is the number of molecules
per unit volume, therefore
P=(~'T + 2) No: 8' app /3. (8.2S)
Combining these,
No:/3Eo = (Er - l)/(Er + 2). (8.26)
This is known as the Clausius-Mosotti equation which relates a microscopic property 0:
with an easily measurable macroscopic property, the relative permittivity Er .
If N is replaced by the expression Nop/M, where No is Avogadro's number,M the
molecular weight in kg and p is the density in kg m- 3 the quantity Noo:/3Eo obtained
by substituting in eqn (8.26) is the molar polarization or polarizability per mole:
NoO: M eEr - 1)
P
3Eo = (Er + 2) .
(8.27)
It must be remembered that this equation is approximate since it does not properly
take account of dipole interactions.
The effect of an alternating electric field on a dielectric will now be considered with
emphasis on low frequency response. The high frequency response is dealt with in
Chapter 10. By low frequencies is meant a region from zero up to about the micro-
wave region of the e.m. spectrum in which permanent dipoles can respond to an
applied a.c. field.
Water is an example of a substance with permanent dipole moments. The dipole
moment of the water molecule arises because of the attractive power of the oxygen
atom for electronic charge, leaving the hydrogen atoms with an effective positive
charge.
Considered microscopically, either in the liquid or solid state, the response of a
polar molecule to an a.c. field will depend on the 'viscosity' of the medium. Ability
to respond to an applied field is usually expressed in terms of a 'relaxation time' T.
Wo = l/T represents the angular frequency at which the viscous forces have their
162 PHYSICAL PROPERTIES OF MATERIALS
maximum effect. Well above this frequency a particular polar molecule can no longer
keep up with the applied field and so is no longer orientated by it, and no longer con-
tributes to the permittivity of the medium. Such 'viscous' effects arise from the
influence of intermolecular forces and lattice vibrations. The idea that there is just one
such frequency at which this effect will occur, that is, a single relaxation time, is over-
simplified. Debye derived a model based on a single relaxation time which describes
the general trend of practical results.
Because of the possibility of a phase lag developing between the applied field arid
the instantaneous polarization, the relative permittivity is expressed in terms of real
and imaginary parts and€; €;',
where
, ."
€r = €r - J€r· (8.28)
As the frequency increases the displacement D changes from being entirely in phase
with the applied field tI to having components both in phase and out of phase with 8
(Fig. 8.11).
D~ out of phase
2
D
D Din phase
• •
Applied field g Applied field g'
(a) (b)
Fig. 8.11 Phase relation of vectors D and". (a) at a low frequency . (b) near the relaxation
frequency.
The Debye equations give the variation of both the in-phase and out-of-phase com-
ponents of €r as a function of angular frequency w for a polarization mechanism with
a single relaxation time r:
(8.29)
€ "- (€S - € .. ) WT •
r - 1 + w 2 r2
Here the factor € .. is the contribution to the relative permittivity at high frequencies
from electronic and ionic polarization and €s is the low frequency value of €r. Eqns
(8.29) are illustrated in Fig. 8.12.
When D varies with time, a displacement current density J flows in the dielectric
due to the fluctuating surface charges, where J =dD/dt in the absence of any ohmic
current; that is, for an ideal dielectric. Thus, referring to Fig. 8.13a, if D is in phase
with 8, J is exactly 90° out of phase.
Energy dissipation will arise in the relaxation process from Joule heating if there is
a component of J in phase with the field giving rise to a non-zero term J . 8. As long
as J is perpendicular to 8, clearly there is no energy loss; but when, as in Fig. 8.13b, D
gets out of phase with tI, J is no longer exactly perpendicular t~ 8 and so has a com-
ponent in the direction of 8.
DIELECTRICS 163
£.
r
Wo= lIT
Angular frequency
Wo
Angular frequency
Fig. 8.12 Real and imaginary parts of the relative permittivity as a function of angular fre-
quency (according to Debye-eqn (8.29».
J J J ~ out of ph.eo
.. -
2
.. >- J ~ out of phase >-
c:: .'!:: c·-
G! '"
E ii
G!'tl
G! '"
E ii
G!'tl
U .. u_
.!!!c .!!!c
.C-
'" ...:::I
e.G!
o'" :;...u
e.G!
U
D
.. Jon phase
Don phase
....
Applied field ~
Applied field g
(a) (b)
Fig. 8.13 Phase relationship of vectors J, D and I. (a) D in phase with 1 and (b) D out of phase
with I.
where w is the angular frequency ofthe field 8. If 8 = 8 0 cos wt, D = €r€O 8 0 cos (wt-
where 6 is the phase angle and J =-€r€O loW sin (wt - IS). Therefore energy dissipated
per second is
21T/W
(-W/21T) J w€o €r &0 sin (wt - 6)&0 cos wt dt.
o
Expanding sin (wt - 6) and multiplying out gives only one term which contributes to
the integral, i.e. the term in cosl wt. Therefore energy dissipated per sec:
21T/W
(w 2/21T) sin 6 J cos 2 wt €r€0&02 dt =(w/2)€r€0 sin c5 &0 2 . (8.30)
o
Figure 8.14 illustrates the relationship between the real and imaginary parts of €r and
the phase angle 6. It follows that for small /j where sin /j ...., tan 6 eqn (8.30) may be
written in the form
(w/2)e;'€0&02 a: tan 6, for small 6.
e"
Tan 8 = er
-~
t"r
er
Fig. 8.14 Relationship of real and imaginary parts of relative permittivity and loss angle 6.
Thus if J lags behind by a small angle /j, as is often the case, tan 6 is a measure of the
power loss and is known as the loss tangent given by tan /j =€;/€; (Fig. 8.14).
At higher frequencies the ionic and electronic polarizations show resonances and
these are discussed in Chapter 10. These occur respectively at approximately 10 12 and
1015 Hz, in the infra-red and ultra-violet regions of the electromagnetic spectrum. In
this region of the spectrum the relative permittivity is measured by the square of the
optical refractive index n. Table 8.3 compares the low and high frequency perrnittivities
and shows the effect of ionic polarization for alkali metal halides.
Orientational polarization occurs much more slowly, typical relaxation times for
polar molecules in liquids vary from 10- 10 to 10- 6 s, that is, in the microwave or radio
regions of the spectrum. Generally, in solids containing polar molecules, the interatomic
forces seriously restrict orientational polarization. In some materials commonly en-
countered in ceramic form, for example barium titanate, ionic polarization is common
and dipole orientation relatively unimportant. In polymers, losses can occur by the
movement of segments of large molecules and this leads to a distribution of relaxation
times.
DIELEcrRICS 165
TABLE 8.3
Compariron of the low frequency relative permittivity and the square of the optical refractive index
for alkali metal halides showing the effect of ionic polarization [1]
F Cl Br I
8.6 Applications
Dielectrics are used mainly as electrical insulators and in capacitors. Oils and wax
impregnated paper are used as insulators in transformers and steatite (a compressed
TABLE 8.4
Dielectric properties of materials
mixture of talc, clay and barium and calcium carbonates) as an insulator for high
voltage terminals. For capacitors, dielectric materials include paper, plastic film and
mica sheet; oxides of aluminium, titanium and tantalum (electrolytic capacitors) and
a wide range of ceramic oxides which are often ferroelectric as well as dielectric. Poly-
mer materials used as commercial dielectrics are polyesters known as Mylar (PET),
polystyrene and polycarbonate. Polystyrene for example is a non-polar compound
with a relative permittivity of 2·55 at 25°C and tan [) less than 2 x 10- 4 over the
range of frequencies 100 to 1010 Hz. The square of the optical refractive index is 2·53.
Table 8.4 gives typical values of permittivity and loss factor for various materials at
room temperature. Fig. 8.15 shows the frequency dependence of €r and tan [) for poly-
carbonate.
~
0.Q12
..!!!.
eo
...>
·S 3.2 c 0.01
...
.;; ~
'§ 3.1 .... O.OOS
0
Q) 1:)
c. ~ 0.006
Q) 3.0
>
.;; III
III
III
a; 2.9 .3 0.004
a:
2.S 0.002
2.7 0 L..---'-_--'-_-'----'-_-'-_-"--_J..
10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 10 10 10 10 2 10 3 10 4 10 5 10 6 10 7 10 8
(a) (b)
Fig. 8.15 Frequency dependence of (a) relative permittivity €r and (b) tan Ii for polycarbonate.
These properties will first be defined and the discussion restricted initially to single
crystal materials. Piezoelectric materials have the property of becoming electrically
polarized in response to an applied mechanical stress. This property, known as piezo-
electricity has an inverse; when an electric stress (a voltage) is applied, the material
becomes strained. The strain is directly proportional to the applied field 8. The
existence of both the direct effect and its inverse is to be distinguished from a similar
property called electro-striction. Here the material becomes strained in an electric
field 8 but in a manner depending on 8, 2 so that there is 110 inverse of this effect. The
difference between the two effects is illustrated in Fig. 8.16.
Apyroelectric material is one which exhibits a spontaneous polarization in the
absence of an electric field and which changes its polarization on heating. If the change
of polarization is M' on raising the temperature by t:..T
M'=A.t:..T, (8.31)
where A. is the pyroelectric coefficient.
Afe"oelectric material is one which exhibits a spontaneous polarization in the
absence of an electric field which may be switched in direction by the application of
a field. Ferroelectrics show a hysteresis in the polarization-field relation.
DIELECTRICS 167
+
(a) Expansion
+
A A
,-,
+
(b)
.,
.2+, Contraction Expansion
- -
+
Fig. 8.16 (a) Electrostriction. Material expands for both directions of field. (b) Piezoelectric
effect. Material contracts for one field direction and expands for reversed field.
32 Symmetry classes
/
11 Centrosymmetric
\
21 Non-centrosymmetric
120 piezoelectric)
10 Pyroelectric (polar)
/ \
11 Non-pyroelectric
/
Ferroelectric (polar)
\
Non-ferroelectric (polar)
Fig. 8.17 Symmetry classification of piezoelectric and pyroelectric materials.
168 PHYSICAL PROPERTIES OF MATERIALS
classes which can have asymmetric properties. All the materials in these twenty classes
are piezoelectric. Materials in ten of these twenty classes have a unique polar axis where
none of the symmetry operations of the crystal point group will turn this axis round.
The existence of a polar axis in a crystal allows the appearance of a spontaneous
electrical polarization and these materials are pyroelectric. A restricted group of
pyroelectrics have the further property of being ferroelectric. Thus, all ferroelectrics
are pyroelectric and piezoelectric. All pyroelectrics are piezoelectric, but the converse
is not true. All piezoelectrics are not pyroelectric and all pyroelectrics are not ferro-
electrics (Fig. 8.17).
8.8 Piezoelectricity
To recapitulate, there are two effects both of which have important practical appli-
cations:
(a) Direct effect. The application of a stress to a crystal produces a strain which
results in a net polarization.
(b) Inverse effect. The application of an electric field produces a strain whose sign
depends on field direction.
These are both linear effects.
If a stress T is applied to a material resulting in a strain S, there are simple relations
between the two involving the elastic stiffness constant c and compliance s for the
material:
T = cS or S = sT. (8.32)
For a piezoelectric material the strain S also results in a polarization and eqn (8.32)
becomes more complicated.
The stress T will produce a polarization P = dT where d is a piezoelectric strain con-
stant. The dielectric displacement, in the presence of a stress therefore contains an
extra term:
D=e8+dT. (8.33)
For the inverse effect the corresponding relation may be written
S =sT +d8. (8.34)
These relationships are complicated by the fact that most crystalline materials are
anisotropic so that electrical or mechanical forces applied in one direction may produce
effects in other directions. This situation may be represented by a tensor notation based
on three convenient axes usually mutually perpendicular. A general stress can be
resolved into six components, three tensile stresses along the axes x, y and z and three
shear stresses about the axes x, y and z.
Consider the plane abcd (Fig. 8.18). A tensile stress acting on this plane in the x
direction is written Txx (or Tu ), the first subscript denoting the direction of the
stress and the second the direction of the normal to the plane. The shear stress shown
acting on the body abcdefgh acts in the y direction around the z axis on planes whose
normals are in the x direction. It is written Tyx (or T21)' Thus in general a stress is of
the form Tij
DIELECTRICS 169
z(3)
Tyx
9
Tyx
a
I
0,x e
y(2)
x{1 )
Fig.8.18 Notation for normal and shear stress.
For a body in equilibrium Tij = T ji . The notation is often simplified by writing tensile
stresses as T 1 , T2 and T 3 , and shear stresses as
T4 = T 23 = T32
Ts = T13 = T31
T6 = T12 = T21
therefore
n
the array of numbers d ik is
2·3o -0·67 o
3 o o 0 0·67
o o 0 o
There are thus only two independent constants dl! and d l4 (d 26 = 2dl!)' As an example
suppose a tensile stress of 104 Nm- 2 is applied in the x direction. What are the
170 PHYSICAL PROPERTIES OF MATERIALS
resultant polarizations in the x, y and z directions?
PI =d l1 TI = (-2·3 x 10- 12 ) (104 ) = -2·3 X 10- 8 Cm- 2,
P2 =d 21 TI = 0,
P3 =d 3I T I =0.
For ammonium dihydrogen phosphate (ADP) the array d ik appropriate to a par-
ticular set of axes is
~
0 0 0 0·17 0
( o 0 0 0 0·17 ) X 10- 11 CN-I.
o 0 0 0 0 5·17
Remembering that the last three columns of the array represent shear around the x, y
and z directions, then if an electric field of 103 V m- I is applied in the x direction
Sk =dlk&l.
Oearly there are no linear strains since d 11 = d 12 = d 13 = 0, but there is a shear strain
about the x axis
S4 =d I4 &1 =(0·17 x 10- 11 ) 103 =0·17 x 10- 8 •
To produce a shear strain about the z axis the field must be applied in the z direction
S6 =d 36 &3·
+
o_ _ _
Transducer
--.JI
Co
R
C
N
QI
U
t:
IV
"0
QI
Q.
E
f, f.
Frequency f
Fig. 8.20 Impedance Z of the circuit of Fig. 8.19 as a function of frequency of an applied a.c.
voltage.
The maximum and minimum shown are called an antiresonance and resonance
respectively and occur at signal frequencies fa andfr. They arise from the resonance
of the mass of the transducer with its stiffness (fr) and from the resonance of the mass
with the electrical capacitance (fa). These frequencies can be measured using the
circuit of Fig. 8.21. The current in the circuit is monitored by a voltmeter across series
resistor R 1 and the frequencies at which the current is a maximum and minimum are
noted. These are the frequencies fr and fa. With reference to Fig. 8.19
Wr 1 = (27Tfr)l = 1ILC, (8.35)
(8.36)
r---------------~~~
V ~____~
J
Oscillator
~ ~
Transducer
I I
Fig. 8.21 Circuit for measurement of resonant and antiresonant frequencies of a transducer.
But
Kl = mechanical energy stored = ! CVl = _C_
total energy stored ! CVl + !CoVl C + Co
172 PHYSICAL PROPERTIES OF MATERIALS
therefore
(8.39)
=( ~ ~}
0 0 0 diS
d, • 0 0 diS 0
d 31 d 31 d 33 0 0
when the poled direction is the z axis.
Application Material
TABLE 8.6
Properties of piezoelectric materials
Quartz (X cut) 8
d 2l =2·25 0·1 550 2·65
d 33 =2·3
Rochelle salt
(45 0 X cut) d 14 =870 0·78 1·77
Barium titanate 11·8 d 3l =-75 0·48 70-90 5·6
PZT5 6·75 d 33 = 374 0·67 290 7·7
Lead
metaniobate 2·9 dD =85 0·42 500 5·8
lithium sulphate
(Y cut) dD =16 0·35 75 2·06
lithium niobate d 33 = -1 0·4 4·63
diS = 68
Ferroelectricity was discovered by Valasek in Rochelle salt in 1921. Since that time the
list of substances showing this property has continued to grow rapidly. Ferroelectric
materials are an important group not only because of the intrinsic ferroelectric
174 PHYSICAL PROPERTIES OF MATERIALS
property, but because many possess useful piezoelectric, birefringent and electro-
optical properties (Chapter 10) which can be exploited in devices. Many early ferro-
electrics and newer ones may be obtained as single crystals, but, as was the case for
piezoelectrics, ceramic ferroelectric materials are becoming readily available.
The intrinsic ferroelectric property is the possibility of reversal or change in orienta-
tion of the polarization direction by an electric field. This leads to a hysteresis in the
polarization P, electric field & relation, similar to magnetic hysteresis. Hence the name
/e"oeiectricity has arisen from the analogy with ferromagnetism. Above a critical
temperature, the Curie point Tc, the spontaneous polarization is destroyed by thermal
disorder, and the permittivity shows a characteristic peak at Tc. These properties are
illustrated in Fig. 8.22.
p
f b
c
I
/
/
________~d~____~----~--------~W
a
(a)
&, p
c
....mo
UI
.~
m
(5
Q..
To To
Temperature T Temperature T
(b) (e)
Fig. 8.22 Ferroelectric behaviour. (a) P, & hysteresis loop. (b) Relative permittivity er versus
temperature showing a maximum at the Curie temperature Te. (c) Polarization as a function of
temperature.
DIELECTRICS 175
Figure 8.22a illustrates a typical ferroelectric hysteresis loop. A direct analogy may
be drawn with B-Hloops in magnetism. For small fields applied to an initially un-
polarized specimen, the P, & relation is linear, but at higher fields the polarization
saturates. When the field is reversed P follows bc, oc being the remanent polarization
for zero field. A reverse field od is required to reduce P to zero. On further increasing
the field in the negative direction saturation again occurs at e and fmally the path eab
is traced. The intercept of where the extrapolated saturated polarization cuts the P axis
is called the spontaneous polarization and od the coercive field.
The explanation of this behaviour is that the specimen consists of many small
regions called domains in which individual molecular dipoles are polarized in the
same direction. The field influences some domains to change their direction of
polarization. The details of this process will be discussed later.
The hysteresis loop is a manifestation of the key property of a ferroelectric, that
of switching of the direction of polarization by an electric field. The f r, T relation
shown in Fig. 8.22b is the simplest observed. The permittivity shows a characteristic
peak near the Curie point T e , and fr may become very large. This is a useful property
ofthese materials. Above Te the curve usually follows a Curie-Weiss law again by
analogy with magnetic materials (Chapter 9). The flO T relation is then of the form
fr - 1 =A/(T - 8),
where A is a constant and 8 is a constant called the Curie-Weiss temperature. This
result can be explained by assuming that there is some internal co-operative action
among the dipoles which modifies the applied field II app by virtue of the presence of
an internal field proportional to the polarization P. The 'actual' field II a is of the
form" app + 1P, where P = 'ENa"a, N being the number of dipoles per unit volume,
a the atomic polarizability and 1 a constant. Thus using eqn (8.14)
P LNa8 1
f -1=---= a 0:----
r follapp fo(8 a -1'ENa8a ) l-1'ENa·
Now if this co-operative effect is temperature dependent such that 1 'E No. -+- 1 as
T-+- Te, there will be a catastrophe as fr - 1 goes to infinity at Te leading to an
idealized form of the behaviour actually observed. Whatever the nature of the detailed
atomic interactions, the existence of a co-operative, temperature dependent polariza-
tion is crucial to any explanation of ferroelectricity. Fig. 8.22c shows how P behaves
as Te is approached. Above the Curie temperature the material is non-ferroelectric
but may of course still be piezoelectric. The change from the non-ferroelectric to the
ferroelectric state at Te , called a phase transition is always accompanied by a change
of crystal symmetry. At lower temperatures other symmetry changes may take place
at specific temperatures and the crystal may remain ferroelectric or change to a non-
ferroelectric phase. This behaviour suggests that the balance between the two phases
of different symmetries is a very delicate one. The minimum energy configuration of
the atomic or molecular units in the crystal structure is achieved by very small shifts
of atomic positions in the change from one state to another, as the temperature changes.
The list of known ferroelectrics is one which constantly increases in length and embraces
a great variety of materials, so that a rationalization is not easy. However there are a
few groupings which provide a degree of classification based on crystal chemistry.
176 PHYSICAL PROPERTIES OF MATERIALS
(1) A large and varied group whose structures contain repeated oxygen octahedra
surrounding another type of ion. The simplest members are based on the mineral
'perovskite' CaTi0 3 [11]. They have the general formula AB0 3 and the best
known is barium titanate, BaTi03 Niobates and tantalates are also included as
well as important mixed compounds such as lead titanate and lead zirconate.
Pyrochlore compounds are based on the mineral 'pyrochlore', CaNaNb 20 6 F,
with the general formula A2 B2 0 7 • An important member is cadmium pyro-
niobate, Cd 2 Nb 2 0 7 • The structure contains Nb0 6 octahedra. The complicated
tungsten bronze structure, based on Ax W0 3 resembles that of the ferroelectric
lead metaniobate, Pbs Nb 10 0 30 , in which Nb0 6 octahedra may be recognized.
(2) A group of tartrates, hydrogen bonded crystals of which the best known is
Rochelle salt, one of the earliest known ferroelectrics. Rochelle salt is sodium
potassium tartrate, NaKC 4H4 0 64H 20, and the group also includes lithium
ammonium tartrate monohydrate, UNH4C4H406H20, and lithium thallium
tartrate monohydrate.
(3) Alkali metal dihydrogen phosphates, the most important being those of potassium
(KDP) KH 2P04 , and ammonium (ADP) NH4H2P04 . The structure ofKDP is
based on a framework of P04 groups linked by hydrogen bonds.
(4) Hydrated double sulphates of guanidinium and a trivalent metal. The best known
is usually abbreviated to GASH, guanidinium aluminium sulphate hexahydrate.
The general formula is (C(NH 2)3)(M(H 20)6XA0 4)2 where M = Al H , Ga 3+,
Cr 3+, y 3+; A = S or Se. These compounds decompose before the Curie point is
attained.
(5) Alums. General formula M1+M3+(R0 4h 12H20; M1+ = NH4 , CH 3 ; M3 + = Fe,
or Y, or In; R = S, Se. An example is ammonium iron alum, NH4Fe(SO)4)212H20.
(6) Triglycine sulphate (TGS), selenate and fluoberyllate. TG8 has the formula
(NH2 CH2 COOH)H2 804 .
(7) Many other miscellaneous compounds
Because of its importance some further properties of barium titanate will be briefly
mentioned.
Its structure is shown in Fig. 8.23. Titanium atoms lie within the oxygen octahedra
and barium atoms squeeze into the interstices between the octahedra. Above the Curie
temperature of 120° C the structure is cubic. Distortions occur at this temperature and
at 5 and -90°C.
Structure: Rhombohedral Orthorhombic Tetragonal Cubic
Temperature (oC): <-90 >-90} > 5} > 120
< 5 < 120
The behaviour of the permittivity and polarization reflects these changes (Fig. 8.24a
and b). The ferroelectricity is thought to arise from the rattling of the Ti4 + ion between
two acentric positions within the oxygen octahedra.
The P, 8. hysteresis loop for a ferroelectric may be displayed using a Sawyer-Tower
circuit (Fig. 8.25a). Fig. 8.25b shows a hysteresis loop for barium titanate. This may
become more complicated near the Curie temperature and appear as a double loop.
DIELECTRICS 177
• Sa
• Ti
Fig. 8.23
o °
Cubic perovskite structure of BaTi03 showing oxygen octahedron.
N
Ie
..~
u
10000 ~ 20
o
'>
';e 8000 ~ 16
'E
~
a:'
Q) 6000 c: 12
c. o
Q) '';:::;
,~ 4000 ~ 8
co 'L
Q; co
a: 2000 (5 4
c.
'"~ o~~~~~~~~~~~----
-'"
0 0 0 0 0 0
0
0 N IX) .q- .q- ~ ~~~~g~o~gg~~
N I I co _ _ ...... 111 --
I I '7 +-'
c:
I 1 I
(al (bl
Fig. 8.24 (a) Relative permittivity and (b) spontaneous polarization of BaTi03 as a function of
temperature.
As well as barium titanate, already mentioned for its piezoelectric applications, other
ferroelectrics can be made in ceramic form relatively cheaply and easily. Again, poling
in an electric field is an essential part of the manufacture of a ferroelectric ceramic. For
example, barium titanate can be poled by applying a field of about 103 kVm- 1 for
5 min at 50°C with the sample immersed in silicone oil. PZT can be similarly poled
at 200°C in fields up to 4 X 10 3 kVm- 1 • Recently, ceramics based on solid solutions of
178 PHYSICAL PROPERTIES OF MATERIALS
P
To X plates
Ferroelectric of oscilloscope
crystal rl.W
--------~~----~-W
Capacitor To Y plates
rl.P
(a) (b)
Fig.8.25 (a) Sawyer-Tower circuit. (b) Hysteresis loop for BaTi0 3 .
N
:I:
~ 800
u'
o
~ 600
...>-
.:;
'E 400
'E
Q;
c.
CD 200
>
'';:::
Qi
a:
'"
o 20 40 60 80 100 PbZr03
PbTi0 3 100 80 60 40 20 o
Composition (mole %)
Fig. 8.26 Relative permittivity of lead zirconate titanate (PZT) as a function of composition
(measured at 25°C, 1 MHz).
TABLE 8.7
Properties of fe"oelectric materials
yf B +++++--
~ ~ >r
+ -
.. + -
+ + + +
(a) (b)
Fig. 8.27 Ferroelectric domain walls (a) 90° wall and (b) 1800 wall.
The small size of domain walls in ferroelectrics con~rasts sharply with dimensions of
the order of 100 A in magnetic materials where the major contribution to the energy
is from exchange (Chapter 9). In ferroelectrics there is a large anisotropy energy. It is
energetically favourable for the dipoles to point in opposite directions across a 1800
wall as quickly as possible, reducing the region in which they are at an angle, to a
minimum. The energy associated with a domain wall is about 10- 2 J m- 2 . The growth
and ordering of domains are responsible for the shape of the ferroelectric hysteresis
Sideways
--
growth
Polarization Polarization
direction direction
Incident
light
-pti
axis No
transmitted
light
Polarizer Analyser
Specimen
Fig. 8.29 Passage of polarized light in the direction of the optic axis in a c domain of BaTi03.
light is quite unaffected by its passage through the domain in this case, so, observed
through the analyser the domain appears dark. The section of the indicatrix (Chapter
10) perpendicular to the polarized beam is a circle, and whatever the direction of
polarization no light is transmitted through the analyser.
If however, in an adjacent domain, the light is not parallel to the optic axis, the
cross section of the indicatrix normal to the beam will now be an ellipse. The light
emerging from the crystal is now elliptically polarized and there will always be trans-
mitted light through the analyser. The intensity will vary with the direction of polariza-
tion of the incident beam, except in the special case where the direction of polarization
is parallel to one of the axes of the ellipse, when no light gets through the analyser. In
general, it will be possible to distinguish the two domains.
Normally the external electric field due to the spontaneous polarization of a pyro-
electric specimen is not easy to detect. The reason is that stray surface charges tend to
neutralize the effect of the polarization. If the polarization is made to change then the
external field can be detected for a short time. The pyroelectric coefficient Ais defined
as the change in polarization per unit temperature change of the specimen:
A = dP/dT.
DIELECTRICS 181
Thus, when the temperature of a specimen is changed it is possible to measure the
change of spontaneous polarization which is the pyroelectric effect. Knowing the
coefficient A the temperature change can be measured. Values for A for various
materials are given in Table 8.8.
TABLE 8.8
Pyroelectric coefficients
LiTa03 yes 19
BaO.27SrO. 73 Nb 2 0 6 yes 280
Gd 2 ( Mo04h yes 5
BLtTi3012 yes 12
GASH yes 14
PbTi03 yes 40
TGSe yes 300
LiNaS04 no 0·8
ZnS (wurtzite) no 0·03
NH4 I0 3 no 3
KLiS04 no 1·5
Tartaric acid no 3
PVC no 3
.
N
'E
N
u
I
o I
~
~ I
0...
.~
~
c: 10
l!l 8
,.
II 5
,. ~
~
";:
til dPs I • :e
6 dT, • '"
, ....%
(5
c.
V>
::J
I I!;
o 4
CD , I
c:
•
---- --
c:
til
2
o
enc. 0 '--_ _-'-_ _ _""'--_ _--'-_ _ ___
-150 -100 -50 o 50
Temperature IOC)
TABLE 8.9
PropertielIofpyroelectric materillllI
Materilll Pyroelectric Relative SpeciFre DenlIity Thermal Curie Figure c;[ merit
coefficient permittivity heat capacity (p kgm m- 3 ) conductivity temperature ()../PlIe: x 109 )
~
)..(10-SC m- 2 deg-1 C) (er ) II (Jkgm- 1 deg-1C) K (Wm- 1 deg-1C) Tc ("C) ~
n
->
l""'
TGS 20-35 42 Bpolar 0·97 x 10 3 1·7 x 103 6 x 10-1 52 1·9-3·3
axis ;iI
BaTi03 20 160 Upolar 0·5 x 103 6·0 X 103 9 X 10- 1 120 0·53
axis ~
41001 polar d
axis 0·10 I:!l
en
NaN0 2 14 7·4 I polar 0·96 x 10 3 2·3 X 10 3 1·0 163 2·3 0
'"lj
axis
PZT ceramics 6-300 300-3000 0·4 x 103 7·4 x 103 1-1.5 320 0·12-1·85 s:
li2S04H2 7·8 10·3 0·4 x 10 3 2·05 X 103 1·7 2·96
liNb03 4 30 Bpolar 0·4 x 10 3 4·64 X 10 3 4·2 1190 0·39 ~
axis ~
751 polar 0·25
>
F;;
axis
DIELECTRICS 183
methods for determining the pyroelectric coefficient of a material. The specimen in
the form of a thin disc may be heated by incident radiation from a hot body and the
initial pyroelectric current is measured. If
incident radiant intensity = W,
area of specimen = A,
thickness of specimen = Z,
density of specimen = p,
specific heat of specimen = s,
then the equation for heat balance is
WA = z ps A dT/dt, (8.40)
where dTldt is the initial rate of rise of temperature of the sample. Since X = dPs/dT
the charge developed is given by
Q=A XdT
and the initial current
1= dQldt =A X dT/dt. (8.41)
Therefore from egn (8.40)
WA =z p sI/A,
i.e.
X =Izps/WA. (8.42)
Hence a measurement of I gives X directly if W is known. If the radiant source is a black
body of known temperature at a known distance from the sample, W can easily be
calculated. A blackened tip of a soldering iron is a convenient heat source in this experi-
ment. An alternative method is to heat the specimen in an oil bath and measure the
total charge released for a given temperature rise. The coefficient X is then given by the
gradient of the graph of charge released versus temperature.
References
Questions
1. The refractive index of water is 1'33, whilst its relative permittivity €r varies with
frequency f as follows
€r 81 3·3 2·8
f(Hz) 0 8·3 x 106 2·5 X 107
Explain these figures.
2. How much power is dissipated per unit volume in polystyrene if a signal of 1 V peak
amplitude at a frequency of 106 Hz is applied to a capacitor having this material as
dielectric? At this frequency the relative permittivity of polystyrene is 2·6 and the
dissipation factor, tan 0 = 7·0 X 10- 5 •
3. Show that the heat dissipation in a dielectric is half its maximum value at the
frequency at which €/" the imaginary part of the relative permittivity, is a maximum.
Assume that the dipoles have a single relaxation time.
4. The polarizability P per kilogram mole of a substance as a function of temperature
Tis given by
P = a + (biT) where a and b are constants.
Calculate the electric dipole moment of each molecule if b has the value 15 m 3 •
5. Ice at _6°C has a static relative permittivity of 80. Its frequency dependence of
tan 0 shows a peak value of 2·93 at a frequency of 104 Hz. Use this data to deduce
the infra-red refractive index of ice and the relaxation time of the dipoles at _6°C.
6. A piezoelectric material is subjected to a tensile stress in the x direction. By consider-
ing the electrical energy stored for polarization in the z direction show that the appro-
priate electromechanical coupling factor K 13 is given by
KI3 =( d 31 2 )1/2
€33 S 11
7. The amplitude of an electric field increases linearly with time from -&0 to +&0'
Compare the current-time response to this field of (a) a loss-free dielectric, (b) a
pyroelectric, and ( c) a ferroelectric material.
9 Magnetic Properties
9.1 Introduction
In Chapter 1 the basic principles of atomic magnetism were described. When atoms
come together to form solids their intrinsic magnetism acts in various ways to give
these solids magnetic properties. Because the balance of magnetic interactions
between atoms is such a delicate one, various types of magnetism are possible and here
the main features are discussed together with some important applications.
In the presence of a magnetic field in vacuum the magnetic induction or j7ux density
B is given by
B=J.LoH, (9.1)
where X is the magnetic susceptibility of the material. B, M and H are vectors; the
unit of B is the tesla or weber m- 2, while M and H have the same units of ampere per
metre. Hence X has no units.
There are five classes into which magnetic materials may be grouped;
(a) diamagnetic,
(b) paramagnetic,
(c) ferromagnetic,
(d) antiferromagnetic, and
(e) ferrimagnetic.
Table 9.1 summarizes features of each.
186 PHYSICAL PROPERTIES OF MATERIALS
TABLE 9.1
9.3 Diamagnetism
(9.4)
where f2 is the mean square distance of electrons from the nucleus. The long diamagnetic
bar ab (Fig. 9.1a) repels flux lines and becomes magnetized by the induced effect with
two north poles together. The stable position which the bar assumes in the field is
shown in Fig. 9.1b.
a
n s
n s
3 E
n s
n s
n s
n s
n s
n s
b
(a) Unstable (b) Stable
Fig. 9.1 Idealized situation of long diamagnetic bar in a magnetic field.
MAGNETIC PROPERTIES 187
9.4 Paramagnetism
In Chapter 1 the source of paramagnetism of atoms has been described. Leaving aside
for the moment the case of ferromagnetic materials which become paramagnetic above
a certain temperature (see below) solids exhibit paramagnetism if they contain para-
magnetic atoms in dilute solution, that is, one such atom or ion must be far enough away
from another, with diamagnetic atoms in between, so that there is no direct magnetic
interaction between them. These paramagnetic atoms form a collection of non-interacting
magnetic dipoles. The direction of these dipoles is normally random because of thermal
vibrations. The magnetization of the solid is thus zero until an external field is applied
when the dipoles tend to line up with the field. The resultant susceptibility varies
inversely as the absolute temperature
X=C/T. (9.5)
M =Nil
,..
(COth p.H _ kT)
kT p.H'
(9.6)
where p. is the magnetic moment of the paramagnetic atoms and N the number per unit
volume.
When p.H/kT ~ 1 the expression in parentheses Simplifies to p.H13kT giving a Curie
law with constant C =Np.2 13k. This is often the case in moderate fields at room tem-
perature. Eqn (9.6) is modified when quantum mechanical effects are considered but
Curie's law is still the main feature of behaviour unless the applied field is large or the
temperature low.
'"m; _
~ --~;,,--_",,::
(e)
Fig.9.2 Distribution of spins in the band model of a metal at T =O. (a) Zero applied field. (b)
Applied field H. (c) Electrons flow to equalize the Fermi level to give excess of spin-up over
spin-down.
188 PHYSICAL PROPERTIES OF MATERIALS
9.5 Ferromagnetism
Ferromagnets become very strongly magnetized in a weak external field and may
possess a spontaneous magnetic moment even in zero field. Ferromagnetism only
exists below a certain temperature, the Curie temperature Te above which the sub-
stance becomes paramagnetic. In this paramagnetic state, a modified Curie law
is obeyed known as the Curie-Weiss law:
X=C/(T- 8), (9.7)
where 8 is a constant. Fig. 9.3 compares the susceptibility-temperature dependence of
a paramagnetic and ferromagnetic material.
x X 1
x
-;-
>
~ ~ ~
;e
(J
ii ii '';::;
a.
Q)
'';::;
Q. Q) X very
E. Q)
c
Cl
01
.
Q) (J (J
(J
'":l large X:C'T_ 6 '":l E
'"
:l
'" '"
0
'"
(J
(J
'';::;
.!:1 Gi
u..
:;::; Q) Qj
Q) c c
c Cl Cl
Cl 01 01
01
~ ~
~
T Tc = 6 T T
Temperature Temperature Temperature
Fig. 9.3 Temperature dependence of magnetic susceptibility x for (a) a paramagnet; (b) a ferro-
magnet above and below Curie point Te; and (c) l/x versus T for a ferromagnet above and below
Te.
M ___ ~A
Magnetic
field H
Fig. 9.4 Magnetization M versus applied field H for a ferromagnet showing typical hysteresis
loop.
The field corresponding to the point C is that required to reduce the magnetization
to zero and is called the coercive force while the magnetic induction appropriate to the
point B is the remanence.
Energy is dissipated when the loop is traversed and the work done is given by the
area of the loop
work done = J H dM
ABCDEA
=..!.. fH dB.
JJ.o
(Compare the work done in compressing a gas p dv. Here the driving 'force' p is the
field H.) The loop can only be traversed in an anti-clockwise direction. In practice a
hysteresis loop may be plotted on a recorder or oscilloscope screen using the circuit
of Fig. 9.5.
Y plates
Fig. 9.S Circuit for displaying a magnetic hysteresis loop of a toroid on a CRO.
190 PHYSICAL PROPERTIES OF MATERIALS
Ci)
.'!:
c:
:::l 50
...
>-
....~ 40
:c...
..e
30
~
c:
0
:;::; 20
IU
N
:;::;
GI
c: 10
Cl
IU
~
0 100 200 300 400 500 600 700
Temperature (K)
Fig. 9.6 Spontaneous magnetization of nickel as a function of temperature.
In general the properties of magnetic materials vary with direction. This is most easily
shown by considering low index crystallographic directions in single crystals. Iron has
a b.c.c. structure and the main crystallographic directions are shown in Fig. 9.7a while
the corresponding magnetization as a function of applied field for these directions is
given in Fig. 9.7b.
The curves show that there are pre/erred directions of magnetization in the crystal,
so it costs more in energy to choose some directions for magnetization rather than
others. The crystal is said to show magnetic anisotropy. The anisotropy of a crystal is
usually expressed in terms of anisotropy constants K. In a crystal such as the one
mentioned above there will be an easy direction of magnetization. Suppose extra
energy E is needed to magnetize the crystal in another direction making LO with the
easy direction. If the crystal is uniaxial, that is one having a single easy direction,
E can be expressed as a series of sin 2 6
(9.8)
Usually only the first term need be considered and K 1 is called an anisotropy constant.
Eqn (9.8) does not contain odd terms in sin 6 because it must reflect the crystal sym-
metry and E must not change sign when 6 changes sign. In a cubic crystal the expression
MAGNETIC PROPERTIES 191
'E 2
<{
<D
o [100]
~ 1.5
~
c
o
''::;
<II
III
:;;
CD
C
g> 0.5
~
[100] a 8 16 24 32 40
(a) Applied field H (x1 03Am- 1)
(b)
Fig. 9.7 (a) Principal crystallographic directions in a cubic crystal. (b) Magnetization versus
field for these directions in a single crystal of iron.
for E involves the cosines of the angles which the magnetization direction makes
with the cube edges. If these direction cosines are al> a2 and a3
'";" 1.5
E
~
o
.-
~
~ 1.0
c
o
''::;
<II
N
.~ 0.5
c
CJ)
<II
~
16 32 48 64 80
Applied field H (x 104Am-l)
Fig. 9.8 Magnetization curve for a cobalt single crystal.
(9.10)
192 PHYSICAL PROPERTIES OF MATERIALS
where Ms is the saturation magnetization for the specified direction at a given tempera-
ture. Eqn (9.10) is obtained by equating the torque due to the applied field to the
torque on Ms due to the anisotropy. Some values of K 1 and HK are given in Table 9.2.
TABLE 9.2
9.7 Magnetostriction
When the magnetic dipole moments in a solid are rotated into alignment, the fields of
the dipoles influence the interatomic spacing. Therefore the shape and volume of a
ferromagnetic solid changes as it is magnetized. The principal change is a positive or
negative reversible strain along the axis of magnetization. This is called the 11'Ulgneto-
striction. The magnetostriction coefficient Ais the fractional change in length Mil
associated with a change in magnetization from zero to saturation.
The magnetostriction is anisotropic partly because the magnetization itself is
anisotropic and partly because the elastic constants are anisotropic. Conversely an
induced strain will result in an anisotropic magnetization.
For iron in the [100] direction A has the value 20·7 x 10-6 and in the [111]
direction -21·1 x 10-6 , the negative sign indicating a contraction.
Fig. 9.10 Domain structure in a cubic crystal (e.g. iron) with magnetization along easy axes.
!/ I
y
,1 /1 "'\ x
1 1800 domain w
__al_I_ ___
-x
(a) (b)
Fig.9.11 (a) Rotation of successive spins to form 1800 domain wall. (b) Coordinate system for
angle (<I> 1 - <1>2) between successive spins.
In magnetic domain walls exchange energy is most important and leads to walls of
the order of 100-1000 A thick. The walls are known as Bloch walls after the man who
did much of the early work on the subject. The boundary energy per unit area is
typically 10- 3 J m- 2 . Notice the coordinate system of Fig. 9.11 indicates that the spins
turn out of the plane of the page so that the azimuthal angle </> is changing. For </> small
eqn (9.11) can be written
by expanding the cosine term as a series and neglecting higher terms. Thus, Eex ex </>2
so if the change in angle is I:l</> for a distance a in the x direction through the wall
I:l</> = a d</>/dx
and
MAGNETIC PROPERTIES 195
where A is a constant which includes the constant J. The total energy 'Y of a domain
wall may now be written
(9.12)
that is, the sum of the exchange and anisotropy contributions to the energy. Eqn (9.12)
can be minimized to find the optimum conditions and this leads to the result that
'Y = 4(AKd l/2
and the effective width of the boundary [j = 1f(A/K 1 )112.
For technological purposes ferromagnetic materials fall into two groups. Those which
are easy to magnetize and demagnetize called soft materials and those which retain their
magnetism and so are more difficult to demagnetize known as hard materials. For
permanent magnets a hard material is needed with a high saturation magnetization, high
remanence and high coercivity. Such a material is characterized by a fat hysteresis loop,
and the shape of the loop in the second quadrant is often used as a measure of the
effectiveness of a material as a permanent magnet. The product BH is plotted as a
function of H and the maximum value of this product which is a measure of the maxi-
mum amount of stored energy is used as an index. Fig. 9.12 shows a demagnetization
curve in the second quadrant of a B-H graph together with the corresponding B-BH
curve plotted in the first quadrant.
1.4
1.0
0.8
(,,)
0.6
:;;
Q)
c: 0.4
Ol
co 0.2
~
Carbon steel magnets were known as early as the thirteenth century. More recently
a variety of compositions have developed which are known widely by their commercial
names such as Ticonal, Alnico and Alcomax. Typical properties are summarized in
Table 9.3. These materials are alloys of iron, nickel, aluminium with other elements,
TABLE 9.3
Properties of hard magnetic materials
TABLE 9.4
Properties of soft magnetic materials
9.11 Antiferromagnetism
Consider the compound manganese oxide, MnO. The Mn2+ ions each have a net moment
of 5 Bohr magnetons. The arrangement of magnetic moments in the lattice of single
crystal MnO is shown in Fig. 9.13. Information about spin arrangements is obtained
from neutron diffraction experiments. Within a given (l11) plane the magnetic
moments of the Mn2+ ions are aligned parallel but alternate (111) planes have their
moments antiparallel. Moments of Mn 2+ ions linked by oxygen ions are also antiparallel.
This arrangement was interpreted as being due primarily to an exchange coupling
between next to nearest magnetic ions which are separated by large intervening oxygen
ions. The direct exchange between nearest neighbour magnetic ions plays only a minor
role. The mechanism of coupling which involves overlap of the electron charge clouds
198 PHYSICAL PROPERTIES OF MATERIALS
of the Mn2+ ions via the oxygen ions is known as superexchange. Consider the simple
line of ions
0 2- - Mn 2T - 0 2 - - Mn 2 + - 0 2 -.
The outer electrons of the 0 2- ion exert their influence through the operation of
Hund's rules and the Pauli principle to force the net moment of the Mn 2+ ions to be
antiparallel. For other magnetic ions having a d shell which is less than half full, ferro-
magnetic alignment can result. For manganese oxide the net magnetic moment of the
compound is zero because of the antiparallel arrangements of moments. The material
is said to exhibit antije"omagnetism. The simplest model of an antiferromagnet is based
on a structure of two interleaved sublattices the individual moments on each being
ferromagnetically aligned but with antiparallel coupling between the two sublattices.
The net moment is thus zero. If the moments of the two sublattices are unequal there
is a resultant magnetic moment and the material is called a jemmagnet. The actual
alignment is a result of competition between superexchange and direct exchange inter-
actions within and between sublattices which may number more than two in com-
plicated cases.
Above a critical temperature, the Neel temperature, TN, thermal energy is sufficient
to disorder individual moments in antiferromagnets and they become paramagnetic.
The magnetic susceptibility of a typical antiferromagnetic oxide is shown in Fig.
9.14. The dashed curves refer to the behaviour of a single crystal where measurements
are made with the applied field parallel (II) or perpendicular (1) to the easy axis of
magnetization; the full curve refers to powdered material. Above TN, X obeys the
following equation:
X = C/(T + 0) where C and 0 are constants. (9.13)
MAGNETIC PROPERTIES 199
-~-------
TN
Temperature T
Fig. 9.14 Susceptibility-temperature curve for a typical antiferromagnetic oxide.
The simplest compounds to show antlferromagnetism are oxides and halides of the iron
group. The oxides are all cubic above the Neel temperature but the structure becomes
slightly distorted in the antiferromagnetic phase. For example NiO contracts slightly
along the [111] cubic axes to become rhombohedral. The halides have rather low
Neel temperatures and become ferromagnetic in large fields. The group encompasses
both dihalides and trihalides.
TABLE 9.5
°
Chromium oxide, Cr2 3, has the hexagonal structure of corundum, Ah 0 3 , with a
unique three fold axis. Ilmenite, FeTi0 3 , has another important structure closely
related to corundum. The difference lies in the ordering of the two different cations
into alternate (111) planes. Ilmenite itself is antiferromagnetic. Other important
compounds have the perovskite, CaTi0 3 , structure (Fig. 8.23) usually in a distorted
form; examples are KCuF 3 and KCrF 3' Some orthoferrites with the general formula
MFe03 also have this structure, for example LaMn03 and LaFe03' In LaMn03 the
La3+ ion occupies the cube comers, the Mn 3+ ion occupies the body centre and oxygen
ions occupy the face centres. Table 9.5 summarizes information on typical antiferro-
magnetic compounds.
Twall
Fig. 9.15 Twalls separating domains (the rhombohedral distortion is exaggerated for effect).
MAGNETIC PROPERTIES 201
9.13.1 Observation of antiferromagnetic domains
There are no intense local fields associated with these domain walls so direct techniques
used for ferromagnets are not applicable. Fortunately the crystallographic distortion
associated with T wall domains particularly enable them to be observed by three
techniques:
(a) Transmitted polarized light. Each rhombohedrally distorted domain is a uniaxial
crystal with the optic axis (Chapter 10) coinCiding with the contraction axis.
The domain therefore exhibits birefringence and domains can be observed by
rotating the crystal between crossed polars (plate 6).
(b) Specular reflection. This method relies on small angular misorientations at the
surface where T walls emerge.
(c) X-ray topography (Chapter 4). Photographs of domains in NiO are shown in
Plate 7 which also illustrates the variation of lattice distortion with temperature
[5].
9.14 Ferrimagnetism
If the net magnetization of magnetic sublattices is not zero, the material exhibits ferri-
magnetism and possesses a net magnetic moment. This moment disappears above a
Curie temperature T c ' analogous to the Neel temperature, at which thermal energy
randomizes the individual magnetic moments and the material becomes paramagnetic.
The most important compounds are ferrites having the formula (M2+02-)(Fe2 0 3 )
where M is a divalent metal ion which mayor may not be magnetic. The structure is
based on that of naturally occurring spinel, MgOAh 0 3 , Spinel occurs in two forms,
normal and inverse, which have identical crystal structures and simple ferrites take both
forms.
•
Metal ion
o
Oxygen ion
(a)
(b)
There is a net magnetic moment per molecule of 5 Bohr magnetons. In a normal spinel
such as zinc ferrite the magnetization of the A sublattice is zero since the Zn 2 + ions are
diamagnetic and so the normally dominantA-B interactions are zero here. The
important interaction is now B-B and this is antiferromagnetic too. Thus the Fe 3 +
ions on the B sublattice arrange themselves antiferromagnetically to give zero moment
at T = o. It follows that the normal spinel zinc ferrite is not ferrimagnetic and the weak-
ness of the B-B interaction is shown by its low Neel temperature of 9 K, below which
this material is antiferromagnetic.
There are other atomic arrangements which lead to a net moment. For example,
there could be unequal numbers of identical moments on the two sublattices. It is also
possible to mix normal and inverse ferrites and achieve a stoichiometric adjustment. If
zinc ferrite is added to an inverse ferrite, the Zn 2 + ions are found only onA sites and
this has the effect of displacing Fe 3 + ions. The net result may be written as follows:
(I - x)Fe~+O(Fe3+M2+)B03 + xZn;tOFe~B03 = (Fet.x Zn2 +)A O(Fe~~+1-xMi~x)B03
inverse ferrite normal ferrite mixed ferrite
Here, a fraction x of a normal ferrite reacts with (1 - x) of an inverse ferrite to form a
new compound, a mixed ferrite. The net moment per molecule of the latter is obtained
by adding the moments remembering that the A and B sub lattice moments are anti-
MAGNETIC PROPERTIES 203
parallel. Thus the net moment per molecule of mixed ferrite = {5(l - x) -
5(1 + x) - mel - x)} J.IB where m is the magnetic moment of the M2+ ion. As the
proportion of zinc ferrite increases the magnetic moment of the A sublattice is diluted
leading to an increase in magnetization of the substance as a whole.
The earliest magnetic material known, lodestone, is an inverse ferrite, ferrous ferrite
(Fe 3 +)A 0(Fe2+Fe3+)B03. The net magnetic moment per molecule is just equal to that
of the Fe 2+ ion, 4J1B. Barium ferrite, BaFe12019, has important commercial applica-
tions; its crystal structure is hexagonal and is derived from the spinel structure. The large
magnetic anisotropy makes this material suitable for permanent magnets. It is one of
a group of hexagonal ferrites with the general formula AB 12 0 19 where A is divalent
(Ba, Sr, Pb) and B is trivalent (AI, Ga, Cr, Fe).
Other ferrimagnetic materials with more complicated structures are the substituted
garnets based on natural garnet, Mg 3Alz(Si0 4 h. The ferrimagnetic garnets have the
formula M3Fe2Fe3012 where M is a rare earth element or yttrium. Yttrium iron garnet
(YIG) is an important material for applications at microwave frequencies. The yttrium
ions are trivalent and diamagnetic, so that the Fe 3 + ions are the only magnetic ions
present. The structure can be analysed in terms of three sublattices comprising 'A' sites
which are occupied by some Fe 3+ ions surrounded octahedrally by oxygen ions, 'D'
sites occupied by other Fe3+ ions surrounded tetrahedrally by oxygen ions and 'C'
sites surrounded dodecahedrally by oxygen ions. Thus in YIG only the 'A' and 'D'
sublattices are magnetic and their moments are arranged antiparallel.
Various substitutions of other ions can be made in YIG and one of the most impor-
tant is when the diamagnetic y3+ ion is replaced by a paramagnetic rare earth ion.
For example, if gadolinium is used, gadolinium iron garnet, Gd 3 Fes 0 12 , is formed.
This has three magnetic sublattices with the moment of the rare earth ion being
coupled antiparallel to the net moment of the iron ions. The structure is represented
by the formula (Gd3)(Fe2)(Fe3)012 with the magnetic moment per molecule =
(3 x 7) + (2 x 5) - (3 x 5) = 16J1B.
Ferrimagnetic materials become paramagnetic above a Curie point T(! and in this
region the inverse susceptibility X-I versus T curve is a hyperbola (Fig. 9.17). The
constant 0 is the 'asymptotic' Curie temperature. The equation to the curve is ex-
pressible in the form
1 T 0 S
-=-+-- - - , (9.14)
X C C T-¢J
where C, 0, S and rf> are constants. In the limit of large T the last term disappears and the
asymptotic equation is given by the remaining terms.
9 Tc
Temperature T
Fig. 9.17 Inverse susceptibility of a ferrimagnetic material above its Curie point.
204 PHYSICAL PROPERTIES OF MATERIALS
~ ~
c
.g., c
0
,!:!
.,
'';::;
;c
N
ic
.,Cl .,Cl
~ ~
To T To
(a) Temperature (b) Temperature T
~ ~
c c
0
0
.,
.;::; .,
'';::;
N
N
.~
-.;::;
Ql
C C
Cl Cl
~
'" ~'"
TABLE 9.6
Properties olle"ites
(-) (b)
Fig. 9.19 (a) Domain patterns in bubble domain material change to magnetic bubbles (cylindrical
domains). (b) An application of a field along c axis normal to a crystal face.
face, to become a set of individual cylindrical domains (Fig. 9 .19b) known as magnetic
bubbles. Sometimes both types may be seen simultaneously (Plate 8).
The stability of a bubble domain depends on minimizing the contributions to its
energy E. These contributions are from (a) the domain wall Ew; (b) the demagnetiza-
tion field within the domain Ed; and (c) the applied field H, EH.
Etota! = Ew - Ed + EH · (9.15)
For a domain of radius r and height h
E total = 2rrrhow - Ed + 2MsHrrr2h, (9.16)
where 0 is the wall energy density and Ms is the saturation magnetization. Differentiating
eqn (9.16) to find the value of r for minimum Etota! and rearranging
oEtotatior / oEd/or
-="-- = ow 2rMs - + H. (9.17)
4rrMsrh 4rrMsrh
(a) (b) (c)
206 PHYSICAL PROPERTIES OF MATERIALS
Written in this way the numerator of the left-hand side represents a force and each
term on the right-hand side represents a field. Term (a) signifies a reduction in wall area
and (b) an increase in size of domain. Thus if the right-hand side is greater than zero a
decrease in energy is achieved and the domain collapses and if less than zero the domain
prefers to expand. Fig. 9.20 shows the competition between these effects graphically
and the radius rB of a stable domain.
Applied field
contribution
Total energy
.
>
Cl _1,
Q)
c:
w
'- ........
I
- .,,"
I Domain wall
I contribution
I
I
Magnetostatic
energy
fB
Radius r
Fig. 9.20 Contributions to eneIgY of a magnetic bubble domain. A stable domain has radius fB'
Bubble domains may also be seen in thin (5 J..Im) epitaxial layers of rare earth iron
garnets, for example Er3FesOl2 (erbium iron garnet or ErIG), holmium iron garnet or
substituted derivatives with complicated stoichiometry such as Y2'18 Ybo'3Gdo,s2Fe3'9-
Gal'l 0 12 , These materials are usually grown on a substrate of non-magnetic rare earth
garnet, gadolinium gallium garnet (GdGaG), (ErGaG), (yAlG), etc.
Hexagonal ferrites BaFel2019 and PbFe12 0 19 have also been used as domain materials.
The commercial importance of magnetic bubbles arises from the fact that the movement
of bubbles can be controlled by currents through permalloy ftlms laid down by photo
techniques on the surface of the magnetic crystal. This allows the storage and mani-
pulation of large amounts of digital information which can be easily accessible. The
presence or absence of a bubble may be interpreted digitally as '0' or '1'. In the futme,
arrays of magnetic bubble material may replace magnetic tape.
Although most of the compounds mentioned have been obtained in single crystal form,
one of the advances which has extended the use of ferrimagnetic materials is their
manufacture in the form of pressed and sintered ceramic powders. Anisotropic pro-
perties may be obtained in such a powder by applying a magnetic field appropriately
during the firing and cooling processes. Being relatively easy and cheap to manufacture,
these materials may be used for all applications of metallic magnets with the particular
advantage of high resistivity and low eddy current losses.
MAGNETIC PROPERTIES 207
There are four main headings under which ferrimagnetic materials, primarily ceramic
ferrites, may be grouped for applications. They are:
(1) magnetically hard ferrites for permanent magnets;
(2) magnetically soft ferrites for transformers and inductors;
(3) rectangular loop ferrites for data stores; and
(4) ferrites and garnets for microwave applications.
TABLE 9.7
Properties of ferrites for permanent magnet materials
SrFe12019
Anisotropic
Isotropic Anisotropic
(BH)max(Jm- 3 ) 8 24
Remanence (Wbm- 2) 0·21 0·38 0·39
Coercivity (Am-I) 14000 13 500 18500
on the permeability JJ.r is represented by treating Ilr as a complex quantity Ilr' - ill/'
to take into account the fact that B is out of phase with the alternating applied field
H. The imaginary part of Ilr is a measure of the dissipation. The behaviour of Ilr' and
Ilr" as a function of frequency is shown in Fig. 9.21 for typical nickel-zinc ferrites.
!J.~
~~
ca
1000 100
.=~;E
Q)
E C-
.- iii 100 10
--
"C+:
lii :s
-~ 0
~!Il
-;;t: 10
Q) ca
a:C-
Frequency (MHz)
Fig. 9.21 Permeability spectra for Ferroxcube IV (Ni-Zn Ferrite NiO: ZnO 25 : 25).
A familiar commercial name for ferrites for these applications is Ferroxcube. For
very high frequencies of 200 to 1000 MHz other compounds known as Ferroxplanas
are used [11]. These are materials called Y compounds (Ba2Me2Fe12022) and Z
compounds (Ba3 Me2 Fe24 0 41 ) where Me is a divalent ion from the first transition
series or zinc or magnesium. These materials which have complicated structures are
hexagonal ferrites with the easy direction of magnetization in the basal plane, and not
parallel to the hexagonal axis as it is in barium ferrite (M compound). There are two
anisotropies associated with this arrangement, one for the movement of the magnetiza-
tion vector in the basal plane and one for movement out of it. The former is low and
the latter high, and Ilr is determined by the lower and the upper frequency limitation
on Ilr by the higher. Fig. 9.22 compares the frequency response of Ilr for Ferroxplanas
with nickel zinc ferrite. Table 9.8 summarizes properties of soft ferrites and Table 9.9
gives information on Y and Z compounds.
,,_-..... h..
/ ----,!).
/\ fl'!"",
, Nickel zinc e "
!_ ferrite /
10 --:ID:=;':f -:7'
~_~:~ e ·1
,. II . ' ,, " \
,.. " ,
I
,/ /. "
0.1
10 100
Frequency (MHz)
(a)
100 10
,
I
I /'
Nickel zinc
'
10 - -
rfIr'
---.-;-;;.;-.=;,..::;--.
,__
e-.
• __ ferrite
,
I ' ".
e-~.~---'.~.
I " ."
n" I " .
rr
I
, I '
"
'
I '
/
/ 100 1000
Frequency (MHz)
(b)
Fig. 9.22 Comparison with nickel zinc ferrite of the frequency response of the real and imaginary
parts of the permeability of sintered samples of Ferroxplanas (a) cobalt Z compound and (b)
magnesium Y compound (12).
Figure 9.24 shows a two-dimensional array of ferrite cores wired as a matrix to store
information. If current corresponding to the field HA /2 is passed through an X and a Y
wire, only where they intersect will the field be of sufficient magnitude to switch that
core should it be in state P0(0). If in state (1) already it will stay in that state. A
negative current corresponding to a field -HD/2 in both wires will cause the core at
the intersection to change from (1) to (0) if in state (1) but will have no effect if in
state (0). By this means information can be read in and out in digital form in three-
dimensional matrix arrays of ferrite cores. In practice a third wire is required to detect
the current pulses due to the flux change when there is a change of state (0) to (1) or
210 PHYSICAL PROPERTIES OF MATERIALS
TABLE 9.8
Ferroxcube A
MnFe204 ZnFe204
48 52 100 0'33 1400 15'9 0'2
79 21 210 0'51 700 39'7 0'8
Ferroxcube B
NiFe204 ZnFe204
36 64 125 0'36 650 31'8 10 3
80 20 400 0'36 45 333'5 10 3
TABLE 9.9
P
c 8
0 A H
D
E Po F
r
r ,... r r
x, f.J iJ .J .J
X2
r r r (
f.J f.J ..; J
:>--- f--
X3
r r r r
f.J f.J f.J ..;
X4
r r r (
f.J W f.J .J
t IHA/2
Y, Y3
Y2 Y4
(1) to (0) in reading out. Also the read-out destroys the information already present and
this information must be written back electronically. A fourth wire is used to inhibit
the reading in of states which are not required.
Ferrites commonly used for this application are manganese magnesium ferrite,
manganese copper ferrite and lithium nickel ferrite. Ideal loops are not attainable in
practice and an actual loop looks like Fig. 9.25. The departure from the ideal is mea-
sured by the ratio of the induction at H m/2 to the maximum induction B m. This
quantity is called the squareness ratio for the loop. Table 9.10 gives some properties
of square loop ferrites.
----~----~~~~-------H
TABLE 9.10
* Hm = 794 Am-I.
/L,
(0)0 0H
lL,
y
(a) (b)
Fig. 9.26 (a) Precession of a magnetization vector M in presence of field H applied in the z
direction. (b) Motion of component of angular momentum vector J in the xy plane.
of the complex permeability. The absorption of energy is indicated by the value of the
imaginary part f..lr". The absorption occurs at resonance, the sharpness of the resonance
being measured by the width at half power points.
,
[1.±
[L'- _ _ ~
Resonant Ho =~
field I~I
Fig. 9.27 Radio frequency permeability of a ferrite to circularly polarized waves. (Ilf± and Il;
are plotted as a function of H with w constant.)
Now suppose that a plane polarized high frequency e.m. wave is travelling along a
piece of ferrite in the direction of an applied field H. This wave is equivalent to two
circularly polarized waves (Fig. 9.28a). It has been shown that these encounter different
permeabilities and since the velocity in the medium v = C/l1r€r and the phase constant
~ = (W/C)l1r€r (where c is the free space velocity of the waves) the waves will travel with
fV\ (]
_~~o)
6 h_(z~d)
0~
zaxis
(Ld
Plane z =0
z=O z= 0
(d)
Fig. 9.28 Phase relationship of two circularly polarized waves h± travelling a distance d in a ferrite
along the direction of the applied field H (z direction). (a) At z = O. (b) h+ at z = d. (c) h_ at z = d.
Cd) Resultant of h+ and h_ at z =d.
TABLE 9.11
Properties of microwave ferrimagnetic materials
Material Initial Saturation Frequency g Curie tem· KdJ1.oMs Linewidth Frequency Relative Dielectric
permeability magneti- of max J1." factor perature Tc AH(Am- 1 ) (GHz) permittivity loss tangent
zation Ms (MHz) (OC) (Er) (tan Ii)
(Wbm- 2 )
N
......
Vl
216 PHYSICAL PROPERTIES OF MATERIALS
different velocities and get out of phase in travelling a distance d. The clockwise
component is retarded by {3+d rad in travelling a distance d (Fig. 9 .28b) while the anti-
clockwise component is retarded by {3_d rad (Fig. 9.28c). The result (Fig. 9.28d) is a
net rotation of the original plane polarized wave through a clockwise angle ({3- - {3+)/2
per unit length of travel in the ferrite.
Faraday devices are best operated in the region of low loss between the resonance
absorption 14' (Fig. 9.27) and the low field losses where the ferrite is not magnetically
saturated. In this region /1:'" > /1~+ and so f3- > {3+. The direction of rotation depends
only on the direction of the applied field and not on the direction of propagation, so
a reflected wave continues to have its plane of polarization rotated clockwise. This is
said to be a non-reciprocal property. The magnitude of the rotation is essentially
determined by the saturation magnetization and relative permittivity of the ferrite. A
typical value is 104 deg m-I •
The most direct application of Faraday rotation is to use waveguides just as one
uses polarizer and analyser in optics, to accept or reject plane polarized waves. Thus,
unwanted reflected signals are prevented in isolators and rotations of the plane of
polarization are achieved in gyrators. More complicated operations are carried out in
phase shifters and circulators. Typical properties of microwave ferrites are listed in
Table 9.11.
References
Questions
where J is the exchange constant, S the total spin quantum number and K the anisotropy
constant. (Hint: assume that 'YK is K per unit volume of domain wall.)
s. Calculate the net magnetization per molecule of
(a) manganese ferrite (Mn 2+Fe 3 +)B(Fe 3 +)A 0 4 , and
(b) yttrium iron garnet (Y~+)A (Fe~}B(Fe~+)C012'
where the arrangement of ions on sublattices A, Band C is indicated by the subscripts.
(Assume a simple antiferromagnetic arrangement of spins on the relevant sublattices.)
10 Optical Properties
10.1 Introduction
When a material is illuminated with light there are several processes that can occur as a
result of the interaction between the light and the material. Some light may be lost
and the energy transformed into heat in which case it is absorbed. Some, however, may
pass through without loss, i.e. transmitted,or scattered back from the surface of the
material, Le. reflected.
Different optical materials exhibit each of these phenomena. Absorption is the basic
mechanism needed for the detection of light, reflection is required for mirrors and
transmission is essential for optical components such as lenses and prisms. It is there-
fore of fundamental importance to understand how these phenomena arise.
The basic interaction which gives rise to all optical properties is that between the
electromagnetic (e.m.) field of the light (particularly the electric vector) and the charges
contained in the material. This is always the basic process but its precise form depends
very much upon the nature of the charges and their distribution, that is whether they
are electronic or ionic, whether they are free as in a metal or bound as in a semicon-
ductor or insulator.
To appreciate the material parameters which dictate the optical properties it is
necessary to understand this basic interaction. To do this requires not just a knowledge
of the charge distribution but also a description of the light. Unfortunately there is no
single way of representing light which provides a suitable description of all the optical
properties of interest, and so it will be necessary to adopt two distinct approaches.
Where propagation is of prime importance, as in transparent media, the optical para-
meter of most importance is the wavelength. For these situations the classical model
of a continuous transverse electromagnetic wave of wavelength Aand velocity C is
adopted. In this chapter the properties discussed will be confined to the region of wave-
length between 0·1 J,Lm (1000 A) in the ultra-violet and 50 J,Lm in the infra-red. It is
worth noting that the visible region occupies the range 0·4-0·65 J,Lm.
Wherever absorption is involved the energy carried by the e.m. wave becomes
important, and for these situations it is necessary to adopt the quantum approach and
to consider the light as consisting of many individual particle-like packets of energy,
i.e. photons. Each photon possesses an energy E where
v and h being the frequency of the light and Planck's constant respectively. In energy
terms the region of interest (0·1-50 J,Lm) corresponds to approximately 12-0·025 eV.
OPTICAL PROPERTIES 219
The various charges which make up a solid will all respond in some measure to the
incidence of an e.m. wave. The electric vector of the wave will force the charges to be
displaced with respect to their initial positions and so produce a dipole. The net
polarization produced will be the sum effect for all charges but will not comprise equal
contributions from every charge. The tightly bound electrons making up the core of
an ion will, if at all, be only very weakly polarizable, whereas the free electrons of a
metal and the weakly bound outer shell electrons of a semiconductor will contribute
much more.
Since the electric vector of the e.m. wave is oscillatory, so will be the induced dipoles,
providing the charges are able to follow the rapidly fluctuating field. The heavy ions
are able to follow the field only at and below infra-red frequencies and so contribute
little to the polarization in the visible region, whereas electrons only experience
difficulty in responding at frequencies corresponding to the ultra-violet part of the
spectrum. If it is assumed that there is a negligible energy loss by the oscillating dipoles,
then they will eventually re-radiate at the same frequency as that of the excitation. In
other words, the presence of the charges and the interaction with the e.m. wave has
not caused any energy loss but has merely delayed its transmission. The material will
exhibit transparency but the 'delaying' action reduces the effective velocity of the
light. The material is said to possess a refractive index n where
velOCity of light in vacuum
n= . (10.2)
velocity of light in the material
The refractive index thus derives from the polarization of the material. The greater the
polarization the more will be the delaying action, and hence the greater the value of n.
It is a result of e.m. theory [1] that for non-magnetic materials
n = (€r)1/2. (10.3)
A medium containing no polarizable charges would obviously have a value of €r of
unity; there would be no delaying action and n = 1. For a material generally, the
greater the density of dipoles the higher will be €r and hence n. The more massive an
1.8
). =0.58 I'm
)(
CD
"tI 1.7
.5
CD
>
~
u
co
....CD
~
a: 1.6
atom the more electrons there are per atom and so it is to be expected that a dense
glass, for example, will contain more electrons per unit volume than a less dense glass.
Variations in density will thus produce a variation in refractive index and this is
shown in Fig. 10.1 for a selection of commercial glasses. Glasses, like most ionic
materials, have a refractive index between 1 and 2 whereas the more polarizable semi-
conductors can have indices greater than this. Table 10.1 shows the values for some
semiconductors, with a few more commonplace materials for comparison.
A discussion of refractive index as defined by eqn (10.2), and the various phenomena
associated with the propagation of light through a material is only meaningful at
frequencies for which the material is transparent. The range of transparency for
different materials will be discussed in section 10.3.
TABLE 10.1
Refractive indices
10.2.1 Birefringence
Ordinary glass is completely isotropic in its properties. Many important optical materials
however, such as single crystal KDP(KH 2 P04 ), or BaTi0 3 are anisotropic. Atomic and
hence dipole density vary with crystal direction and as a result the polarization is
electric field direction dependent. It follows that if €r is dependent upon direction then
so is the refractive index. However, the refractive index for a given light path in a
material is determined by the value of €r associated with the value of II which is in a
direction perpendicular to this path, since the e.m. wave is a transverse wave, with its
electric vector II oscillating in a plane perpendicular to the wave velocity direction
(Fig. 10.2). So that if €r is direction dependent within this perpendicular plane, the
refractive index will be dependent upon the plane of polarization of the light. This is
an important phenomenon and one which is utilized in many optical devices. It is
termed birefringence and occurs naturally in some materials such as KDP and BaTi0 3 •
Birefringence may be enhanced by the application of external forces such as electric
and magnetic fields and even induced by the same means in otherwise non-birefringent
materials. Birefringence might not seem dramatic since n only changes by the order of
2 per cent (in KDP n ranges from 1-47 to 1·51) but the discussion in sections 10.5 and
10.6 will show this to be a very significant change.
OPTICAL PROPERTIES 221
z
Fig. 10.2 Birefringence: the velocity of light in direction z, and hence the refractive index n,
depends on the value of €r in the plane perpendicular to z.
10.2.2 Dispersion
In Chapter 8 it was shown that when a material possesses permanent dipoles the per-
mittivity is not a constant at low frequencies but varies with frequency as a result of
relaxation processes. These have no effect in the optical region but the processes of
induced dipole formation do. As a result of the frequency dependence of Er in the
optical region, the refractive index is dependent on frequency. This is referred to as
optical dispersion and is an essential property for many applications in optical instru-
mentation, since without it the prism in a prism spectrometer, for example, would be
ineffective in splitting white light into its various wavelengths or colours.
Optical dispersion is-of particular importance for the transparent glasses such as fused
Si0 2 and arises from the interaction between the e.m. wave and the bound electrons
and ions which make up the glass_ As a consequence of the forces of binding, each
charge can be characterized by a natural resonant frequency, Vo given by eqn (10.4),
where m is the mass of the bound charge and G is the force constant.
vo 2 = G/4rr'lm. (lOA)
The mechanics of the forced oscillation of such charges is treated in many texts [2,3].
For frequencies well below resonance the induced dipoles have little difficulty in
following the applied field whereas above it there will be very little induced polariza-
tion since the variations in the electric field will be too rapid. The resonant frequencies
calculated from eqn (1004) are such that for electrons they lie in the ultra-violet and
for ions they lie in the infra-red. For a glass in the visible region the refractive index
will therefore be primarily due to the bound electrons since the frequency of the light
will be well above that of resonance for the ions, but below that of the electrons.
As the optical frequency is increased towards the electron resonant frequency, the
greater will become the amplitude of the excited electron motion. As a result the
amplitude of the oscillatory polarization produced will increase with frequency with a
consequent increase in the refractive index. The increase of n with v (decrease of n with
A) is known as normal dispersion and is shown in Fig. 10.3 for fused silica. It is found
that the dependence upon A can be represented approximately by
n'l =A + B/A'l + C/A4 , (10.5)
222 PHYSICAL PROPERTIES OF MATERIALS
1.54
1.52
~
"
.s
CD 1.50
~
f!
~1.48
1.46
where A ~ B ~ C In addition to the normal dispersion shown in Fig. 10.3 there will
be a corresponding region of normal dispersion on the low frequency side of the ionic
resonances, which will occur in the infra-red.
In the immediate vicinity of a resonance frequency the refractive index decreases
with frequency. This property is not normally utilized for instrumentation and is termed
anomalous dispersion (see Fig. 10.4).
(b) v
Fig. 10.4 A schematic representation of the frequency dependence of (a) refractive index and
(b) absorption, for a material containing permanent dipoles, bound ions and bound electrons.
The complete spectral variation of optical dispersion resulting from bound electrons
and ions is shown schematically in Fig. 10.4. The true picture is much more complicated
because of the many possible resonant frequencies that exist, particularly in the ultra-
OPTICAL PROPERTIES 223
violet. For comparison with the discussions of Chapter 8, Fig. 10.4 also shows the
behaviour at much lower frequencies where the relaxation of permanent dipoles be-
comes important. It might be thought misleading to think in terms of refractive index
for these low frequencies, in which case the vertical axis can be considered as (€r)1/2.
The most important point about Fig. 10.4 is that every polarizable charge contributes
fully to €r and hence n, at frequencies well below its particular resonance frequency.
As frequency is increased, first the permanent, then the ionic and finally the electronic
dipoles become unable to respond. There is a general tendency for n to decrease with
increasing frequency, finally reaching the value of 1.
10.3 Absorption
The complete absorption spectrum will derive from the effect of all the energy loss
mechanisms that can result from the interaction between the light and the charges con-
tained in the material. The possible absorption processes comprise three types, not
necessarily all present in every material. The processes depend on whether the charges
are:
(a) tightly bound as are the inner shell electrons of the lattice ions, or the lattice
ions themselves, ,
(b) loosely bound such as the outer shell electrons of the lattice ions, or
( c) free as are the conduction electrons in a metal or semiconductor.
Charges of type (a) are those giving rise to optical dispersion and are characterized by
a resonance frequency given by eqn (10.4). By arguments similar to those given for the
dielectric relaxation process in Chapter 8, excitation by light of frequency near reson-
ance leads to a pronounced absorption (energy dissipation). The presence of both
bound electrons and ions thus leads to absorption in the ultra-violet and infra-red
regions of the spectrum. These absorptions extend over relatively narrow bands of
wavelengths and are shown (schematically) in Fig. 10.4 for electronic and ionic charges.
The absorptions corresponding to the ionic resonances are referred to as Restrahlen
absorptions.
The loosely bound charges of type (b) are best described by the electronic band
theory (Chapter 1) and absorption results from interband excitation. It can only occur
between occupied and unoccupied states of energy. If these states are separated by an
energy M, absorption will only occur provided the photon energy equals M (see Fig.
10.5), i.e.
V = Mlh (10.6)
hence
A = helM.
This fundamental absorption is characterized by a very broad spectrum since the broad
energy bands give a wide choice of transitions. The most important transitions for non-
metallic materials are those between the valence band and the conduction band. Every
material is characterized by an energy gap Eg so that a minimum photon energy exists
which will give excitation resulting in absorption. From eqn (10.6) this leads to a
critical wavelength
A=helEg (10.7)
at which there will be a sharp edge in the absorption spectrum. To the long wavelength
side of this edge the material will be transparent.
224 PHYSICAL PROPERTIES OF MATERIALS
Conduction
band
11
IIe
c::
Il.E
Eg
1j
CD
W
10.3.2 Semiconductors
Semiconductors possess charges of all kinds, bound and free and so exhibit some of
the optical properties characteristic of each. The most important charges are those
OPTICAL PROPER TIES 225
TABLE 10.2
The transparent window of some ionic TTIIlterials (jlm)
associated with the valence band and their excitation by interband transitions to the
conduction band. All semiconductors will possess a characteristic absorption edge
given by eqn (10.7) corresponding to the onset of interband transitions, on the long
wavelength side of which they will be transparent. Table 10.3 includes a range of semi-
conductors, their energy gaps and absorption edges. A typical example is ZnSe
which appears yellow in transmission since its edge occurs in the visible region whereas
InSb is opaque since its edge occurs in the infra-red. The net result of the interband
transitions is to give rise to free electrons and holes, which mayor may not recombine
at a later time.
The exact nature of the absorption edge depends very much on the particular semi-
conductor and does not necessarily occur at exactly the same photon energy as the
thermal energy gap E g • The reason for this lies in the availability of conduction band
energy states relative to the valence band, and requires reference to the energy-
momentum (E, k) diagram. Appendix I shows that the theoretical E, k dependence
for a one-dimensional array of regularly spaced atoms is parabolic. However, a real
TABLE 10.3
Semiconductor absorption edges
Eg (eV) A (jlm)
does not necessarily lie at the same value of k as the maximum in the valence band.
Some semiconductors (for example, GaAs) possess a band structure in which the
minimum in the conduction band does lie directly above the maximum in the valence
band, and these are called direct gap semiconductors (Fig. 10.6). Others (for example,
Ge, Si) have the minimum displaced, and these are called indirect gap semiconductors
(Fig. 10.7). The thermal energy gap Eg in each case is defined as the minimum energy
>-
Cl Conduction band
Q;
t:
Q)
...
<\I
...
1~
t:
Q)
0
Q.
t:
...e
0
Q)
iii
difference between bands irrespective of k, but the onset of intense interband absorption,
i.e. the absorption edge, is represented by the minimum energy difference for a con-
stant k. The reason for this is that the most likely optical transition is one for which
there is a negligible change in k. Momentum must always be conserved and since the
photon momentum is very small in comparison with electron momentum, the most
likely transitions are represented by vertical lines in Figs 10.6 and 10.7, i.e.
(10.10)
where k j and kf are the initial and final momenta of the electron taking part in the
transition. These are referred to as direct transitions. In Fig. 10.7 a smaller energy
OPTICAL PROPERTIES 227
10
I
.§.
..
c:
Gl
~ 10
o
..g...
u
103
Co
o
~ 10
0(
1 0 I:--~---:-I~--:-L:--~,:",
0.6
Photon energy h" (eV)
photon is required to excite an electron from the maximum of the valence band to the
minimum of the conduction band but eqn (10.10) is not obeyed and it is necessary
for a third body to take part in the photon-electron interaction for momentum to be
conserved. A lattice phonon may be absorbed and this less likely process is known as
an indirect transition.
Hence the absorption edges of indirect gap semiconductors possess a second weaker
absorption edge to the long wavelength (low energy) side of the first edge due to the
possibility of indirect transitions. Fig. 10.8 shows this for Ge.
In addition to the mechanism of interband transitions, absorption also occurs in the
ultra-violet due to electronic polarization. This can be very intense but since it occurs
at shorter wavelengths it does not influence the position of the absorption edge. If
the semiconductor is a compound one with a degree of ionicity in its bonding, there
may be an ionic resonance in the infra-red with an accompanying restrahlen absorption
band. As a result of thermal excitation there are many free carriers in a semiconductor,
1.5
I
E
'"o
x
.. c:
Gl
1.0
·u
~o
u
c: 0.5
..;::;o
...Co
oIII
.c
0(
both electrons and holes. These also absorb and for very narrow band gap materials,
such as InSb, the carrier concentration and hence the absorption in the infra-red can
be large (Fig. 10.9).
Except where expressly mentioned, the discussion of semiconductors so far has
related to single crystals. If the long range crystalline order is missing, as in amorphous
semiconductors, the optical properties are modified [7]. There is still an absorption
edge with intense absorption on the short wavelength side of it, but the edge itself is
normally less abrupt than for single crystals. The dependence of the absorption
coefficient upon photon energy is found to be exponential in the region of the edge.
This might be due either to a smearing of the density of states at the edges of the
conduction and valence bands into the otherwise forbidden energy gap or the broaden-
ing effect of localized electric fields on the energy gap.
The structure to the high energy side of the edge is also smeared in the amorphous
state. This is thought to be due to a breakdown of the momentum condition of eqn
(10.10). Due to the atomic disorder the electrons cannot be characterized by well
defined values of k and it is possible for non-vertical transitions to occur. These
transitions are referred to as non-direct.
The optical absorption in the single crystal and amorphous state of a typical com-
pound, CdGeAs 2 , is shown in Fig. 10.10.
--,,,
,,
',Amorphous
....
_-- ........
-
.....
....
o 4 6 8 10 12
Photon energy (eV)
10.3.3 Metals
A metal differs from all other materials by the presence of a very large concentration
of free electrons. These electrons lead to a free carrier absorption which, according to
eqn (10.8), is especially large at infra-red wavelengths. As the wavelength decreases
from the infra-red the free carrier absorption decreases, but at ultra-violet wavelengths
absorption increases again due to the possibility of interband transitions. Dependent
upon the energy band structure of the particular metal, the interband and free carrier
processes may overlap to give absorption throughout the optical region, or there may
be a small 'window' where absorption is much less.
10.3.4 Impurities
In ionic materials, the presence of foreign ions can give rise to absorption in the other-
wise transparent window of the host material, the absorption increasing with impurity
concentration. The process of absorption is a localized one. For example in Tl doped
KCI, the Tl + ion substitutes for K + and introduces localized energy states into the for-
bidden energy gap of the host material. The absorption will not be limited to a very
narrow wavelength region, however, since at any instant all ions will not be in identical
energy states. There is a strong dependence of these energy states on the distance of
OPTICAL PROPERTIES 229
the impurity ion from the nearest host ions. This dependence can be represented as in
Fig. lO.1la where the ground and first excited states are shown. As a result of thermal
vibration, ion-impurity separations and hence the energy states are constantly changing.
First excited
Configurational coordinate -
(mean separation of impurity ion
(a) from nearest host ions)
Conduction band
e e
>-
~
II ..trio.,
~
G) _
hv
-+....-
......_ _....,,&.:()...-_ _ _...... Donor level
Acceptor
c level
...o
'0G)
iii
Since absorption transitions occur in a time short compared to the time taken for
the lattice ions to move, they are represented as vertical lines. At a temperature T K
the ions in the ground state might be anywhere betweenA t andA 2 and so the possible
absorption transitions can range from A t B 1 to A 2B2 , giving rise to an absorption band
centred on the wavelength equivalent to A oB o. Reducing the temperature reduces the
range of likely transitions and so narrows the absorption band.
230 PHYSICAL PROPERTIES OF MATERIALS
The effect of impurities upon crystalline semiconductors depends on whether they
act as donor or acceptor atoms in the host lattice and the magnitude of the electron or
hole ionization energy (Le. position in the forbidden energy gap). Optical absorption
will occur giving rise to impurity ionization if the impurities are not already ionized.
This will be the case for wide gap materials at room temperatures and below, but not
for narrow gap materials unless the temperatures are very low. This photo ionization
of impurities creates free electrons and holes (Fig. 10.11 b). The spectral position of
these extrinsic absorption bands can be chosen for a particular host material, by using
different impurity atoms. Table 10.4 indicates how various impurities could be used
in germanium to extend its use as an infra-red detector, based on absorption and sub-
sequent photo conduction.
TABLE 10.4
Performance of doped germanium detectors [9, 10]
None 77 1·4
Fe 77 3·6
Ni 77 5·7
Mn 77 7·8
Au 60 8·8
Hg 30 14
Cu 15 30
Zn 6 40
B, In 2 120
F_IF .b:<P
, CD
..
c:
,
/
I \ CD
\
+ + ti: _ _ iii
'- \ e/eI e e_
' ... _ c:
....
....oCD
Q.
+ + c:
....e
u
CII
iii
+ + I//, Valence band ~
///////////~ W
(a) (b)
Fig. 10.12 The F centre. (a) The lattice configuration. (b) The F -band transition in KCI.
OPTICAL PROPERTIES 231
energetically removed, to leave the vacancy which then subsequently traps a free
electron. The resulting localized centre (Fig. 10.12) behaves very much like an impurity
ion with an energy level structure which can be represented by Fig. 10.11a or Fig.
10.12b. The associated characteristic F·band absorption arises from the excitation
between the ground and first excited states.
The F centre is by no means the only type of colour centre. More complicated centres
consisting of aggregates of F centres may be produced by heating or irradiating with
light a sample containing simple F centres. The resulting centres give rise to a more
complicated absorption spectrum (Fig. 10.13).
Photon energy (eV)
2 1.5
6
.........- (a)
--(b)
5
I
~
··· ...
~4 :F.
CQ)
:!l 3
o~~ .....
~ ..... ~~ .....
~ .....
~~~~ .....
~--
10.3.6 Excitons
For both impurities and colour centres, optical absorption is produced by a transition
within an existing imperfection. Light can also be absorbed in an insulating material by
a transition which actually generates an imperfection called the exciton [13].
232 PHYSICAL PROPERTIES OF MATERIALS
Conduction band
/////1(//(1//(////«(((
>
~
CD == ====~ ~ -_-_-_-_-_-_-_-~ }EI~~~?Sn
Ii
c:
c.
eo
C)
"0
o c:
c. eo
c: ID
o....
....
(,J
.!!?
~j~55~za
UJ
(a)
Fundamental edge
'I
Jo 3
Fig. 10.14 Excitons. (a) Exciton energy levels. (b) Exciton absorption in CU20 at 4 K [14].
10.4 Reflection
The illumination of a material with light gives rise to the polarization of charges near
to the surface and if the energy of the polarization is not dissipated by interactions
with the lattice atoms, the light is re-radiated, as from an oscillatory dipole (section
10.2). However, the re-radiated light from the polarized charge could well be emitted
into the direction from which the incident light had come. If the surface contains a
OPTICAL PROPERTIES 233
high density of easily polarized charges which readily interact with the incident light a
very large proportion of the beam will be reflected. This arises since even the re-
radiation which is directed further into the material will interact again with a high pro-
bability of being re-radiated back towards the surface. The processes are not simple
but the overall effect will be the reflection of a portion of the incident light.
It is a result of electromagnetic theory that the proportion of an incident light beam
which is reflected normally at the surface of a material of refractive index n, is given
by the reflection coefficient
R = {(n - 1)/(n + 1)}2. (10.11)
A material like glass does not present a high density of easily polarizable charges to
a beam of visible light and so reflection is small (~4 per cent). On the other hand, a
metal possesses a very dense gas of free electrons and penetration by the light is very
small. Thus, the metal acts as an almost perfect reflector.
It is found that whatever the material, for wavelengths at which there is intense
absorption there will usually be a strong surface reflection. This occurs for example in
the visible region for a semiconductor where there is interband absorption and in the
infra-red for ionic materials at the ionic resonance frequencies.
For a transparent isotropic medium the refractive index is related to the relative per-
mittivity according to the relation €r =n2 , where €r is a scalar quantity. For an
anisotropic medium it must be a matrix €ij where, using the notation of Chapter 8
Dj =€o€ij8j: Pi =€O(€ij -1)8j
This implies that an electric field applied parallel to the x direction will induce not only
a polarization Pi parallel to itself given by €O(€l1 - 1) IJ 10 but also polarizations
paralleltothey andz directions given by P2 =€O(€12 -1)81 andP 3 =€O(€13 -1)81
respectively. Since the matrix is symmetric it is possible to fmd a set of axes within
the crystal for which only the diagonal terms are non-zero. These axes are called the
principal axes and the diagonal terms the principal permittivities €1, €2 and €3
TABLE 10.5
Room temperature birefringence in some uniaxial crystals
A(/lm) nz nx a.n
Crystals of lower symmetry possess two optic axes and therefore are said to be
biaxial; examples are Rochelle salt, gypsum and feldspar. These axes are not coincident
with any of the principal axes and nx , ny and nz therefore have different values_
The application of Maxwell's electromagnetic theory equations to an isotropic
medium shows that light of any polarization can be propagated in any direction_ For
anisotropic crystals, however, this is not so, and it can be shown [2,15] that as a result
of birefringence only two states of plane polarization can be propagated for any crystal
direction. These planes of polarization are at right angles to each other and except
when propagation is parallel to an optic axis they possess different refractive indices.
The values of n and the states of polarization for a particular direction of propagation
can be determined from a three-dimensional spatial representation of the refractive
index, known as the index ellipsoid or optical indicatrix_ Its form can be derived as
follows. The energy density W in a dielectric is given by
W=~D-8 (10.14)
and hence from eqns (10.12) and (10.13)
DI2 D22 D3 2
W=-- +-- +--. (10.1S)
2EoEI 2EoE2 2EoE3
A constant energy (W) surface then takes the form, in a three-dimensional space where
dielectric displacement D is the coordinate, of an ellipsoid.
Dividing both sides by Wand rewriting DI2 /2Eo W as Xl , etc_ leads to a simplified
expression
x 2 y2 Z2
1 =-+-+-. (10.16)
n x 2 n y 2 nz 2
This equation represents an ellipsoid of semi axes nx , ny and nz _ For a uniaxial crystal
nx = ny =1= nz and the optical indicatrix for such a crystal is shown in Fig_ 10.1 S.
OPTICAL PROPERTIES 235
x
Fig. 10.15 The refractive index ellipsoid or optical indicatrix for a uniaxial crystal.
It is normal to refer to the index nx(= ny) as the ordinary refractive index no and to
n z as the extraordinary refractive index ne. Hence for uniaxial crystals
(10.17)
For propagation in a general direction OP (Fig. 10.15) at an angle e to the optic axis
(z axis) the electric vector always lies in the plane at right angles to OP (shown shaded).
The refractive index varies from no to a maximum value n(J ; n(J lies between ne and no
and depends on e. For () =0, n(J =no and for () =1rj2, n(J =ne. The two plane polariza-
tions which can be propagated correspond to the maximum and minimum refractive
indices given by the indicatrix construction. Hence for propagation parallel to the optic
axis, there is no birefringence since the section of the ellipsoid perpendicular to this
direction is a circle. For propagation in the x direction perpendicular to the optic axis,
birefringence will be a maximum and the permitted polarizations will be parallel to
the y axis with refractive index no and parallel to the z axis with refractive index ne.
is the extraordinary or E-ray. It does not obey Snell's law and its index will lie between
no and ne (hence the origin of extraordinary index). The two rays are shown in Fig.
10.16 for a uniaxial crystal which is cut at various angles to the optic axis, light being
incident normally. If wavelets are emitted at 0, the ellipse and circle represent relative
shapes of the wave fronts for the ordinary and extraordinary wavelets. The shapes of
these wave fronts follow from the index ellipsoid remembering the phase velocity
v = c/n.
>- >-
I"' I"'
>-
~
I OUJ
~
~"
0
~v
o~%,
/"'b+~
O -ray
wavefront "E- ray
wavefront
Crystal Crystal
9~
aXIs
Air Air
C
Q) E
cQ) E
"u
"0
"'
Q)
.0
"0
'0
c
"'
Q)
.0
.E
(a) (b)
Fig. 10.16 Double refraction in a positive uniaxial crystal, i.e. ne > no for the crystal surface
cut at (a) an arbitrary angle, and (b) parallel to the optic axis.
It can be seen that in general (Fig. 1O.l6a) the two rays, both linearly polarized,
are spatially separating. This is 'double refraction', and given an adequate thickness of
material the rays can be completely separated. For the crystal cut parallel to the optic
axis (Fig. 10.16b) no spatial separation of the rays is achieved for normal incidence
although they do travel at different velocities.
Double refraction can be utilized for producing plane polarized light, as for example
in the calcite Nicol prism. The crystal is cut such that the extraordinary and ordinary
rays are spatially separated (Fig. 10.17). The ordinary ray is totally reflected at a thin
layer of Canada Balsam cement between the two halves of the prism, this cement has a
refractive index intermediate between that of calcite for the E- and O-rays. The angle
E-ray
• •
t
axes and the ellipsoid must now be represented by the more general equation
This modified ellipsoid will have a new set of principal axes relative to which eqn (10.18)
can be rewritten in the form of eqn (10.16). In general, however, these new axes will
not coincide withx,y or z.
The effect of a particular field upon the index ellipsoid of eqn (10.16) is represented
by the changes it produces in the six terms (l/n 2 );, where i = 1, 2, ... , 6. This gives the
ellipsoid the form of eqn (10.18).
0 0 0 -'11 0 0
0 0 0 0 0 0
'41 0 0 '41 0 0
0 '41 0 0 -'41 0
0 0 '63 0 -2'11 0
(a) (b)
Fig. 10.18 The electro-optic tensor for (a) ADP and (b) quartz.
TABLE 10.6
Electro-optic data, ,(x 10-12 mV-l)
Output
Input
light beam
light beam
Analvser
Polarizer
-----~
Optic aKis
(10.23)
It can be seen that the combination of varying the electric field and using the crystal
between crossed polarizers has the effect of modulating the intensity of the light beam.
This has application to optical shutters [17,18] . Some values for Vl/2 are given in
Table 10.7.
The electro-optic effect described is known as the longitudinal effect, since the
applied field is parallel to the direction of propagation. Although simple in principle,
TABLE 10.7
Electro-optic modulation data
When an e.m. wave represented by 8. = 8. oe- iwt interacts with an outer electron of an
atom, at a frequency well below resonance, a polarization P is produced which, provid-
ing the electric field intensity is low, will be in phase with the excitation and pro-
portional to it, i.e.
(10.29)
where Xl is the susceptibility (Xl = €r - 1) and is in general a tensor.
Should the electric field be very large however (~1010 V m- l ), comparable with the
electric fields associated with the atoms, the polarization will not be linearly pro-
portional to 8., but will contain higher order or non-linear terms, i.e.
P=€OXI8. oe- jwt + €oX2(8. oe- jwt )2 + €oX3(8.oe-iw~3 +.... (10.30)
The first, second and higher order susceptibility terms become progressively smaller,
but at high fields the presence of the power terms in 8, give rise to a significant con-
tribution to the polarization. The most important effect is that as a result of non-
linearity, the polarization contains components characteristic of higher harmonic
frequencies of the fundamental exciting angular frequency w. Rewriting eqn (10.30)
in the following form shows that the second order term produces a component of the
polarization characterized by an angular frequency 2w:
(10.31)
The fields necessary to observe non-linear effects can only be produced in a laser
beam. As a result of the appearance of the higher harmonics in the polarization, some
of the atoms will, after irradiation, emit photons of twice or even three times the
exciting frequency. The number of these photons will however be small in comparison
with those still of the fundamental frequency.
The production of light of angular frequency 2w by a non-linear material, from a laser
beam of angular frequency w, is known as second harmonic generation and can be
readily achieved in the non-centro symmetric crystals such as quartz, ADP and KDP
[15,19]. Centro symmetric crystals produce a zero effect because X2 is zero.
Optical absorption can give rise to many secondary effects, the majority of which are
useful as means for the detection of light. The principal processes are photo emission,
photoconduction and photoluminescence.
242 PHYSICAL PROPERTIES OF MATERIALS
eCfi hv
Fermi
level
Fermi
level
~
/.Conduction bang ~
all////////// / /Valence band~
/////////~://%
(a) (b)
Fig. 10.20 Photoelectric emission from (a) metal and (b) a semiconductor.
Substituting for the constants in eqn (10.32) indicates the emission will only occur
for wavelengths shorter than the critical wavelength Ao where
1'24
AO =-;;; microns. (10.33)
Common metals possess work functions from 2 eV upwards and are sensitive, there-
fore, only to visible and ultra-violet radiation (see Table 10.8).
For intrinsic semiconductors emitted electrons must originate from at least the
depth of the valence band and so in this case (Fig. 10.20b) the photon energy must
exceed the sum of the electron affinity Ea and the band gap E g , i.e.
(10.34)
Many semiconductors are characterized by photoemission only in the visible and near
* This is not strictly true since with the Fermi distribution there will be some electrons above
the Fermi level at T> 0 K. There will therefore be a tail in the excitation spectrum to wavelengths
greater than AO'
OPTICAL PROPERTIES 243
ultra-violet regions but there are some composite evaporated materials which are
sensitive in the near infra-red, for example Ag-O-Cs is sensitive to 1·2 Jlm.
It might have been expected that the narrow band gap materials like InSb would
have exhibited photoelectric emission when illuminated with infra-red radiation, but
this is not so, due to their large electron affinities.
Quantum efficiency depends on a combination of the probability of an electron
interacting with a photon and the probability of the electron, once excited, reaching
the surface and escaping. In any electron-photon collision it is necessary to have con-
servation of momentum but for free electrons this cannot occur unless a third body
(an ion or phonon) takes part. The momentum of a light photon is not sufficient to
provide the momentum consistent with the energy of the electron, and the deficiency
has to be met by the third body. The conduction electrons in a metal are free and
since the chance of a photon-electron collision occurring in the presence of a third
body is small, the quantum efficiency will be small, of the order of 10- 3 per cent.
A second factor which reduces the efficiency in a metal is the very limited range
that the excited electron possesses. Collisions with impurities and lattice phonons
occur to degrade the kinetic energy of the electron such that unless the initial excita-
tion occurs within about 30 A of the surface, there is little chance of escape.
In semiconductors the excited electrons come from bound locations in the valence
band and so there are no momentum problems. Furthermore, once excited, the electrons
experience few interactions with conduction electrons since the number of these is
relatively small. An interaction that can occur, however, is that between an excited
valence electron and an unexcited one, but the importance of this process depends on
the relative sizes of the electron affinity Ea and the band gap E g. If Ea < E g, (Fig. 10.20b)
a photon of just sufficient energy to produce emission (i.e. hv = Ea + Eg) will give rise
to an excited valence electron with kinetic energy Ea. insufficient to produce further
electron-hole pairs. As a result these electrons will have a large range, perhaps 250 A,
and the materials will have a high quantum efficiency. A material of this type is CS 3 Sb
for which Ea is 0-45 eV and Eg is 1·6 eV giving a quantum efficiency as high as 25 per
cent. Properties of this and other materials are shown in Table 10.8.
TABLE 10.8
Photoelectric emitters
If Ea > Eg the excited valence electrons will have sufficient kinetic energy to produce
further electron-hole pairs. This process and possible subsequent ones would degrade
the kinetic energy, and even if no other collision occurred would remove the possibility
of overcoming the surface barrier forces. For this case the range can be as low as 30 A
244 PHYSICAL PROPERTIES OF MATERIALS
and the quantum efficiency is consequently low. A material of this type is Na3 Sb for
which Ea is 2·0 eV and Eg is 1·1 eV; its quantum efficiency is only 2 per cent.
Figure 10.21 shows the spectral response of quantum efficiency of some com-
mercially available photodetectors for the visible and near infra-red regions.
24
~
>-
(.)
c: 16
cP
·13
....cP
~
12
E
.ac:
II>
:::l
8
d
0
0.8 1.0 1.2
Wavelength (I-lm)
Fig. 10.21 The spectral variation of quantum effidency for a range of EMI photocathodes [20].
10.8.2 Photoconductivity
The last section showed that for photo emission to occur, relatively large photon energies
were necessary, and as a result there are no known photoemitters for wavelengths
beyond the near infra-red region of the spectrum. In semiconductors, however, rela-
tively small photon energies are capable of producing electron-hole pairs internally, and
hence of increasing the carrier concentrations and the conductivity of the material. This
photo-induced conductivity can be used for light detection from the ultra-violet through
into the infra-red spectral region.
The conductivity of a semiconductor can be written in the form
a = nel1n + pel1p, (10.35)
where nand p are the free electron and hole concentrations, I1n and I1p are the electron
and hole mobilities, and e is the electron charge. Under illumination a change will occur
which can be expressed as
(10.36)
where both tl.n and Ap will depend upon the light intensity.
For intrinsic semiconductors the photo response only occurs for wavelengths shorter
than a threshold corresponding to the energy gap, and by a suitable choice of material
a wide range of spectral sensitivity can be achieved. For GaP the threshold is at 0·55 11m
since Eg is 2·24 eV, whereas, for InSb it is at 6·9 11m since Eg is 0·18 eV. This shows
that narrow gap materials are infra-red sensitive but it is important to remember that
any photo-induced increase in conductivity must be distinguishable from the dark
OPTICAL PROPERTIES 245
thermally produced conductivity. For this reason it is necessary to cool infra-red photo-
conductive detectors to nitrogen or helium temperatures, in order to reduce the dark
conductivity to an acceptable level. Not only does this improve the signal to noise
ratio but it changes the spectral response, since Eg is sensitive to temperature. As an
example, for PbTe the threshold is 3·4 J,tm at 300 K but is 5·9 J,tm at 20 K.
The spectral response of photoconductivity of an intrinsic semiconductor detector
can be extended to longer wavelengths by doping with an impurity. The impurities
put donor or acceptor states in the otherwise forbidden energy gap and these can be
ionized to produce photo conduction. Unlike the intrinsic photoconduction, only one
carrier is excited in extrinsic photoconduction (Fig. 10.11 b). Fig. 10.22 shows how
germanium, whose normal threshold is 1'4 J,tm at 77 K, can have its spectral response
extended to 9 J,tm by the inclusion of various acceptor impurities.
,.. ....
~
I
I
I
Mn'
I
I
J
,
I
I
Ni
Pure materials are characterized by small 1"(-10- 6 s) due to the high chance of electron-
hole recombination, and hence are relatively insensitive. Careful doping however can
suppress recombination and hence increase T and the sensitivity. An example is that of
CdS, in which the addition of iodine has the effect of increasing T to 10- 3 s. The
process is represented in Fig. 10.23. Iodine, substituting for sulphur, acts as a donor,
Conduction band
6-
o 0 0 0 0 0 0 0 0 0 0 ,-
hv
~
o 0 0 0 0
o 0 Cd 2+ Vacancies
Fig. 10.23 Model for the enhancement of electron lifetime in CdS as a result of doping with
iodine.
giving CdS N-type properties. To fit into the divalent ionic lattice the iodine atom
should accept two electrons. However, it displays a strong affinity for one but the
second is only weakly bound and easily removed thermally. In this state the singly
ionized iodine represents an imbalance in the charge of the CdS, and to compensate
for this, Cd 2 + vacancies are generated at a rate of one per two iodine donors. These
vacancies act as efficient hole traps and so upon photo excitation the holes are readily
trapped, but the electrons have an enhanced lifetime due to the now reduced chance
of meeting a hole and recombining. The conduction in this case is purely N-type.
The response time of a photoconductor is a measure of the time for the photocurrent
to reach apprOximately 70 per cent of its final value following the onset of illumina-
tion. For quick response times, short rather than long lifetimes are required so that an
end to illumination is followed rapidly by an end to photo-induced current. Response
times however are governed not only by any recombination centres present, and their
1270
/ Recombination
Q)
o
; 1290
ti
"iii
Q)
II:
1300
TABLE 10.9
Commercially available photoconductors
in the exact location of the excited ions at any given instant. This will range from C 1
to C2 where the energy difference above the minimum is of the order of kT. Follow-
ing this relaxation, the excited ion may de-excite to the ground state by luminescence
emission. The transition will be instantaneous and represented by vertical lines in
Fig. 10.25 but will vary between different ions over the range C1D1 to C2 D 2 •
t
>-
Cl
~
CD
c:
w
Ground
state
Configurational co ordinate -
(mean separation of impurity ion
from nearest host ions)
Fig. 10.25 Photoluminescence by an impurity ion in an insulator, e.g. KCl: Tl.
The discussion has so far been directed solely to photoluminescence, the excitation
being produced by optical absorption. However, there are many other means which
can produce identical excited states in insulators or semiconductors and which can
subsequently give rise to luminescence. These are mentioned briefly.
Cathodoluminescence occurs when excitation arises from the impact of high energy
radiation or particles for example from 'Y-rays, X-rays, electrons, ions or nucleons.
Scintillation counters based on this effect are used for nuclear particle and radiation
detection and employ insulators such as NaI: T1 or perspex doped with anthracene.
The phosphorescent screen of a cathode ray tube might employ ZnS doped with Ag
or Cu.
Electroluminescence occurs when excitation is initiated by an electric field. The
forward bias voltage across a semiconductor P-N junction, for example, enables
minority carriers to be injected into the two sides of the junction. Some of these may
recombine radiatively and if the emission is in the visible region of the spectrum a
controllable light source is obtained. GaP emits red light and is commonly used as a
light emitting diode (LED) in visual displays.
Other types of luminescence are thermoluminescence produced as a result of heat-
ing, triboluminescence from mechanical strain, chemiluminescence from chemical
reactions and bioluminescence as a result of biological action.
250 PHYSICAL PROPERTIES OF MATERIALS
10.9 Lasers
In thermal equilibrium there will always be more ions in the lower energy state, for
if E2 - El is 2 eV, then at 300 K,NdN2 is very large, being 1034 and virtually all
ions would be in the ground state.
Should the ions be illuminated by photons of energy (E2 - Ed, three processes
can arise, each of which is represented in Fig. 10.27. If ions are initially in state E 1,
optical absorption can follow and the ions are excited into state E2 • The number of
absorption processes will depend upon the number of incident photons and the
number of ions Nt in state El and it is likely to be large sinceNI is large.
£2 - - - - . . . - - - - £2------ £ 2 - -........- - -
hv
~
hv hv hv hv
~ '''\1'\1''_ ~
Should ions be initially in the higher state E 2 , either by prior thermal or optical
excitation, they can return to the lower state by one of two possible processes. First
by spontaneous emission of a photon. In the absence of any external influence such as
the incidence of photons, this process is the only one that can occur, and the photons
are emitted randomly in direction and time. The rate at which spontaneous emission
can occur depends on the lifetime of the excited state, being slower the longer the
lifetime.
The second possible de-excitation process occurs as a result of the excited ion being
stimulated to emit, by a photon of just the same energy as that of the de-excitation.
As a result, a negative absorption or increase in the number of photons is produced,
since in each stimulated emission two photons emerge where there was initially one.
OPTICAL PROPER TIES 251
The important factor about this process is that the emitted photon possesses the same
direction and phase as the stimulating photon.
The transition probability for an ion absorbing in state EI can be shown to be
exactly the same as that for an ion in state E z being stimulated to emit. In normal
circumstances, however, NI is vastly in excess of N z and so absorption predominates
for a given incident light beam intensity. If, however, an artificial situation is produced
in which there are more ions in the higher energy state (i.e. population inversion) an
incident beam of photons of energy E z - E I would be more likely to cause stimulated
emission than absorption. An increase in the number of photons would therefore be
produced and such Light Amplification by the Stimulated Emission of Radiation is the
basis of LASER action. In order to build up an appreciable amplification it is neces-
sary to maintain the population inversion and to retain the gradually increasing number
of photons within the lasing medium. Photons will tend to escape from the system
but if the medium is placed between parallel mirrors, a highly directional beam will be
produced in which the light is continuously traversing the lasing medium. If light
losses due to poor reflections and scattering are minimized, and the mirror separation
d is such as to cause successive reflections to be in phase, i.e.
d = nA/2, (10.40)
thus giving a resonant cavity, amplification will result. The output from the laser can
be taken through one end mirror, by having it semi-silvered, and it will be a very
intense, monochromatic, coherent directional beam.
>-
at
~1..
c;
II
hv
.J\.IV\I\r
hv
o Pumping ~ 0
~
Emission 6943 A
Fig. 10.28 Schematic representation of the energy level structure of the Cr 3+ ion in Alz 0 3 and
its utilization in the ruby laser.
The ruby laser is used to provide either a pulsed or a continuous output but it
requires efficient cooling because of the energy released to the crystal by the non-
radiative relaxations to state 2. Table 10.10 gives a few of the many solid state lasing
systems that have been developed.
252 PHYSICAL PROPERTIES OF MATERIALS
TABLE 10.10
Laser systems
A1203 Cr 3 + 0·6943
CaF2 Nd 3+ 1·046
CaF2 Sm2+ 0·7085
CaF2 Ho 3+ 2·09
CaW04 Nd 3+ 1·06
Cawo 4 Ho 3+ 2·05
(b) Gas
GaAs 0·84
GaP 0·72
GaAsxPl-x 0·64-0·86
InP 0·91
InAs HO
PbS 4·27
PbSe 8·53
SnxPbl_xTe 9·5-28
20
17.5
:> _1
/
.!!.
I
>- I
~ 15
Q) I
c: I
w
I
I
: Collision with
I I walls of tube
I I
I
J
I
I
I
I I I
t
Fig. 10.29 Schematic representation of the energy level structure of He and Ne and its utilization
in the He-Ne gas laser.
N p
.;~
~:r:::;I~~mffff~ '.,m' level
(al
N P
Recombination
U~l
Bias
~ voltage
_Q) ...
W
0 c:
Q)
Q, Q)
WP;1$;1&;
(b)
junction
+
(c)
Fig. 10.30 The semiconductor diode laser. (a) The P-N junction without bias. (b) The P-N
junction with bias. (c) The physical construction of the laser diode.
254 PHYSICAL PROPERTIES OF MATERIALS
within the conduction and valence bands respectively. This is because of the heavy
doping and the semiconductor, in this condition, is said to be degenerate. A very large
electron concentration exists in the conduction band and a very large deficiency of
electrons (Le. large concentration of holes) in the valence band. If a forward biasing
field is applied, of amplitude the order of the energy gap (1'5 eV in GaAs), the energy
diagram will be as shown in Fig. 10.30b. Because of the much reduced potential
barrier, a very large current flows, but more importantly there will exist in the thin
(~1 J.!m) junction region a population inversion between electron states in the con-
duction band and those in the top of the valence band. Providing the semiconductor is
of a direct band-gap type, so that the probability of radiative recombination is high,
there will be a very good chance of obtaining stimulated emission following an initial
recombination by spontaneous emission.
Amplification will only occur if the stimulated emission is retained within the junctio
region and this can be arranged by fabricating the device as shown in Fig. 1O.30c. The
light is contained in the plane of the junction by cleaving the opposite faces to be
near parallel. The high refractive index for the III-V compounds ensures a sufficient
reflection coefficient at the surface (~ 30 per cent in GaAs, since n =3'3) for lasing
action. Because of the large current and power dissipated in the device, it is operated
in a pulsed mode, the biasing field being applied as a pulse of the order of microseconds
in length. The output light pulse is obtained through the end faces of the crystal. The
most common material used is GaAs for which emission occurs at 0·838 J.!m. This and
others are shown in Table 10.10.
References
1. Bleaney, B. I. and Bleaney, B. Electricity and Magnetism, Oxford University Press (1968).
2. Fowles, G. R. Introduction to Modern Optics, Holt, Rinehart & Winston (1968).
3. Longhurst, R. S. Geometriazl and Physical Optics, Longmans (1975).
4. Wang, S. Solid State Electronics, McGraw-Hill (1966).
5. Dash, W. C. and Newman, R. Phys. Rev. 99,1151 (1955).
6. Spitzer, W. G. and Fan, H. Y. Phys. Rev. 106, 882 (1957).
7. Davis, E. A. and Mott, N. F. Phil. Mag. 22,903 (1970).
8. Goryunova, N. A., Gross, E. F., Zlatkin, 1. B. and Ivanov, E. K. J. Non-Cryst. Solids, 4,57
(1970).
9. Newman, R. and Taylor, W. W. Solid State Phys. 8,49 (1959).
10. Harvey, A. F. Coherent Light, Wiley-Interscience (1970).
11. Schulman, J. H. and Compton, W. D. Color Centers in Solids, Pergamon (1963).
12. Avery, A. J. Ph.D. Thesis, Southampton University (1967).
13. Dexter, D. 1. and Knox, R. S. Excitons, Interscience (1965).
14. Baumeister, P. W.Phys. Rev. 121,359 (1961)
15. Yariv, A. Introduction to Optiazl Electronics, Holt, Rinehart & Winston (1971).
16. Jenkins, F. A. and White, H. E. Fundamentals of Optics, McGraw-Hill (1957).
17. Karninow, I. P. and Turner, E. H. Appl. Optics 5,1612 (1966).
18. Spencer, E. G., Lenzo, P. V. and Ballman, A. A. Proc. IEEE 55,2074 (1967).
19. Hulme, K. F. Rep. Progr. Phys. 36,497 (1973).
20. E.M.I. Photomultiplier Tubes Selection Chart.
21. Hornbeck, J. A. and Haynes, J. R. Phys. Rev. 97, 311 (1955).
22. Curie, D. Luminescence in Crystals, Methuen (1963).
Questions
1. Simple dispersion theory shows that for an ionic material the refractive index due
to electronic polarization in the visible region of the spectrum can be represented
Ne 2 1
n 2 = 1 +-- 2 2'
Eorn CWo - W )
OPTICAL PROPER TIES 255
where Wo is the electronic resonance angular frequency, and N is the polarizable electron
concentration.
Show that for frequencies below resonance, the variation of n is according to eqn
(10.5).
2. Assuming the dispersion of fused silica varies according to eqn (10.5), use the data
in Fig. 10.3 to compute the values of A and B (assume B > C).
3. The F centre can be considered as an electron orbiting a centre of effective charge
+ e. Use the simple Bohr model of the atom to predict the energy of the F centre
absorption in KC1, assuming the absorption transition to be Is - 2p. The refractive
index for KCl is 1·49. How does the value compare with the observed value of2·20 eV?
4. A crystal of calcite is cut in the form of a rectangular slab, the slab being 1 cm
thick, and the slab face being parallel to the optic axis. A beam of light of wavelength
0·5 J,lm is at grazing incidence on the slab face so that its direction is perpendicular to
the optic axis. Given that no = 1·665, ne = 1·489, calculate the separation of the
points on the rear of the slab at which the E- and O-rays emerge.
5. Light of wavelength 0·5461 J,lm from a mercury lamp is incident normally on a
crystal of uniaxial quartz (no = 1·54, ne = 1·55). If the optic axis is parallel to the
crystal surface and the incident light is linearly polarized with its electric vector at
45° to the optic axis, calculate
(a) the phase difference between the E and 0 rays after 1 mm of quartz, and
(b) the thickness of quartz necessary to produce circularly polarized light (Le. to
produce a phase difference of rr/2.
6. In question 5 the crystal is replaced by one of KDP, with optic axis perpendicular
to the surface. If an electric field is now applied parallel to this axis and the direction
of polarization of the incident light is arranged to be at 45° to one of the new induced
principal axes, calculate the voltage necessary to produce circularly polarized light
(no = 1·511,r63 =9·5 x 10- 12 mV-I).
7. The temperature dependence of the energy gap of a semiconductor can be
expressed as
Eg(T) =Eg(O) + aT.
If for InSb,Eg(O) = 0·26 eV and a = -2·8 x 10- 4 eV K- I ,
(a) calculate the maximum wavelength able to be detected by intrinsic photo
conduction at 77 K;
(b) calculate the ionization energy of the donor impurities necessary to be intro-
duced to extend the ability of InSb to detect out to 10 J,lm; and
(c) discuss the usefulness oflnSb as a detector at temperatures greater than 77 K.
11 Superconductivity
11.1 Introduction
This chapter is about the properties of some materials at low temperatures. What is
meant by a low temperature? Normal room temperature is about 300 K and the
absolute zero of temperature at 0 K is infinitely far below that because it is unattain-
able. Between these limits there are a number of fairly easily attainable standard
temperatures, the boiling points of liquids (Table 11.1) which at normal temperatures
are encountered as gases.
TABLE 11.1
Boiling points of liquids spanning
a range of low temperatures
Attention has been drawn above to one feature of a superconducting material, that is,
of having immeasurably small electrical resistance. A point of great importance is that
there is more to the super conducting state than being just resistanceless. To make this
clear some features of the resistanceless state are first considered and then contrasted
with the behaviour of a material in a truly superconducting state.
Consider then a material which has zero electrical resistance. According to the
predictions of classical electromagnetic theory its state can be characterized by saying
that its electrical field vector I is zero. There is, in addition, a direct relation between
an electric field and the rate of change of a magnetic flux density B producing it. This
Maxwell law is written in vectorial form as
curl I = -(dB/dt). (11.1)
Eqn (11.1) merely expresses Faraday's law for the relation between e.m.f. and rate of
change of flux through an area. Hence for the resistanceless state since I = 0
dB/dt = 0; whence B = constant. (11.2)
Oassical theory, then, predicts that the magnetic flux density threading the resistance-
less state will always be constant while that state exists. If this is so, the simple experi-
ment illustrated in Fig. 11.1 can be conducted to show this. Fig. 11.la shows a material
in the form of a cylinder in a resistanceless state A in the absence of a magnetic field.
A magnetic field is then applied transversely and the material is in state B because eqn
(11.2) states that since B = 0 initially in the sample it must remain so, even in the
presence of a magnetic field. The magnetic lines of induction avoid the specimen to ensure
that this happens. But suppose the starting point is as in Fig. 11.1 b where a magnetic
field is applied when the sample is in its normal state (Le. with resistance) say at room
temperature (C). If the temperature is now lowered sufficiently the sample is assumed
at some point to become resistanceless (D). Again the rule that B is constant means
that the flux of B through the sample is unchanged in the transition from the normal
to the resistanceless state. If now the applied magnetic field is switched off the flux
still threads the sample to keep B constant (E).
Suppose now that the experiment is repeated for a sample which displays super-
conductivity. The result is different and forces us to draw a clear distinction between
the hypothetical resistanceless state and the real superconducting state. Referring to
Fig. l1.la again the sample is now in the superconducting state (B = 0) and as before
a magnetic field is switched on. The transition to state B occurs. The result is the same
as for the resistanceless sample. Fig. 11.lc shows however what happens in the second
part of the experiment. In C the material in its normal state (at room temperature
again) is threaded by magnetic flux in the presence of an applied magnetic field. The
temperature is lowered until the material becomes super conducting. Immediately the
magnetic flux is expelled from the specimen (D). B is no longer constant but instead
258 PHYSICAL PROPERTIES OF MATERIALS
(a)
o B
A
(b)
c D E
(c)
c D
o E
Fig. 11.1 Resistanceless cylinder in a transverse magnetic field. (a) Field applied in resistanceless
state B. (b) Field applied in normal state C, temperature lowered until resistance disappearsD, field
removedE.
Superconducting cylinder in a transverse magnetic field. (a) Field applied in resistanceless state
B.(c) Field applied in normal state C, temperature lowered until superconducting state reached D,
field removed E.
B = 0 characterizes the state of the sample. When the magnetic field is switched off
with the sample in the superconducting state, B stays zero (E). The expulsion of
magnetic flux on entry into the superconducting state is known as the Meissner effect.
It will be necessary later to make some reservations about the size of magnetic field
and the type of material but this experiment shows that the fundamental characteristic
of one type of superconductivity is B = 0 and not resistance R = O. Such a material is
a perfect diamagnet whose magnetic susceptibility X =-1 corresponds to B =0 since B =
J.LoH(1 + X). Materials showing this behaviour are known as Type I superconductors.
For the moment discussion is confmed to this type.
11.3 Superconductivity
Metals which display superconductivity differ from other metals in that at temperatures
below a critical temperature Tc the electrical resistivity is immeasurably small and the
SUPERCOND UCTIVITY 259
material is then in the superconducting state. This effect is shown in Fig. 11.2 which
also contrasts the behaviour of a normal metal for which at very low temperatures the
remanent resistivity is characteristic of residual impurities. The resistance of a super-
conductor is believed to be zero rather than just very small. It is possible to initiate a
current in a closed superconducting loop and apparently it will flow indefinitely, un-
diminished, without a source of e.m.f. Such a current is termed persistent.
f-I~
Q: 0
M
Q:
CD
U
Normal metal
...co
C
IJ)
"iii
~
CD
>
".;:;
co
Qj
a:
o 10 20
Temperature T (K)
Fig. 11.2 Temperature dependence of the resistance of a normal and superconducting metal.
TABLE 11.2
Data on some superconductors
Zn 0'88 4 x 10 3 I soft
Ga 1'09 4 x 10 3 I soft
Al 1'20 8 x 10 3 I soft
Th 1'37 1'3 x 10 4 I soft
In 3-40 2'3 x 104 I soft
Sn 3'72 2'5 x 104 I soft
Hg(P) 3'95 2'7 x 104 I soft
Hg(a) 4'15 3'3 x 104 I soft
Ta 4"48 6'7 'x 104 II soft
V 5'30 1'05 x 105 II soft
Pb 7"19 6'5 x 10 4 I soft
Nb 44% Ti 10'5 9"6 x 10 6 II hard
Nb 9'5 l'56x10 s II soft
Nb 25% Zr 10'8 5'6 x 106 II hard
NbN 16 8 x 106 II hard
V3 Ga 16'5 2"8 x 10 7 II hard
V 3Si 17 1'3 x 10 7 II hard
Nb3A1 18 II hard
Nb3Sn 18'5 1'6 x 10 7 II hard
Nb 3Alo·s Ge O·2 20'9 II hard
Nb3Ge 22'3 II hard
260 PHYSICAL PROPERTIES OF MATERIALS
The normal to superconducting state transition is an extremely sharp one, the drop
from finite to zero resistance occurring over a temperature range of 0·05 K for a pure
material. Superconductors are characterized by Te and values of Te for some are given
in Table 11.2. Important as Te is, cooling below it will not necessarily create a super-
conducting state. As has been observed, magnetic induction is probably more funda-
mental than temperature. It is found in fact that the presence of a magnetic field can
prevent the formation of the superconducting state even below Te. Above Te a given
material can never be superconducting but below it there is, for every temperature, a
critical value of magnetic field He which, if exceeded, will destroy the superconducting
state. This magnetic field tends to a maximum value (H 0) at the absolute zero of tem-
perature. The general behaviour can be represented as a plot of the critical field He
versus temperature and is shown for lead in Fig. 11.3. The region enclosed by the
coordinate axes and the curve is the superconducting region and that outside is
the normal region.
-
0;- 7
E
.,<t.
0
';(
J: " 5
"0
..
~
o
~
Cl
4
3-
IV
E 2 Superconducting
lead
o 5 10
Temperature (K)
Fig. 11.3 Temperature dependence of critical magnetic field He for lead.
The Meissner effect has shown that in the presence of a magnetic field less than He,
B = 0 within a superconductor. An explanation of how this comes about is required.
Closely linked with the problem is the behaviour of a superconductor which carries a
SUPERCONDUCTIVITY 261
current supplied from an external source such as a battery connected across it. This
resistanceless current is 'Called a supercurrent. Since B = 0 in this case too, it implies
current density J = 0 inside the superconductor. Since the current cannot flow inside
it must flow as a surface current. One view of the Meissner effect then could be that
a surface current flows in this case also to make B =0 inside by opposing any magnetic
field which would produce a fmite B in the bulk. An alternative and equally valid view
is to treat the material as having zero relative magnetic permeability, i.e. tJr = O.
Surface supercurrents can flow then either as a screening current to make B = 0 or
as a transport cu"ent driven by an external source. It is reasonable to suppose that to
avoid an infmite current density these supercurrents will flow in a region of finite
thickness near the surface. This is found to be so and the magnitude of the current and
of the magnetic field associated with it falloff exponentially with distance from the
surface. B is not zero in this region and the magnetic field decreases with distance x
into the superconductor according to
H=Ho e- xl\ (11.4)
where H 0 is the value of magnetic field at the surface and A is a characteristic length
known as the penetration depth; A is the distance for H to fall from Ho to Hole
(Fig. 11.4).
Hs - Surface value
The presence of surface currents and associated magnetic fields will obviously have a
profound effect on the properties of thin fllm superconductors, or indeed of any super-
conductors whose dimensions are comparable to the penetration depth. Some values
of A are given in Table 11.3. The small values of A show why these bulk Type I
superconductors exhibit ideal diamagnetism.
Equations to describe the superconducting state were developed by F. and H. London
by modifying Maxwell's electrodynamic equations. The London equations successfully
predict the distribution of current and magnetic field and A occurs as a characteristic
parameter in them [2].
262 PHYSICAL PROPERTIES OF MATERIALS
TABLE 11.3
Penetration depths
Superconductor A(O K)
Hg 7 x W- 8 m
In 6'4 x 10-8 m
Ph 3'9 x 10-8 m
Sn 5'0 x 10-8 m
One of the most fruitful qualitative models of a Type I superconductor is the two-
fluid model. In this model, the superconductivity of a region of superconductor at a
temperature T < Tc arises from an electron fluid which comprises two distinct types
of electron. There are those which have superconducting properties, and move without
resistance, the superelectrons and those with normal properties, normal electrons. This
fluid representing all the conduction electrons in a solid contains both types in a relative
proportion which is temperature dependent. At Tc all the electrons are normal and as
the temperature decreases an increasing proportion become superelectrons until at
T = 0 all are superelectrons. The transition from one type to the other is to be regarded
as a thermal excitation, from a super or perfectly ordered state (of zero entropy) to the
normal condition. The proportion in the super state is known as the order parameter W
which depends on temperature below Tc as follows:
(11.5)
Although this model proves extremely useful in thinking about superconducting
behaviour it must not be taken too literally.
-----~----~:
H=Ha
-=------
H='Jv2 Ha
Fig. 11.5 Superconducting sphere in applied field Ha showing enhanced field at equatorial
positionsAB.
SUPERCONDUCTnnTY 263
must be the same as that just inside. Now within any material there will be an intrinsic
demagnetizing effect which either adds to or subtracts from the applied field. For the
(diamagnetic) superconductor this internal field is additive so that the total internal
field Hi is given by
(11.6)
where n is a demagnetizing factor which takes different values for differently shaped
specimens. For a sphere it has the value 1/3. Thus from eqn (11.6) the internal field at
A is (3/2)H•. As a result the external field at A must also have this value.
Now suppose that H. is increased until the field at A reaches the critical value Hc.
The material of the sphere near A will want to become normal and so flux will pene-
trate at the equator. As an immediate consequence this material, now paramagnetic,
will be subject to an internal field which has reverted to a value H. less than the
critical field H c. Clearly this is a paradoxical situation. It is resolved by the specimen
adopting a completely different pattern of flux within it, as shown in Fig. 11.6. In this
Normal state
Superconducting state
Fig. 11.6 The intermediate state.
state the equatorial situation is once more stabilized, the flux having penetrated in the
form of laminae leaving superconducting material at A. Moreover, the magnetic energy
of the specimen is lowered by allowing flux to penetrate. It costs more energy to keep
it out than to let it in. If the applied field is further increased more laminae will develop.
This state of a Type I superconductor in which flux has partially penetrated is called
the intermediate state. There are a number of experimental techniques by which the
intermediate state may be directly observed. Examples are mapping with tin powder [3]
or using magneto-optics [4] .It is found that the size of the laminae is of the order of
a millimetre.
In general the intermediate state consists of regions of normal material interwoven with
superconducting regions. According to the two-fluid model, there will be a high pro-
portion of superconducting electrons in the superconducting regions, the proportion
depending on the temperature. In the normal regions there are none. At the boundary
between a normal and superconducting region it is unreasonable to suppose that this
transition occurs abruptly. The proportion of superelectrons is measured by the order
parameter Wand Pippard [5] was the first to recognize the idea of a transition length
associated with W, a length in which W changed from its value in the super conducting
region to zero in the normal region. This distance is known as the coherence length (~)
and is about 10-6 m.
264 PHYSICAL PROPERTIES OF MATERIALS
I
I
I
G"----------i--------I;~~
I S
I: '-J~;"
__ c (Po,;';"
magnetic
Gn _ _ _ _ _ _ _ .'.. _ _ _ _ _ _ 2 contribution)
I lJ.oHc2 (Negative
I - - ordering
I
I
_ _ 1- contribution)
(b)
Fig. 11.7 (a) Free energies in normal and supereondueting states. (b) Equilibrium of normal and
supereonducting states in presence of a critical field He.
SUPERCONDUCTIVITY 265
The situation near the boundary is more complicated because both the magnetic and
electron ordering contributions vary with distance and the algebraic sum is not
necessarily zero. The characteristic length which governs how quickly the magnetic
contribution to G rises from the value Gn to Gn + !lloHc 2 is the penetration depth A
while for the order parameter it is the coherence length ~. Thus two distinct cases
arise: (a) when ~ > A and (b) when A <~. These are illustrated in Fig. 11.8(a) and (b)
G ______________
G
~
Normal \super~ondueting
(a) (b)
G G
G. Gn G.
-----+~------------~--- ~---+~------------~~
(e) (d)
Fig. 11.8 Magnetic and ordering contributions to Gibbs free energy near a normal-superconducting
boundary in equilibrium for (a) ~ > Aand (b) ~ < A, and their resultant for (c) ~ > A, (d) ~ < A.
while the variation of total free energy as a function of distance from the boundary is
shown in Fig. 11.8(c) and (d). For ~ > A the surface energy is positive while for ~ < A
it is negative. Fig. 11.8a describes the situation for Type I superconductors and so it
costs energy to form normal-superconducting boundaries. This factor determines the
pattern of boundaries in the intermediate state. Case (b) will be discussed in the next
section.
-M
...
"
c
o
III
N
II
C
III
III
:E
~~ j~ ~
Applied magnetic field H
Fig. 11.9 Magnetization of a superconductor as a function of applied field for different specimen
geometries: (1) a sphere, (2) cylinder in transverse field, and (3) cylinder in longitudinal field.
of Fig. 11.10, assuming zero demagnetization factor. When about 10 per cent indium
is added to give a lead-indium alloy the behaviour is no longer characteristic of Type I.
The magnetization curve (2) is completely different and characteristic of Type II
materials.
-M
..
c
o
III
N
ic
III
III
:E
Hel He He2
Applied magnetic field H
Fig. 11.10 Effect of impurity on magnetization curve of a superconductor: (1) pure lead, and (2)
lead-indium alloy.
Above a critical field called Hcl penetration occurs in ftlamentary form so that the
material remains superconducting. Increase of field causes further penetration until
H =H c2, an upper critical field at which complete penetration occurs and the super-
conductor becomes normal. The condition of the specimen between Hcl and HC2 is
called the mixed state. HC2 is generally much higher than Hc for a Type I superconductor
(Table 11.2).
i.e.
hence
(11.10)
The critical current Ic will decrease linearly with increase of the applied field until it
reaches zero at Ha = Hc /2. Similar considerations apply to pure Type II superconductors
268 PHYSICAL PROPERTIES OF MATERIALS
_H_8 -~-~-
*"""""
--------~~~~-------
fi----=---:
~1 ~ :
Section "x-x"
(b)
Fig. 11.11 (a) Wire carrying current I subjected to transverse field Ha' (b) Cross section of wire
showing fields at equatorial positions on the surface.
for Ha <Hcl ' that is when the superconductor is not in the mixed state. The current
carrying capability is little different from the Type I case. For Ha > Hc!> pure materials
have a small critical current which decreases to zero near Hc2 ' This is not the case for
impure Type II materials in the mixed state. The behaviour of the critical current in a
wire as a function of a magnetic field applied transversely is shown in Fig. 11.12 for
both pure and impure Type II material.
The latter is able to carry a much higher critical current, typically 108 Am- 2 flowing
through the bulk of the solid. The relative values of the currents in the mixed state shown
in the figure differ by several orders of magnitude. What is the reason for this remarkable
difference in current carrying ability? The important features are the presence of the mixed
state and the requirement for an impure material, impure in the sense of possessing
metallurgical faults (dislocations) and/or chemical impurity. The explanation lies in
the interaction of the current along the wire with the fllaments of the mixed state
aligned transversely to the current direction as illustrated in Fig. 11.13.
The filaments and their associated magnetic flux experience a Lorentz force F of
the form I x B acting in the plane of the paper. In pure Type II materials this force can
cause the fllaments to move against the viscous drag of their surroundings. This requires
work to be done, the energy coming from the currents, and is a manifestation of
electrical resistance. The material is then no longer superconducting. When the material
is very impure the resistance to fllament motion is strong because the impurities pin
SUPERCONDUCTIVITY 269
Force
Fig. 11.13 The mixed state. Normal filaments threading a current carrying wire subjected to a
transverse field.
the fIlaments. A much larger current is now required to provide the Lorentz force for
migration and the critical current is correspondingly greater. Type II materials which
are deliberately heat treated and cold worked to introduce dislocations so as to improve
Je are known as hard superconductors. In contrast, materials which are relatively dis-
location free are called soft and include Type I and some Type II.
may be appreciated from the fact that from the time of the discovery of supercon-
ductivity in 1911, almost forty years elapsed before any hope of a solution appeared.
A satisfactory theory was formulated in the 1950s based on the ideas of Frohlich,
Cooper, Bardeen and Schrieffer [7].
(a)
EF E
L--
I
I
I
I
I
:.-- 21:
I
(b) I
I
I
I
I
I
I
I
I
EF E
Fig.l1.14 Density of states as a function of energy (a) normal metal and (b) a superconductor.
For superconductors, however, those electrons which give rise to supercurrents are
in an even lower energy state. An energy gap appears to exist in the density of states
distribution when the temperature is below the critical temperature Tc (Fig. 11.14b).
The steep rise in the density of states shown in the diagram is a consequence of the
appearance of the superconducting energy gap and has been confirmed experimentally
[8] .
The gap of width 2€ is positioned at the Fermi level and the density of states is
exceptionally large immediately above and below it. Experimental evidence for this
energy gap has come from specific heat measurements [9] and from optical data,
particularly infra-red absorption spectra [10]. The latter have shown a sharp absorption
edge. Above the frequency corresponding to this edge photons are strongly
absorbed. Such measurements are consistent with the excitation of electrons
SUPERCONDUCTIVITY 271
by photons across a gap just as in a semiconductor photo detector, but here the gap
width is very small, about 10-4 eV. The effect of the gap, then, can only be observed
at low temperatures. The width of the gap is also found to be temperature dependent
decreasing to zero at Te and remaining zero for T> Te (Fig. 11.15).
1.0r--_ ___
o 1.0
Reduced temperature
The evidence for a gap in the energy spectrum also suggests the possibility that
electrons in the energy states below the gap may have different properties from those
above and perhaps be responsible for superconductivity. If thermal excitation across
the gap is to be small, kT must be less than 2€ (-10- 4 eV) indicating that temperatures
of about 1 K are required. Also the specific heat measurements indicate that the super-
conducting state has a lower entropy than the normal state and so is in some way
more ordered. Since the superconducting-normal transition only involves changes in
electronic properties, other properties being little affected, it is reasonable to suppose
that the greater degree of order is associated with the electrons. A picture emerges
therefore in which ordered electrons below the energy gap are in some way responsible
for superconductivity while those above the gap possess normal properties.
found to depend on the isotopic massM of the particular element and given by
constant
{l 1.1 I)
.../M .
This result suggests that lattice vibrations play an important role and that indeed inter-
actions of type (b) are significant. It was the nature of this Frohlich interaction that
took so long to elucidate. It is represented schematically in Fig. 11.16.
E E
+k
E
- Electric field E
- Electric field
s'
Fig. 11.17 E-k diagrams for electrons in: (a) normal metal with zero electric field; (b) supercon-
ductor with zero electric field; (c) normal metal in applied electric field; and (d) superconductor in
applied electric field.
At temperatures above 0 K some electrons will exist in energy states above the
Fermi level, distributed according to Fermi-Dirac statistics. For the normal metal
successive states near the Fermi level are only about 10-7 eV apart and so even at
low temperatures some of the higher states will be occupied. In a superconductor
an energy gap of 10-4 eV has to be surmounted and very few electrons will exist in
states above the gap near 0 K. Those that do will be single unpaired electrons possessing
normal properties. As the temperature increases the number of paired electrons de-
creases and the thermal effect on the lattice will cause the attractive interaction to be
less favourable and hence the energy gap to become smaller. The disappearance of the
energy gap coincides with the critical temperature Te. This so-called BCS theory closely
predicts the temperature dependence of the energy gap shown in Fig. 11.15.
The effect of an electric field applied to a normal metal is to impose a general drift
motion on the free electrons. This drift, which is opposite to the field direction be-
cause of the negative charge of the electrons, is illustrated by the E-k diagram of
Fig. 11.17c. This results in an increase in the total energy of the electrons resulting
from the kinetic energy associated with the drift velocity. If the electric field is
removed the state represented by Fig. 11.17 c will not perSist as the drifting electrons
will exchange momentum with the lattice ions. The total energy decrease and the
274 PHYSICAL PROPERTIES OF MATERIALS
general state of motion will rapidly return to the equilibrium state represented by
Fig. 11.17a.
If a field is applied to a superconductor the effect on the paired electrons must
be considered. In zero field each electron in a pair will have equal and opposite mo-
mentum and an applied field will increase the momentum of one of a pair by the same
amount as the momentum of the other is decreased. Each pair will therefore drift as a
unit as shown in Fig. 11.17d. When the field is removed the state of motion of the
electrons will only return to the symmetrical E-k distribution of Fig. 11.17b if it is
energetically favourable to do so. To lower the energy of a particular pair (say aa' in
Fig. 11.17 d), the scission of the pair bond would be necessary and this would require
the energy of interaction to be supplied equal to the energy gap. In a superconductor
below Tc the kinetic energy of drift saved by splitting up a pair is insufficient to offset
the energy required to break the bond. The asymmetric E-k distribution of Fig. 11.17d
persists therefore and the current still flows after the electric field is switched off.
The specimen exhibits superconductivity as a persistent current. There is zero resistance
because the electrons cannot be scattered by the lattice below Tc owing to the need to
break a pair bond and this is not energetically favourable for the reasons described.
TABLE 11.4
Properties 0/ superconducting magnet materials
References
1. Kittd, C. Introduction to Solid State Physics, Wiley (1971).
2. London, F. and London, H. Proc. Roy. Soc. A149, 71 (1935).
3. Faber, T. E. Proc. Roy. Soc. A248, 460 (1958).
4. de Sorbo, W. and Healy, W. A. Cryogenics 4,257 (1964).
5. Pippard, B.Physica 19,765 (1953).
6. Trauble, H. and Essmann, U. J. Appl. Phys. 39,4052 (1968).
7. Saunders, G. A. Con temp. Phys. 7,192 (1965).
8. Hebel, L. C. and Slichter, C. P. Phys. Rev. 113, 1504 (1959).
9. Phillips, N. E. Phys. Rev. 134, A385 (1964).
10. Richards, P. L. and Tinkham, M. Phys. Rev. 119,575 (1960).
11. Reynolds, C. A. Serin, B. and Nesbitt, L. B. Phys. Rev. 84,691 (1951).
12. Williams, J. E. C. Superconductivity and its Applications, Pion (1970).
13. Catterall, 1. A. Con temp. Phys. 12,495 (1971).
14. Bitter, F. Rev. Sci. Instr. 10,373 (1939).
15. Snowdon, D. P. Sci. Am. April 84 (1972).
16. Harding, J. T. Jet Propulsion, Lab. Tech. Rept. No. 32-897.
17. Ittner, W. B. and Kraus, C. J. Sci. Am. July, 124 (1961).
18. Buck. D. A. Proc. IRE 44. 482 (1956).
19. Crowe, J. W. IBM J. 1,295 (1957).
20. Burns, L. 1., Alphonse, G. A. and Leek, G. W. IRE Trans. Electron Comput. EC-I0, 438 (1961).
21. Geballe, T. H. Sci. Am. November, 22 (1971).
Questions
1. Given the data of Table 10.1 calculate the critical current density for a 1 mm dia-
meter wire of lead at (a) 4·2 K and (b) 7 K. A parabolic dependence of He upon T may
be assumed.
2. A superconducting Nb solenoid requires a current of 20 A in order to produce a
magnetic field of 6 x 104 Am-I. If the solenoid is immersed in liquid helium (4'2 K)
and H 0 for Nb is 1'56 x 105 Am-I, estimate the minimum diameter wire that may be
used.
3. The BCS theory of superconductivity shows that the super conducting energy gap
at 0 K is given by 3·5 kTe • Calculate the maximum wavelength of microwave radiation
that will be absorbed at 0 K in (a) lead and (b) aluminium.
4. Aluminium is characterized by the following properties: Tc = 1·20 K; H 0 =
8 X 103 Am-I; atomic wt = 26·98 kg; density = 2698 kg m- 3 and it is monovalent.
A cylindrical sample of diameter 5 mm is subjected to a magnetic field parallel to
its axis, at 0 K.
(a) Calculate the energy of magnetization in a 2 cm length when the magnetic field
is almost sufficient to bring the return of the normal state.
(b) What is this energy expressed per atom?
(c) The BCS theory states that the superconducting energy gap at 0 K is given by
3·5 kTc. Compute the proportion of conduction electrons in aluminium which
take part in the pairing interaction at 0 K.
S. Survey the values of Tc for the elemental superconductors and see if there exists
any well defmed pattern in its variation with the number of valence electrons per atom.
Appendix I Electrons in Solids
One fruitful model of a solid, particularly a metal, is that the electrons inside it are
confined by a potential well of the same volume as the solid, the potential being
supposed constant. The potential well is the three-dimensional analogue of the one-
dimensional well shown in Fig. AI. The problem is then to fmd the permitted states
v=o
v=-vo
x=o x= L
of a single electron in this well. These states are then assumed to be filled by all the
electrons treated as non-interacting particles except for the effect of the Pauli
Exclusion Principle. The quantum states of a single electron in the well are found by
solving the Schrodinger equation (AU) with appropriate boundary conditions.
(ALl)
The boundary conditions are 'IT =0 at x =0 and x =L (and similar conditions for the
y and z directions). These conditions are not strictly true unless the well is infinitely
deep but are a good approximation. The electron may be regarded as a wave character-
ized by a wave function l/I (xyz) and a wave number k (k = 27T/wavelength (A)). Since
the de Broglie relation gives the momentum of the electron as p = tzk, the kinetic energy
of the electron =p2/2m = 1i 2k 2 /2m =E - (- Vol, i.e.
h2 k 2
E+ Vo = 2m ' (AI.2)
APPENDIX I 279
so that the SchrOdinger equation becomes
li 2 1i 2 k 2
-11 2 1/1+-1/1=0. (AI.3)
2m 2m
With the specified boundary conditions the solution has the form
Irrx mrry nrrz
./, = A sin - sin - - sin- (AI.4)
'I' L L L '
where I, m and n are integers.
This solution arises by fitting into the box a three-dimensional sine wave satisfying
the boundary conditions. The latter require the rejection of otherwise possible
solutions and lead to a set of quantized states of differing wave number k. The argu-
ments of the sine function makes 1/1 = 0 at x = 0 and L, y = 0 and Land z = 0 and L if
I, m and n are integers. The set of possible states for the electron in the box then have
k values given by kx = Irr/L, ky = mrr/L, k z = nrr/L with
(AI.5)
The group of quantum numbers I, m, n takes successively higher values, the lowest
state having 1 = 0, m = 0, n = O.
The energy spacing between successive levels is clearly of the order of (112 12m) . rr2/L 2
(eqn (Al.2». For a crystal of volume 10- 6 m 3 this spacing is about 10- 15 eV near the
bottom of the well.
Information about the number of states in a given range of wave number k or energy
E may be obtained as follows. Consider a three-dimensional k space in which the co-
ordinate axes are kx, ky and k z (Fig. A2a). A point in this space represents the
dk
(a) (b)
Fig. A2 (a) Spherical shell embracing momenta between k and k + dk in the positive octant of
three-dimensional k space. (b) Cube of k space occupied by the lowest state of volume (rr/L)3.
280 PHYSICAL PROPERTIES OF MATERIALS
and this occupies a volume (nIL)3 in k space. Since these states are very numerous,
dividing the volume of the spherical shell between the vectors k and k + dk by the
volume occupied by one state gives us the result. But only the states in the positive
octant need be considered since the k's in all other octants merely introduce a change
of phase of the wave function and produce no new states. Therefore, the number of
states having wave number between k and k + dk is
1 4nk2 dk Vk 2 dk
dN(k) = 8"
(nIL)3 =~, (A 1.6)
N(k) = - V
2n2
f k
k 2 dk =Vk- .
6n 2
3
(Al.7)
o
This number should be multiplied by 2 to take account of spin by the Pauli Exclusion
Principle. Eqns (Al.6) and (Al.7) can easily be expressed in terms of kinetic energy by
using eqn (Al.2). Then from eqn (Al.6), the number of states per unit wave number
range per unit volume of crystal
where n =N(k)
V
In terms of kinetic energy, the number of states per unit energy interval per unit
volume of crystal
dn dn dk k2 m
dE =dk . dE =2n 2 • fz2 k (Al.8)
Fermi E,F
laval
EF =3·12 eV and so TF =37 000 K. This extremely high effective temperature at the
top of the electron distribution can never be utilized for obtaining heat energy because
the lattice of ions which forms the backbone structure of the solid can never become
so hot. The reason is that the lattice ions can only exchange energy with the electrons
by causing some of them to be excited to higher energy levels. The only energy avail-
able is that of the lattice itself which is at room temperature, say, and this energy is
about 0·025 eV per ion. Only electrons near the top of the distribution will be affected
and the energy they receive will be very small and so in turn will be the energy they
can return to the lattice when they drop back to their ground state levels. A similar
argument shows that only a small proportion of the electrons present will contribute
to the specific heat of the solid. This prediction contrasts with the classical kinetic
theory picture where all the electrons in the solid contribute thermal energy of about
(3/2) kT. The classical result is not in accord with the experimental results whereas the
quantum result is.
When energy is supplied to the electron distribution at T = 0 K electrons near the
top of the distribution are excited to nearby empty levels (Fig. A4). The probability
fiE) T=Q
04-------------y.-=-=~~~~----
E
Fig. A4 Probability!(E) of occupancy, by an electron, of a level of energy E at various tempera-
tures according to Fermi statistics.
Metal Li Na K
The above model is able to explain satisfactorily some features of metallic behaviour
but the assumptions on which it is based, particularly the neglect of Coulomb inter-
action between electrons, means it is not very realistic. Moreover, one of the major
features of a real crystal is the presence of a spatially varying and periodic electric
potential arising from the presence of lattice ions. The next step must be to consider
how this potential can be included in the model. One way is to extend the previous
treatment by considering a periodic series of cubical potential boxes through which
the electron is free to move as an independent particle. The wave function 1/1 of an
electron will then be determined by a set of periodic boundary conditions which again
lead to a set of quantized states.
A one-dimensional representation of the actual potential in which a 'free' electron
moves in the crystal is shown in Fig. AS. The deep attractive wells mark the positions
.. x
V(x)
+ + + +
Fig. AS Potential experienced by an electron in traversing a line of regularly spaced ions.
of the positive ions which are made up of a nucleus plus tightly bound inner shell
electrons. These ions are regularly spaced a few Angstroms apart, the lattice spacing.
It turns out that the electron moving in such a potential can be described by a plane
wave modulated by a factor depending on the potential but essentially the state of the
APPENDIX I 283
electron is again characterized by its wave number k. The momentum of the electron
in these circumstances is called the crystal momentum. The possible values of k are
quantized by the periodic boundary conditions. Each component of k has the form
kx = mrlL, where n is an integer and 1/L is the reciprocal of the crystal dimension in
the appropriate direction. These states are so densely packed in k space as to be quasi-
continuous. When the wavelength of an electron travelling through the crystal is such
that it will be diffracted by a particular set of atomic planes, the Bragg diffraction
condition is satisfied:
n"A. = 2d sin 8,
where n is an integer, "A. the electron wavelength, d the lattice spacing and 8 the angle
of incidence. This condition implies that k = mr/(d sin 8). At this k value the travelling
wave representing the electron can be considered as having been converted to a stand-
ing wave. The solution of the SchrOdinger equation shows that at this particular k
value, there is a gap in the energy spectrum where no real solutions exist, in fact
where k is imaginary.
In one dimension the situation is as shown in Fig. A6. Here the energy of an
electron moving in the x direction is plotted as a function of its wave number. Between
k =0 and k =rrld there is a quasi-continuous set of states (drawn as a continuous
E I
~ -------------------~ I
E2 - - - - - - - - - - - - - - - - - - IP
Fig. A6 One-dimensional E-k diagram for an electron in a periodic potential of spatial period
d. (Region kx = 0 to kx = Trld is the first Brillouin zone.)
line) for the electron. When kx = rrld it encounters an energy barrier of height (El - E 2 )
preventing further increase of wave number unless this amount of energy is available to
excite the electron over the barrier. There are no permitted states for the electron
between P and Qat kx = rrld.
The region OP is said to constitute the first Brillouin zone and by extending the
principle through higher k values and to three dimensions, the states of an electron in
a crystal may be grouped into a series of such zones separated by energy gaps. When
an electron meets such a gap it needs additional energy to move into the next zone.
Within a zone an electron can only move if there are empty states which it can occupy
on gaining energy. Just as in the potential well model the lower states will be filled.
Only the occupied zones of highest energy and the empty zones immediately above
them are usually of interest. The electrons near the top of the Fermi distribution will
be of greatest interest. Another way of representing the distribution is to draw a series
284 PHYSICAL PROPERTIES OF MATERIALS
of energy bands instead of Brillouin zones without specifying k. Fig. A7 shows a set of
such bands. This band picture is that arrived at in Chapter I, from a different starting
point.
The shaded bands are supposed full of electrons and the upper one empty. The
highest fully occupied band is the valence band and the next empty (or partially
occupied) band the conduction band. Each band contains many quantum states closely
spaced in energy. The gaps between bands will vary widely for different materials and
for different directions in a crystal of a given material, and they may disappear
altogether so that bands completely overlap. Depending on the material considered
and therefore on the state of occupancy of the levels in these bands the Fermi level
may be found in the conduction band, in the gap, or in the valence band.
The anomalous behaviour of an electron near a zone boundary, that is near the top of
an energy band is very different from free electron behaviour. These unusual properties
can be deduced from the E-k diagram using some simple classical ideas. From the
de Broglie relation the wave number of an electron of velocity u may be written
hk = mu, whence
u =hk/m (AI.9)
hlkl
If E = - - , it follows that
2m
(A 1.1 0)
Comparison of eqns (AI.9) and (AI.l 0) show that the slope of an E-k curve is a
measure of the electron velocity.
Newton's laws applied to the motion of the electron give
force F =-d dk
(mu) =Ii - . (Al.l1)
dt dt
Also using eqn (Al.9)
F du hdk
-=-=--
m dt mdt
APPENDIX I 285
Using eqn (AI.l 0)
1t
dI
I
I
I
L -_ _ _ _ _ _ +-__ kx I
I
1t I
d
(a) (b)
Fig. AS (a) Variation of dE/dkx(o: velocity) with kx (0: momentum) for an electron in the first
Brillouin zone. (b) Variation of d 2 Eldkx 2 (0: l/mass) with kx for an electron in the rust Brillouin
zone. (One-dimensional 'crystal'.)
and d 2 E/dk 2 within a zone. Since the electron mass can change depending on the
curvature of the E-k curve, the mass of an electron in a crystal is called its effective
mass. It arises because of the influence of the crystal potential on the motion of the
electron. The appearance of a negative effective mass is another way of interpreting
Bragg diffraction of the electron wave. When an electron near the top of a full band
having a negative effective mass is excited to a partially empty band the vacancy it
leaves behind 'the absence of an electron of negative effective mass', may be treated
as a particle of positive charge and positive mass. This entity is called a hole. The
motion of a hole is an alternative and convenient way to describe the motion of all
the remaining electrons in the band.
APPENDIX II
Periodic chart of the elements
II III IV V VI VII VIII
r--
1 1 2
H H He
1·00797 1·00797 4·0026
2
3 4 5 6 7 8 9 10 8
Li Be B C N 0 F Ne
6·939 9·0122 10·811 12·01115 14·0067 15·9994 19·9984 20·179
2
11 12 13 14 15 16 17 18 8
8
Na Mg AI Si P S CI Ar
22·9898 24·305 TRANSITION ELEMENTS 26·9815 28·096 30·9738 32·064 35·453 39·948
,/ \
19 20 21 24 25 26 27 28 29 30 31 33 34 35 36 2
22 23 32 8
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr 18
8
39·102 40·08 44·956 47·90 50·942 51·996 54·9380 55·847 58·9332 58·71 63·54 65·37 69·72 72·59 74·9216 78·96 79·909 83·80
2
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 8
18
Rb Sr V Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe 18
85·47 87·62 88·905 91·22 92·906 95·94 101·07 102·905 106·4 107·870 112-40 114·82 118·69 121· 75 127·60 126·9044 131·30 8
(99)
2
55 56 57 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 8
18
Cs Ba *La Hf Ta W Re Os Ir Pt Au Hg TI Pb Bi Po At Rn 32
132·905 137·34 138·91 178·49 180·948 183·85 186·2 190·2 192-2 195·09 196·967 200·59 204·37 207·19 208·980 18
(210) (210) (222) 8
87 88 89
tAc *Lanthanum Series
Fr Ra 2
58 59 60 61 62 63 64 65 66 67 68 69 70 71 8
(223) (226) (227) 18
Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Vb Lu 32
140,12 140·907 144·24 150·35 151·96 157·25 158·924 162·50 164·930 167·26 168·934 173·04 174·97 9
(147) 2
I \ .. ____ __ ! _______
'"
______
t Actinium Series
, 2
90 91 92 93 94 95 96 97 98 99 100 101 102 103 8
18
Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr 32
32
232·038 (231) 238·03 (237) (244) (243) (247) (247) (252) (254) (257) (257) (255) (256) 9
--- 2
APPENDIX III LIST OF THE ELEMENTS
1 HYDROGEN H 1·008 1 I - 1
2 HELIUM He 4·003 2 - 2 2
3 LITHIUM Li 6·939 2 1 I 4 3
4 BERYLLIUM Be 9·012 2 2 II 5 4
5 BORON B 10·81 2 2 1 III 6 5
6 CARBON C 12·01 2 2 2 IV 6 6
7 NITROGEN N 14·01 2 2 3 V 7 7
8 OXYGEN 0 16·00 224 VI 8 8
9 FLUORINE F 19·00 2 2 5 VII 10 9
10 NEON Ne 20·18 2 2 6 - 10 10
11 SODIUM Na 22·99 2 2 6 1 I 12 11
12 MAGNESIUM Mg 24·31 2 2 6 2 II 12 12
13 ALUMINIUM AI 26·98 2 2 6 2 1 III 14 13
14 SILICON Si 28·09 2 2 6 2 2 IV 14 14
15 PHOSPHORUS P 30·99 226 2 3 V 16 15
16 SULPHUR S 32·06 2 2 6 2 4 VI 16 16
17 CHLORINE CI 35·45 2 2 6 2 5 VII 18 17
18 ARGON Ar 39·95 2 2 6 2 6 - 22 18
19 POTASSIUM K 39·10 2 2 6 2 6 - 1 I 20 19
20 CALCIUM Ca 40·08 2 2 6 2 6 - 2 II 20 20
36 KRYPTON Kr 83·80 2 2 6 2 6 10 2 6 48 36
37 RUBIDIUM Rb 85·47 226 2 6 10 2 6 - - 1 I 48 37
38 STRONTIUM Sr 87·62 2 2 6 2 6 10 2 6 - - 2 II 50 38
39 YTTRIUM Y 88·91 2 2 6 2 6 10 2 6 1 - 2 lila 50 39
40 ZIRCONIUM Zr 91·22 226 2 6 10 2 6 2 - 2 IVa 50 40
56 BARIUM Ba 137·3 2 2 6 2 6 10 2 6 10 - 2 6 - - 2 II 82 56
57 LANTHANUM La 138·9 2 2 6 2 6 10 2 6 10 - 2 6 1 - 2 Ilia 82 57
58 CERIUM Ce 140·1 2 2 6 2 6 10 2 6 10 2 2 6 - - 2 Ilia 82 58
59 Pr 140·9 2 2 6 2 6 10 2 6 10 3 2 6 - - 2 Ilia 82 59
PRASEODYM I UM
NEODYMIUM Nd 144·2 2 2 6 2 6 10 2 6 10 4 2 6 - - 2 Ilia 82 60
60
Chapter 1
Sproull, R. L. Modern Physics, Wiley (1964).
Chapter 2
Barrett, C. S. and Massalski, T. B. Structure 0/ Metals, McGraw-Hill (1966).
Stille, J. K. Introduction to Polymer Chemistry, Wiley (1962).
Chapter 3
Gilman, J. J. (Editor). The Art and Science o/Growing Crystals, Wiley (1963).
Chapter 4
James, R. W. X-Ray Crystallography, Methuen (1928).
Phillips, F. C. Introduction to Crystallography, 3rd edn, Longmans (1963).
Kelly, A. and Groves, G. W. Crystallography and Crystal De/ects, Longmans (1973).
Chapter 5
Smallman, R. E. Modern Physical Metallurgy, Butterworths (1970).
Nielsen, L. E. Mechanical Properties 0/ Polymers, Van Nostrand (1962).
Brydson, J. E. Plastics Materials, Iliffe (1969).
Chapter 6
King, A. L. Thermophysics, Freeman (1962).
Goldsmid, H. J. The Thermal Properties 0/ Solids, Routledge & Kegan Paul (1965).
Morse, P. M. Thermal PhYSics, Benjamin (1964).
Reif, F. Fundamentals 0/ Statistical and Thermal Physics, McGraw-Hill (1965).
Chapter 7
Mott, N. F. and Jones, H. Theory o/the Properties 0/ Metals and Alloys, Dover (1958).
Bulman, P. J. Trans/e"ed Electron Devices, Academic Press (1972).
Mott, N. F. and Davis, E. A. Electronic Processes in Non Crystalline Materials, Oxford
University Press (1971).
Mott, N. F. Metal-Insulator Transitions, Taylor & Francis (197 4).
Chapter 8
Zaky, A. A. and Hawley, R. Dielectric Solids, Routledge & Kegan Paul (1970).
Cady, W. G. Piezoelectricity, McGraw-Hill (1946).
Nye, J. F. Physical Properties o/Crystals, Clarendon Press (1957).
Burfoot, J. C. Ferroelectrics, Van Nostrand (1967).
Fatuzzo, E. and Merz, W. J.Fe"oelectricity, North-Holland (1967).
BIBLIOGRAPHY 293
Chapter 9
Tebble, R. S. and Craik, D. J. Magnetic Materials, Wiley (1969).
Morrish, Physical Principles of Magnetism, Wiley (1965).
Soohoo, R. F. Theory and Applications of Ferrites, Prentice-Hall (1960).
Thourel, L. The Use of Ferrites at Microwave Frequencies, Pergamon (1964).
Chapter 10
Garbuny, M. Optical Physics, Academic Press (1965).
Jenkins, F. A. and White, H. E. Fundamentals of Physics Optics, McGraw-Hill.
Fowles, G. R. Introduction to Modern Optics, Holt, Rinehart and Winston (1968).
Kruse, P. W., McGlauchlin, L. D. and McQuistan, R. B. Elements of Infra Red
Technology, Wiley (1961).
Moss, T. S. Optical Properties of Semiconductors, Butterworths (1959).
Yariv, A. Introduction to Optical Electronics, Holt, Rinehart and Winston (1971).
Baldwin, G. C. An Introduction to Non Linear Optics, Plenum Press (1969).
Lasers and Light, Readings from Scientific American, Freeman (1969).
Chapter 11
Parks, R. D. Superconductivity, Dekker (1969).
Lynton, E. A. Superconductivity, Methuen (1969).
Rose-Innes, A. C. and Rhoderick, E. H. Introduction to Superconductivity, Pergamon
(1969).
Williams, J. E. C. Superconductivity and its Applications, Pion (1970).
General
Materials-Readings from Scientific American, Freeman (1967).
Hutchinson, T. S. and Baird, D. C. The Physics of Engineering Solids, 2nd edn, Wiley
(1968).
Kittel, C. Introduction to Solid State Physics, 4th edn, Wiley (1971).
Elliott, R. J. and Gibson A. F. An Introduction to Solid State Physics, Macmillan
(1974).
Azaroff, L. V. and Brophy, J. J. Electronic Processes in Materials, McGraw-Hill (1963).
Table of Physical Constants
Units
1 Debye = 3·33 X 10- 29 Coulomb m
leV =1·602 x 10-19 J
1 Angstrom A = 1 X 10-10 m
1 micron/J.m = 1 x 10- 6 m
Answers to Questions
Chapter 2
1. Covalent; Van der Waals; metallic; ionic
3. 4; 2; 74 per cent; 68 per cent
4. ./ 8/3
s. 1: 0·414
8. 3·92 x 10- 10 m
9. About 10- 7 m; 1 in 103
10. 1·33 X 10- 4 rad
Chapter 4
1. 9°9'; 13°1'; No
2. 3·15 X 10- 10 m(b.c.c.)
3. 2·50 x 10- 5 K- 1
4. 19°49'
s. 252 K
6. 10·7 mm
Chapter 5
1. 10·7 mm
2. K =EG/3(3G-E)
3. 0·43
4. (111) [10f]; (111) [011]; (111) [101]; (111) [011]; 2·45 x 106 Nm- 2
5. 1·49 x lOs Nm- 2
7. Yes; {100}
Chapter 6
1. e- S •7
3. Electrons 6·83 x 10- 3 J mol- 1 deg- 1 K
Lattice 0·058 J mol- 1 deg- 1 K
Chapter 7
1. (a) 8·37 x 102s m- 3
(b) 7·47 x 10- 4 ms- 1
296 PHYSICAL PROPERTIES OF MATERIALS
Chapter 8
2. 5·06 x 10- 9 W
4. 5·23 X 10- 30 Cm
s. 1·48; 9·6 x 10- 5 s
7. (a) i = dq/dt a: dlt../dt = constant
(b) No response
(c) Current pulse
Chapter 9
1. 3·71p.B
2. 0·61p.B
3. 9·12 X 108 Am- 1
S. (a) 5P.B (b) 5P.B
Chapter 10
2. A = 2·097; B = 1·03 X 10- 14 m2
3. 2·07 eV
4. 1·55 x 10- 3 m
S. (a) 115 rad (b) 0·014 x 10- 3 m
6. 4·17 kV
7. (a) 5·21p.m (b) 0·12 eV
Chapter 11
1. (a) 1·7 x 108 Am- 2 (b) 1·36 x 107 Am- 2
2. 1·16mm
3. (a) 0·57 mm (b) 3·43 mm
4. (a) 1·58 x 10- 5 J (b) 4·2 x 10- 9 eVatom- 1 (c) 2·3 x 10- 3 per cent
Index
Compounds-continued Defects
ionic, 20 dislocations, 36, 37
mixed valence, 140 Frenkel,35
transition metal, 139-141, 197-216 mterstitial, 35
Compression testing, 89 line,34,36
Condensation polymerization, 62 plane, 34, 39,40
Conduction band, 18, 130-134,284 pomt, 34, 35
Conductivity, electrical-see Electrical Schottky, 35, 36
properties vacancy, 35
ConfIgUrational coordinate diagram, 229, 248 Deformation, plastic, 90, 91
Cooper pair, 272 Degree of freedom, 106
Coordination, number, 29 Dendritic growth, 45, 57
octahedral, 202 Density of states
tetrahedral, 202 m amorphous materials, 148, 151
Correlation, 140 m metals, 112,280
Coupling coefficient, electro mechanical, m superconductors, 270
170-172 Diamagnetism, 12, 186
Covalentbondmg,19,21,34 Diamond structure, 31
unsaturated,19, 22, 34 Dielectric constant, 155
Covalent solid, 30, 34 loss, 162-164
Crack,97-98 properties of materials, 153-183
Creep, 102 relaxation, 161-165
Critical current m a superconductor, 260, Diffraction electron, 78-80
267-269 neutron, 76-78
Critical magnetic field m a superconductor, X-ray, 65-76
259-260 Dipole domam, 138
Critical temperature m a superconductor, mduced electric, 153, 156
258-260 magnetic, 11, 13
Cross Imkmg, 64 electric, 153-166
Cross slip, 95 moment electric, 153
Crowe cell, 276 permanent electric, 157
Cryotron, 276 Direct band gap semiconductor, 226
Crystal field, 12 Direct optical transition, 226
Crystallography Dislocations, 36, 56
crystal directions, 26 climb of, 95
crystal growth-see Growth cross slip of, 95
crystal morphology, 27 edge, 37, 38,93
crystal planes, 26 Frank-Read source, 94-95
crystal structure, 23 jog, 95-96
crystal symmetry, 23-26,167 loops, 93
crystal systems, 24 multiplication of, 94
X-ray, 23, 24, 65-69 observation of, 85
Crystal momentum, 283 screw, 37-38, 93
Cubic system, 26, 27 slip, 90-92
Curie Law, 158, 187 stackIDg fault, 39
Curie Temperature, 114, 143, 174, 188, 190, Disordered structures, 146,148
203 Dispersion hardenmg, 102
Curie-Weiss Law, 175, 188 Dispersion of e.m. radiation normal, 221
Curmg,64 anomalous, 222
Czochralski method for crystal growing, 46, Distribution function, radial, 80-81
48 Domams antiferromagnetic, 200
dipole, 138
ferrimagnetic, 205
Davisson-Germer experiment, 79 ferroelectric, 175, 179
De Broglie relationship, 1 ferromagnetic, 192-195
Debye magnetic bubble, 205-206
characteristic temperature, 111 observation in ferroelectric materials, 180
dielectric relaxation equations, 162 observation m ferromagnetic materials,
rmgs m electron diffraction, 79 195
-Scherrer method of X-ray diffraction, 73 Domam walls m ferroelectric materials, 179
theory of specific heat, 109, 110 m ferromagnetic materials, 194
unit of electric dipole moment, 153 Donor impurity, 132
INDEX 299
Double refraction, 235-237 valence, 18
Drift velocity of an electron, 124 Energy gap in a superconductor, 270
Drude theory of metallic conduction, 124-125, Epitaxial growth, 55-56
127 Epitaxy, 55
Ductile fracture, 98 Etching, chemical, 85
Ductile material, 89 Etch pit, 85
Exchange interaction, 13, 14, 179, 194
Edge dislocation, 37, 38, 93 Exchange constant, 14, 15
Effective mass of charge carriers, 131, 142, Exciton, 231
284 Exclusion Principle, 6, 7,19,21, 117
in III-V compounds, 136 Expansion coefficient, 115
Einstein theory of specific heats, 109, 110 thermal,114
Elastic modulus, Bulk, 89 Extended states, 148, 149
Shear, 89, 92 Extraordinary ray, 236
Young's, 89, 90 Extrinsic semiconductor, 131
Elastomer, 103
Elasto-<>ptic effect, 240 Face centred cubic, structure, 28, 30
Electric dipole, 153-166 Faraday effect, 195,216
displacement, 154 Faradays law, 257
moment, 153 Fatigue, 99
susceptibility, 155 Fault, stacking, 39
Electrical conductivity, 124 Fermi level, 17, 18, 117,128,131,132,151,
polarization, 140, 153-164, 241 242,270,272,273,280-281
properties of, Fermi statistics, 281-282
alloys, 129, 130 Ferrimagnetic materials, 198, 201-216
amorphous materials, 146, 150 applications of, 206-216
magnetic semiconductors, 143 domains in, 205
materials, 124-152 hysteresis in, 208-212
metals, 125-130 Ferrites, 143, 186,201-216
metals under pressure, 130 applications of, 206-216
semiconductors, 130-139 hexagonal,206
transition metal compounds, 141 microwave, 212-216
Electrical resistance, 124 rectangular loop, 208
resistivity, 124 soft, 207
Electro luminescence, 249 Ferroelectric materials, 166-167,175-180
Electro mechanical coupling coefficient, anisotropy energy in, 179
170-172 ceramic, 177-179
Electron band, 16-18,282-284 domains in, 175, 179-180
diffraction, 78 domain walls in, 179
drift, velocity, 124 observation of domains in, 180
cloud,2,5,6,14 phase transitions in, 175
magnetic moment of, 10 Ferromagnetic materials, 16, 186, 188-197
mobility, 125 applications of, 196-197
pairing in a superconductor, 272-274 B-H product for, 196-197
scattering, 128, 131 domains in, 192-195
spin,5 domain walls in, 194-195
Electronic polarizability, 156, 158 hard,196
Electronic polarization, 156, 158 observation of domains in, 195
Electronic structure of the elements, 8 soft, 197
Electro optical coefficient, 237 Ferromagnetism, weak, 205
Electro-<>ptic effect, linear, 237-240 Field critical magnetic, 259-267
non-linear, 241 Field ion microscopy, 84
quadratic, 240 Figure of merit for a pyroelectric material, 183
Electro-<>ptic modulation, Floating zone method of crystal growth, 49
longitudinal, 237-239 Fluorescence, 247
transverse, 240 Flux density, magnetic, 185
Electro-<>ptic properties of materials, 237-241 Fluxed melt method of crystal growth, 53
Electro-<>ptic tensor, 238 Fracture, 96-99
Electrostriction, 166 brittle, 96-98
Elements, list of, 287-291 ductile, 98
Energy bands, conduction, 17 fatigue, 99
theory of, 16-18,282-284 Frank-Read source, 94-95
300 INDEX