by G. I. Epifanov, D.Sc.

Moscow Institute of Electronic Engineering

Translated from the Russian by

Mark Samokbvalov, Cand. Se.
Preface

Ten years nave passed since the first Russian edition of this textbock
was pblished. In this time solid State physics has developed rapidly
as the scientific background of numerous front-line branches o
technology,. absorbing new discoveries and the ores. This has been
considered in preparing the new. edition.
At the same time coilege curricula have been changed to improve
the basic preparation of versatile engineers, especially in physics
and mathematics. This too had to be reflected in this book.
Also, the years that have elapsed .since the first edition have seea
much comment, some critical, and manv proposals from Soviet and
foreign readersfrom coilege teachers and students, teachers of
vocational and secondary schools, engineers and scientists. The
author is grateful for al the comment and proposals.
There was a need therefore to revise the book compietely.
As in the first edition, the presentaron of material has foilowed
the aim of elucidating the physical natura of the phenomena ais-
cussed. But, where possible, the qualitative relations are also pra-
seated, often though without rigorous mathematics.
The manuscript was reviewed in detaii by Prof. L. L. Dashkevich,
Dr. of Technical Sciences, and Prof. I. G. Nekrashevich, Honored
Scientist of the Belorussiaa Republic. It was perused by Prof.
L. A. Gribov, Br. of Mahematical and Physical Sciences, Assistant
Prof. Y . B. Zernov, and Z. S. Sazonova. The author extends sincere
thanks for their efforts and criticism, which he took into aecount
when revising the manuscript.
The author is also indebted to Snior Lecturer F. Zh. V"ilfr Cand.
of Technical Sciences, and Assistant-Prof. Yu. A. Moma, Cand. o
6 rreiace

Technicai Sciences, or manuals used in this textbook on supercon-

ductivity, Gunn enect, ana principies of operation cf impulse an
high-frequency diodes, and to Z. I. Epifanova or ail her work in
preparing the manuscript.
The author will be most grateful or comment and proposals that
might improve this book. They should be sent to the publishers.

G .I .E .
 Conents

Preface

1 Bonding. The Infernal Strucfure

of Solids

i The van der Waais forcea

2 The ionic bond
3 The covalent bond
4 The metallic bond
o The hydrogen bond
3 Comparison between bonds o various kinds
' ~ Forcea of repulsin
s Crystal lattice
9 Notation usea to describe sites, directions, and planes in a
crystal
10 Classication of solids based on the natura of bonds
11 Polymorphism
12 Imperections and eects of the crystal lattica

2 Mechanical Properfies of Solids

13 Elastic and plstic deformations. Hookes law

14 Principal laws governing plstic flow in crystals
15 Mechanical twinning
1S Theoretical and real shear strengths of crystals
17 The disiocation concept. Principal types of dislocations
v 13 Forces needed to move dislocations
19 Sources of dislocations. Strengthening of crystals
5 20 Brittle strength o solids
21 Time dependence of the strength o solas
- 22 Methods of increasing the strength of soiids
8 Cante nis

3 Eements o Physical Sfatlsfics

23 Methods used to describe the State of a macroscopic System 84

-24 Degenerate and nondegenerate ensembles 88
25 The number of States for microscopio partiles 91
26 Distribution function for a nondegenerate gas 94
27 Distribution function for a degenerate fermion gas 96
28 Distribution function for a degenerate boson gas 103
29 Rules for statistical averaging 105

4 Thermai Properfies of Solids

30 Normal modes of a lattice 107

31 Normal modes spectrum of a lattice 110
32 Phonons 112
33 Heat capacity of solids 115
34 Heat capacity of electrn gas 120
35 Thermai expansin of solids 122
36 Heat conductivity of solids. 126

5 The Band Theory of Solids

37 Electron energy levels of a free atom 133 -

38 Collectivization of electrons in a crystal 136 -
39 Energy spectrum of electrons in a crystal 138
40 Dependence of electrn energy on the wave vector 142 -
41 Eective mass of the electrn 147
42 Occupation of bands by electrons; Conductors, dielectrics, and
semiconductora 151
43 Intrinsic semiconductora. The concept of a hole 153 '
44 Impurity semiconductora 156
45 Position of the Fermi Tevel and free carrier concentraton
in semiconductora 159 >
46 Nonequilibrium carriers 166

Elctrica! Conduciivfy of Solids

47 . Equilibrium State of electrn gas in a- conductor in the ab-

sence of an electric Seld 169
48 Electron drift in an electric aeld 170
43 Relaxation time and mean free path 171
50 Specic conductance of a conductor 173
Conten S*

51 Electrical conductivity of nondegenerate and degenerate

gases 174
52 Wiedemann-Franz-Lorenz law 17S*
53 Temperatura dependence o carrier mobility 177
54 Electrical conductivity of pura metis 133-
55 Electrical conductivity of metal alloys 184
56 Intrinsic conductivity of semiconductora toS-
57 Impurity (extrinsic) conductivity of semiconductora^ 190
58 Deviation from Ohms law. The eSect of a strongfield 193
59 The Gunn effect 1S5
60 Photoconductivity of semiconductora 196
61 Luminescence , 203-
62 Fundamentis of superconductivity 207

7 Magnetic Properties of Solids

63 Magnetic field in magnetic materials 004

64 Magnetic properties of solids 225-
65 Magnetic properties of atoms 232.
66 Origiu of diamagnetism 238
87 Orgin of paramagnetism 240-
68 Origin of ferromagnetism 247
89 Antierromagnetism 254
70 Ferrimagnetism. Ferrites 255-
71 Magnetic resonance 25T
72 Fundamentis of quantum eiectronics 259-

3 Contac Phenomena

73 Work function 265-

74 Contact oi two metala 26a
75 The metal-semiconductor contact 271 -
76 Contact hetween two: semiconductora o dierent types oi
conductivity 27a Vi
77 Physical principies of semiconductor p-n junction devicss 2s a
78 Fundamentis of integrated circuit eiectronics (microeiectron-
ics) 299-

9 Thermoeiecric and Galvanomagnafc

Phenomena

79 The Seeheck edect 302

80 The Peitier eSect 307
10 Contente

81 The Thomson eSect 310

82 Galvanomagnetic phenomena 310
83 Practical applicatioas of thrmoelectric and galvanomag-
netic phenomena - 315

Appendices

I Derivation of the Maxwell-Boltzmann istrihution unction 317

II Derivation of the Fermi-Dirac istrihution unction 318
III Derivation of the Bose-Einstein istrihution unction 320
IV Tables 321

Glossary of Symbols and Notations 322

13ibliography 326
Index 329
1
Bonding.
The Interna! Structure of Solids

Matter can exist in the solid State uuly because there are forces
of interaction acting between the structurai partiles when the
latter are brought sufcientlv cise together. For the solid to have
a stahle structure the forces of interaction between the partiles
should be of two types: of attraction, to prevent the partiles from
moving away from each other, and of repulsin, to prevent the
partiles from merging.
Let us discuss briey the nature of these forces.

1 The van der Waas forces

The most general type of boad existing between anv two atoras,
or molecules, is due to van der Waals forces. Those forces were
first introduced to explain the equation of State of real gases, the
van der Waals equation:

{ f + y r ) { y - b ) = RT ( 1 .1 )

wkere the correction terms a/V1 and k account, respectively, for

the effect of the forces of attraction and repulsin acting between
the molecules of a real gas. These forces maniest tkemseves in an
almost ideal form in the interaction between the molecules with
saturated Chemical bonds (0.2, H ,, N2, CK4, etc.), as well as between
the atonas of inert gases, making possible their existence in the
liquid and solid States.
As a general case, the van der Waals bond is madc up of the disper
sin, orientational and induction interactions. L ets consider them
separately.
-Dispersin interaction. Consider the simplest example of two
interacting helium atoms shown in Figure 1.1. The electrn density
distribution in a helium atom is spherically symmetrical and for
 12 Solid Siaie Physics

this reason its electric moment is zero. But this means only that
the average electric moment o the atora is zero. At every moment
of time the eiectrons occupy particular points in space, thereby
preating instantaneous rapidly changing electric dipoles. When two
helium atoras are hrought together, the motion of tlieir eiectrons

Figure 1.1 Dispersin interaction. The interaction between helium atoms

is due to the correlation in the motion of eiectrons resulting in the
appearance of instantaneous dipoles: (a) correlation resulting
in 'attraction; (b) correlation resulting in repulsin.
Attraction Repulsin

Heitum atoms Helium atoms

Instantaneous dipoles Instantaneous dipoles

(a) (b)

becomes correlated and this laads to the orces of interaction. The

orces can he of two types. Ir the motion of the eiectrons is correlated,
as shown in Figure 1.1 (a), the instantaneous dipoles attract each
other: if the correlation is as shown in figure l.l(h ),'th e resulting
interaction is repulsin. Since the realization of the arrangement
of Figure l.l(a) leads to a reduction in the energv of the system,
...this arrangement is more probable and is realized more frequently.
in enect the cause of the cnstantlv existing forc of attraction
between helium atoms.
The bonds aiscussed above that owe their existence to a correla
tion in motion of the eiectrons in adjacent atoms are t-enned disper
sin forces. They were ftrst calculated bv F. London in 1930. 'The
calculation was based on the following model: the instantaneous
electric aipole of one atom causes the other atom to be polarized
and it becomes an induced dipole leading to the realization of the
arrangement in Figure i.l(a ), which corresponds to attraction. The
calculation had as its final resuit the following expression for the
energv of the dispersin interaction of two partiles:
r; 3 a 2S=rc
(1.2)
Bonding. Interna! Sfructure of Sods 13

where a is the polarizability of the partiles1*, Eexc their energy

of excitation, and r the distance between them.
Orientational interaction. Should tbe molecules possess a constant
dipole moment M, that is, should they be polar molecules, an
electrostatic interaction would be established between them tnding

Figure 1.2 Orientational interaction of polar molecules.

ooo
ooo
to rrange the molecules in a strict order (Figure 1.2), since that
order corresponds to the mnimum energy of the system. The correct
orientation of the molecules is disturbed by thermal motion. There-
fore the energy of the system due to the mutual orientation of the
molecules is strongly dependent on temperature. At iow tempera-
tures, when the orientation of the molecules is perfect, the interac
tion energy is determined from the expression
M-
tfor = 2ns 0r3
(1.3)

where r is the distance between the molecules, and eu the permittivity

of free space.
In the high temperature range the energy of interaction of polar
molecules, as had been demonst-rated by W. H. Keesom, is of the
following form:
T J __________ (A / \
Of 24xHlk-^Tr*
O rrZ o V ,^ T ri Va ' ' * /

The type of interaction discussed above is termed orientational

interaction.
% Induction interaction. Lastiy, in case of polar molecules o high
polarizability an induced moment due to the action of the fieid
of constant dipoles may be established (Figure 1.3; the induced
dipoles are shown by dotted iines). The energy o mutual attraction
due to the interaction of the rigid dipole of the rst molcula and
the induced dipole of the second molecule, as has been shown by
Debye, is inependent of temperature ana is givea by the expression
aM1
Um = (1.5)
"Sif r6
1 Let as recall the physical meaning of a. The charges in the molecule are
displaced under the action of an esternal eld of intensiy %. Thie Iead3 to
a dipole moment M proportionai to %: M a.%, the propprtionality facior a
being the polarizability o i the molecule.
14 Solid State Physics

whcre, as before, M is tbe constant dipole moment of tbe molecule,

and a its polarizability..
Such interaction is termed induction, or dejormation, interaction,
In general, when two molecules are brought cise together all
three types of interaction may be established, tbe interaction energy
being the sum of the energies of the dispersin (U), orientational
(/r), and induction (U[n) types of interaction:
= U + U 0T+ la

Figure 1.3 Induction interaction o molecules (dotted lines show

the induced diles).

Table 1.1 shows the relative magnitude (in percent) of each of

those components of the total bonding energy for water, ammonia,
hydrogen chloride and carbn monoxide. The data presentad in
Table 1.1 show the induction interaction for all the substances
to be weak. Three quarters or a half of the bond energy in substances
made up of polar molecules is due to the energy of orientational
interaction; while in materials made up of nonpolar molecules almost
all of the bond energy is due to the dispersin interaction.
T a b ie 1.1

Tvpe of interaction
Substance
Dispersin Induction Orientational

W ater 19 4 77
Ammonia 50 5 45
Kydrogen chloride 81 -t/ 15
Carbn monoxide 100

Table 1.2 shows the vales of the bond energy for some molecular
crystals held together by van der Waais forces.
T a b le 1.2

Substance b Substance
(103 J/mole) (103 J/mole)

Nen (Ne) .9 Argn (Ar) 8.4

Nitrogen (N ). 6.6 Carbn monoxide (CO) 3.4
Oxygen ( 0 2) 8.2 Methane (C K J 10.8
Bonding. Infernal Siruciure o Solids 15

2 The ionic bond

Atoms that occuoy places in the Mendeleev oeriodic table next
to inert gases tend to assume the electronic conguration of thes?
gases either by giving awav or accepting electrons. The val ence
electrn o alkali metis, which immediately follow the inert gases,
moves outside the closed shell and is only weakly connected with
the nucleus. The halides, which immediately precede the inert

Figure 1.4 Dependence of energy

U of nteracting ions on the distance
between them: 1 energy o
attraction, 2 energy of repulsin.
3 totalenergy of interaction.

gases, lack one electrn to complete a stable shell-characteristic of

an inert gas. Therefore they exhibit high afnity to an excess elec
trn.
Such atoms, that is, typical metis and halides, are bonded in the
following way. First a recharging of the atoms takes place. The
electrn from the metal moves over to the halide. This turas the
metal atom into a positively charged ion and the haioid atona into
a negativelv charged one. These ions internet according to the Cou
lomb iav/ as two opposite charges. Such a bond becarae kaown as
an ionic bond.
The energy of attraction of two ions separated by the distance r is
a2
U att ( 1. 6)

where q is the charge of the ions.

The curve 1 in Figure 1.4 shows the dependence of f7aU on r.
As r decreases the ahsolute valu o the energy increases monotoni-
caily, tending to infinitv as r-*-0 . The forc of attraction tends
to bring the ions together as cise as possihle. This, however, is
prevented by the torces of repulsin, which begin. to make themselves
feit at small distances and rise very rapidly with the decraase iu
distance. The repulsin energy Unp is shown in Figure 1.4 by the
16 Solid Safe Physlcs

curve 2. Max Born and other scientists expressed the repulsin

energy in the .orm
.(1.7)

w'here B and n are constants.

The resulting-interaction energy of two ions is

( 1. 8 )

This energy is shown'in Figure 1.4 by the curve 3 which has a

minimum at r = r0; the depth of this mnimum determines the
bond energy b \, and r0 determines the distance betw.een the ions
in the molecule. Making use of the fact that in equilibrium (at r = r0)
the forc of attraction, F att = (dUatt/dr)r=sro, equals the forc
of repulsin, F rep = {dUTSp/r)r = To, we can easily express (l.S) as

4rte0ru ( 1 n)
(1.9)

The energy of the lattice made up of N sucn molecules is

7,lattice NA 4rce0r0 ( 1- 10)

where ,4 is the Madeiung constant, which takes account of the energy

of interaction of the given molecule with all its neighbouring mole
cules in the crystai.
Table 1.3 shows by way of an example the experimental vales
of the bond energy of some ionic crystals and its vales calculatea
with the aid of (1.10). The discrepancias do not exceea 1-2 percent,
which is proof of good agreement between theory and experiment.
T a b le 1.3

U \j (103 3/mole)
Crystai .*,... ------:------------ , .
experiment theory

Sodium chioride (NaCl) 752 754

Potassium iodine (KI) 650 630
Rubidium bromide (RbBr) 635 645
Caesium iodine (Csi) 595 5S5

3 The covalenf bond

The ionic and van der Waals bonds are unable to account or the
existence o such compounds as H a, 0 2, N2, etc., as well as for bonds
in atomic crystals of the diamond type. Evidently, atoms of one
kind cannotjorm oppositely charged ions by changing the distribu-
Bonding. Infernal Sfructure o Solids 17

tion of valence electrons, as was the cas8 in the metal-haiide interac-

tion. On tke other nana, the bond iu the H3, 0 2, and N , molecules
is much stronger than that which could be attributed to the van der
Waals forces. For such compounds the role of the van der Waals
forces is that of a smail correction to the bond mainly responsibie
for the strength of the compounds. This bond became known as the
covalent bond.
Let lis consider the nature of this type of bond using the hydrogen
molecule as an example.

Figure 1.5 Calculating covalent bond between hydrogen atoms:

A, B hydrogen atoras; a, b their nuclei; 1 electrn of atora A;
2 electrn of atora B\ rou r5 3 distances of electrons from their
nuclei; r12 distance between electrons; ra2 ' distance of electrn 2
from nucleus a; rbl distance of electrn 1 from nucleus ; r distance
between nuclei.

Suppose that two hydrogen atoms are at a rather great distance r

from one another. The atom A consista of the nuclens a, and the
electrn 1 and the atom B consists of the nucleus b and the electrn 2
(Figure 1.5). Since the density of the electrn cloud which describes
the electrn State in an atom falls on very rapidly as the distance
from the nucleus increases, the probabilities to discover electrn 1
near nucleus b and electrn 2 near nucleus a are very smail. Calcula-
tion shows that for r 50 each electrn can visit the other nucleus
on the average only once in 1012 years. Because o that atoms A and 3
may be regarded as isolated atoms and the energy of the system
mane up of such atoms may be taken to be equal to 2E 0, where
is the energy of the isolated atom in the gxound State.
When the atoms are hrought closer together, the probability o
the electrons going over to nuclei other than their own increases.
For r 2 A the electrn clouds begin to overlap noticeabiy and the
transtioo frequency rises up to 1014 s -1. If the atoms are hrought
still closer together, the frequency of the electrn exchange rises
so that it becomes meaningless to speak of electrn 1 as beiongmg
to the atom A and of electrn 2 as belonging to atom B. This corres-
18 Solid Siafe Physics

ponds to a new state that is not characteristic for a system made

up o two isolated atoras. A reraarkable property of this new state
is that the electrons in it belong simultaneously to both nuclei,.
in other words, are. collectiuized.
The collectivization of the electrons is accompanied by a changa
in electrn density distribution- \p 2 and .in the energy of the system
as compared to the total energy 2 0 of the isolated atoms. The dotted
lines 1 in Figure 1.6 show the electrn cloud density of the isolated
atoms, the thin solid line 2 shows the total density obtained by simple
superposition of the .electrn clouds of isolated atoms, and the t-hick

Figure 1.6 Electron density distribution in a system of two hydrogen

atoms: 1 electrn density distribution in isolated hydrogen atoms,
2 electrn density resulting from simple overlapping of-electrn clouds
of isolated atoms brought together to within a distance of r, 2 actual
electrn density distribution in a hydrogen molecule.

solid line 3 the actual density distribution along the axis ioining
the nuclei a and b brought about by. the collectivization of the
electrons. The .figure shows that the collectivization of the electrons
results in the electrn clouds being drawn into the space between
the two nuclei: at a small distance from the nucleus outside this
space the density of the clouds falls off, as compared with the density
in isolated atoms, at the same time rising in the space between the
nuclei above the sum of the densities of isolated atoms. The appear-
ance of a state with an electrn cloud of increased density that filis
the space between the nuclei ahvays results in a decrease in the
system s-energy and in the appearance of forces o attraction between
the atoms. Speaking guratively, we may say that the electrn
cloud formed in the space between the nuclei by a collectivized
pair of electrons drav/s the' nuclei together, striving to bring them
as ciase together as possibie.
Such is the qualtative pictre of the origin of the covalent hond.
Quantitative calculatfoas of the h^/drogen molecule were first
Bonding. infernal Sfrucfure of Soiids 19

perormea by W. H. Heitler and F. Londoa in 1927. 'Those calcula-

tions have shown that a system consisting of two closely spaced
atoms of hyarogen can have two saergy vales depeading on the
direction of the electrn spins in the atoras:

. U, = 2E, + r 1 11)
( .

when the spins are antiparaliel, and

7 k -2 E+7=t 1-12)
when the spins are paxailel. Here 2E 0 is the total energy of the two
isolated atoras, K is the energy of the electrostatic interaction o the
eiectroas with the nnclei, of the electrons with one another, and o
the nuclei. Another ame for it is Coulomb energy. Its sign is nega
tive. By A we denote the energy of exchange interaction due to the
atoras exchanging electrons. This is the additional energy that ap-
pears as the result o the change in the electrn density distribution
in the process of the formation of the molecule. Its sign is negative
and its absolute valu is much iarger thaa K ( .4 | | K ); S is
the overlap integral whose valu lies within the limits 0 ^ 5 ^ 1.
The State with the energy U%is termea symmetric and. with L \
antisymmetric..Since both K and A are negative and 5 ^ 1 , the
energy of the system in the symmetric State is less than the energy
of two isolated atonas
l \ < 2 E, (1.13)
This corresponda to the appearance of forces of attraction. Since-
the absolute valu of the exchange energy .4 is considerably greater
than that of the Coulomb energy K the decrease in the system s
energy is mainly due to .4. For this reason the forc of attraction
that appears' between the atoms is ' termed the exchange forc.
For. the same reason. that is, beca use A i j K |, the formation
of the ant.isymmetric State lea.ds to.. an increase in the systems ener
gy. This corresponds to the appearance of repulsive crces.
Figure 1.7 shows the dependence of Us ana fF^on ra. where r'is
the distance between the atoms, and a 0.529 is the radius cf
the rst Bohr orbit (the Bohr radias). The zeroth energy level has
been xed at 2E 0. Figure 1.7 shows that in the antisymmetric State
the systems energy rises steadily as the atoms. are brought closer
toget-her (curve. I), this corresponding-to the mutual repulsin of
the atoms. Therefore a hydrogen molecule cannot be formed in such
a state. l a the symmetric State (curve 2) the systems energy at first
ralis as the distahce r between the atoms decreases, attaining its-.
mnimum valu a t r = r,.. As the distance r decreases strll further;
the energy. begins.' to rise because f he appearance-of strong repulsiva-
20 Soiid Stafe Physics

forces. The existence o! a mnimum on the potential energy curve

makes the existence of a stable system o two hydrogen atoms, that
is, a hydrogen molecule, possihle. To destrov this system work must
be performed equal to the depth of the potential well, Ub.
Calculation provides the following vales of- Ub and r0: 7b =
= 4.37 eV, r0 = 0.735 A; the experimental vales are f7b =
= 4.38 eV, r0 = 0.753 A . The agreement between theory and ex-
periment is quite good.
Table 1.4 shows the vales of the bond energy for some covalent
compoundsthe molecules of H 2, N2, 0 2, COand for diamond,
Silicon and germanium crvstals in which the bonding is due to
covalent forces.

R g u r e 1.7 n te ra c tio n en erg y

of tw o h y d ro g en ato m s:
1 a n tisy m m e tric S ta te ,
2 sy m m e tric State.

The data show the covalent b.ond to be a verv powerful oner, its
energy being as high as i0ft-106 J/'mole.
Table 1.4

Gss uh Crystal
(105 J/mole) (105 J/unole)

Carbn monoside (CO) 10.8 Diamond (C) 6.3

Nitrogen (N.) 9.5 Silicon (Si) 4.4
Oxygen (0 ?) 5.0 Germanium (Ge) 3.5
Hydrogen ( 2) 4.4

Characteristic properties of the covalent bond, which distinguish

it from the bonds of other tvpes, are its saturabiiity and irec-
tionality. i
Saturabiiity means that each atom can form covalent bonds
only with a limited number of its neighbours. This means that each
hydrogen atom can form covalent bonds only with one of its neigh
bours.. The electrn pair constituting such a bond has antiparaiiel
Sonding. Infernal Sfructure of Soiids 21

spins and occupies one quantum eell. A third atom in this case
instead o? being attracted will be repelled.
The valence bond is formed in the direction of the greatest density
of the electrn cloud corresponding to the valence electrons. In this
case there is mximum overlapping of the electrn clonas of the
bonding electrons, which implies that the valence bond is directional.

4 The metaic bond

There is a special group of substances, called metis, that occupy
places at the beginning. of every period of the Mendeleev table.
The formation of the metallic bond cannot be explained by the
presence of the ionic or the covalent bond. Indeed, the ionic bond

Figure 1.8 Electron density

distribution in the aluminium
lattice obtained by X-ray
photography.

appears only between atoms having diSerent affinities to the addi-

tional electrn, fr instance, between the atoms of a metal and a
halide. Evidently, such bond between kindred atoms of a metal
having identical affinity to the electrn is impossible. On the other
hand, metallic atoms do not have enough vlenos electrons to form
valence bonds with their nearest neighbours. For instance, the'copper
atom has one valence electrn and can form a valence bond STy
with a single atom. But in the copper lattice every atom is surround-
ed by twelve neighbours with which it mus be connected by lines
of forc. This points to t-he fact that in metis there is a special type
of bonding known as the metallic bond. Let us consider the nature
of this bond.
In metallic atoms the external valence electrons are rather weakly
coupled to the nucleus. In the liquid and solid States the atoms come
so cise together that the valence electrons are abie to ieave their
respective atoms and wanaer thraughout the lattice. This leads to
an extremelv homogeneous distribu tion of the negative charge in the
crystal lattice. This conclusin is supported by direct experiments.
Figure l.S shows an experimental curve o the electrn density
22 Sod Stae Physics

distribution between tbe sites o! the aluminium. lattice obtained

by means of X-ray photography. Most part of tbe distance between
the sites the electrn concentration reruains constant. Only. quite
cise to the sites it rises sharply because of the presence of intemal
shells of the aluminium atoms.
In the lattice of a metal the bond is due to the nteraction of the
positiva ions with the electrn gas. The electrons movingbetween
the ions compnsate the forces of repulsin existing between the
positivelv charged ions and draw them closer together. As the distance
between the ions becomes smaller the density of the electrn gas
rises and this leads to an increase in the forc drawing the ions to
gether. On the other hand, in this- case the repulsive forces acting
between the ions tend to move them away from each other. When
the distance between the ions becomes such that the forces of attrac-
tion are compensated by the forces of repulsin, a stable lattice is
formed.
It appears that the metallic bond is somewhat similar to the
valence bond, since they are both based on the collectivization of
external valence electrons. However, in case of the valence bond
only atoms that form pairs of nearest neighbours take part in the
collectivization of electrons, and the respective electrons always
reruain between the atoms. In case of the metallic bond all atoms of
the crvstal take part in the collectivization of electrons, and the
collectivized electrons are no longer localized near .their respective
atoms but move frseiy inside the lattice.-

5 The hydrogen bond

The hydrogen. bond is formed between an atom of hydrogen and an
extremely electronegative atom, for instance, an atom of oxygen,
uorine, nitrogen, chlorine. Such an atom attracts the bonding
electrons and becomes negativeiy charged; the hydrogen atom after
losing the bonding electrn assumes a positive charge. The hydrogen
bond is a resuit of electrostatic attraction of those charges.
As a tvpical example we may cite the hydrogen bond 'between
molecuies o water (Figure 1.9): The OH bond between an oxygen
atom of one molecule and a hydrogen atom of another behaves as
a tiny dipole with a 5 charge on the oxygen atom and a -f-d charge
on the hydrogen atom. The forc of attraction between those charges
is the cause of the hydrogen bond shown by dots in Figure 1.9. The
attraction is enhanced by the sm all dimensions of the hydrogen atom
that enabie it to come cise to the electronegative atom. Sfcill,
this distance r0H = 2.76 is much greater than the radius of the
covalent bond HO, ra, in the water molecule itseif, which is equal
to 0.95 A. This is quite natural sinca the encrgy of the covalent
Bonding. Infernal Struciure of Solids 23

bond is about an order of magito de higher than that of the hydrogen

boad. Its valu for water is (21-25) X 10: J/mole.
The hydrogen bond is the cause of association of molecules of
liquids (water, acids, spirits, etc.), which results in greater viscosity,
higher boiling point, ahnormal thermal expansin, etc. Water may
serve best to illustrate _this. Should. there be no hydrogen bonds
between the molecules of water, its boiling point at atmospheric

Figure t.9 Hydrogen bond between water molecules.

H\ X - 5 +5
103 0 ...............
V
X H oh- 0 * * * * * t\
h/ ,0

/

pressure would be not -f-TOO C but 80 C and its viscosity /would

be lower by almost an order of magnitude. When water is heated
above 0 C, the hydrogen bond is destroyed. Since the hydrogen
bond is resoonsible for the loose structure of associated comolexes,
in which the water molecules are rather far apa.rt (2.76 A), the
destruction of such a loose structure should result in an. increase
in the density of water. On the other hana, an increase in the temper-
ature of water and a corresponding increase in the intensity of
thermal motion of its molecules should lead to thermal expansin
and a decrease in its density. Experiment shows that in the tempera-
ture range 0-4 G the first factorthe increase in density due to the
disruption o the hydrogen bondsis the prevalent one. Because
of that within this range the density of water rises upon heating.
Above 4 C the other factorthermal expansinprevaiis. This is
why when water is heated above 4 C its density decreases, as is the
case with other (normal) liquids.

6 Comparison between bonds

o various kinds
The van der Waals bond is the most universal one. It exists in aii
cases without exception. At the- same time this is the weakest,
having an energy of the order of 104 limle. Ideally, it operates
between neutral atoms, or molecules, with cioseri inner electrn
shells. SpeciScallyv the van der Waals forces are responsibie for the
24 Solid Sfaie Physics

existence o the liquid and solid States of inert gases, hydrogen,

oxygen, nitrogen and many other organic and inorganic compounds.
They also, as we w ill see later, fonn bonds in many of the molecular
valence crystals. Because of low energy vales of the van der Waals
bond all structures based on it are unstable, volatile and have low
melting points.
The ionlc bond is a typical Chemical bond very frequent among
the inorganic compounds such as metal-halide compounds, metallic
oxides, suldes and other polar compounds. The ionic bond is also
a feature of-numerous intermetallic compounds (carbides, nitrides,
selenides, etc.). The energy of the ionic bond is much higher than
that of the van der W aals bond and may be as high as 106 3/mole.
Because of that solids based on the ionic bond have high sublima-
tion heat vales and high melting points.
The covalent bond is extremely widespread among organic com
pounds. but is also present in inorganic compounds, in some metis
and in numerous intermetallic compounds. This bond is responsible
for the existence of valence crystals of the diamond, germanium and
other types, as will be riiscussed below. The energy of the covalent
bond is also high ( ~ 1 0 fi J/mole), which stems from the fact that
the solids with this type of bond have high melting points and high
heats of sublimation.
The metallic bond formed as a result of the collectivization of
the valence electrons is characteristic of typical metis and numerous
intermetallic compounds. The order of magnitude of the energy of
this type of bond is comparable to that of the energy of covalent
bond.
Lastlv, the hydrogen bond, aithough relatively weak, still plays
an important part in nature.
It should be pointed out th a tin real solids no .types of bonds
discussed above ever exist purely by themselves. Practically, there
is alwavs a superposition of two types of bonds or more. One of them,
as a rule, plays a dominant part in determining the.steucture and
the properties of the solid.

7 Forces of repulsin
For the formation of a stable system of interacting atonas or mole-
cules, together with forces of attraction there should be forces of re
pulsin, which wouid prevent the complete merging of the particles.
The origin of the forces of repulsin is rst of all the interaction
of the nuelei each of which carries a considerable positive charge.
The energy of this interacion, U'Tep, depends on the distance be-
tween the nuelei and on the degree of screening by their internal
electrn. shelis.
/
Sonding. Inferna! Sfrucfure of Solids 25

The following expression for U'Tsp may be obtained rom quantum

mechanical calculations:
UnPCCe-rfa (1.14)
where r is the distance between tbe nuclei, and a 0.529 the
Bohr radius.
Ths dependence of 6Qep on r determines tbe nature of the forces
of repulsin: they 'attain enormous vales at short distances and
fail oa abruptly as r increases. For instance, when the distance
between a proton and a hydrogen atom decreases from r = 2a to
r = a2 (4 times), the repulsive energy increases almost 300-fold.
The repulsive forces due to the interaction of the nuclei pay
a dominant role when light atoms, whose nuclei are rather pooriy
screened by the electrn shells, are brought together. In all other
cases the dominant part is played by repulsin due to the overlapping
of closed electrn shells of the atoms being brought together. Consid-
er by way of an example the interaction between a chlorine ion
with a closed 3p shell and a sodium ion with a closed 2pshell. When
the ions are brought together to a distance at which the 3p and 2p
shells overlap, the number of electrons in each of them begins to
exceed that which is compatible with the Pauli exclusin principie.
Because of that some of the electrons are forced to go to higher
energy leveis (for instance, 3d or 4s). This resulto in an increase
in the systems energy and, consequently, in the appearance of
forces of repulsin. Quantum mechanical calculations show the
energy of such repulsin to an exponential dependence on the
distance, as well:
C/fop OCe~r/p (1.15)
where p is a constant usually obtained experimentally.
Often the repulsin energy is expressed in the form (1.7). This
expression gives a less steep decline of f7rep with the increase in r
and is less c-onsistent with experiment than (1.14) or (t.T5) hut is
nevertheless widely used by researchers.

8 Crysial lafrics
No matter what the origin of forces appearing when particies are
brought together is, their general nature is the same-(Figurel.lO(a)):
at comparatively large distances forces of attraction F at increase
rapidly as the distanca between the particies - decreases (curve I );
at small distances forces of repulsin Frep come into being and wth
a further decrease in r they increase much more rapidly than F ltt
(cur/e 2). At the distance r = r0 the repulsive forces counterbalance
the, forces of attractian, the resuitant forc- of interaction F vanshes
26 Sod Siete Physics

Figure 1.10 Interaction between

approaching partiles:
(a) interaction orces, 1 forc of
attraction, 2 repulsive forc, '
3 resultant forc; (b) interaction
energy.

(curve 3), and the energy attains

its minimum valu Uh (Fig
ure 1.10(b)). Beca use of this the
partiles brought together to a
distance r0 are in a State of
stable equilibrium. By the same
reason the free partiles should
arrange themselves in a strict
order at a distance r0 from one
another thus forming a body
with a reguiar internal struc-
turea crystal. Such a structure
will remain stable until the abso-
lute valu of the bond energy
remains greater than the energy
of thermal motion of the parti
les. The partiles constituting
the crystal cannot freeely leave
'their equilibrium sites because
this involves an icrease in their
energy and leads to the appearance of forces tending to
re tura them to their equilibrium sites. One may say that the partiles
are fixed in their equilibrium sites. The only form of-motion allomad
to them are random vibrations around^ their equilibrium sites.
To describe the regular internal structure of crystals one may conve-
nientlv use the concept of the crystal lattice. There are lattices of two
types the translational Bravais lattice and the lattice with a basis.
Bravais lattice. From the geometrical point of view a regular
periodic arrangement of partiles mav 'be' describea with the aid of
a iranslation. Figure, l.ll( a ) shows a lattice obtained with the aid
of translation of a partile aiong the three axes: OX over the sec-
tions a, 2a, 3a, . . ., ma, . . OY over the sections b, 2b, 3b, . . .
. . ., nb, . . .; OZ over the sections c, 2c, 3c, . . ., pe, . . . (m, n, p
are integers). The position of any partile in this lattice is aescribed
by the vector
r = /na -f- nb + pe (1.16)
Sondng. Infernal Sfrucura of Solids 27

The vectors a, b, c are termed the translation vectors an their nume-

rical vales the translation periods.
A lattice built with the aid of translation of any site along the
three directions is termed a Brovis lattice. The smallest parallelepi-
ped built on the vectors a, b, c is termed the unit csll of the crystal
(Figure l.ll(b )). The shape and the volume of ali the unit cells

Figure 1.11 Crystal lattice: (a) Bravais lattice, (b) unit cell
of Bravais lattice.

comprising the lattice are idntica!. All cell-apexes are cccupied by

identical atoms or groups of atoms and ;are therefore equivalent.
They are termed lattice sites.
To discribe a unit cell, six quantities should generally be stated:
three edges of the cell (a, , c) and three angles between them (a,
p, 7 ). Those quantities are termed the parameters of the unit cell.
Often the sections a, b, c are used as units o length in latticss
instead of the metra. They are termed axial units.
Unit cells with partiles only at the vrtices are known as.prinzi-
tive cells. There is only one particle to each such cell.
In some cases to express the lattice symmatry more fully the unit
cells are built so that they contain partiles not only in their aperes
but at other points as well. Such cells are termed complex cells.
The most widespread types o such cells are (Figure 1.12) the oody-
centered (BC), the face-cenisred (FC), and the base-centered (iraseo)
cells. It may be shown that such cells may easily be reduced to pnm-
itive cells. Because of that they are Bravais-type cells. ^ .
A lattice with a basis. Not every type of lattice may be obtained
by translation of a single site. Figure 1.13 shows a two-dimensional
lattice with a generai-type basis. It may. easily be seen that it is
impossible- to describe the unit cell o such a lattice in ternas of
28 Sod Sfate Physics

a single-site unit cell. Such a lattice may be imagined as consisting

o two Bravais lattices, 1 and 2, each determined by the basis vec-
tors a and b and inserted into each other. The relative displacement
of the lattices is described by an additional basis vector A. The
number of such basis vectors may be arbitrary.

Figure 1.12 Typical crystal structures: (a) base-centered (BaseC);

(b) _ body-csnter8d (BC); (c) face-centered (FC).

The lattice of this type is termed the lattice with a basis. It may
be built with the aid of the same translations as can be used to build
an3r of the Bravais lattices that make it up. However, in this case
we shall have to transate not one site but several sitesthe basis,

Figure 1.13 Two-dimensional

Sublattice 1 lattice with a basis: A basis
vector.

defined by the totaiity of the- basis vectors. Thus, the two-dimen

sional lattice shown in Figure l.i3 may be ohtained by a translation
of the basis made up of two sites: O and O' . An example of a three-
dimensional lattice with a basis is the diamond lattice shown in
Figure 1.14{a). t may be obtained by inserting one FCC (face-centered
cubic) lattice into another FCC lattice displaced aiong- the space
diagonal by one-fourth of its length. Figure 1.14(b) shows a tetra-
hgdron designated by a dotted. iine in Figure 1.14(a). It may be
Sonding. Infernal Srructure of Soiids 29

Figure 1.14 Diamond lattice: (a) spatial arrangement o atoms in the

latice; (b) tetranedral pattern of atom3 in the lattice.

seen from Figure 1.14(b) that in the diamond lattice every atona is
surrounded by four nearest neighbours in the apexes of the tetra-
hedron whose edge is a !2.

9 Nofation used to describe sites,

direcions, and planes in a crystai
Let us mention briev the notation conventionaliy used to describe
sites, directions and planes in a lattice, the Millar indices.
Site indices. The- position of any lattice site relative to the chosen
origin of coordinates is deftned by three of its coordinates (Figure

Figure 1.15 .Indices o.a crystai lattice

site.

forxn:
x ma, y nb, z = pe
where a, b, c are the lattice narametars, and m,. n, p are integers.
30 Sod Sfate Physics

Should we use lattice parameters as units of length aiong the

respective axes we would obtain tile lattice coordinates sim ply in
the form of numbers m, n, p. Tbese numbers are termed site indices
and are written in the form [[mrcp]]. For a negativo ndex tire minus'
sign is written above tbe ndex. For instance, ior a site witb coordi
nates x = 2a, y = ib, z = 3c tbe indices are written in tbe
form [213].
Indices of direction. To describe a direction in a crystal a straigbt
line passing through tbe origin is cbosen. The position of tbis is
uniquely dened by tbe indices of tbe rst site tbrougb wbicb it
passes (Figure 1.15). Therefore tbe indices of the site are at the
same time tbe indices of tbe direction. Tbe usual notation for a di
rection is [mnp]. The indices of direction are. by definition, tbe
three smallest integers tbat describe tbe position of tbe site nearest
to the origin whicb lies on tbe given direction. For instance, tbe
indices of the direction tbat passes through the origin and the site
[[435]] are [435]. Figure 1.16 shows the principal directions (crystal-
lographic orieatations) in a cubic crystal and their notation.
Plae indices. The position of a plae is dened by tbe cboice of
three sections A, B, C it cuts of when it interseets witb the three
coordnate axes. The procedure of nding the indices of sucb a plae
is as follows.
The sections ABC are expressed in axial units and the reciprocal
quantities are written as HA, 1IB, IC. A common denominator is
found for all tbe three fractions. Suppose -it is D. Then tbe integers
h = DIA, k D IB , l = DIC will be tbe plae indices. They are
written in the form (hkl).
Determine, for example, the indices of a plae that cuts of tbe
sections A = 1/2. B = 2, C 1/3 on the X, Y, Z axes respectively.
The ratios HA -f- H B -f- IC = l/(l/2) -f- 1/2 h- 1/(173) = 2 ^
-r- 1/2 -f- 3. The common denominator is Z5 = 2, The indices of the
plae are h = 2/(l/2) = 4, k 2/2 = ! ,- / = 3/(l/2) = 6. The plae
is denoted (416). Figure 1.17 shows tbe principal planes of tbe cubic
lattice.
It may easily be sbown tbat in a cubic crystal the distances be-
tween the planes belonging to a given family may be expressed with
tbe aid of tbe indices of tbese planes in tbe following way:

where a is the lattice parameter. Tbis formula shows tbat tbe greater
are tbe plae indices tbe shorter is tbe distance between tbe planes.
To denote tbe planes in a hexagonal crystal a four-axis coordnate
system is used (Figure 1.18): three axes (oj, a,, a 3) make angles
of 120 witb one another and lie in tbe base of a hexagonal prism,
the .fourth axis, c, being perpendicular to- the base plae. Every
3onding. intornai Sb-uctur of Soids 31

Figura 1.16 Indices of principal directions in a cubic crystai.

Figure 1.17 Indices of principal planes in a cubic crystai.

32 Soiid Siae Physics

plae is denoted by four ndices: hkil. The additlonal label i occupies

the third place and is calculated with the aid of h and k: i
= (h + k). The base plae parallei to the axes alf a2, a 3'has

Figure 1.18 Indices of principal

ptanes in a hexagonal crysta.

the indices (0001). The planes parallei to the lateral faces of the
prisni have indices of the (1010) type. There are three such planes
(not parallei to one another). They are tenned first-ordsr planes.

10 Ciassificafion of solids based

on the naure of bonds
The nature of the crystal structure is primarily dictated by the
nature of bonding forces acting between the structnral partiles
(atoms, ions, molecules) which make up the solid. In accordance
with the ve existing types of bonds there are five principal types
of crystal lattices: ionic, or coordination. lattices, with the ionic bond
playing the main part; molecular lattices, with the van der Waals
forces mainiy responsible for the bonding;-atomic lattices. with bonds
of a dist'inbtly covalent type; metallic lattices, w m *characteristic
metallic bonds; and lattices with the kyrogen bond.
Let us analyze from this viewpoint the crystal structure of Chemi
cal elements and of tynical Chemical comoounds (see Appendix IY,
Table 1).
The Chemical elements may be roughly divided into four classes
according to the type of crystal structure. The analysis may best
be started with Class IY.
Class IY. This class ineludes all the inert gases. In the process
of compression and crystallization of these gases only comparativeiy
weak van der Waals forces act between the atoms, which have spher-
ically symmetrical electrn shells. Acted upon by these forces the
symmetrical atoms draw together to form a most tightly packed
 i

Sonding. Infernal Sfruciure of Solids 33

face-centered cubic lattice. Every atora is surrounded by 12 nearest

neighbours. The number of nearest neighbours is usually termed
the coordination number of the lattice.
Class III. The Glass III includes Silicon and carbn from the short
periods of the Mendeleev periodic t-able, gennanium and tin from
Group IVB, and all the elements from Groups VB, VIB, VIIB.

Figure 1.19 Crystal structure of sheraical elementa crysiailizing

in accordance with the (8N)-rule: (a) elementa of Group IVB;
(b) elementa of Group VB; (c) elements of Group VIB, (d) elementa
of Group VIIB.

The crystallization of the elements of those classes proceeds in con-

formity with the (8 iV)-rule: every atom in the lattice is surround
ed by 8 N nearest neighbours, N being the number of the group
to which the element belongs. Thus diamond, Silicon, gennanium
and gray tin are elements of Group IV of the Periodic Table. There-
fore the coordination number of their lattices shouldvbe 8 4 = 4.
They all ochave a tetrahedral lattice in which avery'aTm is sur
rounded by 4 nearest neighbours, as is shown in Figure 1.19 (a).
Phosphorus, arsenic, antimony and bismuth beiong to Groug V.
Their coordination number is 8 5 = 3. Every atom has 3 nearest
neighbours lying in a plae, as shown in Figure l.lS (b ). Their lattice
has a laminate structure, with the atomic iayers bonaed by van der
W aals orces.
Selenium and tellurium beiong to Group VI and have a coordina
tion number 2. Their atoms orm long spirai-shaped chains with
each atom having two nearest neighbours (Figure 1.19(c)). The chains
are hondea by van der Waals forces. Lastly, iodine belongs to
Group VII (Figure l.lS(d )) and has a coordination number 1. The
atoms in the iodine lattice are arranged in pairs bonded by van der
...... ^ aa k .fo rc e s. This ezplains the high volatility of iodine.
3
34 Solid Srae Physics

Such nature of the crystal structure of Chemical elements whose

crystallization conforms to the (8 iV)-rule is quite understandahle.
The atoms of Group IV elements have 4 electrons in their outer
shell. They lack 4 additional electrons to make up a stable 8-electron
configuration. They make up the dficit by exchanging electrons
with 4 nearest neighbouxs, as shown in Figure 1.19(a), forming
a strong covalent bond with each of them. Accordingly, every atom
in the crystal lattice of those elements has 4 nearest neighbours.
In the same way the electrn shells of Groups V, VI, VII are complet-
ed to contain 8 electrons.
Many Chemical compounds crystallize in crystals with covalent
bonds. Quartz S i 0 2 may serve as a typical example. In the quartz

F ig u r e 1.20 Structure of quartz S i 0 2 crystal.

crystal every Silicon atom is surrounded by a tetrahedron of oxygen

atoms (Figure 1.20) bonded to the Silicon atom by covalent bonds.
Every oxygen atom is bonded to two Silicon atoms thereby joining
two tetrahedrons. In this way a three-dimensional net of SiOSi
bonds is formed, and the hardness and the melting point of the
resulting crystal are high.
It may be of interest to note that the S iOSi bonds may be
arranged into a one-aimensionai chain. Such compounds described
by the common formula
R R R

. . . O Si O Si O S i O . . .
! ! !
R R R
n
where R is an arbitrary organic group, are tenned silicones. The
number ;i in a chain m ay be as high as severa! million. The chains
may be joinea together with the aid of the lateral groups R. In this
way new materials, silicone resins, are formed. Because of the high
strengtho the S OSi bonds and of the high flexibility cf silicone
Sond'tng. infernal Strucfure of Solids 35

chains such resins retain their properties at much lower and much
higher temperaturas than natural rubbers. This fact enables them
to be used for thermai sbielding of space sbips and aircrat, as weil
as in extreme arctic conditions,
Class I. This is the most populated class which contains metis.
Sinca metallic lattices are made up not of atoms but of ions, having
tile spherical symmetry of the atoms of inert gases, it is to be expected
that metis too should crystallize into the same tightly packed
lattices as the inert gases. Indeed, the following three types of crystal
lattices are characteristic for metis: the face-centered cubic lattice
with the coordination. number 12 (see Figure 1.12), the hexagonal
close-packed (HCP) lattice with the coordination number 12 (see
Figure 1.18) and the body-centered cubic lattice with the coordination
number 8 (see Figure 1.12):. The latter is the least ciosely packed
metal lattice.
Class II. The Chemical elements belonging to Class II are in a sense
intermedate between metis and the Class III elements, which
crystallize in conformity with the (8 iV)-rule. For instance, the
Group IIB elements Zn, Cd and Hg are metis and one would expect
them to have a typically metallic lattice with a high coordination
number. Actually. Zn and Cd crystals are a special modication
o the compact hexagonal lattice in which every atom has 6 nearest
neighbours instead of 12, as required by the (8 V)-rule. These
atoms occupy the base plae. In the case of mercury the (8 iV)-rule
is observed even more strictly: its crystal structure is a simple
rhombohedral in which every atom is surrounded by 6 nearest'
neighbours. Boronan element of Group IIIB has a lattice that
may be descrbed as a deformed lattice .with 5 nearest neighbours.
.This too agrees with the (S .V)-rule.
The ionic bond, as was stat-ed above, plays one of the main pars
in the world of inorganic compounds. in particular, in numerous
ionic crystals typicallv represented by the rock salt crystal NaCi
(Figure 1.21). In such crystals it is impossible to pick out single
molecules. The crystal should be regarded as a ciosely packed system
of positive and negativa ions whose positions altrnate so that the
elec-trostatic attraction between the nearest neighbours would be at
its mximum. With the most favourable relation between the radii
o the positive (M +) and the negative (X~) ions which exists in the
NaCl crystal the ions touch one another (Figure 1.22(a)) and the
closest possible packing is achieved, in which every ion is surrounded
by 6 nearest neighbours of the opposite charge. When the ratio of the
radii of-the ions and X " is less favourable (Figures i.22(b, c)) c o s
tal structures with other coordination numbers, 4 or 8, are formed.
Ionic compounds of the M X 2 type, such as CaCL and N a0, have
still more complex lattices. But the principie upon which they are
buil rmains the same: the ions are nacked so as to be surrounded.
36 Solid Stafe Physlcs

by ions of the opposite sign in accordance with the formula of the

compound and the ratio of their radii.
Finally, let us consider crystals featuring the hydrogen hond.
A typical representative of such crystals is ice. Figure 1.23(a) shows
a two-dimensional diagram of the arrangement of water molecules
in an ice crystal: each molecule is sarrounded by four nearest neigh-
bours a distance r0H 2.76 away from it with whom it forms
hydrogen bonds. In space the molecules occupy the vrtices of a

Figure 1.21 Structure of rock salt NaCl crystal.

j
regular tetrahedron (Figure 1.23(b)). The combination of such tetra- . |
hedra forms the regular crystal structure of ice (Figure 1.23(c)).
The structure is very loose and this is the cause of the ahnormally
low density of ice. Upon melting, some ( ~ 1 5 percent) of the hydro
gen bonds are disrupted and the packing density of the water mole
cules rises somewhat with the resultant rise in the density of water:
the density of ice at 0 C is 916.8 kg/m3, and the density of water
at this temperatura is 999.87 kg/m3. ....
It may be of interest to note that if there was no hydrogen hond
the melting point of ice would be 100 C instead of 0 3C.
It should finally be stressed again that the hydrogen hond plays
an extremely important part in vital biological compounds: protein
molecules owe their heiical shape exclusively to the hydrogen hond;
the same type of bonds holds together the double helixes in the DNA.
I t is no exaggeration to claim that lile on our planet would have
assumed radically diSerent formsif any at allwere hydrogen
bonding not present in water and in the proteins and nucleic acis
that compose living cslls and that transmit hereditary traits.2

2 G. C. Pimentel and R. D. Spratley,. Chemical Bonding Clarifie Through

Quantum Mechantes, Holden-Day, San Francisco (1969), p. 261
5ondng. Inferna! Strucfure of Soiids 37

Figure 1*22 ESect oi relativa dim8nsions of icns on theirpacking in

the lattice. ....

Figure 1.23 Crystala with aydrogen bond: (a) plae diagram of

arrangement of water molecules in an ice crystal; (b) spatial arrangement
of water molecules in an ice crystal; (c) crystal structure of ice
33 Solid State Physics
ii
j
Table 2 of Appendix IV shows the general classication o solids. 5
The upper left crner contains typical metis with collecti-vized 1
eIectrons'{silver, copper) and the upper right comer typical. valence ]
crystals with distinctly localized electrn bonds. The extreme right-
hand part contains crystals with van der Waals bonds. Such elements
as Silicon and germanium occupy an intermediate position between
the metis and the valence crystals. At absolnte zero they are typical
valence crystals; however, as temperature rises the valence bond is
gradually disrupted and they begin to exhibit metallic properties.
Such solids as sulphur, phosphorus, and selenium occupy an inter
medate position between the valence crystals and crystals with
the van der Waals bond.
The lower left crner of the aiagram contains alloys of the NiCu
type with the characteristic metallic bond and the lower right crner
ionic crystals (sodium chloride). Intermedate positions between
them are occupicd by numerous intermetallic compounds of the
Mg3Sb type featuring the ionic bond (Mg3Sb 2 corresponda to a
a Mg^--Sb-3 compound). Intermedate position between the ionic
and the valence crystals is occupied by such compounds as SiO,
and SiC, with bonds of partially ionic nature made possible because
of electrn displacement. Compounds of the FeS and the TiO (tita-
nium dioxide) tj'pe occupy an intermediate position between ionic
crystals and crystals with the van der Waals bond.
There are a great many crystals in which ionic or covalent bonds
act in atomic planes while the bonds between the planes are of the
van der Waals type.

11 Polymorphism
Some solids have two or more crystal structures each of which is
stable in an appropriate range of temperaturas and pressures. Such
structures are termed polymorphic modications, or polymorpks,
and tile transition from one modihcation to anotner, polymorphic
transforrnation.
It is the practice to denote polymorphic modications by Greek
letters: the modification stable at normal and lower temperatures
is denoted by a; modications stable at higher temperatures are
denoted by the letters 5, y, , etc. The polymorphism of tin may
serve as a classical exampie. Below 13.3 C the stable modication
of tin is a-Sn, which has a tetragonal cubic lattice of the diamond
type. This is the so-called gray tin. It is brittle and may easily be
ground to powder. Above 13.3 C a-Sn transforms into j-5n, which
has a body-centered tetragonal lattice. This is the familiar white
metallic tin, a rather ductile metal. The transforrnation from {J-sa
to a-Sn is accompanied by a considerable increase in snecific volume
Soning. internai S'iruciure of Solas 33

(by about 25 percent). Long ago when many things were raade
of fin, tbe perplexing phenomenon of growing bulgas on tbem and
their subsequent destruction following excessive cooling was attri-
butsd to a mysterious metal disease, the tin plague.
Many other Chemical elements also exhibit polymorphism: carbn,
iron, nickel, cobalt, tungsten, titanium,boron, berillinm, and others,
as wll as many Chemical compounds and alloys.
An interesting and a practically important case of polymorphism
is that of carbn, which exists in the forms of diamond and graphite.

Figure 1.24 Crystal stnicture of graphite.

Tais case deserves a more dstailed study. In the diamond lattice

every atom is surrounded by 4 nearest neighbours occupying the
vrtices of a tetrahedron (see Figure 1.14) with whooi it is bondad
by strong covalent forces. The length of the bond is 1.544 A and
the energy per bond is about 3.5 X 103 J/mole.
The graphite lattice is of the pattern characteristic of the Group VB
element lattices: the carbn atonas form two-dimensional layers in
which every atom is surrounded by 3 nearest neighbours with whom
it is bondad by covalent bonds (Figure 1.24). The length of the
bond is rn = 1.42 , that is, less than in the diamond lattice,
therefore the formar is stronger. The distance between the layers is
much greater than the length of the. GC bond, being equal to
r3., = 3.6 . Only weak van der Waals forces can act at such great
distances and the layers are held together by them. The energy of
this bond is (4-3) X 103 J/mole.
Such a great difference in the natura o the bonding forces in the
diamond and graphite structures should evidently manifest itself
in a great difference n their properties, which is actuallv the case.
40 Salid Srafe Physics

Diamond is extremeiy hard and strong and is an ideal abrasive.

Grapbite slides easily along tbe planes held together by weak van
der Waals forces. Tberefore it is used to advantage in making lead
pencils and as dry lubricant. Tbe electrons in diamond are hel
securely between tbe atoms orming bonds. Ligbt o tbe visible
part of tbe spectrum is unable to knock ont sucb electrons and tbere
fore is not-absorbed in diamond. Becanse of tbis diamond is an ideal
transparent crystal unable to conduct electric current (a dielectric).
In grapbite one of tbe four valence electrons of tbe carbn atom
is actuallv collectivized by tbe atoms orming tbe layer. Sucb elec
trons can easily be moved by tbe action of an externa! electric field,
making grapbite a two-dimensional conductor. Tbe presence of
mobile electrons explains ligbt absorption (tbe gray coiour of
grapbite) and its characteristic metallic glitter.
In normal conditions grapbite is a somewhat more stable moaifi-
cation tban diamond, altbougb tbe diference in the energies of
tbose modications is quite sm all oftbe order of 2 x 103 J/mole:
C ( d i a m o n d ) C (grapbite), A7 = i . 88 X 103 J/mole
Still, sucb afdifference is enougb to bring about a sufciently
rapid transformation of diamond into grapbite when beated above
1000 C in tbe absence of air.
Tbe density of diamond is greater than tbat of grapbite (3500
and 2250 kg/'m3 respectively), whicb is due to a loose packing o tbe
atomic layers in grapbite. Tberefore at greater pressures diamond
becomes more stable and grapbite less stable. At sufciently bigh
pressures diamond becomes more stable than grapbite. In sucb condi
tions by raising the temperatura to increase tbe m bility of tbe
carbn atoms we may bring about tbe transformation of grapbite
into diamond.
Tbe conditions for sucb transformation to proceed at a practical
rate were calculated by tbe Soviet pbysicist O. I. Leipunskii. He
writes: Firstly, graphite sbould be heatjj^tq. at least 2000 C for
tbe carbn atoms to be able to move frm place to place. Secondly,
it must be subjected to very bigb pressure, not less tban 50 000 at-
mospberes.3 Tbese conditions -were rst acbieved by tbe scientists
of tbe General Electric Research and Development Canter, who in
1954 succeeded in producing tbe first syntbetic diamonds in tbe form
of dark unsigbtiy crystals, tbe largest being 1.5 mm long. Subse-
quently tbe syntbesis of diamonds was mastered in Sweden, tbe
etberlands, and J apan.
In tbe Soviet Union tbe production of syntbetic diamonds on
a comercial scale began in 1961. The pressure in tbe process is of

3 Leipunskii, O. I.: Quoed frota I. I. Shafranovskii, D iam o n d a, Nauka ,

Moscow (1964). -
Sondtng. niernal Structure of Sotids 41

the order of 100000 atm and the temperature about 2000 C. Synthel-
ic diamonds produced by tbis process are bardar and stronger tban
natural diamonds and their industrial use is about 40 percent more-
efcient tban tbat of natural ones.
Anotber material of extreme hardness bad been synthesized in
a process similar to tbat of tbe diamondtbe cubic boron nitride-
BN, wbicb became known as borazon. It is barder tban diamond and
may be beated up to 2000 C in atmospberic conditions. In is hexago
nal modication boron nitride is similar to grapbitea white powder
oily to tbe toucb.
From tbe tbeoretical point of view all soiids sbould exbibit poly-
morphism provided the range of tbeir stability is not limitad by
tbe processes of melting and sublimation. Tbe existence of polymor-
phism is a direct consequence of tbe variation of tbe strength and
tbe nature of tbe bonds in tbe crystal lattice caused by tbe changas-
in intensity of atomie motion and in tbe distance between them as
a result of heating or of application of external pressure to the-
crystal. Cise to absolute zero tbe stable structure sbould be tbat
with tbe strongest bonds possible for tbe given atomie ensemble.
In the case of tin, which belongs to Group IV of the. Mendeleev
periodic table, sucb structure is tbe diamond structure, in wbicb
every atona is bonded to 4 nearest neigbbours by strong colvalent
bonds. However, as the temperature is raised, tbose bonds, because
of tbeir strict directionality and rigidity, are easily destroyed by
tbermal motion, and already above 13.3 C the flexible, metaliic
structure formed witb the aid of collectivized electrons becomes
more favourable. Tbis bond has its own stable crystal structure.
tbe tetragonal body-centred lattice.
Tbe transition from one modication to anotber is accompanied
by tbe liberation or absorption of laten beat of transformation
and is thereore a pbase transition of tbe rst kind. Sucb a transition
involves tbe transformation of tbe crystal lattice and tbis fact togeth-
er witb a low mobility of atoms in soMdsmakes possible a pr-acti-
caliy innitely long existence of a modication tbermodynamically
unstable in particular conditions. Diamond wbicb can exist ages
witbout turning into grapbitetbe stable modication in normal
conditionsis a striking example of tbis point.
Polymorpbism is very important for practical purposes. Eeafc
traament of steels to obtain various properties, tbe production
of stainiess steels, tbe treatment of various ailoys to obtain the
necessary properties are all to a large extent based on tbe use of
polymorpbism.
42 Solid Sraie Physics

12 imperfections and defecfs

of the crysfal lattice
Mosaic structure. Numerous data obtained in the study of the struc
ture of real crystals point to the fact that their internal structure is
essentially not the same as that of an ideal crystai. To begin with,
real crystals have a mosaic structure-. they are made up of regular
blocks which are only approximately parallel to one another. The
-dimensions of the blocks vary from 10-8 to 10~8 m and the angles
between them from several seconds to tens of minutes. Because of
the dierent orientation of adjacent blocks there is a transition
layer between them in which the lattice changes its orientation

Figure 1JZ5 Crystai lattice deects: (a) Frenkei deects;

<b), (c) Schottkv deects.

-
1

1
2 t; 3

2 3 4 \
>- .4

(a) (b) (c)

gradually from that of the urst block to that of the second. Therefore
in this layer the lattice is deformed as compared with that o an
ideal crystai.
Lattice deformations are even greater near the 'grain boundaries
in a polycrvstal. since the orientation of adjacent grnins may difer
by as much as tens of degrees. The grain and block boundaries carry
excess free energy, which iucreases the rate of Chemical reactions,.
of polvmorphic transformations, of difusion, etc. They also serve
as effective carrier scattering centres responsible fot a considerable
part of the solids (metal or semiconductor) electrical resistance.
Frenkei deects. The distribution of energy among the atoms of
a solid is very nonuniform, as is the case with the moiecules of a gas
or liquid. At any temperature there are atoms in the crystai whose
energy is many times greater or less than the average energy corre-
sponding to the law of equipartition of energy. The atoms that at
a given instant of time have enough energy can not only move a
considerable distance away from their position of equilibrium, but
can also surmount the potential barrier set up bv the neighbouring
atoms and move over to new neighbours, to a new cell. Such atoms
acquire the capabiiity, so to say, to evaprate" from their lattice
sites and to condense in its internal cavities, or interstitials (Fig
ure 1.25(a)). This process results in the creation o a vacant site (a va-
Sonding. Interna! Sructure of Soiids 43

caney) and of an atom in tile interstiiial position (a displaced atora).

Such lattice defeets are termed Frenkel dejeets.
Calculation shows the equilibrium. concentration of interstitial
atoms n? at a given temperature to be
nP = ANe~E?/h-BT (1.18)
where E? is the formation energy of the interstitial whose order of
magnitude is several electrn volts, N is the number of sites in the
given volume, and A is an integer (usually about i) indicating the
number of identical interstitial positions per one lattice atom.
Both the interstitial atoms and the vacancies do not remairi
localized in one place but difuse through the lattice. The diSusion
of a displaced atom proceeds by the motion of tbis atom from one
interstitial position to another, and the difusin of a vacancy by
a relay process in which the vacancy is filled by neighbouring atoms
(Figure 1.25(a)): when atom 2 moves into vacancy 1 the vacancy
moves over to site 2, when atom 3 moves to the now vacant site 2
the vacancy moves to site 3, et,c.
Schottky defeets. In addition to internal evaporation there is also
a possibility of a partial or eren complete evaporation of an atom
from the surface of a crvstal. Complete evaporation means that the
atom leaves the crystal surface and joins the vapour phase (Fig
ure 1.25(b)). Partial evaporation means that the atom leaves the
surface laver and arranges itself on top of it (Figure 1.25(c)). In
both cases a vacancy is produced in the surface layer. But when an
atom from the interior of a crystal occupies a vacancy, the latter
is pulled into the crystal and diuses there. Here there are no dis
placed atoms to correspond to the vacancies, since the latter are pro
duced without the simultaneous transition of atoms to interstitial
positions. Such vacancies are termed Schottky defeets. Calculations
show the equilibrium number of vacancies nsCfl in a crystal of A
sites to be
(1.19)
where Esco is the energy of formation of a single vacancy. It is
somewhat lower than E?. For instance, for aluminium it is equal
to 0.75 eV. Substituting this valu into (1.19), we obtain nsCh ~
^ 1013 m~3 at T = 300 K; at T = 923 K, that is, cise to the melting
point of aluminium (Tm = 933 K), nsci 1025 m '3. Such vales
are characteristic of- all metis at temperatures cise to their melting
points.. ; -
The energy of formation of the Frenkel defeets is approximately
equal to the sum of formation energies of the vacanc 3r and the inter
stitial atom.
The Frenkel and Schottky defeets play an important part in many
processes in crystals. They act as carrier scattering centres reducing
44 Solid Sfate Physics

tlieir mobility. They can also act as sources o carrier production,

that isr play the role of donors and acceptors (usually the latter).
They can also appreciably afect optical, magnetic, mechanical,
and thermodynamic ptoperties of crystals, expecially of thin semi-
conducting films and fine crystalline specimens (because defect
concentration in them is usually much greater tban in bulk speci
mens).
mpurities. Impurities are one of the most importaat and most -
common type of detects in the structure of real crystals. Contempo-
rary refining methods are unable to guarantee absolute purity of
materials. Even the most pur materials contain up to 10-9 percent

Figure 1.26 Deormations oi crystal Iattice of solid Solutions:

(a) intersticial; (b) substitution.

of impurities, whieh corresponds to a concentration of about 1017 im-

purity atoms per cubic metre of the material. To illustrate this
negree of purity we would like to cite an equivalent exampie of one
grain of rye contained in about 10 tons of wheat.
The impurities contained in the crystal may, depe'nding on their
nature, be in the form of dissolved atoms or in the form of inclusions
of various dimensions. In the procesa of dissolution the impurity
atoms enter the interstitial positions between the atoms of the crystal
or substitute some of them in their sites.. The solid solution of the
first type is termed the interstitial solution '(Figure 1.26(a)) and that
of the second the substitution solution-(Figure 1.26(b)). Because of
a diference in the physical nature and the dimensions of the impu
rity atoms from the atoms of the crystal, their presence results in the
deformation of the crystal Iattice.
Impurities may appreciably affect chemical, optical, magnetic,
and mechanical properties of solids. They are eSective carrier scatter-
ing centres, being the cause of electrical resstanos that does not
vanish at absolute zero temperature. In semiconductor crystals the
impurities create new energy levels leading to the appearance of
impurity conductivity. Calculations show that a periectly pur
Silicon should have a specic resistance of the order of 2000 ohm-m.
Bonding. Infernal Sfrucfura of Solids 45

Active impurities contained in it in a concentration of 10"9 perceat

reduce the resistvity- to several units. Technically pur germanium
was for a long time regarded as a metal because its resistivity was
of the same order as that of metis. Only perfect renng methods
that brought impurity concentration in germanium down to 10~7-
10_s percent made it a typical semiconductor.
Int9resting results were obtained in tbe course of investigations
into tbe properties of extremely pur metis. Tbus thoroughly
puried iron turned out to be chemically inert and immune to corro
sin even in conditions of tropical humidity. Titanium, cbronium,
bismuth, tungsten, molybdenum, wbich had a reputation for brittle-
ness, turned out to be ductile even in conditions of extreme cooling;
tin puried to contain no more tban 5 X 103 percent impurities
is so soft tbat it bends under its own weight like dougb.
Some striking results were obtained in debydration experiments:
materials dried so as to contain negligible amounts of residual
moisture change tbeir properties in a marked degree. Tbus dried
oxyhydrogen gas does not explode at bigb temperatures; carbn
monoxide does not hura in oxygen; sulpburic acid does not react
witb alkali metis, etc. Tbe Englisb cbemist H. B. Baker sealed
11 tborougbly puried individual cbemical compounds in glass tubes
togetber witb pbospboric anhydride (a powerful absorber of water).
The tubes-were opened 9 years later in conditions that precluded
tbe admission of moisture. Tbe results were startling: the boiling
point of all tbe compounds rose appreciably. For instance, tbe boiling
point of benzol turned out to be 26 C higher tban tbat specified
in tables, that of etbyl alcohol was 60 G bigher, tbat of bromine
was 59 G bigher, and tbat of mercury was almost 100 C bigher.
Subsequent experiments carried out by otber investigators substan-
tiated tbose results. More tban tbat, it was established tbat vsry
dry materials not only changa tbeir boiling point but melting point
and otber properties as well.
Despite substantial progress in tbe eld of production of ultrapure
materials .there is a growing demand for b'ecter purication methods.
and presently there will be a need for materials witb impurity con-
centrations of no more tban 10 10-i0~13 percent. Tbis applies in tbe
rst instance to materials used for thennonuclear fusin apparatus,
to microelectronics materials, as-well as to matarais used in otber
brancbes of industry. Sucb materials are not only difcuit to produce
but also difcuit to keep pur, especiaily if tbey nave to be processed
before use. To illustrate bow easy it is to make a mistaba wbile
working witb sucb materials we wouid like to cite a case told by
tbe weil-known Germn pbysicist Weraer Heisenberg. When a target
was irradiated' witb a flux of neutrons in a mass spectrometer, gold
nuclei were detectad. Tbis enect vanisbed after tbe experimentar
took oir and put away bis gold-rimmed eyegiasses.
 F v le c a n ic a ! P r o p e r tie s o f S o lid s

The mechanical properties strengtn, hardness, ductility, wear-re-

sistance are the most characteristic of the properties of solids.
Thanks to those properties the practical use of solids as construction-
ai, building, electrotechnical, magnetic and other materials without
which the growth of economy is impossible has become so widespread.
The very ames of the periods of human culture reect the ames
of the solids whose mechanical properties made a qualitative leap
in the process of development of human society possiblethe Stone
Age, the Bronze Age, the Iron A.ge.
This chapter deais briefly with modern physical concepta concern-
ing the mechanical properties of solids, the laws of their plstic ow
and destruction, the physical nature of strength, and prospects or
the development of materials with unique mechanical properties.

13 Elastic and pasfic deformaflons.

Hooke's law
When a crystal is subjected to an external extensin load, the dis-
tanees between the atoms become greater and the atoras are displaced
rom their equilibrium positions in the crystal. This destroys the
equiiibrinm between the orc-es of repulsin and attraction character
istic of the equilibrium State of the atoras in the lattice andresults
in the appearance of internal forces tending to retara the atoms to
their initial equilibrium positions. The valu of those forces per
unit. cross-sectional area of the crystal is termed stress. Let us
calclate it.
It was shown in Chapter i that the encrgy o interaction o parti
les 1 and 2 in a solid is a function of the distance r between them.
This can be described by the curve U(r) scfcematicaily shown in
Figure 2.1(a). When partile 2 is displaced rom its equilibrium
position to a distance x, that is, when the distance between the
partidas is increased to r r 0 -f- a:, the particles energy grows
45
Mechanical Properites of Solids 47

and becomes U (r). The change in energy U({z) = U (r) U (r0>

can be found if we expand U (r) into a Taylor series in powers of z:
d \
( 2 . 1)
dr 1 o
Leaving only the quadratic terna of the seriesandTtaking into-
aucount the fac that (.dU /0r)o at point O' is zero, we obtaiu

( 2. 2>

where p is the rigidity of the bond.

We obtained an approximate expression for the change in energy
of a particle brought about by its displacement from its equiiibriura

Figure 2.1 Variation of (a) interaction energy and (b) interaction forc with
the displacement o a particle from equilibrium position ;by a distance x.

position to a distance z. This expression is an approximation hecause-

we let only the quadratic t-erm in the expansin (2.1), neglecting'
higher-order ternas. Graphically this dependence is exuressed by
a paranoia shown in Figure 2.1(a) by a dotted line.
The forc which appears between partiles 1 and 2 when the dis-
tance between them is changed by x is equal to
dU(x)
/ = (2.3)
dx

It follows fronai(2.3) that the forc is linearly depencent on x

and is directed towards the position of equilibrium, as indicated
bv the minus sign. It is wel'l known that a body acted upon by suck
a forc oscillates harmonically. Therefore such. forc is termed harm on
io, the same term being applied to the approximation (2.2) (harm on
io approzim atiori). Figure 2.1(b) is a schematic diagram of the / (x)
dependence for small vales of x. It is a straight line.
Now let us imagine that a tensile load F is applied to a rod with
a cross-sectional area S and a leagth L . This load changes the distance
between the neighbouring atomic planes 1 and 2 by the amount x

Figure 2.2 Uniaxial extensin of a rod by an exernal forc F:

1 and 2 are the schematic representation of adjacant atomic planes.

I 2

causing thereby an extensin of the rod by L L (Fijare 2.2). It wili

be counterbalanced by the internal forc r ;nt equai numerically to
F *1 = f t r = (N
$x2.4)
where N is the number of partiles in the atomic layer of area S .
The stresses a whicn appear in the extended rod will be

s s pz (2.5)

where c N $S . Multiplying and dividing the right-hand side

of (2.5) by the distance between the atomic planes, r0f we obtain

a = crn = ns (2.6)
ro
where
V
E = cr0 = j3r( (2.7)

is termed the elasticity m odulas, or Y oungs modulus, and

s = x/r0 (2.S)
is the relativa change in the lattice parameter in the irecticn of t -
external forc F .
Multiplying the numerator and the denominator of (2.8) by the
number of atomic layers N' contained in the sampie of length L ,
Mechanica! Prcperries of Solids 49

ve obtain
xN' _ A
s (2.9)
r0N' ~ L

Henee is the reiative elongation of the sampie under the action

of the esternal tensile load.
It follows from formula (2.6) that as long as the harmonio approxi-
mation remains valid, that is, as long as the forces acting between
the partiles displaced in relation to each other as a result of the
deformation of the body remain linear functions of the displacement,
the stresses o which appear in the body will remain proportional
to the reiative deformation of the body:
a Es

The elasticity modulus E serves as the proportionality factor.

Formula- (2.6) expresses the well-known Hooke's law. It is valid
only as long as the harmonio approximation is valid, that is, only
for very small reiative deformations e.
The physical meaning of the elasticity modulus is quite evident
from formula (2.6). Putting s = i, we find that a = E . Henee the
elasticity modulus is numerically equal to the stress which is capabie
of causing an elongation AL L of the sampie, proviaed Hookes
law remains valid and the sampie is not destroyed. No real material
except rubber can stand such deformations.
Table 2.1 shows the vales of the elasticity modulus of some
metallic crystals.
Table 2.1

E (GPa) G (GPa)
Sabstance
mximum mnimum mximum minimum

Alumimum 77 64 ; v.v- 29 25
Copper 194 68 77 31
Iron 290 135 118 60
Magnesium 514 437 184 171
Tungsten 400 400 155 155
Zinc 126 65 497 278

It follows from data presented in Table 2.1 that the elasticity

modulus of solids is very large (of the order of 1010-10u Pa), which
is an indication that the honding forces in those boaies are very
strong.
For some crystals the valu of the elasticity modulus depends
appreciably on the direction in which the latiee is deformad.
Table 2.1 shows the vales of E for airections in which it is at its
4-0835
 50 Soiid Sfafe Physics

mnimum and at its mximum. For some crvstals the ratio 5 ma:c /-min
may be as high. as 3, pointing to a high degree o anisoiropy of such
crystals. - i
The elasticity modulus depends onlv on the nature of the atoms
(molculas) making up the body and on their mutual arrangement.
It can be.changed only by a- substantial change in composition or
internal structure of the solid. However, even in such cases the
changes in E are relatively small. Thus, high concentra-tion alloying,
heat treatm ent, coid rolling, etc. of Steel result in great improvement
" in its hardness and in other mechanical properties but only in negli-
gible (up to 10 percent) changes in its elasticity modulus; alloying
copper with zinc up to 40 per cent leaves the elasticity modulus

Figure 2.3 Typical extensin curve

o a ductile metal: cy yield
stress, Eres residual (plstic)
deformation. OAelastic deiormation
regin, A3 plstic (deiormation
regin.

practically unchanged, although other properties experience a pro-

found change.
We have discussed the tensile stress. However, alX the considera-
tions and the results obtained remain valia for other types of defor
mationcompression and shear as well. In the latter case one
should malee use of the shear modulus G, whose vales are also pre-
sented in Tabla 2.1.
When the externa! load is steadilv increased, stress a and defor
mation s increase steadilv too (Figure 2.3). At some stress a r , char-
acteristic of the specic crystal, the crystal is either destroyed or
the direct proportionality between a and s ceases and a residual
(plstic) deformation s res sets in wnich remains after the externai
load has been removed. The rst case is that of a brittle material
and the second o a ductile one. The stress a y at which a noticeable
plstic flow in the body sets in is termed the yield stress and OA
and .4 5 are the regions of the elastic and plstic deormations re-
spectively. , -
la brittle matarais the elastic limit coincides with the tensile
strength, and their destruccin begins beiore a noticeable plstic
flow sets in. In ductile metis, on the other hand, the elastic limit
Mec'nanical Properties of Sods 5!

and theyield stress are, as a rule.much lower'than the tensile strength,

and these materials are destroyed only after a substantial plstic
deormation has taken place.

14 Principal laws governing plstic flow

in crystals
Residual deformations occur in all cases when the stress in ductile
crystals tested for extensin and compression exceeds the yield
stress. However, neither extensin or compression can by - the en
selves be the causes of such' deformations. An increaso in the length

Figure 2.4 Crystal deormation by a shear forc F: (a) inicial

uostressed crystal; (b) elast'ic deormation caused by shearrng stress
not exceeding elastic limit; (c) early stases of plstic shear (slip)
in the S plae.caused by stress exceeding elastic limit: (d) external
forc is removed, residual deforma tipa (residual shit of one part
of the lattice in relation to another) remains.

*- - - ** - < *
I .!
**- s. *
i i T i
**
i
**** -f
i i i
ra
t t r t
mu f

(a) ib) c id!

of the crystal can only result in an increase in the distance between

the atomic planes perpendicular to the acting forc. When these"
planes are drawn far enough apart, it may be that the forces oi
attraction shall no longer be able to compnsate for the external
load and the crystal will break. Compression can only araw the
atomic planes closer together until the repulsive forces appearing
between the atoms are able to counterbalance the externa! load.
Deormation in this case is ideally elastic and cannot lead to irre
versible displacement of parts of the lattice.
Plstic deormation may only be the result of shearing stresses,
which are able to shit some parts of the crystal in relation to the
others without destroying the bonds between them. Such displace
ment is termed slippin g. It lies at the basis of the plstic ow process
in crystalline materials. Figure 2.4 shows how residual deformations
orignate and develcp in crystals (Figure 2.4(a)) acted upon by
a shearing forc F. As long as the elastic limit is not reached the
crystal experiences elastic deformations (Figure 2.4(b)) with the
 52 Soiid Siae Physics

tangential stresses growing in proportion to the relative shearing

dsiorraation y (H ooke's law ):
r Gy (2.10)
where G is the shear modulus. After the crystal is relieved from'exter-
nal load the atoras return to their initial positions. Wlien the elastic
lixnit is exceeded, one part of the crystal shilts in relatioa tcnanother
(Figare 2.4(c)) by one or more atomic distances along denite planes
S termed slip p lan es. When the external load is withdrawn, the
elastic stresses in the lattice vanish. However, one part of the crystal
remains displaced in relation to another (Figure 2.4(d)). Such small
irreversible displacements that proceed along numerous slip planes
sum up to produce the residual deformation of the c o s ta l as a whoie.
The degree to which a crystal can be subjected to plstic deforma-
tions is determined, first of all, by the nature of the bonding forces
acting between its structural elements.
The covalent bond with its rigorous directionaiity is appreciably
weakened already by small relative displacements of the atoms.
Shear destrovs such bonds even before the atoms are able to establish
them with other neighbouring atoms. On account of this the valence
type crystals (such as diamond, Silicon, germanium, antimony,
bismuth, and arsenic) are incapable of plstic deformation. Outside
the elastic deformation range they experience brittle destruction.
The metailic bond, which does not exnibit any directionaiity,
on the other hand, remains practically unchanged- as a result of
relative tangential displacements of the atoms. This makes very
great (some thousand atomic distances) relative displacements of
some pars of the lattice possible, resulting in a high degree of auc-
tiiity of crystals of this type.
The ionio bond occupies an intermedate position between the
metailic and covalent bonds. It is iess directional th-an the covalent
bond but not so flexible as the metailic bond. Typical ionic crystals
such as NaCl, C aF,, and KC1 are aimost as brittle as the valence-
tvpe crystals. At the sam lime silver chloride crystals are rather
ductile.
Slipping takes placa in c o s t is along denite. crystallographic
pa nes and directions, usuaiiy along- the closest-packed planes and
directions. This is because the closest-packed planes and directions
are the strongest since the interatomic distances in them are the
shortest and bonding is at its mximum. On the other hand, the
distance between such planes is the greatest (see (1.17)); on account
of this the bonding between them is at its mnimum. Slipping along
such planes and directions results in the mnimum disarrangement
in atomic order and is therefore the easiest to accompiish.
The combnation of the slip plae and the slip direction, which
lies in it, orms the slip system. In the face-cantered cubic lattice the
Mechanical Properiias of Sods 53

slip plae coincides with the octahedral plae (111) and the s-lip
direction with the direction of the body diaconal [ l i l i . In hexagonal
crystais the S S slip plae coincides with the base plae (0001) and
the X slip direction with one of the three axes lying in the base
nlane (see Figure 2.5, where P is the esternal'deforming load).
Numerous experimenta nave shown that the crystal begins to
slip in the givsn slip system only after the shearing stress x acting

Figure 2.S Slip planes ana directions in a crystal.

Crystai before deformador! Crvs:a aiter riefonnadon

in this system reaches the critical valu xcr termed the critical shear
ing stress. Table 2.2 shows the vales of critica! shearing stresses
for some pur metallic single crystais.

Table 12

Impurity Slip Tcr ;-

Metal
content (lO-4) plne direction (1Q7 Pa)

Cadmium 0 .4 (000 i) [100] 0.05S

Copper 10 (111) (101 ] 0.1
Magnesium 5 (0001) [100] Q.083
Nickel 20 (til) [101] 0.58
Silver 1 (111) [101] 0.08
Zinc 4 0001) [100] 0.004

It foilows from the data of Table 2.2 that for' the most ductile
single crystais the critical shearing stress does nct exceed 106 Pa.
The critical shearing stress depends to a large extent on the prior
deformation of the crystal rising with the increase Tn. the latter
54 Solid Saie Physics

This pnenomenon became known as strengthening, or coid working.

Thus a 350 percent preliminary deformation of the magnesium single
crystal increases xcr nearly 25 times. Even greater is the effect of
cola working on the cubic crystais aluminium, copper, nickel,
etc.
The strengthening of crystais is a witness to the fact that irrevers
ible processes involving the relative displacement of atoms and
of parts of the crystal take place. This results in changes of the
internal energy of the crystais. Experimental study of this phenome-
non has proved that the changes in the internal energy of solids
in the process of their plstic deformation do, indeed, take place.
Table 2.3 shows the mximum amounts of energy that are accumulat-
ed by various metis in the process of their plstic deformation.

Tabie 2.3

Metal Q (J/kg)

Aluminium 4400
Bras 2000
Copper 20OO
Iron 4SOO
Nickel 3t20

Should this energy be transformad into heat it would sufce to

heat the metal by several degrees.
Since the accumulation of energy in the crystal in the process of
its plstic deformation involves irreversible displacements of the
atoms and of parts of the crystal, this energy is, in effect, the energy
of residual stresses remaining in elastically deformed parts of the
crystal iattice.
Because of a higher valu of internal energy in a coid worked
crystal it is less thermodynamically stabie than the annealed crystal.
This gives rise to processes that tend to bring the crystal to the
equilibrium State. Relaxation and recrystallization are two such
processes.
R elaxation consists in the dissipation of internal stresses, with
the atoms of the deformed parts of the Iattice returning to their
regular positions. This process does not involve visible changes in
the crystal structure and results in a parta! or complete removal
of the strengthening obtained as a.result of plstic deformation.
Being a diffusion-controlled process relaxation proceeds at a rate
that strongly depends on temperature and on the latent heat of
defect formation. Metals with a low melting point (such as tin,
lead, cadmium, zinc) have com-parativelv high self-aiffusion rales
Mechanical Properfies of Solios 55

already at room temperaturas. Accordingly, their relaxation rates

at room temperatures are quite noticeable. At the same time there
is practically no relaxation at room temperature in the metis with
a high melting point; but the relaxation rate rises sharply as the
temperature is increased (the relaxation process goes as far in 1 min
ute at 315 C as it would in a hundred years at room temperature).
Another process that also results in the disappearance oi the
hardening in a coid worked crystalthe recrystallization process
proceeds intensely at temperatures of the order of one quarter of the
melting temperature of the metal (on the absolute se ale). In contrast
to relaxation, which produces no visible changes in the crystal
structure, recrystallization involves nucleation and growth of new
crvstals free from internal stresses. The nucleation o such crystals
falces place in the rst instanca in the most deformad parts of the
lattice, where much of the excess free energy is concentrated. In this
way a complete change in the microscopic structure of the crystal
takes place with the crystal generallv going over from the single
State to the polycrystaliine one. In the process of recrystallization
the latent heat accumulated in the deformed crystal is given o 2 in
the form of heat.

15 Mechanical twinning
PIastic][deformation may also take the form of twinning, which is
a process of step-by-step relative displacement of atomic planes
parallel to the twinning plae by a fixed distance equal to a fraction
of the lattice parameter. Figure 2.6 shows the diagram of twinning
of the crystal A EC D A . The area A BCD A is the undeformed par
of the crystal, B E C B is the part where twinning has taken place,
and BC is the twinning azis. Tile positions of atoms before twinning
are denoted by crosses. The plae passing through the twinning
axis and separating the regin of twinning from the undeformed
part of the crystal is termed twinning piarte.
Figure 2.6 shows that twinning results in the displacement of the
atoms of the plae 11 relative to the twinning plae B C hy a fraction
of interatomic distance in the twinning direction. The plae 22 is
displaced relative to the plae 11 by the same fraction of interatomic
distance, tile displacement relative to the twinning plae being twice
as great. In other words, every atomic plae parallel to the twinning
plae is displaced in itself by a distance proportional to its distance -
from the twinning plae. As a resuit, the atoms in the twinned re- .
gion assume positions that are mirror reflexions of the positions in
the undeformed part of the crystal in the twinning plae.
Twinning, in the same way as slipping, may take place oniy along
specic crystallographic planes. In case of a face-centsred cubic
 56" Sod Sfae Physcs

crystal tais is the (112) plana, in casa o a hexagonal closs-packed

crystal tais is the (1012) plae, etc. For twinning to take place the
tangential stresses should exceed some critical valu. The procesa
is a very rapid one and is usuaily accompanied by a characteristic
crackie.
Because oaly negiigihle reiative displacements of the neighbourin*
atomic planes are involved in the process of twinning it caaaot
resuit ia a great residual aeormation. For instance, a complete

Figure 2.6 _ Twinning in a co stal:

sign ' denotes initial aornic
positions in the twinning regin.

transition of a zinc crystal to the twinned rom brings about only

a 7.39 percent elongation. For this reason in crvstals capable of
plstic flow by means o siipping, twinning is responsible only for
a negiigihle ractioa of the total plstic aeormation. In- contrast
to that, negiigihle defermation that precedes destruction of the
vaience crystais, in which siipping cannot take place, is due to
twinning. In hexagonal crystais unfavourably oriented in reiation
to the external forc twinning and subsequent reorientation of the
crystal may resuit in appreciable residual deormations produced
by the normal siipping process.

16 Theorefica] and real shear sfrengths

o crystais
Shear is the principie mechanism of plstic ow in crystais. For
a long time it was presumen that such shear takes piacs by means
of a rigid ispiacement. of one part of the crystal in reiation to another
simultaneously along the entire slip plae SS- (Fignore 2.7).
L et us make a rough estmate o the tangential stress needed to
produce such shear.
The atoms of two adjacent parallel planes in an undeformed lattice
oecupy equilibrium sites corrssponding to the mnimum oi the
Mechanical Properties of Solids 57

Figure 2.7 Diagraru o rigid shcar:

(a) equilihrium position o atoms
in atomic planes adjoining the si ip
plae; (b) gradual shift of ODe
plae in relation to another caused
by esternal stress x; (c) iower
atomic plae as a whole displaced
by one interatomic distnce
in relation to the uppcr plae.

Figure 2.3 Variation of resstanos to shear in the process of dispacement

of one part of the lattice in relation to another.

potential energy (Figure 2.7(a)). The forces acting between them

are zero. As one atomic plae is displaced relative to the other tan-
gential stresses t appear. They resist the shear and tend to bring
back the original equilihrium State (Figure 2.7(b)). If we assume
the dependence of those stresses on the dispacement to be sinusoidal
(Figure 2.S), we shall be able to express the resistance to shear in
the forra
; r A .sin ~ ~ (2.11)

where x is the dispacement of the atoms from their equilihrium

positions,- b a the interatomic distnce in the slip) plae, and A
is a constan!. For small displacements sin 2zixlh 2nxlb and there-
fore
t = A 2nzJb (2.12)
On the other hand, for small displacements Hookes lawAs vaiid:
x = Gzd (2.13)
where G is the shear modulas, and d the distnce between the planes.
From (2.12) and (2.13) we obtain A (b/d) GJ(2x). Therei'ore
_G_sin if_ 12 141
53 Sod Sfafe Physics

The mximum, valu Tcr o the angential stress t is attained for

x b 4 and this is assumea t be the theoretical strength:

T cr= T l r (2-15)
Setting b = d, we obtain

- ,= - (2 . 1 6 )

Henee the critical shearing stress should be equal to about one

tenth o the shear modulus. A more rigorous consideraron o the
nature o the bonding forces acting between the atoms leads to
a negligible correction factor. The minimum valu that was obtained
for xcr is .<3/30. Table 2.4 shows experimental and theoretical vales
o xcr or several metis.
Tabie 2.4

TCr (10 Pa)

Tcr (10* Pa) theory
Metal G (10 Pa)
exper ime nt
C/2.-I G/30

Cadraium 0.06 2640 420 ss

Copper 0.10 4620 735 154
Iron 2.90 6900 1100 230
Majjnesium 0.08 1770 230 59
Nickel 0.53 7300 1240 260
Silver 0.06 2910 459 97
Zinc 0.09 3780 600 126

A comparison of these figures shows that the real shear strength

o crystals is some 3-4 orders of magnitude less than' the theoretical
valu. This points to the fact that shear in crystals does not take
place bv means of a rigid relative displacement of atomic planes
but bv means of a mechanism involying. the displacement of a com-
paratively smail number of atoms at a time. The understanding
of this fact led to the evolution of the dislocation theory of plstic
flow of crystals.

17 The disiocaicn concepf.

Principal fypes of disiocaions
The dislocation theory of plstic flow assumes that the slipping
process starts always at imperections in the crystal structure and
develops along the shear plae by means of a gradual motion of this
imperiection which at a time involves only a limited number of
atoms. Such imperections are termed isiocations.
Mechanical Properfies o Solids 59

Edge dislocatioiis. Suppose gliding took place in the crystal K

in the plae A BC D in the direction of ths vector b involving the
area A H ED (Figure 2.9). The atomic planes on both sides rom the

Figure 2.9 Shear that creates an edge dislocation. Shear took place
onty in regin AHED o slip plae ABCD. Boundary H E is the edge
dislocation.

Direction of dislocation
motion

slip plae A H E D are displaced in relation to one another by the

distance b in the slip direction. The boundary H E separating the
area A H E D , where slipping took place, rom the area H B C E , where

Figure 2.10 Arrangemcnt of a-toms in the plae perpendicular to dislocation

line HE (see Figure 2.9). Dislocation occupies the regin in ,vhich lattice
atonis are displaced from their equilibrium sites (bounded by a circle):
O dislocation centre; (a) positive dislocation; (b) negative
dislocation.

M
999 9 9 9 9 9 rr 1
7 i i
-f * 9 t----f - f 9
1 1. i * F 7^K | T.
Y i LOj i , ,L.
u j Y*mi 4 r i> t i.
i l
T l i\ 1 4 H* t * - f * -i-J -.
4 I*rT t;
9O-
! 1 :i 1 . 1 i'i 1 1 1 i iT T
i r i S-J. i l i
\
iaf6- -- -a i a- 1
-
r r i ] i i
- 3 ---- * 4----*5----* *

1) u
\b i

slipping has not yet taken place, constitutes an edge dislocation

and the vector b is termed the Burgers vector. It describes how far
slipping has proceeded in the area A H E D .
Figure 2.10 shows the arraagement of atoras in a plae perpendicu
lar to the dislocation line. As a result of the shit wliich took place
60 So!id Sfae Physics

over the area A H E D the upper part o the lattice contains one atomic
plae (plae OM) more than the lower. Because o that the atomic
rovr 1 lying above the shear plae contains ona atom more thn the
row 2 below this plae. The interatomic distances in the upper row
near the point O (the dislocation centr) will accordingly be shorter
than the normal valu (the lattice is contracted), while the interatom
ic distances in the lower row near the point O will_be longer(the
lattice is extended). As the distance to the left or to the right, and
up or down, from the dislocation centre O increases, the deformation

Figure 2.11 Formation o a screw dislocation: (a) shear which

produces the screw dislocation. It took place in the BCD plae.
3ounriary AD is a screw dislocation; (b) arrangement o atoms around the
screw dislocation. Plae o drawing is parailei to siip plae. White
circles denote atoms o the plae lying immediately above the slip
plae, biacl: circles denote atoms
f the place lying under the slip plae.

o the lattice gradualiy-vsubsides and at an appropriate distance

from O in the c n ^ ta l ntirmar disposition o atoms is restorl. How-
ever. in the direction perpendicular to' the plae o the diagram the
dislocation passes through the entre crystal or through a consid
erable part o it.
Thus a eature o the edge dislocation is lite existence o an excess
atomic plae OM in some part o the ciystal. Therefcre the process
o formation of such a dislocation mav be imagined as that o pulling
the lattice apart and inserting an additionai atomic plae in it.
Such plae is termed exira. p la e. I the plae is inserted into the
upper part o the lattice, the edge dislocation is assumed to be posi
tiva (Figure 2.10(a)). But if the extra piane is inserted into the
lower part o the lattice, the dislocation is assumed to he negative
(Figure 2 .0 (b . A dislocation whose Burgers vector is equal to the
Mechanicai Properties o Solids 81

Figure 2.12 Explaining the origia of the acrsw dislocatioa:

(a) arrangement of atoxns in a screw dislocatioa; (b) atom a moves
towards atoms b, c, d, s, etc. constituting the screw dislocatioa
along a spiral.

iattice parameter is called the unit dislocation. When a unit disloca-

tion passes through a cross section of the crystal, one part of it
shifts in reition to the other by a distance b. The motion of a posi
tiva dislocation to the left causes the same shift of parts of the Iattice
as a motion of a negativa dislocation to the rigkt (Figures 2.10(a, b)).
Screw dislocations. Suppose an incomplete unit shift is made in
the crystal K in the direction of the vector b over the area ABCD ,
as shown in Figure 2.11(a); AD is the boundary of the area that
experienced' the shift. In Figure 2.11(b) the white circles denote
the atoms of the plae immediately above the slip plae and black
circles the atoms of the plae below the slip plae. In the undeformed
part of the crystal to the left of AD the atoms of those planes are
arranged one on.top of the other; therefore the black circles coincide
with the white (this is shown by white circles with a black point
in the centre). In the right-hand part of the crystal, where the shift
has covered one interatomic distance, that is, to the right of E H ,
the atoms of the planes discussed above are also arranged one on
top of the other. However, in the narrow strip A D E H the atoms of
the upper plae are displaced in relation to those of the lower plae
the more the farther - away -fchey are from the boundary,A D . This
displacement resulta in a local aeformation of the Iattice, which
became known as the screw dislocation; the boundary AD is termed
dislocation axis. The origin of the terna screw dislocation mav be
easily understood from Figure 2.12: the motion of the atom a to-
wards the atoms b, c, d, e, etc. (Figura 2.12(a)) lying in the plae
of the screw dislocation proceeds, as may be seen from Figure 2.12(b),
along a spiral. A distinction is made between right and left screw
dislocations (Figure 2.13); the motion of both in opposite directions
results in a shift in one direction.
Gomparing Figures 2.9 and 2.11(a), we see that in contrast to the
edge dislocation, which is perpendicular to the Burgers vector, b,
the screw dislocation is parailel to it. The motion of the edge aisl-
62 Soiid Sfafe Physics

catin is in the direction of the Burgers vector b, and the motion of

ths screw disiocation is in the direction perpendicular to it.
Recentiy experimental methods for direct observation of disloca-
tions have been developed. Figure 2.14(a) shows a schematic diagram
of an electrn micrograph of a thin lm of platinum phthalocyanine
and Figures 2.14(b, c) an electrn micrograph of a copper sulphide
crvstal obtained with the aid of a special procedure. Dark stripes
on the micrographs are the traces of the atomic planes, which in
platinum phthalocyanine are arranged at a disance of 12 trom

Figure 2.13 Right and leit screw dislocations.

one another and in copper sulphide at a distance of 1.88 . The

micrographs distinctly show the extra planes which terminate inside
the crystal and orm edge dislocations.
Figure 2.14(d) shows an ptica! micrograph of a decorated screw
disiocation in a C aF2 crystal. The method of decoration as used for
transparent crvstals consists in the precipitation along their disloca-
tion coros of impurity atoms, which make the disiocation visible in
an ptica! microscope. The striking agreement hetween those pictures
and the theoretical concepts as set out in Figures 2.10 and 2.12 is
vorthy of admiration. Points of exit of dislocations oa the crystal
surace may be detected with the aid of etching. When a crystal
is etched in a specially selected etch, the parts of the crystal where
the lattice is most deformen dissolve more readily because the atoms
in those parts possess an excess energv and are chemically more ac
tive. The places of exit of dislocations on the crystal surace are just
such parts. Figure 2.15 shows a photograph of an etched germanium
surace. The dark patches are the points o exit of dislocations.
Mechanica! Properties of Sods 63

' Figure 2.U Ofaservation of

disiocations in an electrn microscope
(a) schematic aiagram of an
electrn micrcgraph of a thin
piatinum phthalocyanine ftim (dark
fnes are atomic traces); (b), (c)
electrn micrograph of a copper
sulfie crystal (dark linos are traces
of atomic planes): (d) screw
dislocaron in a CaF; crystal
obtained by decoration method.

Figure 2.13 Etch pits on germanium suraca. Dark poiats along the grain
boundary are points of ercit of disiocations.
4 Sod State Physics

18 Forces needed to move disiocaions

Suppose there is a positive dislocation with the centre at (9, bounded
at points a and k, and lying in the plae S in which slipping is
possible (Figure 2.16(a)). In equilbrium the forc with which the
lattice acts on the dislocation is zero. This may easily be seen from
the roller model shown in Figure 2.17. The struc ture of the upper
row of rollers which normally occupy recesses between the rollers

Figure 2.16 Calculating the forc needed to move a dislocation:

(a) regin of positive dislocation in crystal; O is dislocation
centre, a and k are dislocation boundaries, S is the slip plae;
(b) forces needed to move an atom in the dislocation regin (forces
applied to atoms equidistant from the dislocation centre are equal
i magnitude and opposite in direction).

F
F,

1 2 3 4 5 6 7 8 9 10

(b>

of the lower row was deformed so that the section A B which pre-
viously contained 6 rollers now contains onlv 5. Such deformation
gives rise to forces which tend to return the rollers 1, 2, 4, 5 to
their stable equilibrium positions (the forces Fi, F ,, F 3, F 4, F 5).
The. forces applied to rollers 1, 5 and 2 , 4 are equal in magnitude
and opposite in direction. Therefore, if the rollers of the upper row
are interconnected by means of an elastic spring acting as a bona
betwreen them, the forces F x and F 5, F.. and F 4 will be mutually
compensated and the system will be in a state of equilibrium.
The same situation occurs in the case of a dislocation schematicaliy
shown.in Figure 2.16(b); the forces acting on atoms of the upper row
occupying positions symmetrical with resnect of the dislocation
centre O are equal in magnitude but opposite in direction (the forces
F b = F j, F c = F , F d = Fh, F e F 2). Therefore the resultant forc
is zero and the dislocation is in a state of equilibrium. If, however,
the dislocation moves a little in the slip plae the symmetrical
arrangement of the atoms in respect of the dislocation centre will
be disturbad giving rise to a forc which resists the motion of the
Mechanicai Properties of Soiias 65

dislocation. It 3 evident from Figure 2 .7 that this forc cannot be

grsat since tbe movement of be roilers 1 and 2 to their new equi-
librium position is to a larga extent tbe result of tbe actioa of tbe
forces excercised on them by tbe roilers 4 and , wbich also strive

Figure 2.17 Roller model of an adge dislocation! Forces applied

to atoms 1, 5 and 2, 4 are eqnal in magnituda- and opposita
in direction.
T
F F2 f , f 5

to occupy positions of stabie equilibrium. Calculations show tbe

tangential stress needed to move tbe dislocation to be equal to
2*b \
(2.17)
i ( VJ )
where G is tbe sbear modulas, v tbe ? oisson ratio, b tbe interatomic
distance, and d tbe distance between adjacent slip planes. The stress

Figure 2.13 Dislocation mechanism of motion of (a) an earth-worm

and (b) a snaka.

t0 is tbe t-beoretical valu of tbe critical shearing stress. Setting

b d and v = 0.3, we obtain t , = 3 X O-4 G. W itbin an order
of magnitude tbis coincides with tbe experimental vales of Tsr.
Tbus tbe tbeory of dislocations resolves tbe contraietion between
the tbeoretical and tbe experimental vales of sbear strength of
crvstals.
Tbe mechanism of motion by means of dislocations is quite fre-
quent in nature. Snakes, worms, and sbellsb move, because tbey
generate dislocations. The motion of an earth-worm beglns witb
5-0885
. 66 Solid State Physics

the ormation of a n . extensin . dislocation near the neck. The

dislocation-subsequently spreads along the body to the tail (Fig
ure 2.l8(a)). In contrast, the motion of most snakes involves the for-
mation of a contraction dislocation near the tail and its motion
towards the head (Figure 2.18(b)).

1 9 Sources f dislocafions.
Sfrengfhening of crysfals
The dislocations in a real crystal are formed in the process of its
growth from the m elt or from a solution. Figure 2.19(a) shows the
boundaries of two blocks growing towards each other. The blocks

Figure 2.19 Forma tioa

i 1 ! _j i n of dislocations at block boundaries:
u n - (a) two'blocks growing towards
i Ll-i-L i i I each other at an angle cp;
1 1 u i T (b) dislocations appearing when'
1 i 1 >1 T the blocks fuse together.
1i i 1 IT T
1 M i-
i i -l L i I 1 1
1i i L
i 1 \ L n i [
i i \ Lu i :
i i
l ii li iX 1 j i
i 1*11
i 1 1 r i .i .
~i i T 1
1 1l. n
i 1!
J 7 n 7 .L i J

(a) (b)

make a smali angle cp between themselves. - As the blocks fuse to

gether, some of the atom ic planes do hot spread through the entire
crystal but termnate at block boundaries. Those are the places
where dislocations are formecT (Figure 2.19(b)). The same, situation-
occurs in the process of fusin of diSerently oriented grains in a poly-
crvstalline sample. Since the block and grain boundaries in real
solids are very extensive, the number of dislocations in- them is
enormous as m aay as 1(T2 dislocations per'sanare* metra can* be
counted in well annealed metis. Ater coid working (rolling, drawr-
ing, etc.) dislocation densities rise to 1013-O16 m-3. Those disioca-
tions accumulate almost the entire energy absorbed by the metal in
the process of plstic deformation.
Vacancy clusters m ay also serve as sources of dislocations in an
undeformed crystal. Figure 2.20 shows an example of the formation
of a positive and a negative dislocation from a clustar of vacancies.
Mechanical Prop9rtes of Sods 6?

The shear procese in a crvstal in response to an applied externa!

forc is, in etfect, the motion of disiocations in the slip planes and
their emergence through the crystal surace. Should only the disid
cations already present in the crystal be responsible for this process.
plstic deformation would lead to their exhaustion and to the per-
fection of the crystal. This is in contradiction with experiencs,
which demonstrates that as deformation grows the lattice does not
become more perfect. In fact, just'the opposite is true: the density
of disiocations grows in the-process. It is-an-established fact that
disiocations responsible for plstic deformation are generated in the
shear process itself by the action of the external forc applied to the
crystal. One such gcneration mechanism"was discovered in :. 1950
by F. C. Frank and W . T. Read. For the purpose of better understancl-

Figure 2.20 Disiocations formed from vcancy clusters:

(a) vacancy cluster in crystal; (b) positive and negativo disiocations
formed from this cluster.

(a) ib)

ing this mechanism let us consider soap bubble formation with the
aid of a tube (Figure 2.21). After the end of the tube has been immersed
in a soap solution a fat-film remains that cioses the tubes orfice.
As the air pressure in the tube is increased, the film swells and passcs
through the stages 1, 2, 3 , 4, etc. Until it- assumes the shape of
a hemisphere (stage 3) its State is unstable: as pressure fails the film
contracts striving to return to the original State 1. Alter the bubble
has passed stage 3 the State of the bubble ckanges: it is now capabie
of evolution not only at. a constant but also at a gradually decreasing
pressure until it leaves the end o the tube. After the first bubble the
second begins to be formed, foilowed by the third, etc.
Now let us discuss the operation of the Frank-R ead source. Figure
2.22(a) shows an edge aisiocation DD in a slip plae; points D
and D' are xed and do not take part in the motion of the disloca-
tion. Disiocations may be anchored at the points of intersection
with other disiocations, at impurity atoms, etc. Under the action
o io u a Mate rnysics

of an external stress t the dislocation starts bending itx the same

way as was the case with the soap film and at some time assumes
the shape of a semicircle (Figure 2.22(b)). Just like the soap lm
the dislocation can continu to bend only if t grows steadily until
it assumes the shape of a semicircle. Its subsequent evolution takes
place by itself and results in the formation of two loops (Figure
2.22(c)), which after meeting at.p oin t CJFigure 2.22(d)) divide the

Figure 2.21 Procesa of formation

of a soap bubbie.

dislocation in two: an external one, which closes forming an external

circle (Figure 2.22(e)), and an internal one, which returns to the
original position D>'. The external dislocation grows until it reacnes
the surace of the crystal and results in an elementary shift; the
internal dislocation having returned. to the initial position DD'
begins agaia to bend under the action of the applied forc and to
growr in the manner described above. Such process may be repeated

Figure 2.22 Operating sequence of a Frank-Read source:

(a) initial position of dislocation DD', (b) acted upon by external
forc the dislocation bends and assumes the shape of a semicircle;
(c), (d) urther symraecric development of the dislocation loop;
(e) formation of external closed dislocation loop spreading across
the crystal and of internal dislocation DD' returning to the original
position.

(d) (O
Machanicai Properfies o Solids 69

any number of times eventually leading to a noticeable shift of one

part of tiie crystal in relation to another in a particular slip plae.
Low sbear strength of crystals is due to the presence of innate
dislocations and to tlie generaion of others in the process of sbear. On
the otber band, it is an estabiisbed fact tbat the crystal is strength-
ened in the process of plstic defonnation accompanied by tbe
growth in the number of defects. The essence of such strengtbening

Figure 2.23 Schematic

representaron of an edge dislocation
surmounting an obstacle: AB shape
of edge dislocation away from
the obstacle D; 1, 2, 3 gradual
bending of the dislocation as it
approaches D and closing of
the newly formed loop behind the
obstacle; A'B' straightening
of the dislocation far away from
the obstacle.

A' ! B'

is tbe interaction of dislocations with each otber and with otber

tvpes of lattice defects causing tbeir motion in tbe lattice to be
obstructed.
Interaction of dislocations. Every dislocation, being tbe canse of
elastic stresses of tbe lattice, creates a forc eid around itself which
may be described by tbe vales of the tangential t and normal cr
stresses at every point. Wben another dislocation enters this eid,
torces begin to act which strive to bring tbe dislocations togetber
or to move them apart. Dislocations of like signs lying in one plae
are repelled, whiie those of opposite signs are attracted. Tais is tbe
reason why, as dislocations are accumulated in a denite plae,
the crystals resistance to sbear is increased and tbe crystal is strength-
ened.
Surmounting of obstacles. Suppose a dislocation wben moving-
in a slip plae under tbe action of tangential stresses x runs into
a stationary obstacle D , for instance, an intersection with some
otber dislocation, an impurity atom, or some otber type of aeect.
Figure 2.23 shows tbe metbod by means of which dislocation A B
could, tbeoretically, surmount obstacle D ; as tbe dislocation an-
proachesD (positions 1, 2, 3) it gradually bends forming a loop tbat
envelops the obstacle. Behind tbe obstacle tbe loop closes and tbe
dislocation A 'B' again becomes a straight line. Figure 2.24' shows
a pnotograph illustrating a case wben a dislocation runs into a sta
tionary obstacle (darle lines represent dislocations decoraied by
70 Solid State Physics

etching). The sim ilarity in th epictures leaves not a trace of doubt.

as to tiie- validitv of the theoretical pattern shown in Figure 2.23.
. In passing around the obstacle, the length of the dislocation and
the deformation of the lattice are increased, which requires addi-
tional work to be performed. Therefore the resis lance to the motion
of the dislocation in the interval where it has to surmount a defect
is much greater than in other parts of the lattice..This is the essence
of the fact that defects strengthen a crystal. The growth in the number
of dislocations in the crystal with greater plstic deformation in-

Figure 2.24 Microphotographs of

a chrornium grain. Dark lines are
etched dislocations (x2000).

creases the number of obst-acles at points of their intersection, which

is the cause of strengthening brought about by plstic deformation.
Impurity atonas have a similar efect: they create local lattice imper-
fections and thereby hinder the motion of the dislocations, with
the result that the cry stals resistance to shear is. increased. Block
and grain boundaries and foreign inclusions in the lattice are espe-
ciaily eective in hindering the motion of the dislocations. They
sharply increase the resistance to the motion o dislocations and
greater stresses are required to overeme their effect. The phenome-
non of strengthening in the process of cold-working, in the process
of introaucing im purity atoras (doping), and in the process of forma-
tion of inclusions (tempering, aging, etc.) is widely used in practice
to improve mechanical properties of engineering materials. This
method enabied tho strength of the materials to be increased from 6
to S times in the last 40 vears.
Mschanicai Praperties o Solids 71'

20 BritHe srengfh of solids

The destruction of solids m ay be of cae of two principal types: of ths
britils and of the plstic, or viscous, types.
Brittie desruction takes place if the tensile strength of the ma
terial is below the eiastic limit. Such a material experiences only
elastic deformation prior to destruction. No irreversible changes
take place in such a material before it breaks down.
In the ductiie materiais the elastic limit is not only below.the
tensile strength but also below the yield stress. Because of that the

Figura 2.25 Calculatiag theoretical

strength oi solids after Poianyi
(explanation in test).

destruction process is preceded bv an apnreciable plstic deforma

tion, which prepares the subsequent destruction process. In this
case strength, being a typical kinetic parameter, is strongly depen
den: on the time the destructiva stress is appid.
To begin with, le us discuss brittie strength of solids.
Theoretical strength of solids. There nave been numerous at-
tempts to calclate the strength of solids on the basis of molecular
interaction in them. The strength c 0 thus calcuiated is tsrme
theoretical strength.
Kere is a glance at some oi the methods of estimating Oj..
Poianyi's method. The simplest method of estimating the strength
of solids theoretically is due to M. Poianyi. Its essence s as:olio\vs.
Suppose a tensile stress cr is applied to a rod of a cross-sectional
area of 1 m2 (Figure 2.25). This stress increases the distance. between
the atomic planes. It is assurae that for destruction to take place
a stress o 0 able to increase the distance between the atomic planes
by a valu of the order of the lattice parameter a should be applied.
The work needed to move an atomic plae a distance a awav from
the neighbcuring plae is assumed to be equal to <70a. It is.further
assumed that this work is transformad into the free anergy of two
72 Solid Sfae Physicsv

new surfaces with a total area of 2 m2 formed as a result of the break-

up, the free energy being equal to 2a, where a is the surface energy
(surface tensin) of the solid. Henee
c 0a = 2a
and the theoretical strength is
a 0 = 2a la (2.18)
For copper a 1.7 J/m 2, a = 3.6 X 10*10 m, and a0 1 x 1010 Pa,
for silver a 1.14 J/m 2. a = 4 X 10"10 m, and ct0 = 0.6 x 1010 Pa.
D eterm ination of cr0 from the heat of suhlimation. The energy equal
to the heat of sublimation Qs is required for the evaporation of
a mole of a solid. For the evaporation of one molecular layer of the
area of 1 m2 the required energy W is a fraction of Qs equal to the
ratio of the mass of this layer m to the molar mass M :
W = Qsm/M
But
m = N s fi, M = iVAu
where u is the molecular weight, iVA = 6.023 x 102U mole-1 the
Avogadro number. and N s the number of molecules per square metre
of the solid's surface.
For an intermolecular distance of a the area per molecule is appro-
xim ately equal to a2 and the number of molecules per square metre
.Vs a~2. Therefore
= Nj_ Qs
N ^a-
Should the assumption be made that the evaporating molecules
loose contact with the solids surface when they are a distance of
the order of the lattice parameter a away from it, we would obtain
for the forc needed to tear away an entire surface layer as a whoie
W Qs 1
cr,O' (2.19)
a N
ct0 is assumed to be the theoretical strength of the solid.
For copper Qs = 3 X 105 J/mole, a = 3.6 X 10_1 m, and a Q
1010 Pa. Similar calcuiations lead to the ollowing resulta: for
iron o0 2.3 X 1010 Pa, for aluminium a9 0.6 X 1010 Pa, and
for silver a 0 0.6 X 1010 Pa.
C alcalatin g a0 from the jarees of molecular interaction. Finally,
let us discuss the method of calculating theoretical strength of
solids from the forces of molecular interaction. Figure. 2.26 shows
the dependence of the potential energy U (a) and the forc of inter
action between the particles / (x) on the distance x between them.
Since it is not easy to determine the exact law governing / (z)T the
Mechanical Properties of Solids 75

practica is to approximate this dependsnca by various functions.

For instance, M. Poianyi and E . Orowaa used the approximatioa
in the feria of a hal of a sinusoid:

/ ( * ) - / : m a*S n -^ - (2.20>

Whea a boay of cross-sectional area o 1 m2 is slowiy tora ia twor

the requirad torce is o = /iYS) where iVs is the number of partida?

Figure 2J Calcuiating theoretical

strength rom forcas of molecular
intaraction (explanaron in test).

per square metre of the cross sectioa. Substituting / from (2.20)

we obtaia
a = cr0 sin j::f :- (2.21)

where cr3 = / max Ars is the theoretical strength o the body.

For small displacaaients reiatioa (2.21) may be rewrittea in the
form
a = a2 xz/c

On the other hand, for small displacsments Hookes law is vaiid:

y = Exic

Equating the right-hand sides of these eqaations, we obtaia

a0 E!(2a) 0.1E (2.22)
Calculations show that a more accarate estmate of the natura:
of the boading in solids resals only in a negiigibie correction to (2.22).
Comparing the vales o theoretical strength a 4 calculated with.
the aid of various methods we see that all of them yieid nearly the-
same rasult whose order a i magnitude is O.I.E. Thereore it may be:
74 So'd State Physics

legitimately assumed that

cr0 0.1 E (2.23)
This is an enormous figure of the order of 10M 010 Pa.
Real (technical) strength- of solids. The strength of real crysals
and solids used for technical purposes is termed r ea l, or technical,
strength g .. Table 2.5 shows the vales of the eiasticity modulus E\
of theoretical strength a 0 O .E , .of technical strength a r, and of
the ratio c a/ o c for some industrial materials.

Table 2.5 ' .

Eiasticity Theoretical Technicat

modulus strength strength
Substance c0 0.1E
E r
(107 Pa) (107 Pa) (107 Pa)

Aluminium 6000 G00 9.0 65

Copper 12000 1200 23 50
Giass sooo 800 S.O 100
Iron 21000 2100 30 70
Rock salt 4000 400 0.5 SOO
Silver 8000 SOO 18 45

It follows from the data of Table 2.5 that the technical strength
o solids s from 2 to 3 orders of magnitude less than their theoretical
strength.
At present there is a general agreement that such. discrepancy
between o 0 and a r is due to the presence of defects in real solids of
various types,. in particular o microscopio cracks which reduce the
strength of solids. This is accounted for by the so-called Griffitk
theory. Let us calclate the technical strength using this theory.
We take a sample in the form of a tliin pate and apply a tensile
stress g to it (Figure 2.27(a)). The density of elastic energy in such
.an elastically extended sample is c ?/(2ii).4
Now let us imagine that a transverse microscopio crack of the
lengtn l running through the entire thickness of the sample has
developed in it. The appearance of the crack is accompanied by the
formation of a free surface S m 215 inside the sample and by an
increase in the samples energy by the amount i\Ux 2 lb a (a. is

4 Indeed, the relative deormation in a sample under stress a is e = a/E,

the ahsoiute deormation AL = sL (L being the length of the sample). The
work periormed by the stress a to extena the sample by AL is {i/2)aSAL ==
= a2SLl(2E) = g-V/23 (S is the crcss-sectionai area and V the yoiume of
the sample). This work is transormed into the eiasticy energy of the sample
ct volume V. Therefore the specic voium density o the elastic energy is a'2!{23).
Mechanicai Properiies o Soiids 75

the free surface energy of the, sample'per unit area). On the other
hand, the formation' of a crack relieves the elastic stress from the
volume V 125 of the sample, whereby its elastic energy is reduced
by the amount AU P 6a2(2E ). The total chango in the energy
of the sample W(i) brought about by the appearance of a crack in
it is --
.?F(Z) = 2ZSa-Z2S-g- (2.24)

Figure 2.27(b) shows the dependence of W on the length i of the

crack. It has a mximum where its derivative vanishes: d W ll =

Figure 2.27 The Grifth theory of calculating the real strength

of solida (explanation in text).

= 25a l5 a z/E = 0. Denote the length of the crack corresponding

to the mximum energy by Zcr. We obtain from the last relation

ZCr = 2 a E o z (2.25)

It may be seen from Figure 2.27(b) that as long as the length l

of the crack remains below the critica! valu cr energy is needed for
it to develop. On the other hand, starting wi.th l lC7 further exten
sin o the sample results in a reduction in its energy. Therefore it
takes place spontaneously with the brittle destruction of the sample
as the nal result.
Henee the technical strength of soiids having microscopio cracks
should be calculated according to the Grifth theory from relation
(2.25):
Y 2aEIL p V clEU /O
r

This result was subsequently veried by many investigators for var-

ious quite dierent methods of appiying loads to the sample. A neg-
76 Solid State- Physics

ligible diSerenc8 wras observed only in case of the numerical coef--

cient p.
Siiould we substitute the vales of a , E , a . for copper (a m
1.7 J/m 2, E = 1.2 X 1011 Pa, and crr = 1 .8 'X 108 Pa) into
(2.26), we would obtain 8 X 10-8 m. Approximately the same
vales of l may be obtained for ther solids.
It follows that for the strength of the solids to be reduced from the
theoretical valu to the valu of the technical strength microscopio
cracks of the order of several micrometers in length should deveop
in them up to the moment preceding their destruction. Many factors
may be the cause of such cracks.
The cracks may be produced in the course of the production of the
solid, especially in the course of its mechanical processing. A proo

Figure 2.28 Formation of a crack near a dislocation pile-up.

Obstacie

-rack l
\ Plae of mximum
Slip plae SS \ /tensile stresses

A
\

of this is, in particular, a signicant dependence of the strength of

the sample on its dimensin?, especially in the small dimensions
range. Thus the strength of a glass lament of 2.5 im in diameter is
almost 100 times that of a massive sample. The explanation is that
as the dimensions of the sample are reduced so too is the probability
of a large crack responsible for low strength appearing in it. Such
dependence of the strength on the dimensions of the sample became
known as the scale factor. The cracks may be the result of a large num-
ber of vacancies merging together.
Figure 2.28 shows a dislocation mechanism of crack production.
Dislocations of a similar sign move in slip plae -SS and meeting
obstacle B begin to accumulate in its vicinity. Large stresses able to
produce cracks l may deveiop at the head of this dislocation
pile-up.
Machanical Properties o Solids 77

21 Time dependence of fhe sirength

of soiids
The theory of strength based on the condition (2.26) and discussed
above describes actually tbe final stage of the destructiva process
vrben tbe body already contains cracks able to cause brittle rupture.
However, tbe initial stages of tbe destructiva process during whch
tbe cracks originate and grow to attain critical dimensions Zcr are
also important. Tbis process is a gradual one and takes time x to
be completed. Tbe time x tbat it takes for tbe destructive process to
develop from tbe moment tbe load is applied to tbe body to the mo-
ment of rupture is termed tbe durability of tbe material.

Figure 2.29 Durability versus stress for alumiuium (/), Plexigls (2),
and silver cbioride (S).
78 Soiid. Sfrara Physics

The rst experiments aimed at investigating durabiiity were car-

ried out by S. N. Zhurkov and-.G. M. Bartenev with coworkers..They
also developed modera notions about the physical natura o dura-
bility. ' -
It was establisbed by experiments that the durabiiity x, the ten-
sile stress cr, and the absolute temperature T are related by the ex-
pression: _1 ..
r = T / - ya)lhs T : . _ (2.27)
where t 0, U0, and y are constants dependent on the nature and-struc-
ture o the body. - -
For T = constant, formula (2.27) may be rewritten in the form
. (2.28)
where A = x0eUo/k'BT, and p = 7/(^3 T).
Formulae (2.27) and (2.28) were tested on a great number of diSer-
ent matcrials (metis, poiymers, haloid compounds, etc.) in an S

Figure 2.30 Activation energjr

of rupture o visese fare at
diSerent temperaturas
(A - 7 6 C, O -r20 sCt + -F80 C)
versus stress.

to 10 order of magnitude range of the vales of x and in a wide range

of the vales of T.
Figure 2.29 shows the dependence of the durabilities r of alumi-
nium (2 ), Plexigls (2 ), and silver chloride (<?) on the applied stress
a at various temperatures expressed in the log t versus a coordnate
system. It may be seen from Figure 2.29 that the. dependence r (a)
in semilogarithmic coordinates is wrell represented by a straight line.
A am ily of such straight Unes obtained for a given material at diSer-
ent temperatures resembles a fan with the apex at some point callea
p o le. It ollows from equation- (2.27) that x will be independen! of
T and that the straight lines lgT(o) at aifferent temperatures will
intersect at one point (at the-pole) only if U'0 yo 0; but in that
case log x = log x0. Henee the pole should be at a distance log x0
below the q-axis.
It is evident from Figure 2.29 that the polea for.all the~materiais
tested lie practically on the same straight line parallel to the a-axis.
 Machanica! Properties of Sods 79

This means that xg is approximately the same for all the matarais.
Experimenta show i t to be of the order of s, that is,
cise to the period of oscillations of atoms in solids.
Let us take the logarithm of (2.27):

' l o g T = I o g t , - h ^ U = logT0+ ^ (2.29)

U = O , - yo (2 .2 9 ')
Moasuring the dependence of log t ou l/2Vor coustant vales of a,
we can determine U for various vales of the stress o experimentally;
the dimensions of U are that of energy and beca use of this it is termed
activation energy of the destruciive procsss. Figure 2.30 shows the de-
pendence of the activation energy of ruptura of visese fibre on stress-
for various temperatures. It may be seen that U is independen o
T and is determined solely by a; for a = 0 the mximum valu o
U is Ua se 40 kcal/mole; fer a stress a 107 X 107 Pa we see that

Figure 2.31 Durability oi

'*(? . V-
A
V75*C
i
r
visese bre at diuerant
temperatures versus stress.

V7\
V "
\
\ 1
WH \ 1
1
V
w 40 SO 120'
o, 10* Pa

U = 0. Figure 2.31 shows that for a 107 x 10 7 Pa a practicaily

'instantaneous ruptura of viscous bre takes place (during the time
r 0), no-matter what its temperature is.
Meticulous experimenta carried out by S. N. Zhurkov with cowork-
ers and by other investigators on a varietv of matarais demonstrat-
ed that-for metis U0 is quite cise to the sublimation energy Qs and
forpoiymers to the thermal destruction energy Qd. Tahls 2 .6 shows-
the values: o U0, Qs, and Qd for some materials. It m ay be seen that
-CY'coihcides either with Q- orwith Qd with a higa degree of accuracy.
Universal val'dity of the dspendence thus obtained merits the
conclusin that the procass oi destruction of a solid is one of a kinet-
i c na ture (that is,,develops in time) and its origin is the same for all
soiids. Modera- notions of the physical 'mechanism of this process-
are-set out below.
The atoms in a solid take part in thermal oscillations with a pe
riod of t0 1012-10 -i3 s. Thermal uctuations from time to time
so Soiid State PHysics

Table 2.6

Activation energy Sublimation Thermal destruc-

Substance of destruction energy Qs tion energy Qa
U0 (10s J/mole) (105 J/mole) (105 J/mole)

Aluminium 2.16 2.2

Nickel 3.48 3.4
Nylon 1.8 1.72
Platinum 4.8 5.1
Poiymetbyl meth-
acrylate 2.18 2.1-2.2
Polvvmyl chloride 1.4 1.28
ilver 2.56 2.72
Teflon 3.0 3.0-3.1
Zinc 1.0 1.08

xesult in the rupture of Chemical bonds. The probability of his proc-

ess depends on the height of the potential barrier of destruction U
and on the temperature T. This probability increases with the rise
in T and the decrease in U. In the absence of esternal stress a the
energy required to break a bond is equal to the energy of the bond
itself. This is the reason why the height of the potential barrier U 0
obtained from experiments in mechanical destruction of solids
turned out to be equal to the sublimation heat of metis and to the
thermal destruction energy of polymers.
The stresses induced in a body reduce the height of the potential
barrier from 70 to U0 yo and thus increase the probability of rup-
ture of the bonds and, consequently, the number of ruptured bonds
per unit voiume.
The formation of submicroscopic' volumes in which the bonds have
been broken and their merging resulta eventually in the nucleation
and development of cracks. When the length of the cracks attains
a critical valu, the body breaks up under the applied stress.' The
higher is the stress cr the lower the activation barrier U0 ya ana
the greater the rate of bond rupture; therefore it takes less time for
the destructive process to develop, that is the less should be the du-
rability of the body. This is exactly what is observed in practice.
From the above point of view the destruction of solids should take
place at any stresses provided the time they act is long enough. But
in that case it is not easy to understand why bridges.and other in-
stallations built many centuries ago and carrying loads all that time
-still remain intact.
To explain this fact we again tum to Figure 2.29. We see that the
lower the temperature th weaker the load dependence of durabilitv
is. This dependence is practically nill at suciently low tempera-
tures. For glasses and metis with a high melting point already room
temperaturas are low enough. Beca use of that their strength is actual-
Mechanical Propeniss o Solids 81

ly a unique characteristic of the material. la all other conditions it

is aot justified to speak of strength 'without mentioning the time dur-
ing which the material is to work under load. Thus industrial prod
ucs made of Plexigls during a years Service can endure loads not
exceeding 30% of their short-time strength; steam turbine blades
working at hign temperaturas are calculated for strength aiways with
account takan of-their durability.

22 Mehods ofjncraasing the srengih

ofsoids
The nucleation mechanism of breaks in continuity and the mechanism
of crack growth. are both greatly influenced by the atomic structura of
solids. Therefore strength is a sructure-sensitive characteristic of
such boaies.
Stresses in' crystals occasion the production of dislocations and
their motion in slip planes. In this way plstic shifts resulting in
plstic deformations ara realized. Meeting impurities, grain and
block boundaries, interceptions of slip planes, etc., the dislocations
lose their mobility and the crystal is hardened. As was mentioned
above, stresses may develop at the head of a dislocation pile-up ca
padle of causing cracks.
To increase the strength of such boaies it is necessary to retar the
production of dislocations and the nucleation and growth of cracks.
This can be done by two methods.
(1) By producing imperection-free crystals free from internal
stress sources, which in the long run cause the nucleation of cracks.
This method has up to the present been realized onlv in the la-
ment type crystals known by the ame of whiskers. They are sin
gle crystals grown under special conditions using the method of decom-
position or reduction of appropriate Chemical compounds, the meth
od of condensation of vapours of pur metis at an appropriate
temperatura in hydrogen or in an inert gas, and the method of eiec-
troplating metis from a solution onto electrodos of extremely small
dimensions. The lament-type crystals are usuaily 2 to'iO mm long
and 5 to 50 um thick.
A striking property of such crystals is that their mechanical para-
meters are extremely high. Their strength turaed out to be cise to
the theoretical strength of solids. Thus, the strength o iron whiskers
is about i . 34 X 10 10 Pa, of copper whiskers about 3 X 10 Pa, and
of zinc whiskers 2.3 X 10 Pa, while the strength of normal samnies
made from those metis is 3 X 108, 2 X 1G8, and 1.3 X. 10 8 Pa re-
spectively.
Filament-type crystals of iron experience only elastic deformation
reaching an enormous gure of the order of 5-6 perceat, after wiiich
8-0835
82 Solid Safe Physics

brittle destruction occurs. Note that in normal iron noticeable pls

tic ow starts already at a deformation of s. 0.01 percent.
The unusually high. mechanical parameters of thefilament crystals
are due to their ideal internal structnre. Such crystals contain prac-
tically no dislocations, are exceptionally pur and their surface is
so perfect that even a magnication of 40000 times fails to reveal any
traces of roughness. Such perfection is mainly due to the condition
of growth of small-size crystals, in which the freezing-in of lattice
imperfections is less probable because it is easier for them to leave
the crystal through a nearby surface.
Because of the absence of dislocations and of other defects in fila-
ment crystals a shift in a slip plae can only take the forra of a rig-
i shift, in which the bonds of all the atoms in the slip plae are
simultaneously broken. Stresses cise to the theoretical stress limit
of the crystals are needed to efrect such a shift and this is what is
observed in practice.
An unnaturally great elastic deformation of the whiskers is due to
the absence of mobile dislocations, which in normal crystals are
responsible for the plstic deformations occurring already at very
low stresses.
Henee the rst method, the method of producing imperfection-
free (in particular, dislocation-free) crystals, holds out a promise of
producing materials of extreme strength cise to the theoretical
strength of solids.
(2) The second method is a direct opposite of the first. It consists
in the mximum deformation of the internal structure of a crystal
through the introduction of impurities, precipitation of dispersed
pitases, great plstic deformation, etc. Such defects hinder the motion
of the dislocations and the growth' of cracks and thus increase the
strength o the m aterial, as was already discussed in' detail aboye.
Science and industry have up to now made use only of this method and
succeeded in attaining with it a strength of the order of 4 X 10*' Pa.
The efrect this had on technology may be inerred from the following
example. The specific weight of a modern aircrat engine is about
kgf per hp; at the turn of the century it was about 250 kgf per hp.
The recent times have witnessed the appearance of composite ma
terials consisting of a m atriz lled with lament crystals. Stainless
Steel, nickel, titanium and other materials are used for the m atriz.
The matrices are lled with tungsten, aluminium oxide, etc. fila-
ments. The results obtained so far hold out a promise of obtaining
by this method in the near future materials [of [5 to 10 times the
strength (especialiy a t eievated tempera tures) of the best steels and
of 1.5-2 times lighter weight.
The strength of amorphous bodies and glass polymers is no less
sensitive to internal structure. The strength of glass and quartz
lamsnts newiy extruded at a high temperature and practically fres
Machanical Properfies o Solids 83

from deects3 is 100 times as high as that of normal specimens and

quita cise to the theoratical valu.
The room temperature strength of unoriented glass polymers is
of the order o 10 Pa. Films and bres made of them having an orient-
ed structuxe have a strength of the order of 10 Pa comparable to
that of high quality steels. With a perfect orientation of the polymer
molcula chain the strength of the needleiike crystals of the polymer
may be as high as 3 X 10 Pa. If one takes into account that the den-
sity of the polymers is cise to unity, one can imagine how great
their valu for technology may be. *
There is a rapidly growing demand on the quality of the materials
for modera Science and technology. Already now there is a need for
materials able to withstand temperatures of several thousand degrees
with the necessary strength characteristics at such temperatures
and without any noticeable plstic deformation at normal loaas.
What are the prospects for such materials?
One of the-easible methods for producing such extra strong and
extra heat-resistant materials was proposed by the Soviet physicist
A. V. Stepanov who pointed to a particular property of such molecu
lar crystals as sulfur. The crystal of sulfur is constructed of mole-
cules bonded by relativeiy weak molecular forces. Because of that
the strength of the crystal and its melting point are low (115 C).
The atoms in the sulfur moiecule itself, on the other hand, are held
together by powerful Chemical nonas. If one would be able to con-
struct a sulfur lattice with the atoms retaining the same bonds th at
act in the moiecule, the result would be an extremely strong crystal
with the melting point of about 34 700 C. Similar modications
could be introduced into other molecular crystals as well. Are there
any real grounds for such projects? The fac that we were able to-
transform soft graphite and hexagonal horon nitride into extra strong,
hard, and high melting point diamcnd and borazon crystals by sub
stitu ing powerful covalen t bonds for weak van der Waals forces-
lenas ground to such hopes. The prospects that will be opened by
such materials are so enormous that any work, no m atter how great,.
put into their production shall be generously rewarded.

5 Since the atomic structure of amorphous boaies is irregular, the tarmi

deect may apply oniy to inolusions (clusters of ioreigar atoms, cracks, inhomo-
geaeities) larga i compared with atomic dimensions.
 3
Eiements of Physical Statistics

Every solid is a systera, or an ensamble, consisting of an enormous

number of microscopio partiles. Sucia systems obey specific siatistical
laws, which are the subject of statistical pbysics, or physical statis
tics.
The present chapter deais briey with the principal elementa of
physical statistics needed to describe the prcperties of solida.

23 Mefhods used to describe the State

of a macroscopic system
There are two methods of describing the State of a system consisting
of a great number of microscopio partiles, the thermodynamic and the
statistical method. Let us discuss them.
Thermodynamic description of a system. In the thermodynamic
approach to the description of the State of a system consisting of an
enormous number of partiles the later is regarded as a macroscopic
system, it being of no interes o what type of partiles it consists.
Such a system is termed a thermodynamic system.
A thermodynamic system may be either closed or open. A olosed
systeni-dbes'mot interact in any wav with.4he surroundings, and an
onen system can enchange heat and/or work with the surround-
ings.
The State of a system in which it can remain inniteiy long is
termed the equ.ilibriu.rn State. It is uniquely determinad by a set of
independent physical parameters, the siais parameters. The principal
State parameters are the volunte oi the system V, the p ressurs p, and
the temperature T . However, often those parameters are inadequate
for a complete oharacteristic of the system. For a system maae up of
several substances one has also to know their concentrations; for
a system in an electric or a magnesio eld the intensities of these
fields shouid be specied; etc.
84
lamenn of Physical Safisics 35

Any changa in a thermodynamic system involving the variation

of at leasi one State parameter is termed a tkermodynamic process.
The sum of all types of energy of a closed system is termed the in
ternal energy (E) o the system. It is made up of the kinetic energy
of the partiles constituting the system, of the potential energy of tne
interaction between the partiles, and of the intemal energy of the
partiles themseives (which shall not be considered here since it is
not subject to change in usual processes).
The intemal energy is a function of State of the system. This means
that there is one and only one definite valu of intemal energy tnat
corresponds to each State no matter how the system arrived at this
State.
Interacting with the surroundings a thermodynamic system may
receive or reject some amounts AQ of heat, may perform work A.A W
or have work performed on it. In all cases the variation in internal
energy of the system, dE, should be equal to the diSerence in the
amount of heat received from outside, AQ, and the work AA per
formed by the system against ekternal forces: ^
dE = LQ &A (3.1)
This is the first law of thermodynamics.
It should be pointed out that in contrast to the intemal energy the
work A.4 and the amount of heat AQ depend not only on the initial
and the nal States of the system but on the wav the State is changed
as well. Since
AA = p d V (^2)
where dV is the variation of the volume of the system the pressure in
which is p, we may write (3.1) in the forra
dE = AQ p dV (3.3)
The second law of thermodynamics maintains that the amount o
heat A(> received by the system in a reversible process results in the
increase of me entropy of the system by
dS = -^- (3.4)
where T is the temperature at which the heat is received. Substitut-
ing AQ from (3.4) into (3.3), we obtain
dE T dS p dV (3.5)
It ollows from (3.5) that the systems intemal energy can be changed
at the expense of work performed or heat exchanged.
However, the systems energy may also change with the change in
the number of particles it contains, for every partile leaving the
system takes away a denite amount of energy with it. Therefore the
86 Sod State Physics

general expression for the iaw of conservaticn of energy (3.5) should

be written in the form
dE = T dS p dV + p dN ' (3.6]
where dN is the variatio of the number of partiles in the system.
Parameter p, is termed the cln.em.ical potential of the system. Its phys-
ical meaning is as ollows. For an isolated systemfof constant volnme
which neither receives or gives away heat, dS AQ/T = 0 and
dV = 0. For such a system
dE = p dN (3.7)
Whence
(3.8)
Henee the Chemical potential expresses the variatio of the energy
of an isolated system of a constant volume brought about by a unit
variatio in the number of partiles it contains.
Let us- consider the conditions of equilihrium of a system whese
total number of partiles remains constant but the partiles can go
over from one body belonging to the system to another. Two electrn
conductors, for instance, two metis, in contact with each other at a
constant temperature may serve as an example of such a system. De
note the Chemical potential of the electrn gas in the hrst metal by
u.t and in the second by lu. Suppose dN electrons fiow from one metal
to another. According to (3.7) this will reduce the energy of the rst
metal by dEl = \ixdN and increase the energy of the second by dE =
u,od^V. For the metis to be in a State of equilihrium the necessary
conaition is
dE1 dE, or \ixN \.dN
Henee the condition of equilihrium is
Pl = U; (3.9)
This condition is va.lid nct only in the case of two electrn conductora
in contact with each other but for any phases in contact with each
other: the solid and the liquid, the liquid and the gaseous, etc. In all
cases the condition of equilibra am is the equality of the Chemical
p o te n tia ls.^
Statistical method of describing a system. To describe the State
of every partile one should specify its t hree coordina tes and three
components of the momentum, Apparently, if one was to write the
equations of motions of the partiles and solve thern, he would be
able to obtain complete Information on the behaviour of the system
and to predict its State at any moment of time. Such calculations,
however, are not only extremelv tedious but, in fac, useless. The
complexity of. the probiem stems from the fact that to describe the
behaviour of the gas molculas normally eontained in 1 m3 one would
 H '!
!emsns o Physicai Sfafisiics 87

have to solve about 102S interconnectea equations o motion and also

take into account the initial conditions, whicK is practically impos-
sible. Should such calculations be carried out, they would be o no
valu since tbe properties o a system in the State of equilibrium not
only are independent of tbe initial vales o tbe coordinates and of
tbe momentum components but generally remain constant in time,
although tbe coordinates and tbe momenta of tbe partiles do cbange.
It follows from bere tbat tbere is a qualitative distinctioTbetween
tbe system and tbe individual partiles and tbat tbe bebaviour of
tbe former is governed by laws difieren! from those tbat govern tbe
bebaviour of individual partiles. Tbese laws are tbe statistical laws.
Tbe following examples are proof of tbeir existence.
Tbe velocity of an individual gas molecule is a random quantity,
wbicb is impossible to predict. Despite ibis fact, in a gas with a
very large number of partiles, on tbe average a distinct velocity
distribution p its molecules may be obsexved. In otber words, on tbe
average a quite denite fraction of tbe molecules bas a speed of, say,
from 100 to 200 m/s, from 400 to 500 m/s, etc.
^ -It is a matter of chance whetber or not a givon molecule shall enter
a specied volume of tbe gas. Despite tbis fact there is a denite reg-
ularity in tbe distribution of tbe molecules ovar tbe volume: equal
elementa of volume contain, on tbe average, equal numbers of mole
cules. '
TSe~situation here is similar to tbat wben a coin is tossed. The
landing of the coin heads or tails up is a random event. Nevertheless.
wben tbe number of times tbe coin is tossed is very great, a quite
denite regularity may be observed: on tbe average, tbe coin lands
heads up balf tbe.number or times..
Such TeguIaritis are tarmed statistical. Tbe principal feature of
statistical laws is tbat they deal with vrobabilities.. They enable pre-
dictions to be made only as t-o tbe probability of some event occur-
ring orsomeresuit being realized. In tbe exampie witb the coin the
predicted probability of the coin landing one or tbe otber side up is
1/2. The resulta of individual tests may, and undoubtedly shaii, de-
viate from those vales tbe more tbe less the number of tests is. If
we toss a coin ve times, tbe nead may fall out any number of times
from 0 to 5. But. tbe greater tbe number of tosses, tbat is, tbe more
numerous tbe ensemble, tbe more accurate tbe statistical predictions
are. Calculations show the relative deviacin of an observed physicai
quantity (for instance, of tbe number of partiles per unit volume)
from tbe average valu M in a system of N noninteracting partiles
to be

or inversely proportionai to Y N ,
33 Sod Sfrata Physicj

As N is increased, the ratio AM /M -> 0. For very great N we have

MIM 1. Thus m3 o air nonnally contains on the average
2.7 x 1028 molecules,. The relative deviation from this number is
on the average equal to
i p t ' ^ 2 x l O - ilo/0
yn
This deviation is so negligibie that there are no Instruments capable
of detecting it. Thereore when dealing with large volumes it *is al-
ways reasonahle to assume that the distribution of molecules over
the volume is uniform.
It should, however, be pointed out that deviations from the av
erage vales are not a possibiliy but a necessity. Snch deviations are
termeb fluctuations.

24 Degenerae and nondegnerafe

ensembles
Microscopio partiles and the ensemble. A1I microscopio partiles
making up an ensemble may be subdivided into two classes accord-
ing to their behaviour: fermions and bosons.
Fermions inelude electrons, protons, neutrons and -other partidas
with a half-odd integral vales of spin: hl2, 3h/2, .... Bosons inelude
photons7~phonons and other partiles with integral vales of soin:
0, ti, 2, ... .
The fermions in an ensemble exhibit marked individualistic
tendencies. If some quantum state is already occupied by a fermion,-
no other fermion shall settle in it. This is the essence of the well-
known Pauli exclusin principie, which goveras the behaviour of
fermions. Bosons, on the other hand, strive for unieation. They
can settle in the same state in any numbers and do it the more readily
the m ore-pop ul a te d the "state already is. ^
Degenera t and' nondegenerate ensembles. Let us discuss the pos-
sibie eneets of the nature of the partiles (their fermionic or bosonic
character) on the properties of the ensemble as a whole.
For the nature to be felt the partiles must meet oten enough.
This means that they must occupy the same state or at least su-
cientiy closely-spaced States.
Suppose that there are G diSerent States which any one of N simi
lar partiles can occupy. The ratio N/G may serve as a measure of
the meeting frequeney. The meetings wl be rare if
- N/G < 1 (3.10J
In this case the number o diferen vacant States is much larger than
the number of partiles: G > N. Evidently, in such circumstances
 Elemsnts of PhysicaJ Sransiics 83

th8 specic nature of fermions and bosons shall not be felt, since every
partile has at its disposal a large number of diferent free States and
tile problem of seyeral partiles occupying the same State actualiy
does not arise. Therefore the properties of the ensembie as a whole
shall not depend on the natnre of the partiles that make it up. Such
ensembles are termed nondegenerata, and- condition (3.10) is the
condition of nondegeneracy.- : -
If, however, the number of States G is of ths same order of magni-
tude as the number of partiles N, that is if
N /G 1 - ----- - (3.11)
then the problem of how "the States should be occupied, whether
individually or collectively, indeed assumes much imprtanos. In
this case the natura of the microscopio particle. is fully revealed in
its eect on the properties of the ensemble as a whole. Such ensembies
are termed deganarata.
Degenerat ensembles are a unique property of quantum objects,
since the parameters of State of such objects only change discretely
with the result that the number of possible States G can be finite.
The number of States for classical objects whose parameters change
continuously is aiways infinite and they can forra only nondegenerata
ensembles.
It should be pointed out that quantum mechanical objects too
may form nondegenerate ensembles provided condition (3.10) is
fullled (see Table 3.1).
Table 3.1
Enseables
Object
degenera te nondegenerate

Classical No Yes
Quantum..., Yes Yes

. Q-Classical and quantum statistics. Physical statistic3 that studies

nondegenerate ensembles is termed classical statistics. It owes much
to J. C. Maxwell and L. E. Boltzmann (the Maxwel-Boltzmann sia~
tistics).
Physical statistics that studies degenerate ensembles is termed
quantum statistics. Owing to the eect of the partiles natura on the
properties of a degenerate ensembie, degenerate ensambles of fer
mions and bosons behave in essentially dierent ways. On this ground
a distinction is made between two quantum statistics.
Quantum statistics of fermions owes much to E. Fermi and P.A.M.
Dirac (this, by the way, expiains the origin o the term fermion').
It is termed the Farmi-Dirac statistics.
90 Soid Sfafa Physics

Quantum statistics of bosons owes much to S. N. Bos8 and A- Ein-

stein (henee the tena boson). It is termed the Bose-Einstein .sta
tistics.
It follows then that quantum statistics deais only with quantum .
objeets while classical statistics may deal both with the classical
and the quantum objeets. If we reduce the number o particles in an
ensemble or increase the number of States, we shall eventuaiiy turn
a degeneris ensemble into a nondegenerate one. In that case the
ensemble shall be described by the Maxwell-Boltzmann statistics
no matter whether it contains fermions or bosons.
Distribution function. What is the connection between the distri-
bution o the particles over particular States and the State o the en
semble as a whole? To specify the State o an ensemble, or instance,
of a gas of particles. one should specify its state parameters. To spec
ify the State of each partile one should specify its coordnales and
momentum components or its energy. which is a function of coordi-
nates and momentum.
'The two types of quantities are connected by the statistical distri
bution function
N\i ,t{E) dE (3.12)
which species the number of particles having an energy rom E to
E 4- dE in the system described by the State parameters u and T.
This function is termed complete statistical distribution function. To
simplify notation the indices denoting the state parameters are usual-
ly omitted.
The complete distribution function may be rspresented by the prod-
uct of the number of States g (E) dE per energy interval dE and the
probability of occupation of those States by the particles. Let us de
note the latter by / (E). Then
N{E) dE Jf(E))g(E). dE (3.13)
The function f(E) is termed simpiy the distribution function. As
was stated before it signifies the probability of the occupation of the
respective States by the particles. If, for instance, 10 of 100 closely
spaccd States are occupied by particles (the total number of parti-
cies in the system being much greater than 100), the probability of
occupation of such States will be equal to 0.1. Since on the average
there is 0.1 of a partile per each state, we may take f(E) to be the
average number of particles in a given state.
Henee the problera or fiading the complete distribution function
of particles over the States is reduced to that of finding the function
g(E) dE, which describes the energy distribution of the States, and
the function f(E) which determines the probability of their occupa
tion.
We start by aetermining the function g(E) dE.
Elements of Physical Sfaisics 91
>EA 7a
25 The numher of sfafes for microscopio
parficies
Concept of phase space of a microscopio partile and quantizatiorr.
In classical mechanics the State of a particle is determined ir its
three coordinates (x , y, z) and three components of its momentum
(ps, pv, p z) are specified. Let us imagine a six-dimensional space
with the coordinate axes x, y, z, p x, py, p z. The State of the particle
at every moment of time will be described by a point (x, y , z, ps,
py, p). Such space is termed phase space and the points {x, y , z, p s,
py, pz) are termed phase points. The quantity
AT = ATyATj, = dx dy dz px dpy dps (3.14)
is termed an element of the phase space. Here ATy = dx dy dz is
an element oi volume in coordinate space and ATP = dpjdpydps an
element oi volume in momentum space.
Since the coordinates and the momentum components of a classical
particle may change continuously, the elements AIV, ATp and with
them the element AT as well can be chosen as small as desired.
The potential energy of a system of noninteracting partidas not
acted upon by an external field is zero. Such partiles are termed
free. For such partiles it is convenient to use a three-dimensional
momentum space instead of the six-dimensional phase space. In
this case the element ATV is simply equal to the volume V in which
the partiles move, because no additionai restrictions are placed on
them.
The divisin of the phase space into elements of volume is not quite
so simple i the particle in question is an electrn or some other mi
croscopio object possessing wave properties. The wave properties of
such partiles make it impossible, in accordance with the uncertainty
principie, to aistinguish between two States, (x, y, z, p x, p v, p z) and
[x + dx, y -f- dy, z -i- dz, px + dpx, py p rdpy, p z -f dpt), it the
product dx dy dz dpx dpv dp is less than Since this product rap-
resents an element of volume in six-dimensional phase space, it
follows from the above that dierent quantum States shall corre-
spond to aiSerent elements of volume in this space only if the size of
those elements is no less than h \ Therefore quantum statistics makes
use of an elementarv cell in -the six-dimensional space with the
volume
A r = A r r Arp= /i3 (3.15)
TKe element of the three-dimensional momentum space for free par
tidas for which ATy = V is
ATp = h W (3. i 5)
92 Soiid State Physics

Each element of this kind has corresponding to it a denite quantum

stata. _
The process of dividing the phase space into ceils of finite size
(,h3 or h3{V) is termea quantization of phase space.
Density of States. We wish to calclate the number of States of
a free partile in the energy interval from E to E -j- dE. To this end
draw two sberes of the radii p and p 4- dp in the momentum space-

Figure 3.1 Calcuiating the number oi States oi a microscopio partile.

(Figure 3.1). There is a spherical layer with the volume of 4npzdp

contained bet-ween the sberes. The number of phase ceils contained
in this layer is
4np2dp
(3.17)

Since-tkef-is^one partile State to ccrrespond to every cali-the

number of States in the interval dp between p and p -f- dp is

g (p) dp = EEL. pzdp\ (3.18)

For free (noninteracting) p anieles

D3
2771*
Using these relations to expresa p- and dp and substituting the results
into (3.18), we obtain
nF 3/2
g (E) dE h' (2m VEdE (3.19)
Eiements of Physicai Stafistics " 93

This is the number of .States of a free partile in the energy interval

{E, E -f dE). Dividing the right- and the left-hand sides of (3.19)
by dE, \ve obtain the density of States, g(E), which species the num-
ber of States of a microscopio partile per unit energy interval:

g(E) = l f - ( 2 m f Z V (3-20)
?
It follows from (3.20) that as E increases the density of States rises
in proportion lo Y E (Figure 3.2). The density of States depenas,
besides, on the partiles mass and increases with m.

Figure 3.2 Energy dependence of density of States.

In case of the electrons each phase cell corresponds, to be e.xact,

not to one but to two States, each distinguished by its spin. They are
termed spin. states. Therefore in case of the electrons the number of
States (3.18) and (3.19) and the density (3.20) should be doubled:

g(p) dp = ^ r 1pzdp (3.21)

g(E) dE = (2m)3/3 V EdE (3.22)

(E) = L (2 my' 3/2 Y e (3.23)

Condition of nondegeneracy for an ideal gas. Integrating (3.20)

with respect to energy from 0, to E, we obtain the number of partile
States contained within the energy interval (0, E):
(2.71)3/2 | 3/2

Setting E = (3/2) i sJ , we obtain

94 Sod State Physics

Substituting this expression into (3.10), we obtain the condition for

nondegeneracy:

where n = N /V is the number of partiles per unit volume.

Consider some molecular gas, for instance, nitrogen in normal con-
ditions. For it n 1026 m -3, m = 4.5 X 1028 kg, and kBT = 4 X
X 10-21 J. Substituting the figures into the left-hand side of (3.24),
we obtain nh3(2nmk-aT)~3/2 ~ 10~s, which is much less than unity.
Accordingly, the molecular gass'are normally nondegenerate and
must be described with the aid of the Maxwell-Boltzmann classical
statistics.
Consider now the electrn gas in metis. For it n 5 X 102S m~3,
m 9 X 10~31 kg. For such vales of n and m the electrn gas turns
out to be nondegenerate only at temperaturas above 105 K; the left-
hand side of (3.24) for such temperaturas diminishes to less than unity
(at T = 105 K it is approximately 0.5). Therefore in practice the
electrn gas in metis is ahvays degenerate and on account of this
should be described with the aid of the Fermi-Dirac statistics.
It ollows from (3.24) that a nondegenerate State of a gas can be
realizad not only.by raising its temperature but by reducing its con
centraron n as well. For rt s; 1022 m-3 the left-hand side of (3.24)
for electrons at normal temperaturas is approximately iO-3 and the
electrn gas becomes nondegenerate. Such (and smaller) concentra-
tions of the electrn gas are found in some 'semiconductors. In such
semiconductora termed nondegenerate, the electrn gas is nondegen
erate and is described by the classical Maxwell-Boltzmann statistics.
Let us trv now to find the distribntion function f(E). The orm of
this function depends in the first instance on whether the gas is de
generate or nondegenerate. In the case of a degenerate gas the impor-
tant point is whether the gas'consists of fermions or bosons.
Let us start with a nondegenerate gas whose distribution function
}{B) is independent o the partiles nature.

26 Distribution function
for a nondegenerate gas /
Appendix I contains an elementary derivaticu of the distribution
function for a nondegenerate gas. It is of the following form:
/ (E) = e(n- E)/h-T (3.25)
(3.25)
Elements of Physical Stattscs 95
/EF &rr / U S l
gas:
TN / A*
(3.25)

Substituting it into (3 25) we obtain:

(3.27)

We would like to remind again that f->&[E) dE expresses the proba-

bility of occupation of the statas in the energy interval (E , E -f dE);
the terna for it is the Maxwell-Boltzmann distribution function.

Figure 3.3 Distribution functions

for nondegsnerate gas:
(a)the Maxwell-Boltzmann
distribution function expressing the
average density of State occupation
by partiles; (b)the complete
Maxwell-Boltzmann distribution
function.

Figure 3.3(a) shows a graph of the function f^(E ). It has a mxi

mum at E = 0 and asymptoiically approaches zero as E-r- co. This
means that the lower energy States nave the greatest probability of
occupation. As the energy of a State increases its probability of oc
cupation diminishes steadily.
Multiplying / M() by the number of States g(E) dEflsee (3.22)],
we obtain the complete distribution function of partidas over the
energy
N (E) Y V 4S (3.2S)
or-
2N
N (E) di r m bx y e dE (3.23')
V (ABA)3
96 Solid State Physics

It is tenned the complete Maxwell-Boltzmann distribution function.

Figure 3.3(b) shows the graph of this function. Because of the factor
]/ E its mximum is displaced to the right of the origin.
Knowing the distribution function we may easily- find
the laws of distribution of the partiles over the momentum,
N(p) dp, and over its components, N{ps1 p v, p z) dp JLpydpz, over
velocity, N(v) dv, and over its components N(vs, vg, uz) dv^dvydv^
over one of the components of velocity, say N(vx) dus, etc. Those
distributions are shown below

N (p )= M (3.29)

(3.30)
N g-(P|+p2+P=)/(27nA3D (3.31)
Pz) =
{2nmkBT)3/2
j3/2^_m(r +Pjx 0|)/(2ft3r) (3.32)
iV(y_, vv, vz) - A (2nj!3r
1/2 g-m|/(2ABD (3.33)

The reader is requested to obtain those results himself.

27 Distribution function
for a degenerate fermion gas
The distribution function for a degenerate fermion gas was first ob-
tained by Fermi and Dirac. It is of the following form:
f?(E) = (es -'/k*T+ i r ' (3.34)
An elementary derivation of this expression is presentad in Appen-
dix II. Herejas bjefore. u denotes the Chemical .potential, which in
the case of a degenerate fermion gas is tenned the Fermi level.
Formula (3.34) shows that for E = x the distribution function
f?{E) = 1/2 at any temperatuxe T # 0. Therefore from the statis-
tical point of view the Fermi level is a State whosa probability of
occupation is 1/2.
The function (3.34) is tenned the Fermi-Dirac function. To obtain
a clear picture of the natura of this function one should consider the
degenerate electrn gas in metis at absoluta zero.
Electron distribution in a metal at ahsolute zero. Fermi energy.
The metal is a sort o a potential trough for free electrn. To leave
it the electrn should have work perormed on it against the forces
retaining it in the metal. Figure 3.4 shows the aiagram of such a
Elemenis o Physical Sfafisiics S7

potential trough. The horizontal lines denote energy leveis which the
electrons may occupy.In compliance with the Pauli exclusin prin
cipie there may be two electrons with opposite spins on each such
level. For an electrn gas of N electrons the last occupied level will
be the Ni2 level. This level is termed the Fermi level for a degenerate

Figure 3.4 Schematic rspresentation

of a metal as a potential trouga
for ree electrons.

electrn gas. It corresponds to the mximum kinetic energy E? an

electrn in a metal may possess at absolute zero. This energy is
termed the Fermi energy.
Thus at absolute zero ali States with the energy E < E? are oc
cupied by electrons and all States with the energy E > E? are
fres. In other words, the probability of occupation of a State with
the energy E < E ? at T 0 K is unity and the probability of occu
pation of a State with the energy E > E? is zero:
/ p {E) i for E < ZF
= 0 for E > E? (3.35)
To obtain this result from (3.34) one should assume that at
T = 0 K .the Chemical potential of the electrn gas measured from
the bottom of the potential trough is equai to the Fermi energy E?:
Ltq = E? (3.36)
Indeed, setting in (3.34) 'iE-w, we obtain
f ? (E) = (elK~E^)lk (3.37)
If E << E?, then g(-2p)4*3r ->-0 at T = 0 K and /? =* 1; i
E > E?, then e(-E~E?vk3T co at T 0 K and /p = 0.
Figure 3.5(a) shows the graph o the Fermi-Dirac distribution func-
tion at absolute zero. It has the shane of a step terminating at
E = E?.
T 0835
S8 Soiid State Physics

Multipiying (3.35) by the number of States g(E) dE [see (3.22)],

we obtain the complete Fermi-D irac distribution junctio at absoluta,
zero: ~ .
N (E) dE = (2m)s/2y~ dE (3.38)
because/F = in the energy interval (0, E?). The graph o? the func-
tion is presented in Figure 3.5(b), where the area of the occupied States
is shaded.

Figure 3.5 The distribution funciion for degenerate fermioa gas:

(a) the Fermi-Dirac distribution functiou for T 0 K,
(b) the complete Fermi-Dirac distribution unction for T = 0 K.

States

Integrating expression (3.38) frona 0 to Ec, we obtain

.V = n2 EV- ' '
whence the Fermi energy E? mny be easilv obtained
r
ih2 / 3/i \ 2/3
i (Ib (3.39)
2.7i \ 8?
in i
where n = N1V is the concentration of electrn gas in the metal.
Knowing the energy distribution function o the electrons, we may
easily calclate the average energy of the electrons at absoiute zero,

-= 3 rp 3k2 f 3/1 V 2/3

= T f = = lte ) (3-40>

Lastly, knowing E ? and E 0, we can calclate the mximum veiocity

v-p and-the efective veiocity vstt (corresponding to average energy)
of free electrons in a metal at absoiute zero:
 S9
Esmanfs o Physicai Sfatistics

Table 3.2 shows the Fenni energy EF, tho average energy E0t
the mximum and eective velocities o ree elecrons^ at absoiute
zero for some metis. The last column contains the Fermi temperatura
determined from the relation
T? = EFk3 (3-42)
This is the temperature at which a molecule in a normal nondogen-
erate gas would have the energy of thermal motion (3/2) k3T equa.
to the Fermi energy EF multiplied by 3/2.

Tabie 2.2

E? l'e S T p

Metal (eV) (10 m/s) (!06 m/s) (104 K)

(eV)
Copper 7. 1 4.3 1.6 1.25 - * ?

Lithiuxn 4.72 1 2.3 1.3 1 O d

Sil ver 5.5 3.3 1.4 1.1 .4

Sodium 3.12 1.9 1.1 0.35 3. i

It may be seen from Table 3.2 that Fermi temperatures are so high
that no metal can exist in a condensad State.
It should be stressed that although the Fenni energy representa
the kinetic energy of translational motion of free electrons it is not
the energy of their thermal motion. Its nature is pureiv quantum nie-
chanical and is due to the fact that electrons are fermions satisying
the Pauli exclusin principie.
Temperature dependence of the Fermi-Dirac distribution. When
the temperature is raised, the electrons becorne thermaily excited and
go over to higher energy levels. This causes a chango in their distri
bution over the States. However, in the rango of temperaturas in.
which the energy of their thermal motion, k^T, remains much less
thn EF only the electrons in a narrow band about approxiraately
k s T wide adjoining the Fermi level may be thermaily excited (Fig
ure 3.6(a); the excited States are shaded). The electrons of the lower
levels remain practical'ly unaffected because the energy of thermal
excitation k3T is not enough to excite thcm (to transfer them to
levels above the Fermi level).
As a resuit of thermal excitation some of tho electrons with an ener
gy less than E? are transferred to the levels with energies greater
than EF and a new distribution of electrons over the States is esfcab-
lished. Figure 3.6(b) shows the curves of the electrn distribution over
the states for T 0 K (curve 1) and for f > 0 K (curve 2). It can
be seen that the rise in temperature causes the original distribution
to smear to a depth of k3T with. a tail BC appearing to the right
of EF. The higher'the temperature the greater tho change in the dis-
100 Sod Sate Physics

tribution function. The tail BC itseif is described by the Maxwellian

distribution function.
The shaded areas in Figure 3.6(b) are proportional to the number of
electrons transferred from the States with E < E ? (the area ADB)
to the States with E > E? (the area E?BC). Those areas are equal
since they represent the same numher of electrons.
Let us make an approximate estimate of this number, AN . There
are NI2 energy levels inside the interyal (0. E?), where N is the num
ber of free electrons in the metal. To simplify the problem we may

Figure 3.6 Temperature dependeace of the Fermi-Dirac distribution

function: (a)ihennal ezcitation of electrons; (b)the Fermi-Dirac
distribution function for T > 0 K.

-------------------------- N

assume that these levels are equidistant, the separation being As =

= E ?/(:V/2) = 2EvY. Only the electrons in the band k3T wide
just belotv EF (Figure 3.6(a)) are thermally excited. There are
k 3TI As = k3TNl(2E?) levels inside this band occupied by
2k3TN!(2E?) = k3T/E? electrons. Assuming that not more than
a half of those electrons go over the Fermi leve!, we obtain the fol-
lowing approximate relation.lqr^AiV;
AjY M . j V (3.43)

At room temperature k3T m 0.025 eV, E? = 3-10 eV, thereore

AiV/'iV <; i percent; at T 1000 K. we nd that AiY/iY 1-2 per-
cent.
Henee in all the temperature range in which the electrn gas in a
metal is degeneris its distribution is cise to that at absolute zero.
Accordingy, only a negligible fracton of the electrons cise to the
Fermi leve! are thermally excited. At room temperature this fraction
is less than 1 percent of the total number of conuction electrons.
The laws governing the distribution of the electrons in metis dis-
cussed above remain valid practically in all cases, because in aii the
;|i riijrnFiflrrrognu Elemenfs of Physical Sfatisiics 101

temperature range in which the existence of metis in the con

densad State is possible the electrn gas remains degeasrate.
Coasider the temperature depeadence of the Chemical poten-
tial. lategrating the complete Fermi-Dirac distribution function
f?{E) g{E) dE over energy, we obtaia the total number N of free
electrons ia a metal:
co oo
N = f /> (E) g(E) d E = j f( 2m)3/2 j E ^ d E {elE~'l)/k*T+ l)~l
o o ----
Geaerally, there is ao analytic expression for this iategral. Approx-
imate calculatioa ia the temperature raage ia which the electroa
gas remains strongly degenerat-e yields the following temperature
dependence of u:

As k3T remains much less than E? up to the melting point of a

metal, the decrease in u with the rise in T turas out to be so smail that
it can often be neglected and the Fermi level can be assumed to coin
cide with E? at any temperature.
One can also calclate the average energy of the [electrons E
in a degenerate electrn gas dividing its total energy, E, =
= Ej?(E) g (E)dE, by the number of the electrons, N:
j? Et _ f 2 3,zd / * Eu-dE \
N ~ J [ ) (-n>ABru-iJ
o 1 o
Aa approximate calculatioa of these integris yields

At T = 0 (3.45) turas into :(3.40).

Liiting of degeneracy. Nndegenerate electroa gas. When the
condition for nondegeneracy (3.10) is fullled, every gas inciuding
the electrn gas must become nndegenerate. Let us discuss this in
more detail.
According to (3.10) the gas is nndegenerate if the average occu-
pancy of the States by the partiles is much less than unity. Since
the distribution function f(E) representa the occupancy of the States,
the condition of nondegeneracy (3.10) may be written in the orm
(E) < 1 (3.46)
The Fermi-Dirac function (3.34) wiii be much less than unity if
the term (E-pO/fea7 will be much greater than unity:
e(S-)/kBT ^ 1 4,
102 Solid State Physics

This inequality should hold for all the States including that with
E = 0:
6- ^ bt ^ i (3.47)
It follows from (3.47) that for a nondegenerate electrn gas in which
condition (3.46) is satisfied u should be a positive quantity con-
siderably greater than kBT:
- u . > kBT (3.48)
The Chemical potential p, should be negative and greater than k BT
in absoiute valu.
If the condition (3.46') is fullled, unity in the denominator of
the Fermi-Dirac function can be neglected and the following expres-

Figure 3.7 Temperature dependence of energv, pressure, and the square

of eective velocity o eiectrons in a metal.

sion for the distribution function of a nondegenerate electrn gas na ay

be obtained:
f{E) = e'L,kQre~E/kBT _ (3.49)
Comparing (3.49) with (3.25), we see that a nondegenerate electrn
gas like every other nondegenerate gas is described by the Maxwell-
Boltzmann distribution function.
The electrn gas in metis, where the fres electrn concentration
is always very high (wlO23 m-3), is always in a degenerate state de
scribed by the Fermi-Dirac distribution function.
A nondegenerate electrn gas is a feaure of the intrinsic [pare) and
ioeakly doped semiconduciors, which are the mainstay of modera
semiconductor electronics. The concentration of free eiectrons in
such semiconductora is substantially ess than in metis, varying from
Ui6-10i,J to 023-i021 m -3 aepending upon the concentration of elec-
tricallv active impurities. The nondegeneracy condition (3.10) re-
mains valid for such concentrations and the electrn gas is nonae-
generate.
In conclusin, to (Ilstrate the drastic aifference in the behaviour
of the ideal, nondegenerate gas obeying the Maxwell-Boitzmann sta-
Eiemants of Physical SaHsfics 1Q3

tistics ana of the degeneris electrn gas described by. the Fermi-Di-
rac statistics, we would like to cite soma of their propertie^
(Tabla 3.3).
Tabie 3.3 '

Gas
P a r a m e te rs
n o n d e g e n e ra t9 d e g sn e ra te

n
0K 0 ' = 4 - ?
3

TK =

OK 0 m/s
'e 0
= :, / 3 i r
TK vets: y vt ~ "eff
m

OK 0 p3 = -=-5 ~ i010 ?a
O
P
RT
TK P= ? ~ p
V

It follows that average energy E, er active velocity vef and pres-

sure p of a nondegenerata ideal gas are uncUons of temperatura that
vanisn at absolute zero. At the same time E, vg, and p of a degen
eris electrn gas are very large alreadv at absolute zero and are prac-
ticaily independent o temperature (Figure 3.7}. This points to the
fact that, as noted above, E, and p of a degenerare electrn gas
are for the most part not of thermal origin, the contribution of the
thermal electrn motion to these quantities being negiigible.

23 DSsfribufion funciian
for a degeneris boson gas
In con trast to the electron s, which sa tisfy the Pauli ex clu si n prin
c ip ie , th e bosons can occupy both the free States and the States alreadv
occupied by other bosons the more readily the greater the occupancy
of th e latter.
The distribution function of bosons over tbe States was rst ob~
tained bv Bose and Einstein. It is of the following torm:

I Bose
(3.50)
104 Soiid State Physics

(An elementary derivacin of this expression is given in Appendix III).

It is termed the Bose-Einstein distribution function. Let us use it to
describe the State of a photon gas.
Suppose that a cavity inside a black body at a temperature T is
filled with equilibrium thennal radiation. From the quantum mechan-
ical point of view this radiation may be regarded as consisting of
an enonnous number of photons constituting a photon_gas. The
photons spin is s = 1. Therefore photons are bosons, which implies
that the photon gas should satisfy the Bose-Einstein distribution.
The photons have some peculiarities as compared with other bo
sons, for instance, the helium nucleus, ;>He.
(1) The rest mass of a photon is zero.
(2) All photons move with the same speed equal to that of light,
c, but can have diSerent energv E and momentum p\ E and p depend
on the photons frequency:
E = hv = ha, p = kv/c = hu/c 7 (3.51)
where K = h!2x and co = 2nv. It follows from (3.51) that
E = pe (3.52)
(3) The photons do not collide with one another. Therefore an
equilibrium distribution in a photon gas can be established only in
the presence of a body capable of absorbing and emitting photons.
The wails of the cavity in which the radiation is contaihed may serve
as an example of such a body. The transformation of a photon of one
frequency into a photon of another frequency takes place in the pro-
cesses of absorption and subsequent emission.
(4) The photons may be generated (in the act of emission) and an-
nihilated (in the act of absorption) in any numbers. Therefore the
number of photons in a photon gas does not remain fixed but depends
on the State of the gas. For specied vales of V and T the photon gas
in a State of equilibrium contains so many photons N 0 as are needed
for the energy of the gas to be a tits mnimum. This makes it possible
to express the condition for the-equilibrium of the photon gas in the
form:
( # ) v . r= <3'33>
Since according to (3.8) (dEldN)V<T u the equilibrium condition
(3.53) means that p =- 0. Henee the Chemical potential of an equi
librium photon gas is zero.
For a nondegenerate gas the Chemical potential is negative and
has a relatively great absoiute valu. The fact that for the photon gas
u = 0 means that such gas is always degenerte.
Setting u = 0 in (3.50), we obtain the distribution function for the
photon gas:
fp (E) = (eE/k3T- i)"1 = (ehu,kBT- i)-i (3.54)
 Elemenfs ai Physical Sfatistics 105

This formula was first obtained by Max Planck and is termed the
Planck formula. Itrepresents the average fraction o photons having
the energy E = hu>. Using this formula, we may easily formlate the
law for the energy distribution in the spectrum of a black body. The
following expression can be obtained for the energy rensity pu of
the radiation of such a body:
2 /0
p(i ~ jw i ^ ' 00'
The readers are requested to derive this formula themselves mak-
ing use of (3.54) and (3.18).

29 Rules for stafisfical averaging

As was already stated, to specify the State of an ensemble one should
specify its State parameters. To specify the State of a partile one
should specify the vales of its coordina tes and m ornen tum compo-
nents.
The problem of going over from the parameters of the individual
partiles to the State parameters characterizing the ensemble invoives
the problem of transition from the dynamical laws describing the
behaviour of the individual partiles to the statistical laws describ
ing the behaviour of the ensemble. To efect such a transition it is
necessary to perform the averaging of the characteristics of motioa
of the individual partiles assuming the chances of all the partiles
belonging to the ensemble to be identical. The state parameters of
the ensemble are expressed in terms of the avsraged parameters of
the individual partiles belonging to the enseable.
To make the rules of averaging apparent let us consider an ensemble
of N identical partiles each of which is capable of assuming one
of the discrete set. of energy- vales: Ev E, . . ., Em. Choose an
arbitrary moment of time and not the energy every partile has at
that moment. We obtain as a result a set of numbers N{EA express-
ing the number of partiles having the energy Et. To determine the
average energy E o the partiles we add up the energies of all of them
and divide the sum by the number of partiles. The total number of
partiles is N 2 f = tiV(J), and the sum of their energies is
i iV(t) E t. Therefore ~

2 X(
(3.5o)
? X (Ei)
 C6 Solid Sfafa Physics

The average valu E obtained in this fashion is terrned an ensemble

uverage.
If the particles energy assumes a continuous set f vales, the
practice is to count the number of partiles having an energy lying
within an in te r v a lo , E -f- dE) instead of the nmber of partiles
having an exact valu of energy. The average energy will then be
oo
\ EN(E) dE
= 4 r ----------- (3.57)
l N(E) dE
b
Such averaging may be performed for any phvsical quantitv M
that is a function of the coordinates and the momenta of the parti
dos making up the ensemble.-If M is continuous,

\ MN(M) d.l
T l = -------------- (3.58)
N(M)dM

Let us determine the average energy of the partiles of an ideal

nondegenerate gas. According to (3.57) and (3.28), ve have

l EX(E) dE
- JL -- ---------------
__________
CC
e E/k-nT 27'!'*> J P
B E""-aL (3,59)
V x W )*
\ N(E) dE
0
The resuits of the calculations of the average vales of the velocity,
of a velocity component, of the efective velocity, and of its compo-
nent for the partiles of an ideal gas are presenten thus:

\ ara /
_ / 3 ar d/ 2
^ - r r r j >
The reader may for the sake of practice perorm these calculations
himsel.
 T h e rm a i P r o p e r tie s o S o iid s

30 Normal modes of a latiice

The atoms of a solid take par in thermai vibrations around their
equilibrium positions. Because of a strong interaction between them,
the nature of those vibrations turas out to be extremely complex
and an accurate description of it presents enornious difhculties. There-
fore approximate methods and various simpiications are used to-
solve this problem.
Instead of describing the individual vibrations of the partiles
the practica is to consider their collective motion in the crystal,
which is a spatially ordered structure. This simpiication is based on
the fact that powerful bonds immediately transmit the vibrations of
one partile to other partiles and a collective motion in the form of
an elastic wave inyolying all the partiles of the crystal is excited
in it. Such collective motion is calied the normal moas of a latiice.
The number of normal modes coincides with the number of degress
of fresdom. which is 3N if N is the number of partiles constituting
the crystal.
Figure 4.1(a) represents a one-dimensional model of a solida
linear chain o atonas separated by a distance a and able to vbrate
in the direction perpendicular to the chain. Such a chain may be re-
garded as a string. If the ends of the chain are xed. the fundamental
mode corresponding to the lowest frequencv oort,tn is represented by
the standing wave with a node at'each end (Figure 4.1 (b ), curve 1).
The second mode is representen by the standing wave with an adrii-
tional node in the centre of the chain (curve 2). The tlrd mode, or
third harmonio, is represented by the standing wave. with two addi-
tional nodes that divide the chain in three equal pars (curve 3), etc.
The length of the shortest wave in such a chain is evidently equal te
twice the distance between the atoms of the chain (Figure 4.1(c)):
W = 2a . (4.11
The corresponding mximum fregueney fflmax is
(0mas. = 2~uilma = (4-2
where v is the velocity of wave propagation(of sonad) along the chain
108 Sod State Physics

This mximum frequency is a parameter of the chain s material

and is determined by the interatomic distance and the velocity of
wave propagation. Should we set a = 3.6 X 10-10 m (the la'ttice
parameter of copper) and v = 3550 m/s (the velocity of sound in
copper) we would obtain coma* ^ 3 X 1013 s-\ which corresponds to
the frequency of atomic vibrations in a solid.

i a Figura 4.1 Normal modas of a linear

chain made up of identical atoms:
(a)linear chain; (b)normal
modes of the chain; (c)normal
modes of the chain corresponding
to shortest wavelength (to highest
frequency); (d)dispersin curves
expressing dependence M normal
mode frequency on wave vector.

(c)

To describe wave processes one usuailv uses the wave vector q

whose direction coincides with that of wave propagation and whose
absolute valu is
q = 2tX (4.3)

It follows from (4.2) thafcrSitA = co/y. Thereores

- J3r>*: r ;"W
q = colv, or co = qv (4.4)

* The phase velocity , which enters (4.4), is itself a unction of the wave
vector q ana. or a linear chain of atoms bonded by eiastic forces is expressed
by the following relation:
sin {gal2)
v v, (4.4')
aa/2
where is the velocity of wave propagation in a continuous string or which
a 0. It follows from (4.4') that or a constant a the velocity u is practically
independent of q and is approximaely v0 only in the range of small q s, where
[sin (qai2)]/(qa:2) se 1. In* this range a increases approximately in proportion
to q (Figure 4.1(d)). As q increases the valu of [sin [qa2)]/(qai2) steadily dimin-
isbs and or q = .n/a tenas to 2in. This causes the dispersin curve u>(q)
to atten out, so that or q rJa- it mus parallel to q; -
Thermal Properies of Solids 109

Figure 4.1{d) shows the dependence on wave vector q of the re-

quency of normal modes in a linear chain mane up of atoms of one
kind. As q increases from 0 to na the requency of the normal modes
rises, reaching the mximum valu cDnas = nula for q na, that
is, for X = 2a. Curves of this type expressing the dependence of the
vbration frequency oa the wave vector (the waveiength) are termed
dispersin curves.

Figure 4.2 Normal modes of a chain mada up of atoms of two kinds:

(a)arraagament of atorr.3 in the chaia;. (b)acoustic modes;
(c)ptica! modes; (d)dispersin curves for acoustic
and optical modes; (e)linear lattice with a oasis in which ptica!
and acoustic modes are present.

L ------- 2 a
. 4
m -----------1
m M
---------- ------

12 12 12 12
id)

Consider now a chain mae up of atoms of two kinds placed in

a regular sequence one after another (Figure 4.2(a)). Denote the mass
of the heavier atoms by M and that of the lighter atoms bv m. Two
types of normal modes can be present in such. a chain, as is shown in
Figure 4.2(b, c). The modes 4.2(b) are quite identical to the modes of
a uniorm chain: the pilases of the neighbouring atoms are practicaliy
the same. Such vibrations are termed acoustic modes, since they in-
clude the entire spectrum of the acoustic modes of the chain. For
them coac = 0 for q = 0. They play a decisiva, part in determining
the thermal properties of the crystals such as heat capacity, heat
conductivity, thermal expansin, etc.
In case of normal modes shown in Figure 4.2(c) the phases of the
neighbouring vihrating atoms are oposita. Such modes can be reafc-
110 Sod Siate Physics

ed as the relative vifarations of two interpenetrating sublatices, each

made up of atoms of one kind. They are termed optical modes, since-
they play a decisive part in the processes of interaction of iightw ith
crystals.
Figure 4.2(d) shows the dispersin curves for the acoustic {!) and
optical (2) normal modes of a chain made up of atoms of two kinds.
In contrast tcran acoustic mode, whose frequency rises with the wave-
vector reaching the mximum valu at qmax = ni(2a), the mximum
frequency of an optical mode corresponds to q 0; with the increase-
in q the frequency of an optical mode falls off to its minimum at
5mas = n!(2a).
The optical vibrations are possible not only in a chain made up
of atoms of difierent kinds, but in a complex chain made up of two,
or more, interpenetrating chains containing atoms of one kind,
as shown in Figure 4.2(e). Unit cell of such a complex chain contains
two atoms. The opticalmodes are the result of the relative vibrations
of two sublattices.

31 Normal modes spectrum of a lattice

One of the principal problems of the theory of lattice vibrations is
the problem of the frequency distribution of normal modes. Consider
now the simplest case of the normal modes in a linear atomic chain
(see Figure 4.1).
The wavelengths of the normal modes in such a chain are
Xn = 2Lln (n = 1, 2, 3, . . ., N) (4.5)

where L is the length of the chain' and N thenumber of atoms in it.

The number of normal modes a having the wavelength [equal to
or greater than Xn will evidently be n:

In the same way the number of standing waves in a three-dimen-

sional crystal of volume V (for instance, in a cube with edge L and
voiume Z 3) having the wavelength equal to or greater than X should

(4.6)

(4.7)
Thermai Fropenies c Solids 111

Differenfciating this expression, we obtain

37
' dz g (co) do = o Zck (4.8)
'TZ'V1
Formula (4.8) expresses the number o normal modas per frequeiicy
interval (co, co ac). The function
37
2.-IV5 T4-9)

determines the density of the normal vibrations in do of the spectrum,

that is their spectrum. The function g(co) is termed speccral distri-
bution function of normal modos.
Since the number of normal vibrations in a lattice is 3N, the func
tion g(co) should satisfy the following normalization condition:

[ g (o ) d o = 3N (4.10)
o
where cq is the mximum frequency limiting the spectrum of nor
mal modes from above.
Substituting (4.9) into (4.10) and integrating, we obtain

3iV (4-T)
2n-t;3
Henee
N \i/3
coD= y ( 6rA y . j (4.12)

The frequency coD is termed the Debye frequency and'the tempera-

ture
0 = ^2 (4.13)

the Debye temperature. Table 4.1 shows the Debye emperatures of

some Chemical eiements and compounds.
At the Debye temperatura the entire vibration spectrum is e:c-
cited in the solid, including the one with the mximum frequency
(0 ]> Accordingly, any fuxther rise in temperatura (above ) shall
not be accompanied by the appearance o" new normal modes. In
this case the role of the temperature is to increase the intensity of
each of the normal modes with a resultan! increase in their average
energy.
Temperatures T P> "are usually referred to as higa temperatares.
Substituting iA from (4.TI) into (4.9), we obtain
 112 Sod State Physics

Table 4.1

Element e (K) Element e (K) Element (K)

Al 418 Fe ' 467 - Pb 94.5

Ag 225 Ge 366 Pd 275
Au 165 Hg (60-90) Pt 229
Be 1160 In 109 Si 658
Bi 117 KBr 174 Sn (gray) 212
C (diemond) 1910 KC1 227 Sn (white)' 189
Ca 219 La 132 Ta 231
CaF, 474 M? 406 Ti 278
Cd . 300 Mo 425 TI 89
Co 445 NaCl 320 V 273
Cr 402 N'b 252 W (379)
Cu 339 Xi 456 Zn 308

32 Phonons
Each normal mode carries with it some energy and momentum. -
Oscillation theory contains the proof o the act that the energy o
a normal mode is equal to the energv of an oscillator with a mass
equal to the mass o the vibrating atoms and the freguencv of the
normal mode. Such oscillators are termed normal.* 1
Denote the energv o the ith mode characterized by the requency
by E ia,m. It is equal to the energv Ein,0 of a normal oscillator of
the same requency co: n.tn = Ein.0- The total energv of the crys-
tal in which all the 3V normal modes nave been excited is
l
3-V
= 2 Ei n.o
1

Kence the total energy oi the crystal made up of N atoros taking

part in coupied vibrations is equal to the energy of 3N independent
normal harmonio linear oscillators. In this sense a system of N atoms
whose vibrations are interconnected is equivalent to a set of 3N nor
mal oscillators and the problem of calculating the- average energy of
such a system is reduced to a simpler problem of calculating the av
erage energy of normal oscillators.
It should be pointed out that normal oscillators have nothing in
common with real atoms except the mass. Every oscillator representa
one of the normal modes of the lattica in which ail the atoms of the
crystal take part vibrating with the same requency co.
The energy of a quantum oscillator is, as is weil known, expressed
by the relation
/2) pm \n o, C 2 ; v . .) (4.15)
Thermal Properfies of Solios 113

where a is the oscillator s vibration frequencv, and n the quannra

number.
Figure 4.3 shows the energy spectrum of a linear harmonio oscil-
lator. It consists of a set of discrete levels spaced at an interval of
rtC.
Since n.m = Em0r the expression for the energy of the normal
modes of a lattice should be (4.15) and the energy spectrum should
coincide with that shown in Figure 4.3.
The minimum portion of energy that can be absorbed or emitted
by the lattice in the process of thermal vibrations corresponds to

Figure 4.3 Energy spectrum of linear

harmonio oscillator.

the transition of the normal mode being excited from the given energy
level to the aajacent level and is equal to
eph = /co (4.16)
This portion, or quantum*, of energy of thermal vibrations of the
lattice is termed a phonon.
The following analogy may help to clear up the point. The space
inside a black body is filled with equilibrium thermal radiatioa.
From the quantum mechanical point of view such radiation is treat-
ed as a gas made up of the light quanta, or photos, whose energy
is s = hia = hv and whose momentum is p Kq /c = h/\, where c
is the velocity of light, and /\ its wavelength.
The eid of elastic waves in the crystal may be treated simarly
as a gas made up of quanta of the normal modes of the lattice, or of
phonons having the energy s? = a hv and momentum
pvh = heov = hX q (4.17)

where u is the velocity of sound, and X the length of the elastic wave.
3-0385
114 Solid State Physics

Froin this point of.view a iieated crystal may_be likened to a box

filied with phonon gas. Tbe analogv may be extended.
Phonons are described by tbe same Bose-Einstein distributioa
function (3.54) as photons:
i 1
7 ( E\= /ph/ABr _ ! = eh 'kBT- i

Depending on tbe intensity of excitation of the normal mode it can

emit a definite number of pbonons. Tbus, if some normal mode

Figure 4.4 (a)phonon energy-distribution function; (b)illustration

of the fact that normal modes excited in the iattice at temperatura T
are mainly those with'Jenergy'Jquanta a < k%T.

was excited to tbe.tbird Level (Figure 4.3), its energy became E 3 =

= (3 172) /too; this means tbat tbe particular normal mode has
generated" tbree identical pbonons eacb with an energy of Hca.
Figure 4.4(a) shows tbe grapb of the phonon energy (frequency)
distributioa function f(E). We see that or a given temperature T
all normal modes in a Iattice up to those with the energy ftco kBT
are excited; practically no quanta of higher frequencies with the
energy > kBT are excited. This is quite evident rom Figure
4.4(b). Horizontal strokes here denote energy spectra of normal modes
with the frequencies i = kBT{8h), 2 = kBTI(4n), 3 =
= kBT/(2), co4 = kBT!n, and 5 = 2kBT/h: the level correspond-
ing to kBT is shown by a dotted line. It follows that for a given tem
perature T the mode with the frequency x is excited approximately
to the 8th leve!. As was stated before, this means that this normal
moda generates eight identical phonons with the energy %<l =
Thermal Properfies of Salids 115

= kBTI8 each. The normal mode with the frequency , is excited

approximately to the 4th level, that with the frequency c3 to the
second, and that with the frequency co4 (whose quantum of energy is
A; = kBT) to the rst.. At the same time the vibration c3 is rarel}/
excited at T because its excitation energy /zc3 is too high. The exci-
.tation of still higher frequencies is a much more rare event. Therefcre
we can say that approximately only the vibrations with frequencies
not greater than co corresponding to the energy neo ks T are excit
ed in a solid at temperatures T < 0 .
By denition, the distribution function f(E) expresses the aver-
age number of phonons having the energy spil = 7ico. Therefore to
obtain the average energy of an excited normal mode, E n.m, of the
frequency co one-has to m ultiply (3.54) by neo:

^n.m = ^ ( ^ /&BT- i r 1 (4-18)

33 Heat capacify of solids/C

The thermal energy of a solid 2?lattice is the sum of the energies of
its normal modes. The number of normal modes per spectral interval
dco is g(co) dco (see (4.8)). Multiplying this number by the average
energy En.^.oi the normal mode, (4.18), we obtain the total energy
of the normal modes in the interval dco

Integrating this expression over the entire spectrum of the normal

modes, that is, from 0 to coc , we obtain the energy of the'thermal
vibrations of the lattice of a solid:
. d
^lattice = V g() da (4.19)

The heat capacity at constant voluma of a solid, Cv , is the change

in the thermal energy of a soiid brought about by a one negree change
in its temperatura. To nd. it one should difierentiate ffuttics with
respect to T :
Cy = d E ^ J d T (4.20)
The fundamental problem in the theory of heat capacity is the
temperature dependence of Cv . Let us rst consider it from a quali-
tative point of view for two temperature rangas: for the range of
temperatures much below the Debye temperature
T < 0 (4.21)
8
115 Salid Stafa Physics

which is tem ed the low temperatura range, and for the range of tem
peraturas above the Debye temperatura .
T> 0 (4.22)

the term for which is the high temperature range.

Low temperatura range. In this range maiaiy the low frequency
normal modes, with the energy quanta co < k3T, are excited. The
approximate valu of the average energy of normal vibrations may
in this case be calculated with the aid of the following method. Ex-
pand the denominator of expression (4.18) into a series leaving only
two ternas:
= ico l)- Seo(1 -i- J 5 L + ... _ i ) '1

Henee in the low temperatura range the average energy of every nor
mal mode increases in proportion to the absolute temperature T:
arnCCT (4.23)
This law is due to the increase in the probability of excitation of
every normal mode with the rise in temperature resulting in an in
crease in its average energy.
In addition to this the rise in temperature in the low temperature
range causes new higher frequency normal modes to be excited.
The approximate number of the latter, z, may be calculated-with the
aid of (4.8). If we assume that at a temperature T all normal modes
up to the frequency ce ~ kaT!n are excited, we get
Br/fi &3r/fi
g (cd) d (2 a r3

It follows that with thfexise in temperature the number of normal

modes increases in proportion to the cube of the absoluta tempera
tura:
z a r 3 - (4.24)
To sum up, the crystals energy in the low temperature range in
creases with the rise in temperatura by means of two mechanisms:
() the increase in the average energy of every normal mode, En.m,
due to the rise in the probability of its excitation, and (2) the increase
in the number of the normal modes of the lattice.
The first mechanism is responsible for the increase in energy pro-
portional to T and the second for the one proportional to Tz.
Therefore the total enect is an increase in the.energy of the lattice
proportional to T4:
"lattice a r l (4.25)
Thermai Properfies of Soiids ___ __ 117

ana a rise in heat capacity proportional to T3:

C yC C P (4.26)
Formula (4.26) is the Debye Tz law, which agraes well with experi-
merit in the low tamperature ranga......
High temperatura range. A s hasbeen aiready stated, all normal
modes of a lattice are excitad at the Debye temperatura, so'that
a further rise in temperatura cannot increase ther number. Therefore
the variation in energy of a solid in the high temperatura range may
only be due to the rise in .intensity of the normal modes, reslting
in an increase in their average energy E a.m- Since E^m CC T the
variation of the energy of the boay as a whole, too, should be pro
portional to T:
^latticeOC T (4.27)
and the heat capacity must be independent of T :
Cy dE^iticJdT = constant (4.2S)
Relation (4.28) is the expression of the Dulong and Petit law,
which is quite well substa.ntiated by experiment.
A rather wide range of temperaturas, the so-called mdium, temper-
ature range. lies between the high and low temperatura rangas.
In this mdium temperature range gradual transition from the
Debve Tz law to the Dulong and Petit law takes place. Calculations
in this range are the most difcult.
To sum up. the following physical picture of the variation o f the
tamperature dependence of energy and of heat capacity of a soid
with the rise in temperature may be presentad.
In the low temperatura range (T 6) the solids energy increases
with the rise in temperatura firstly because of the increase in the pro-
bability of excitation of every normal mode, that is because of the
increase in its average energy/, E^m, which is proportional to. T,
and secondly because naw= normal modes are drawn into the process
causing the bodys energy to increase in proportion to J 3. The energy
of the lattice, as a whoie, rises in proportion to Ti and the specic
heat in proportion to Tz.
As the temperature approaches the Debye temperature the second
mechanism gradually beccmes inoperative and F attlce becomes less
dependent on T , causing a deviation from the Debye Tz law.
At the Debye temperature the entire spectrum of normal modes is
excited. Therefore the second mechanism has no part to play; only
the rst mechanism operates here causing the energy to rise in pro
portion to T and the heat capacity Cv to remain independent of T
(the Dulong and Petit law).
The qualitative laws of the variation of Cv with T obtained from
the studv of physical processes in soiids-may be substantiated by
118 Soiid State Physics

more rigorous quantitative calculations. To this end let us tura to

(4.19) and try to calclate the lattice energy as a function of.tem-
perature more accurately.
Substituting g (o ) from (4.14) and Z n>m from (4.18) into (4.19),
we obtain

E lattice J hu3d( ( ^ ^B7 _ i)-l (4.29)

One can introduce the dimensionless quantity x = h(/k3T and

rewrite (4.29) in the forra

ian,ct = 9;V=9(-g-) j ***

e- 1
(4.30)
o
where 0 is the Debye temperature.
W e will consider the high and the low temperature ranges separa-
tely.
Low temperature range ( T <C)- In. this range we can substitue
innity for the lim it of integration in (4.30). Taking into account
co
that - - - - - = ^ , we obtain
J sx l Id
0
Eu m ' . ? j - N k ae ( ^ - y < X T i (4.31)

Dierentiating (4.31) with respect to temperature, we obtain

c v = ^ f - i V k Q (-|~)3ccr3 (4.32)

W e have arrived at the Debye T3 law in accordance with whieh the

heat capacity of a lattice vares in the low temperature range as the
cube of the temperature.
High temperature range. For such temperatures the vales of x
are small and henee it is possible to drop all but the first two terms
of the expansin ex = 1 -f- x -f- . . Then
e/r
^latuce = 9 A % 0 ( - g - ) 4 j x-dx = 3NkBT<XT (4.33)
o
The heat capacity o the crystal is

Cv dE]-*ce = 2>NkB= constant (4.34)

For a mole of a mcn.atomic substance N = N 6.023 X i0 "3

m ole-1 (Avogaros number), N Ak3 R 8.31 J/(moie-K) (the
Thermal Properies of Soiids 119

gas constant) and

Cv 3R 25 J/(mole-K) (4.35)
Formula (4.35) expresses the Dulong and Petit law, which was
formulated by them in 1819.
The solid line in Figure 4.5 shows the theoretical temperatura
dependence of the heat capacitv o soiids, the points being experi-
meutal vales for silver, diamond, aluminium, copper, and rock

Figure 4.S Tempera ture dependence of heat capacity of soiids. The solid
Une is the theoretical Debye curve.

sait. The agreement between theory and experimenl is quite satis-

factory not only from the qualitative but also rom the quantitative
point of view.
Knowing the temperature dependence of the energy of a iattice,
we can easily nd at least the qualitative dependence of the concen-
tration of the phonon gas on temperature, that is, the number of
phonons npt2 exciten in a unit of voiume of the crvslal.
The concentration of the phonon gas in the low temperature range,
in which iattice CC T4 and the phonon energy /ico m kBT CC T, must
be proportional to Tz:
w phocr (4.36)
120 Solid State Physics

n the high. temperture range, wiiere lattlce a T and the phonon

energy altains the mximum valu of /cd ~ kBQ independent of T,
Lhe concentration7 of the phonon gas should be proporcional to T:
ripijCCT (4-37)

34 Heaf capacity o electrn gas

The metis, in addition to ions which constitute the lattice and v
brate around their equilibrium positions, contain also free electrons
the number of which per unit volume is approximately the same as
that of the ions. For this reason the specic heat of a metal should be
the sum of the heats! capacity of the lattice Clattice calculated in the
previous paragraph and of the electrn gas Ce:

If the electrn gas was a normal classical (nondegenerate) gas,

every electrn would have an average energy 3kBTI2 and the energy
of the electrn gas per mole of the metal would be

=1> = A 'a-| - : b = ^ - R T

and its heat capacity would be

C = - ~ VAi B = - L * (4.40)

The total heat capacity of the metal in the high temperture*range

would in this case be
Cv = Cuttice - f C f ] = ~ R 37 J/(mole-K)

Actuallv, the heat capacity of metis, as well as that of dielectrics,

in the high temperatura .range, where the Dulong and Petit law is
valid, is equal to Cv ^i&#/tiid-e-K), a proof that the contribution
of the electrn gas is negligible.

7 To be exact, to calclate the concentration o the phonon gas one must

know the average energy of the phonons Sp^ both in the low and the high tem-
perature ranges since the. lattice energy is equal to the product o the average
phonon energy and their concentration. The caicuiation of yields
s.^ T-k-^T/5 (4.38)
for the low temperatura range, and
. ph = 2Bd/3 (4.39)
for the high temperture range.
This justifies the temperatura dependence of /ipp espressed by formulae
(4.36) and (4.37) and obtained by quaiitative methods.
Thermal Properies of Solids 12T

This situation incomprehensible from the point of view of clas-

sical physics found its natural explanation in-quantum theory.
Indeed, as was demonstrated in Chapter 3, the electrn gas in
metis is a degenerate gas described by the Fermi-Dirac quantum
statistics. As the temperatura is raised, not all the electrons are-
thermally excitad; only a negiigible fraction of them, AiV, occipying
States cise to the Fermi level (see Figure 3.6) are thermally excited.
The number of such electrons is approximately expressed by the re-
lation (3.43).

where E? is the Fermi energy. For copper at T ~ 300 K and E?

7 eV we have AN/N .002, that is, less than one percent.
Every thermally excited electrn absorbs an energy of the order
of kBT just as a partile of a normal gas does. The energy absorbed
by the electrn gas as a whole is the product of kBT and the number
of thermally excited electrons AN:
Ee ^ k BT A N ^ N k BT ^ f - ' (4.41 >

The heat capacity of the electrn gas is

7. b

A more accurate calculation yields the following expression:

Y 7. ^3^ (4.43)
Ce -N/v3 2 p
we obtain

* kf (4.44}
C f)

It follows from (4.44) that the ratio of the heat capacity of a degen
erate electrn gas to that of a nondegenerate monatomic gas is ap
proximately equai to the ratio of kBT to E? . At normal tempera-
tures the ratio r<,kBT!EB < 1% . Therefore
Ct < O .O iC f} (4.44'}
Henee, because of the degeneracy of the electrn gas in metis
even in the higa temperatura range only a sin all portion of the fres-
electrons (usually less than one percent) is thermally excited; the
rest do not absorb heat. This is why the heat capacity of the electrn
gas is negiigible as compared to that of the lattice, and the heat ca
pacity of a metal as a whole is practically equal to the iatter.
The situation is aierent in the low temperatura range cise to-
absoiute zero. Here the heat capacity decreases in proportion to T%
122 Sotid rafe Physics

-with. the fall in temperature and cise to absolute zero inay prove
to be so small that the contribution of the heat capacity of the elec
trn gas, Ce, which decreases xauch more slowlv than C^iattice (Cs CC
<X T), may become predominant. Figure 4.6 shows the temperature
dependence of the lattice and electrn components of specic heat
of an alloy (20 percent vanadium and 80 percent chromium) whose
Debye temperature is 0 = 500 K. It may be seen frota Figure 4.6_
that cise to absolute zero the heat capacity of the electrn gas is

Figura 4.6 Temperature dependence of Clattice and Ce of an alloy.

consisting of 20 percent vanadium and SO percent chromium.

much greater than that of the lattice (Cat*tce < Ce), the siga of the
inequality remaining the same up to T 8.5 K. At T > 8.5 K.
the siga is reversed, the inequality becoming st-ronger with the rise
ia T. Already at T 2 5 K the heat capacity of the alloy is mainiy
do to that of its lattice (at T = 25 K the heat caDacity is C iattice
S 10 Ce).

35 Thermal expansin of solids

T o explain the elastic properties of solids. in Chapter 2 we nave
introduced the harmonio approximation according to which the
elastic forc acting on a particie displaced from its equilibrium posi-
tion is proportional to the displacement (see (2.3)] and its potential
energy is proportional to the square of the displacement [see (2.2)];
this fac is representad hy a parabola (the dotted line in Figure 2.1).
The immediate result of the harmonio approximation was the
H ookes law, which describes the elastic deormatioa of solids. The
 Thermai Properrias of Soiids 123

same approximation was used in Chapter 3 as a basis or calculating

the thermai vibrations of a lattice and-constructing the theory of
heat capacity of a lattice, which is in fair agreement with experiment.
However, tbe harmonio approximatioa was unable to expala
such well kaown pheaomeaa as, for iastance, the thermai expaasioa
of soiids, their heat conductivity, etc.
Indeed, let us turn to the dependence of the potential energy of
iateractioii of the partiles of a solid on the distaace betweea them

Figure 4.7 The origin of thermai

expansin of soiids (explanation
in text).

(Figure 4.7). At absolute zero the partiles occ-upy positions r0 cor-

cesponding to- the minimum interactioa energy UQ (at the bottom
of the potential trough abe). Those distaaces determine tile dimensions
of the body at absolute zero.. As the temperature rises the partiles
begin to vbrate arouad their equilibrium positions O. For the sake
of sim plicity let us assume that partile 1 is xed and oaly partile 2
is vibratiag. The kinetic energy of the vibrating particle is at its
mximum Ek when the particle passes its equilibrium position O.
In Figure 4.7 the energy Ek is measured upwards from the bottom of
the potential trough. When particle 2 moves to the left, its kinetic
energy is used to overeme the repulsiva forces acting from particle 1
and is transormed into the potential energy of the partiles inter-
action. The displacement to the left stops when all the kinetic ener
gy Ek is transformad into the potential energy. In the extreme left
position of particle 2 displaced by the dist-ance xx the potential ener-
gys increment is U (%) Ek and its valu (t\ U \xx)). When
particle 2 moves to the right, its kinetic energy is spent to overeme
the forces of attraction to particle 1 and is, as in the previous case,
transformed into the potential energy of the particles interaction.
At point B displaced from the equilibrium position by the aistance xz
the entire kinetic energy is transformed into the potential energy.
the iat-er increasing by U (x; ) = Ek to become (U 0 U (xE\).
If the vibrations of particle 2 were purely harmonio, the forc / (x)
caused by its displacement from the equilibrium position by a dis-
124 Soiid Siete Fhysics

tance x would be strictly proportional to ibis dispiacement and

directed towards the equilibrium position:
/= fh (4.45)

The change in the particles potential energy U (x) would in this

case be described by the parabola a be' (Figure 4.7) whose eqnation is

U (x) = prV2 (4.46)

This parabola is symmetric about the straight line bd passing paral-

le to the axis of ordinates at a distance oi r0 rom it. Therefore the
displacements xx and x would be equ'al in magnitude and the centre
of AB would coincide with the equilibrium position O. In this case
heating a body would not bring about its expansin, for a rise in
temperatura would result only in an increase in the particles am-
plitude of vibrations, the average distances between them remainfng.
unchanged.
Actually, the potential curve abe is, as may be seen from Figure 4.7,
not symmetric about the straight line bd, its ieft branch ba.rising
-much more steeply than the right branch be. This means that the vibra
tions of the particles in a solid are avh.axm.onic (not harmonio). To
account for the asymmetry of the potential curve an aditional trra
gz?IZ expressing this asymmetry (g is a proportionality factor)
should be introduced. Then (4.45) and (4.46) will assume the fol-
lowing formt
U (x) = Per/2 - *73 (4.45')
i (x) = dU/ax (4.46')

When the particle 2 is displaced to the right (x > 0), the term gr73
is subtracted from 2 and the slope of the branch be s less than *
that of the branch be \ when the dispiacement is to the left (x < 0 ) ,
the 'term gar73 is dded to r/2 and the slope of ba is greater than
that o ba'.
Because of the asymmetric nature of the potential curve the right
and left displacements of particle 2 tura out to be difersnt. the former
being greater than the latter (Figure 4.7). As a result, the central
position of particle 2 (point G\) no longer coincides with its equi
librium position O but is displaced to the right. This corresponds to
an increase in the average distance between the particles by x.
Henee heating a body should result in an increase in the average
distances between particles and the body should expand. The cause
of this is the anharmonic nature of the vibrations of particles mak ixing
up the solid due to the asymmetry of the dependence of the parti-
ces interaction energy on the distance between them.
Let us estmate the valu oi the thermai expansin coefcient.
Thermal Properies of Solios 125

rThe average vala of the forc caused by the dispiacement of

partile 2 from its equilibrium position is
/ = px + gx2

"When the partile vibrates freely, / = 0; therefore gx1 = (5x. Henee

v e obtain ^ __
x = gx2/f> (4.47)
The expression for the potential energv of a vibrating. partile
correct to the second order of magnitude is (4.45) and its average
valu is U (x) $xz!2. Henee
x 2 2U (x)/{3
Substituting this into (4.47), we obtain
x = 2g~(x)/p-
In addition to potential energv U (x) a vibrating partile has
kinetic energy E k such that U (x) = Ek. The total average energy
of the partile is E = E<- 4- U (x) = 2 U (x). This fact makes it
possible to rewrite the expression for x in the following form:
x = g~/F
The relative linear expansin, that is, he ratio of the variation of
the average distance between the partiles, x, to the' equiiibrium
distance between them, r0, is equal to

and the linear expansin coeicient is

t ...... g dE
(4.48)
di ~ 3-r0 dT ~ y'Cv
where
X = ^/(P2ro) (4-49)
and cv is heat capacity per partile.
Tlius the thermal exnansion coefncient nroves to be proportional
to temperature. Figure 4.8 shows the temperature dependence of cv
and a. It can be easily seen that both are interreiated.
In the high temperature range the energy of a partile engaged in
linear vibrations is kBT and its heat capacity is cv k3. Therefore
the thermal expansin coefncient o t a linear atomic chain will be
^B
es = 7-cy
1%
126 Sod Sfafs Physics

Substituting the vales for g, k3, 6, and r or various solids, we

obtain a valu of the order of 10"4- 0 -5 for a, which is in fair agree-
ment with experiment. Experiment also supports the conclusin that
in the high temperatura range cc is practically independent of tem-
perature (Figure 4.8).
In the low temperature range behaves in a way similar to that
of cv : it decreases with the fall in temperature and tenas to ze.ro as
absolute zero is approached.

Figure 4.8 Temperature dependence of linear expansin coefcient a

and of heat capacty cv of copper.

In conclusin we would like to remark that for metis a formula

similar to (4.48) was first proposed by E. Grneisen in the form

(4.50)

where x is the metal's compressibility, V the atomic voiume, and y

the Grneisen constani equal to 1.5-2.5 depeading on the metal.

36 Heat conducfivify of solids

Heat conductivity of dielectrics (lattice heat conductivity). The sec-
ond efect causea by the anharmonic nature of atomic vibrations
is the thermal resstanos of solids. There could be no such resistance
should the atomic vibrations be strictly harmonio propagating
through the lattice in the form of noninteracting elastic waves. In
the absence of nteraction the waves vvrould be able to travel without
scattering, that is, without meeting any resistance, like light in
vacuum.
If we were to set un a temperature difference in such a crystal,
the atoms of the hct end vibratins with large amplitudes would traas-
Thermal Properfies cf Scds 127

mit their eneigy to the neighbouring atoms and the ront of the heat
wave would travel along the crystai with the veiocity of sound. -
Because the wave would meet no resistance there would be a consid
erable heat flux aven for an innitesimally small temperature diner-
ence and the heat conductivity of the crystai would be infinitely
great. -
The nature of atomic vibrations in real crystals at temperatures
not too low is anharmonic, as indicated by the second term in[(4.45'}.
The anhannonicity destroys the independence of the normal modes
of the lattice and causes them to interact, exchanging energy and
changing the direction of their propagation (through mutual scatter-..
ing). It is just those processes of interaction between.the elastic
waves that make possibie the transfer of energy from the modes of
one frequency to the modes of another and the establishment of
thermal equiiibrium in the crystai.
a . The process of mutual scattering of normal modes is convenienty
described in tenas of phonons, the thermally excited crystai being-
regarded as a box containing phonons. n the harmonio approxima-
tion, in which the normal modes are presumed to be independen!,
the phonons make up an ideal gas (a gas of noninteracting phonons).
The transition to the anharmonic modes is equivalent to the intro-
duction of an interaction between phonons, which may result in
the splitting of a phonon into two or more phonons and in the or-
mation of a phonon from two other phonons. Such processes are termed
phonon-phonon scattering. Their probabiiity, as is-he case of all
scattering processes, is characterized by the eljectiue scattering croes
>section Cp,. Should the phonon be, from the point of view of the
scattering processes, represented by a sphere o the radius rph then .
Opjj = n r ^ .'T h e phonon-phonon scattering may take place oniy
if the phonons approach to within a distance at which their efective-
cross sections begin to overlap. Since the scattering is due to the anhar-
monicity of the atomic vibrations, numerically described by the
coefncient of anharmonicity g, it would be natural to assume that
the phonon efective cross section radius is prportional to g and'
a PhCC g-.
Knowing the efective scattering cross section apil, we can calclate
the mean free path A,ph of the phonons, that is, the average distance
the phonons travel between two consecutive scattering acts. Galcu-
lations show that

cc
^ph^ph a pno"

where nPh is the phonon concentration.

In the kinetic theory of gases it is proved that for gases the heat
coniciiuiiy is
= A./Cy/3 (4.52)
 123 Solid State Physics

where ?. is the mean fres path of the molecules, y their thermal ve-
locity, and Cr the heat capacity of the gas.
Let us apply this onnuia to the phonon gas substituting for Cv
the specific heat, of the crystal (the phonon gas), for X = 7ph the
mean free path of the phonons, and for y the veiocity of sound (the
phonon veiocity). W e obtain the following expression for the, lattice
heat condiLctivity:
^"lattice vX^CylZ (4.53)
Substituting Xpb into (4.53) from (4.51), we obtain

^iatticeOC-^- - (4.54)
n the high temperature range, in accordance with (4.37), nph CC T;
.henee
la t t ic e OC-^p- (4.55)

Since Cv in this range is practically independent of T, the lattice

thermal conductivity should be inversely proportional to the absolute
temperatura, w h ich is in qualitative agreement with experiment.
Formula (4.55) also inciudes the anharmonicity factor g and the
sound veiocity v, which depend substantially on the rigidity of the
bonds between the partiles of the solid. Bonos of lesser rigidity
-corres pond to lower v's and to greater g's, since the weakening of the?
bonds makes for greater thermal vibration amplitudes (for a specied
temperature) and for greater anharmonicity. 3oth those factors
5 should, according to (4.55), bring about a reduction in c/Tiattice-
This conclusin is supported by ekperiment. Table 4.2 presents the
vales of sublimation heat Qs, which is a measure of bonding energy,
and of the lattice heat conductivity 5Tattice for some covalent crystals
with the diamond lattice: diamond, Silicon, and germanium.
Table 4.2 -J '

Substance Qs (105 J/mole) <Wlattice(w /(m-K))

Diamond 71.23 550

Silicon 46.09 137
Germanium 37 54

We see that the decrease in the bond energy from the valu of
diamond to that of Silicon and, finally, germanium is accompanied
by a noticeable deersase in the lattice heat.conductivity. A more
' detaiea analysis shows that /TlattiCe is also strongly dependent of
the mass M of the partiles, being iess for greater M s. This to a large
extent acco.un.ts for the act that the lattice heat conductivity of the
Thermai Properies of Soiids 29

lighter elements occupying the upper part of the Mendeleev perodic

tabie (B, G, Si) is o'the order of tens or even hundreds of watts per
metre per kelvin, the corresponding vales for the elements of the
middle part of the Mendeleev table being several watts per metre per
kelvin, and that for the heavier elements occupying the lower part
of the Table even to several tenths of a watt per metre per kelvin.
A striking feature is that the lattice heat conductivity of crystals
with light partiles and rigid bonds may be very high. Thus at room

J C .lattice '
I0IJ W/ f m- K)
''
Figure 4.9 Temperature dependence
of heat conductivity of syntnetic
sapphire.

temperature ^uttice of diamond is greater than the total heat con

ductivity of the best heat conductiva metal, siiver: vCAg = 407
W /(m -K).
At temperatures below the Debye temperature there is a sharp
drop in phonon concentration with a fall in temperature leading to
a sharp increase in their mean free pata, so that ai 7 ^ 0/20
it becomes comparable with the dimensions of the crystal. Since the
crystal surace usually is a poor reflector of phonons, any further
* decrease in temperatura does not lead to an increase in Xph, for the
latter is determinad by the crystals dimensions only. The tempera-
ture dependence of the lattice heat conductivity within this range
of temperaturas is detennined by the temperature dependence of the
heat capacity Cv . Since Cv CC T 3 in the low temperature range,
^'lattice too should be proportional to T3:
^ la t tic e ccr3 (4.56)
This result is also in qualitative agreement with enperiment. Fi
gure 4.9 shows the temperature dependence of thermai conductivity
of syntnetic sapphire. In the low temperature range- ^atucs is
indeed apprpximately proportional to Tz.
90835
 'i iitM.
130 Solid State Physics

As the temperature rises so does the concentration o phonons nph, J

and this should per se cause S^iattice to rise. However, an increase in -
nph is accompanied by an increase in the phonon-phonon scattering
and a consequent decrease in the mean free path of phonons XPh,
which should result in a decrease in S^iatttce- For low npb the first^
factor should be the predominan't one and Tiattice should rise with T.
However, starting with a denite concentration nph the second factor
should assume primary importance and ^iattics after passing through
a mximum should fall with the rise in T. This decrease in the high
temperature range is approximately of the 1IT type.
Amorphous dielectrics in which the size of regions with a regular
structure is of the order of interatomic distances should present a
similar picture. Phonon scattering on the boundaries of such regions
should be the dominant factor at all T s and thereore Xpb should be
independent of T. Because of that the heat conductivity of such di
electrics should be proportional to T3 in the low temperature range
and independent of T in the high temperature range. This is just
what is observad in experiment.
However, at present the theory is unable to predict not only the
exact vales of lattice heat conductivity, e/$Tlattice but oven its
order of magnituoe.
Heat conductivity of metis. In metis, in contrast to dielectrics,
heat is transported not only by phonons but by electrons as well.
Thereore generally the heat conductivity of metis is the sura of the
lattice heat conductivity S^attice (conductivity due to phonons)
and the heat conductivity v\ of the free electrons:
zfi, Vi lattice ~T~V
/VY.+ __ /VT/* |
e

The heat conductivity of the electrn gas, can be calculated

with the aid of (4.52). Substituting into this formula the heat capac-
ity of the electrn gas, Ce, the electrn velocity, pF, and their mean
free path, Xe, we obtain
/, 9
q Ue/trAe/ O (4.57)
.bstituting Ct from (4 .4 3 ) into (4.57), we have

(4.53) >
mnv p
Let us make a qualitative estmate of the temperature aependence
of heat conductivity of pur metis.
High temperature range. Practically, of all the quantities contained
in the right-hand side o (4.58) only Xft decenas on T. For pur metis
at temperaturas not too low, Xe is determined by electron-phonon
scattering and, all other conditions being equ'al. is inversely propor
tional to phonon concentration: Xpb CC i/n pb. In the high temper
atura range. according to (4.37), nPb CC T. Substituting this into
Thermal Proparties of S?i<s 131

(4.53), we obtain
%t constant (4.5S)

Henee the heat conductivity of pur metis in the high temperatura

range should be independent of temperature. This is an experimental
fact. Figure 4.10 shows the experimental curve depicting the tem-

Flgure 4.10 Temperature dependence of heat conductivity cf copper.

^Clattice, 10^ W/fm-K)

perature dependence &C for copper. It follows that above SO-OO K

tbe heat conductivity of copper is practically independent of tem
perature.
Low temperature range. The phenon concentration in this range
is 7ip CC T3; therefere Ae CC 1!T3. Substituting into (4.58), we
obtain
a f-1 (4.60)

Consequently, in the low temperature range where the Debye T3

law is true, the heat conductivity of metis should be inversely propor-
tional to the square of the absoluta temperature. This conclusin too
is in general supported by experiment (Figure 4.0).
Very low temperature range. Glose to absoluta zero the phonon
concentration in a metal hecomes so small that the main part in
electrn scatteri'rrg processes is taken over by tbe impurity atoms,
which are always present in a metal nomatterhowpure it is. Inthis case
the mean fres path of an electrn, Xe CC 1IN[ (N\ is'the concentra
tion of impurity.atoms), is no longer.ependent on temperature and
*
132 Sciid Safe Physics

the heat conductivity o a metal becomes proportional to T :

Ss CC T .(4.61)
which is an experimental act.
Let iis estmate the magnitude of for metis, making use of
(4.57). For typical metis* Ce s 0.01CV 3 X 104 J/'(m3-K ),
v ? s 10 m/s, and Xe s 10-3 m. Substituting into (4.57), we obtain
dre s ]102 W /(m -K). Thus for metis may be as high as hun-
dreas of watts per metre per Icelvin. This is substantiated by experi-
ment. Table 4.3 shows the room temperature heat conductivities for
soma typical metis and for one alloy, constantan, which consista
of 60 percent copper and 40 percent nickel.
Table 4.3

Metal <5ST(W/(m-K)) Metal (W/(m-K))

Silvsr 403 Aluminium 210

Copper 384 Nickel 60
Gold 296 Constantan 23
It follows that for pur metis SrC can indeed be as high as hundreds
of watts per metre per kelvin.
Let us also estmate the contribution of the lattice heat conductivity
of a metal, making use of (4.53) and (4.57):
^lattice Ovt'A.ph
/-e Cgfj.g
v being the phonon (sound) velocity. For pur metis CJCV s
ss 0.01, v 5 X 103 m/s, kph 10-9 m, yF- 10 m/s, and Xe s
s 103 m. Henee v^iattice/^e ~ t X 10"2.
t follows'ythen that the heat conductivity of typical metis is
almost entirely due to the heat conductivity of their electrn gas,
the contribution of lattice heat conductivity being a few percent.
This picture may, hQwev^>4:Q,tally change when we go,,ayer to
metallic alloys, in which impurity scattering is the principal electrn
scattering mechanism. The electrn mean free path for which such
scattering is responsible, A., is inverseiy proportional to the impurity
concentration Nl (A,s OC l/iVj) and for high iVjs may become com
parable with the phonon mean free path (Xplx s Xs). Naturaily, in
such] a case the electrn contribution to heat conductivity may be-
come of the same order of magnitude as that of the phonon contribu-
tion: STe s ^lattice- This too is supported by experiment. Table 4.3
gives the heat conductivity of constantan. It is much less than that
of nickel or copper. This proves the fact that electrn scattering in
constantan is mainiy due to lattice defeets causea'by impurity atoms.
W e also not that and SShattce measured by R. Bennaa in con-
stantan proved- to be of the same order of magnitude. - ; -
 5
The Band Theory of Solids

The theory of free electrons was the first successful attempt to ex

pala the electric and magnetic properties of solids (primarily of
metis). It was based on the assumption that metis contain free
electrons capable of moving around the metal like gas molecules in a
vessel. The theory of free electrns was successful in explaining such
phenomena as the electric and the heat conductivities, thermionic
emission, thermoelectric and galvanomagnetic eSects, etc. However,
this theory proved incapable of dealing with such properties of solids
as are determined by their internal structure. It could not even ex-
plain why some bodies are conductors and someinsulators.
The next stage in the progress of the electrn theory has been the
band theory of solids, which is outlined in this chapter.

37 Electron energy levels of a free atom

The State of an electrn in an atom is determined by four quantum
numbers: the principal n, the orbital l, the magnetic m, and the
spin a numbers. In a hydrogen atom the principal quantum, number, n,
describes the steady-state energy of the electrn:
E (n) = Rnz (5.i)

where R 13.6 eV is a universal constant, called a rydberg.

The orbital quantum number, l, describes the orbital angular mo-
msntum of the electrn, p:
P, =* Y 1)
(h = h!2jt, with h the Planck constant). The quantum number l
may assume only the following integral vales:
Z = 0, 1, 2, 1)
n vales in al!.
134 Solid State Physics

The magnetic quantum number, mh describes the orientation of

the orbital angular momentum with respect to some specied direo
tion H (Figure 5.1(a)); the orientation of p with respect to H may
only be such that its projection onto this direction is a mltiple of h:
P ih (5.3)
The number m ay assume the following set of integral vales:
mt = _Z , _ ( i _ i), . . 0, 1, 2.......... I (5.3')

21 -f 1 vales in all.
Lastly, the spin quantum number, a, describes the orientation of
the intrinsic angular momentum (the spin p4) of the electrn with

Figure 5.1 Orientatioas of orbital](a) and spin (b) angular momsnta

with respect to H.

H H

respect to specied direction H (Figure o.i(b)): the vector p 4 may only

be oriented with respect to H so that its projection onto H is equal to
(5.4)
only the vales 1/2 and 1/2 being ailowed for a.
The States Vvith orbital quantum number l 0 (the vales of
other quantum numbers being irrelevant) are termed s staies; those
with l = , are termed p States; with l 2, d staies; l = 3, / States;
etc. Electrons in those States are termed s-, p-, d-, /-, etc. electrons.
In contrast to the hydrogen atom the energy of an electrn in manv-
electron atoms depends aot only on n but on l as well: E E (n, l).
Only discreta vales of n and l being ailowed, the energy snectrum
of electrons in atoms may assume only discrete vales too; the spec-
The Band Tnoory of Solids 135

Table 5.1

Atomic energy ' Total number of Splitting of levis into

levels and g = (2Z-f-i) electrons on a g = (2Z4-l) sublevis
their noiation level: n=2(2Z-f-t) when degeneracy i3 lifted

2
1
E ( 3, 2) 3d 5 10 3d----- 0

---- 1
E ( 3, 1) 3p 3 6 3p----- ---- 0

E (3, 0) 3s 1 2 3*----- ---- 0

1
E ( 2, 1) 2P 3 6 2P----- -----0

E (2, 0) 2s 1 o 0
E (i, 0) Ir 1 2 is----- 0

trum consists of a set of allowed levels E (n, l) separated by forbid-

den energy intervals. Table 5.1 shows a diagram (not to scale) of
the first three groups of such levels.

Figura 52 Electron ciouds of the 2p State.

All the s levels are nondsgeneris. This means that everyone of

thera corresponds to a single electrn State in the atona. l a compli-
ance with the Pauli exclusin principie there may be two electrons
with opposite spins in snch a state.
The p levels are thres-fold degeneris: there is not one but three
States with diSerent magnetic quantum numbers m to correspond
to each of them. For l = 1 those vales ara ml 1, 0, -{-i. Fig
ure 5.2 shows the shape o electrn ciouds corresponding to thcse
136 Solid Sfafa Physics

statas. Since there may be two electrons per State, the total number
of electrons in the p State is six.
The degeneracy of the d levels is ve-fold, since the allowed vales
of the magnetic quantum number for l = 2 are ml 2, i, 0,
- f l , -f-2. This level can accommodate 10 electrons.
Generally, a level with the orbital quantum number l is a (21 -f 1)-
fold degenerate one and can accommodate 2(2Z -f- 1) electrons.
When a fres atom is placed in a strong esternal eld, the degener
acy vanishes and every level splits into (21 -f- 1) closely spaced sub
levis, as shown in the last coiumn of Tahle 5.1.
The eSect of an esternal eld on dinerent atomic levels is not the
same. The splitting of the levels of inner electrons, whose interac-
tion with the nucleus is strong, is so small that it may be neglected.
As the shell radius is increased the energy of interaction of the re
spective electrons with the nucleus becomes smaller and the enect
of the esternal eld becomes more noticeable. The eect of an external
eld is most pronounced for the energy levels of outer electrons,
whose bonds with the nucleus are relatively weak.

38 Coecfivizafion of electrons
in a crystai
The interatomic -distances in solids are so small that every atom
nds itself in a strong eld of the neighbouring atoms. To gain in-
sight into the. eSect this eld exercises on the energy levels, consider
the following idealized example.
Arrange N sodium atoms in the pattern of a three-dimensionai
lattice having the shape of a sodium crystai but with interatomic
distances so great that the interaction between the atoms can be
neglected. In this case one can legitimately assume the energy. States
in every atom to be the.same as in an individual sodium atom. Fi
gure 5.3(a) shows the energy. diagram of two such atoms-.. Each of
them has the appearance of a spindle-shaped potential trough inside
of whch the levels ls, 2s, 2p, 3s, . . . are shown. The ls, 2s, and 2p
levels in a sodium atom are fully occupied, the 3s is occupied to one
hall, and the levels abcve 3s are empty.
As shown in Figure 5.3(a) the individual atoms are separated by
potential barriers o the width r a, where a is the lattice constant.
The height of the potential barrier is not the same for electrons
occupying dierent levels. It is equal to the height measurea from
hose levels to the zero level 00. The potential barrier prevens the
electrons from moving reely from one atom to another. Calcula-
tions show that for r m 30 A the average transtion rate of a 3s~
eecltron from one atom to another is once in every 103a vears.
Tha 3and T'neory o( Solids 137

Figura 5.3 Variatioa of el8ctronic stasa in approachmg atom3:

(a)energy diagram o! aodiuin atoms placad at a distance much greater
than the sodium lattice parameter; (b)energy diagram of sodium atoms
brougbt together to a distance of the order o the lattice param9ter.
138 Solid- rata Physics

The upper part of Figure 5.3(a) shows a cualitativa plcure of the

space distrihutiou of the density p = Axrhjpb* o the prohahility of
detecting electrons at a distance r from the nucleus. The maxima of
those curves correspond approximately to the radii of 3ohr orhits
of such electrons.
Now make the lattice contract uniormly so that its symmetry
remains unaffected. As the atoms are brought closer the interaccin
between them increases and for an interatomic distance equal to the
lattice constant a it turas into one characteristic of the crystal. Fig
ure 5.3(b) depicts that situation. We see that the walls of the poten-
tial troughs separating neighbouring atoms (they are shown in the
figure by dotted lines) partly overlap to create new walls shown by
solid lines that are lower than the zero level 00.
Namely, the efrect of the reduction of the interatomic distances is
two-fold: to reduce the height and the width of the barrier, the iatter
to the valu of the order of a. The height of the reduced barrier is U\
for ls-, U3 for 2s-, and U 3 for 2p-electrons, the original 3s levels of
the sodium atoms lying above it. This fact makes it possible for the
valence electrons of this level to move practically unhindered from
one atom to another. The nature of the electrn clouds of valence
electrons (shown in the upper part of the figure also points to the
same conclusin: their overlapping is so complete that the density
of the resulting cloud is practically uniform (Figure 5.3(b)). This
corresponds to their complete collectivization in the lattice. Such
collectivized electrons are usually termed fres electrons and the to-
tality of them, the electrn gas.
A drastic reduction in the width and the height of the potential
barrier brought about by the decrease in interatomic distances makes
it possible for the electrons occupying other atomic levels besiaes
the valence levels to move inside the crystal. Their motion takes place
by means of tunneling through the barriera that separata neighbour
ing atoms. The narrower and the lower those barriers are, the mora
complete is the collectivization of the electrons and the greater is
their freedom.

39 Energy spectrum of electrons

in a crystal
In the same way as the main goal of the theory of atoms is to describe
the electrn States in an atom and calclate the allowed energy levels,
one of the main problems of the theory of the solid State is to deter
mine the energy spectrum of the electrons in a crystal. One may ob-
tain a qualitative idea about this spectrum using the following
approximate gmethod to treat the behavicur of electrons in a
crystal.
Tha Band Theory of Sollds - . __

To approximately describe the motion of an electrn ia a crystal

we may use the following Schrodinger equatioa:

V ^ + - |? - ( - t O i> = 0 (5.5)

'where U is the poteatial eaergy, E the total energy, aad m the mass
of the eiectroa.
For electroas sufficieatly stroagly bouad to the atoms the potea
tial eaergy may be written ia the forra
7 = f 7 a + 6t/ (5.3)
where 7a is the electroas eaergy ia aa isolated atora. Ia a crystal
it is a periodic fuactioa with a period equal to the lattice parameter,
since there is a recurrence ia the valu of eaergy as the eiectroa moves

Figure 5.4 Periodio rariation

of eiectroa poteatial er.srgv
in crystal.

from one atom to aaother (Figure 5.4); 5 U represeats a correctioa

that takes accouat of the eSect of aeighbouriag atoms on this eaergy.
If oae aeglects the correctioa 5 U ia (5.6), that is, coasiders oaly
the zero approximation, oae should take the wave fuactioa Tj>a aad
the eaergy a(n, L) of the eiectroa ia aa isolated atom as the wave
fuactioa aad the eaergy of the eiectroa ia a crystal: tp = -qoa, T =
= E ^ n , i), where n aad l are the priacipal aad orbital quaatum
aumbers, which determine the eaergy of the electrn ia an atom.
Ia this case the difference betweea a crystal and an isolated atom
is that in an isolated atom a specied eaergy leve! E &{n, l) is unique,
but in a crystal consistiag of N atoms there are N such Ievels
(Figure 5.5). In other words, every eaergy level of aa isolated atom
is iV-old degenerate in a crystal. Such degeneracy is termed irzm-
positional.
Now let us estmate the correctioa 5U in the poteatial eaergy (5.6).
As the isolated atoms are brought together to form a lattice each
acom increasiagly feels the eld o its neighbours with whom it inter-
acts. As we have already seen, such interaction results ia the liftiag
o- degeneracy including traaspositional degeneracy. Because of
that each level nondegenerate ia an individual atom spiits up lato A
closelyfspaced sublevis to form aa ener?y band.
140 Soli Sae- Physics

Ir every level in an isolated atom was (21 -f 1) times degenerate,

the corresponding band shall ccntain N (21 -f- 1) sublevis. Accord-
ingly, the s level produces the s band consisting of N sublevis and
capable of carrying 2N electrons; the p level produces the p band
consisting of 3N sublevis and capable of carrying 6N electrons; etc.
The separation between the sublevis in an energy band of an ordi-
nary crystal is very sm all. A crystal of a volume of one cubic metre
contains 1028 atoras. For a band of the order of i eV wide.the sepa
ration between the sublevis is about 10~28 eV. This separation is* so

Figure 5.5 Each atomic energy

leve! E a in a systam consisting
of N atoms is repeated N times
(is JV-fold degenerate).

negligible that the band may be considered to be practiealiy contin-

uous. However, the fact that the numher of leveis in a band is
nite is very important for the determination of the distribution
of electrons over States.
The mximum efect of the lattice eld is on the external valence
electrons. Because of that the State of such electrons in a crystal
experiences the greatest change and the energy bands formed by
their energy leve-ls are the widest. The interna! electrons, which
are strongly bohf'i:h''their nuclei, are only slighily perturbed by
the presence of neighbouring atoms and accordingly their energy
bands in the crystal are almost as narros as the leveis of isolated
atoms. Figure 5.6 shows a schematic diagram of energy band forma-
tion from discrete atomic leveis.
Thus in a crystal there is an allowed energy band to correspond
to each energy level o an isolated atom: the Is band to correspond
to the Is level, the 2p band to the 2p level, etc. The allowed energy
bands are separa ted by forbidden energy bands E g. As the energy of
an electrn in an atom is increased so is the width of the respective-
energy band, the width of the forbidden band being reduced.
Figure 5.7 shows the energy band structure of lithium, beryllium,
and elements having the diamond lattice (diamond, Silicon, ger-
The Band Theory of Scids 141

tnanium). In the lithium crystal (Fig. 5.7(a)) the spiitting of the s

ievei is a narrow on8, the spiitting of the 2s level being wider so that a
suhciently wide 2s'energy band is ormed. In the beryilium crystal
{Fig. 5.7(b)) the 2s and the 2p bands overlap to form a mixed, or
the so-called hybrid, band. The situation is quite similar in case of
the other elements of the main subgroup of Group II of the Mendeleev
periodic table.
The pattern of band formation in crystals with the diamond lat-
tice (Figure 5.7(c)) is somewhat diferent. In this case the bands
formed from the s and the p levels overlap and split into two bands, so

Figure 5.6 Sckematic represeatation
of energy band formation
in a crystal rom discreta atomic
levels.

Figure 5.7 Formation of bands from atomic levels: (a)in lithium

crystal (band 2s is only kal-Slled); (b)in beryilium crystal
(lled band and frea 2p band overlap to form a partiaiiy ied
hybrid band); (c)in diamond-lattica elementa of Group IV (the arrows
denote tha spin of the electrons).

E z E E

(a (b) (c)
142 Solid Stat Physics

that each of them contains iour States per atom: oce s State and three-
p states. Those bands are separated by a forbidden bacd. Tbe lower
band is tenned the valence band and the upper tbe conduction band_

40 Dependence of electrn energy

on ihe wave vector
J.t was demonstrated in the preceding section that the pattern of
the electrn energy spectrum in a ^rystal is of a band type. Consider
now the dependence of the electrn energy E on its momentum p
inside each band, that is, the shape of the E (p) curves. The momen-
tum dependence of E is tenned the dispersin law, or dispersin rela-
tion.
Turn now to the simplest case of a free electrn moving along the x
axis and described by the following Schrodinger equation:

o (5.7)
II
x-
where
E = -Z - (5.8)
^ 2/71
since a free electrn has only kinetic energy.
Formula (5.S) is the dispersin relation for free electrons, which
expresses the momentum dependence of E. It may be rewritten in
the following forro. According to the de Broglie formula
h _ h
(5.9)
P= I TJ{2E) = k
where X is the waveength of the electrn, and
k = 2nk (5.10)

The vector k coinciding in direction with the direction o the electrn

wave propagation and equal in magnitude to 2rJX is termed ivaue
vector of the electrn. Substituting p from (5.9) into (5.8), v e obtain

(5.11)

Formula (5.11), which expresses the dependence of the energy of

a free electrn on its wave vector, is just another form of writing
the dispersin relation (the dispersin law) for such electrons.
It follows from (5.8) and (5.11) that the dispersin law for free
electrons is quadratic and for one-dimensional motion of the electrn
takes the form of a quadratic parabola shown in Figure 5.8(a).
The Band Theory of SoUds

The solution of equation (5,7) is a traveliing plane'wave

. ' \p = Ceikx (5.12>
where C is the waves amplitude.
As is well known, the square of the modulas of the wave function is
proportional to the probahility of detecting the electrn in a speci-

Figure 5.8 Motion of a iree electrn: (a)dependence of snergy on wave

vector (dispersin curve); (b)square modulus of wave function
proportional to the probahility of the electrn being at point z.

ed regin of space. As raay be seen from (5.12), for a fres electrn

this probahility is independent of the electrons^ position^since
th 2 = (5.13>
which means that for a free electrn every point in space is equivalent
and the probahility of detecting it is everywhere the same (Fig
ure 5.8(b)).
The case of an electrn moving in a periodic field of a crystal
formed by regularly arranged ions is diSerent (Figure 5.9). The proba-
bility of detecting it in a specied point of the crystal should be a
periodic function in x, since positions displaced from one another
by a mltiple of the lattice constant a (for ihstance, the positions D ,
D ' , and B in Figure 5.9) are equiprobable for the electrn. The posi
tions inside. a period a (for example, inside D F D ') are, however, all
difierent. This means that the amplitude of the wave function ic(rV
of an electrn moving in a periodic eld does not, as in the case of
a free electrn, stay constant but changes perioaically, or it mav b-
said to be modulated with a period. equal to the lattice parame-
ter a. Denote this amplitude u (x). Then the wave function of the-
electrn moving in a periodic eld of a c o s ta l in the direction of the xr
axis may be expressed in the foilowing form:

ip (x) u (x) e (5.14),

144 Solid State Physics

where u (x - f na) = u (x), where n is an arbitrary integer. Relation

{5.14) is termed the Bloch function. The specic forra of this functioa
is determined^by the potential energy U (x), which enters into the
Schr5dinger equation (5.5).
There shouid be a corresponding change in the dispersin relation
for electrons moving inside a crystal as compared with that for free
electrons. Firstiy, as we have already seen, the energy spectrum of
such electrons assumes a band pattern: allowed energy bands formad
from corresponding atomic levels a are separated by forbidden energy
bands. Secondly, calculations show the electrn energy inside each

Figure 5.9 Motioa of an electrn in periodic eld. The square modulus

of the wave function that describes the probability of the electrn
being at point x of the horizontal axis is a periodic function
of coordnate x, the period being equal to the lattice parameter a.

band to be a function of the wave vector k, which for a one-dimen-

sional crystal (an atomic chain) with the parameter a is of the form

E (k) = E k -f C. 4* 2A eos ka (5.15)

where F a is the energy of the atomic level from which the band was
formed; C is the_ displacement of this level due to the eect of the
eld of neighl3*M}iag.,a.toms; A is the so-called exchange integral,
which takes into account the newly created probability for an elec
trn to move from one atom to another owing to the overlapping of
the atomic wave functions (see Figure 5.3(b)). The exchange integral
is the greater the greater the overlapping of the wave functions,
that is, the greater is the exchange rate of the electrons of neigh-
bouring atoms. For s States A , <C 0, for p States Av > 0, therefore
it is reasonable to write out relation (5.15) individually for the s
and the p bands. For the s bands
E s {Jc) E \ 2AS eos ka (5.15)
and for the p bands
E p (k) Ep -r 2AVzos ka (5.17)
The 3ard Theory cf Soiid 145

where E\ = F a, - f C3, E'v F ap -f- Cp, and A s and Ap are the afa-
solute vales o tile exchange integris for the respective States.
Figure 5.10 shows dispersin curves E (k) for the s and p bands
drawn to satisfy equations (5.16) and (5.17). For the s States E ,
at k 0 assumes its minimum valu E s = E\ 2AS. As k
increases, eos ka decreases and the valu qf E s (k) rises reaching its

Figure 5.10 Dispersin curves for an electrn moving in a periodic

eid: the lower curve corresponds to the s band, the upper curve
to the p band; dotted lines are paraboias expressing the E(k) ependence
in the centre and oa the boundaries of the Briilouin zone.

mximum E s ma, = E's -j- 2A , at k = rea. In the interval of vales

of k from 0 to xla, E } (k) changes in a similar fashion. The width
of the ailowed s band from E 3 mln to F s mas is
F F , E t max F jrnln 4.4j (5.1o)
As may be seen, it is determined by the absolute valu of the exchange
integral, which, in its turn, decenas on the overiappng of wave func-
tions o aeighbauring atoms. The shape of the curve E s (k) is that of
an overturned bell.
For the p States E v mln = E'v 2Ap corresponds to k =
and Ep max Ep 2P to k = 0 . The width of the p band
h-Fp = Ep mai np =4Ap (5.19)
as in the previous. case, is determined hy the absoluta valu of the
exchange integral A p. As a rule, the higher the atomic level the
greater the overiapping of the corresponding wave unctions in the
crystal, the greater the valu of the exchange integral, and the wicer
the eneigy. band formed from this leve!. For this reason high atomic
leveis are the origin of wiae energy bands separated by narrow for-
bidden bands (see Figure 5.6).
10-0835
 145 Sod State Physics

The intervals of k inside which the electrn energy E (k) as a period-

ic function of k completes its fuil cycle are termed Brillouin zones.
For a one-dimensional crystal (an atomic chain) the rst Brillouin
roe lies between k = na and k = irla and is 2nla wide (Fig
ure 5.10). In the vicjnity of a dispersin curveas extremum, that is,
in the vicinity of k = 0 and k = n la (the centre and the boundary
of the rst Brillouin zone), eos ka can be expanded into-a power series
in ka (k is measured from 0 if the extremum is in the centre of the
Brillouin zone and from n la if it is on its boundary) leaving only
two terms of the expansin
eos ka = 1 (ka)2!2 + . . .
Substituting this into (5.16) and (5.17). weobtain E s (k) = E STnln -4*
-f A s (ka)2, E p (k) = E v m2s Ap (ka)2. The mnimum of the
dispersin curve E (k) is termed the bottom of the energy band and
the mximum the top of the band. Therefore we may rewrite the sought
for relations in a more general form:
-Shottom (&) = ^min + & i^a)Z (5.20)
for the bottom of the band, and
Exop( k ) = E mX- A '( k a ) 2 (5.21)
for the top of the band.
Henee cise to the top and the bottom of an energy band the por-
tion of the electrn energy that depends on the wave vector is pro-
portional to the square of the wave vector, measured in the way indi-
cated above, and to the exchange integral that determines the width
of the band. The parabolas corresponding to equations (5.20) and
(5.21) are shown in Figure 5.10 by dotted lines.
The E(k) dependente for real crystals is, as a rule, much more
complex than that expressed by formula (5.15).
Figure 5.11(a) shows the dispersin curves for the bottom of the
conduction band (curve 1) and the top of the valence band (curve 2)
of Silicon. We see that the bottom of the conduction band, D , of
Silicon is not in the centre of the Brillouin zone but near its boundary
in the direction [100]. The valence band is bounded by a curve in the
shape of a parabola with its apex at B in the centre of the Brillouin
zone. However, despite such complexity of the dispersin curves,
the quadratic dependence of E (k) expressed by formulae (5.20) and
(5.21) remains valid for both band-edges in this case.
The width of the orbidden band, or energy gap, is determined by
the mnimum gap between the valence and the conduction bands;
in Figure 5.11(a) this is denoted by E g.
Frequently, when making a simplied analysis of the energy-band
structure of semiconductors, instead of the actual dispersin curves,
which bound the valence and the conduction band, use is made of
The Band Theory o Soiids

two parallel lines, one drawn tangentially to the bottom o the con
duction band and the other to the top of the vlenos band (Fig
ure 5.1 l(b)). The first line is taken to represent the lower boundary

Figure 5.11 Band pattern of Silicon: (a)dispersin curves E(k) bounding

the conduction band (curve 1) and tbe vaienca band (curve 2); the snergy
minimum o the conduction band is at point D in the [100] direction,
energy mximum of the valence band is in the Brillouin_ zona centre B ;
the distance between minimum D and mximum B is the forbidaen band
width E s ; (b)schematic representaron of energy band pattern of Silicon.

Conduction

(the bottom) of the conduction band and the second the upper bound
ary (the top) of the valence band. The separation between the lines
is equal to the forbidaen band width E g.

41 Effecfive mass of Ihe electrn

The de Broglie formula establishes the following relation between
the momentum o a ree electrn and its vave vector:
d = nk

The velocity of the electrn's translational motion is

v = p/m .(hTii) k (5.22)
Dinerentiating (5.11) with. respect to k, we obtain
,_m i E
^ K dk
10
148 Soiid State Physies

Substituting this into (5.9). and (5.22) we obtain

m. dE 1 dE
P Ylk m n dk
(5.23)

Expressions for the momentum and or the velocity of translational

motion written in this form turn out to be vaiid not only for free
eiectrons but or eiectrons moving in a periodic crystal eld as well.
Momentum p is in the latter case termed qua.simoTnen.tum of the elec
trn.
Set up an external eld S in the crystal. This eld acts on the elec
trn with a forc

imparting to it an acceleration
. _ dv__ J___ ldE_\ _ 1 <PE dk
^ dt h dt \ dk ) h dk- dt
The work performed by the forc F during the interval dt is

dW = Fv dl = ^n- ^dk
Ldt

This work is spent on increasing the electrones energy by an amount

dE = {Fh) (dEidk) dt. Henee F/h = (dkldt). Substituting into the
right-hand side of the expression for /, we obtain
F d-E q% d-E _ q% (5.24)
- d k h - dkz ~ me{

Formula (5.24) establishes the relation between the electrons

acceleration j and the external forc.e F with which an external eld %
acts on it. Henee it is an expression of Newtons second law. It fol-
lows then that the electrn acted upon by an external forc moves in
a periodic crystal eld on the average in the same way as a free elec
trn .WQuld move if its mass were

(5.25)

The mass mzt is callen the effective mass of the electrn. Having attri-
buted to the electrn in a periodic crystal eld a mass mef, we may
now regara it as being free and describe its motion in an external
eld in the same way as we would describe the motion of an ordinary
free electrn.
However, the eective mass, which embraces all the detaiis of
electrn motion in a periodic crystal eld, is a very particular quan-
tity. To begin with, it may be positive or negative, many times
larger or many times smaller in magnitude than the eiectrons rest
. mass m. Let us make a more detailed study of the problem.
The Band Theory- of Solids 149

For electrons cise to the bottom of a band tbe energy is ottom =

= -^min ~r A (ka)2, the second derivative with respect to k being
cPE/dk2 = 2A a2. Substituting into (5.25), we obtain the following
expression for the efective mass of the electrn, which we shall de
note m^:
(5.26)

Since A > 0, we see that mn > 0. Henee electrons cise to tbe bottom
of an energy band bave a positive efective mass. For tbis reason tbey
bebave normally in an external field, accelerating in thedirection
of the acting forc. The difference between such electrons and free
electrons is tbat tbeir mass may be quite diSerent from tbe rest mass.
It may be seen from (5.26) tbat tbe greater A is, that is, tbe wider
the allowed band, the less tbe efective mass of tbe electrons occupying
States cise to the bottom of tbe band is.
For electrons cise to the top of tbe band tbe energy is Sop =
= m a s A' ( ka)z, the second derivative of E with respect to k
being cEidkr = 2^4 'a2, and the efective mass, which we denote m'n ,
is
h-
(5.27)
TTln
2A'a-
It is a negative quantity. Such electrons bebave abnormaily in an
external eld set up in a crystal: they are accelerated in the direc-
tion opposite to tbe acting forc. Tbe absoluta valu is, as before,
determined by tbe widtb of tbe energy band: is the smaller the
wider the band is.
Let us dow find what is responsible for such a strange behaviour
of the electrn in a crystal.
In case of a free electrn all the work W peronned by an external
forc F is spent to increase tbe kinetic energy of the electrons trans-
lationaT motion:
77 = kz

Diferentiating 2Tk t-vice with respect to k, we obtain drEyJdkr = az!m.

Substituting into (5.25), we find that msS = m. Henee the efective
mass of a free electrn is simply its rest mass.
The situation may be quite diSerent in a crystal wThere tbe electrn
has not only kinetic energy but potentiai energy as well. When the
electrn moves under the action of an external forc F, some of the
work peronned by tbis forc may be transfonned into kinetic ener
gy E {, the rest being-transfonned into potentiai energy U, and W
E'c -~ U. In this case tbe result will be a smaller increase in kinet
ic energy and, consequently, in electrn velocity tban in the case
150 Sod State Physics

of a free electrn. The electrn, figuratively speaking, gains weight

and moves under the action of the forc F with a smaller acceleration
than a free electrn would.
Should the entire work of the external forc be transformed into
potential energy 7, that is W U, there would be no increase in
the kinetic energy and in the Velocity of the electrn and it would.
behave as a partid a of infinite effective mass.
Finally, if in the course of the electrons motion not only the entire
work of "the external forc F but some of the kinetic energy that
the electrn had initially too shall be transformed into the potential
energy. so that U = W 4- 2?, then as the electrn moves along the
crystal its velocity shall diminish, it shall be accelerated, behaving
as a partile with a mgative effective mass. Just such is the behaviour
of electrons occupying States cise to the top of the conduction band.
However, a situation is possible in a crystal when in the course
of the electrons motion under the action of an external forc F
not only the entire work of t-his forc but some of the electrons poten
tial energy, sav U ', shall be transformed into its kinetic energy so
that E l = W 4- U '. The E l and the velocity v of such an electrn

Figure 5.12 Depeadence o electrn energy , group velocity v,

and eective electrn mass mett on k.

E
 The Bartc Theory o Soiids 151

shall rise quicker than those of a res electrn. It looses weight as

compared with the freo electrn, so that its eective mass mza < m .
The aforesaid is illustrated in Figure 5.12, which shows the nature
of the variation of the total energy of the electrn E (k), of its trans-
lational velocity y (k), and of its eective mass mett with the rise in k
from 0 to rca.
Cise to the bottom of the band (k = 0), as long as the electrons
energy E (k) rises approximately in. proportio~to k1, the velocity
of the electrons translational motion v OC dEldk increases in pro-
portion to k, the acceleration remains positive. and the eective mass
;neff CX (drEdk )~l retains its constant positive valu mn. At the
pomt of inlection C of the curve E{k)- the second derivative d^Eldk-
vanishes and the rst derivative reaches its mximum valu. There-
fore, as this point- is approached, metf * co and v *- ymax. After
the point of inflection dEldk starts to decrease causing a decrease in y;
henee acceleration becomes negative, which, for the direction of
the external-forc F remaining unchanged, is equivalent to a change
in the sign of the eective mass from the positive to the negative.
If the curvature of the curve E(k), which is proportional to drEikr,
also changes, this would lead to a change in the absolute valu of
metf OC (d-Edk2)~l. Near the top of the band E(k) again becomes
a quadratic function of k, and the eective mass assumes a constant
negative valu m'n.

42 Occupation of bands by eisefrons.

Conducors, dielecfrics,
and semiconductors
Each energy band contains a limited number of energy levels. In
compliance with the Pauii exclusin principie each level may be occu-
pied by no more than two electrons. With the limited number of
electrons in a solid only the lower energy bands will be lled.
According to the nature of band occupation by electrons all soiids
can be classied into two large groups.
The rst group ineludes bodies in which tbere is a partially lled
band above the completelv lled lower bands (Figure 5.13(a)). Such
bands are formed from partially lled atomic levels as, for instance,
in the case of alkali metis. A partially lled band may also be the
result'of overlapping of lled and empty or partially lled bands, as
is the case with beryllium and alkali-earth metis (Figure 5.13(b)).
A partially lled band is a feature of metis.
- The second group inciudes bodies with empty bands above compiete-
ly iie ones (Figures 5.13(c, dj). Typical exampies of such bodies
are the elements of Group IV of the. Mendeleev periodic tablecar
i bn in the diamond modification, S i l i c o n , germanium and gray tin,
152. Soiid Sfate Physics

which has the structure of diamond. This group also ineludes many
Chemical compoundsmetal oxides, nitrides, carbides, haiides of
alkali metis, etc.
According to the band theory of solids, the eiectrons of the outer-
most energy bands have practically the same freedom of movement,
no matter whether the solid is a metal or a dielectrc. The-motion
takes place by means of tunneling from atom to atora. Despite this
fact, there is a dierence of many orders of magnitude in the electric

Figure 5.13 Occupation of bands by eiectrons: (a, b)there is

a partialiy filled band above the filled band; (c, d)there is an empty
band above the lled band.

properties, in particular in the electrical conductivity, of the bodies

of both types: in metis a m 107 (ohm-m)"1, and in good dielectrics
a < 10-1*1 (ohm-m)-1.
To gain insight int-o the mechanism of electrical conductivity of
solids let us discuss the behaviour of the eiectrons of partialiy and
completely lled energy bands in an external eld.
Set up an external eld $ in the crys.tal. This eld acts on every
electronHvith-a torce F = q S that tenas to disort the symmetry
in the velocity distribution of the eiectrons so that tbose moving
against the forc are decelerated and those moving in the direction of
the forc are accelerated. Since such acceleration and deceleration
inevitably entail a change in the electrons energy, this is equivalent
to the electrons transition to States with higher or lower energies.
Such transitions may, evidently, take place only if there are unoccu-
pied States inside the band to which the eiectrons beiong, that is. if
the band is not completely lled. In this case even a weak electric
eld is capable of imparting to the eiectrons the necessary additional
momentum that will take them to nearby ree levels. A prevailing
motion of the eiectrons in the direction opposte to that of the eld
will be set up in the solid resulting in an electric current. Such solids
should be good conductora, which is actually the case.
The Band Theory of Solids 153;

Now let us imagine that the vaience band of the crystal is com-
pietely iled and is separated from the nearby empty band by a wide-
energy gap E g (Figure 5.13(c)). An external fieid appiied to such.
a crystal is incapable of changing the uature of electrn motion in the-
valence band because it is unable to lift the electrons to the empty
band lying above it. Inside the vaience band, which has no free lev
is, the fieid may only cause the electrons to chango places and this-.
does not distort the svmmetry of the electrn distribution over veloc-
ities. Therefore in such solids an external fieid is incapable of induc-
ing directional motion of the electrons, that is, an electric current, and
the eiectrical conductivity of such solids should be practically zero.
Henee for a body to have high eiectrical conductivity it must have-
in its energy spectrum some energy bands only partially iled with
electrons, as is the case with the typical metis (Figure 5.13(a, b)).
The absence of such bands in solids belonging to the second group-
makes them nonconductors despite the act that they contain electrons
weakly bonded to individual atoms.
The solids of the second group are conventionally subdivided into-
dielctrica and semiconductor according to the width of the forbid-
den band. Dielectrics inelude solids with a relatively wide forbl'ddea.
band. For typical dielectrics E ? > 3 eV. For diamond E g 5.2 eV;.
for boron nitride E = '4 .6 eV; for ALO-., E = 7 eV; etc.
Semiconductors inciude solids with a relatively narrow orbidden
band (Figure 5.l3(d)). For typical semiconductors E ? < 1 eV. Thus
for germanium E, = 0.66 eV; for Silicon E s = 1.08 eV: for indium.
antimonide E s = 0.17 eV; for gallium arsenide E = 1.43 eV; etc.
Let us consider this class of solids in more detail.

43 lnfrinsic semiconductors.
The concept of a nole
Intrinsrc semiconductor. Semiconductors containing a negligible-
amount of electro-active defeets (chemical and crystallographic>
are termed intrinsic semiconductors. They inelude some pur Chemical i
elements (germanium, siiieon, selenium, teiiurium, etc.) and numer-
ous chemical compounds such as gallium arsenide (GaAs), indium-
arsenide (InAs), indium antimonide (InSb), siiieon C a rb id e (SiC),.
etc.
Figure 5.14(a) shows a simplined schematic aiagram of an intrinsic.
semiconductor. At absoluta zero its vaience band is completey iled
and the conduction band, which is a distarme E s aheve the vaience-
band, is empty. For this reason at absoluta zero the intrinsic semicon
ductor, same as a aielectric, has zero conductivity.
However, as the temperatura increases the electrons of the vaience-
band become excited and some of them raceive enough energy to sur-
154 Solid State Physics

Tuount the forbidden band and go over to the conduction band (Fig
ure 5 .14(b)). This resulta in free conduction electrons appearing in
the conduction band and in free electrn levels capable o accepting
valence band electrons appearing in the valence band. When an e s
ternal eld is applied to such a crystal, a directional motion o the
electrons of- the conduction and the valence bands is established,
resulting in the appearance of an electric curren.~ The crystal be-
comes conducting.

Figure 5.14 Intrinsic semiconductor: (a)at .absolute zero the

valence band is completely filied by electrons and the conduction band
is completely empty; (b)at temperatures above absolute zero part
-of the electrons from the valence band are excited to the conduction band;
liles appear in the valence band and free electrons in the conduction
band (white circles denote holes and black circles denote electrons);
c is the bottom of conduction band and v is the top of valence band.

Cond uct ion Conduct i on

band ba nd

Valonee band | ...

Val onee hand|

(a) (b)

The narrower the forbidden band and the higher the crystals
temperature the groater the number of electrons going over to the
conduction band and, eorrespondingly, the greater the crystals
electrical conductivity. For instance, for germanium with E g =
0.66 eV the concentration of the electrn gas in the conduction
band already at room temperature is as high as nl 1019 m -3 and
specic resistance is as low as 0.48 ohm-m. At the same time
for diamond with E e = 5.2 eV nx at room temperature is only about
104 m -3 and pt 10s ohm-m. However, already at T = 600 K the
electrn gas concentration in diamond increases by many orders of
magnitude and its specic resistance becomes as low as that of ger
manium at room temperature.
Two important conclusions may be drawn from the above.
(1) The electrical conductivity of an intrinsic semiconductor is
-an excited conductivity: it appears only as a result of the action of
some external factor capable of imparting sufcient energy to the
The Bae Theory of Solids 155

electrons of the valence band to move them over to the conduction

band. Such factors may be heating of the semiconductor and irradia
ron with light or with ionizing radiation.
(2) The divisin of materials into semiconductors and dielectrics
is essentially a matter of convention. Diamonda dielectric at room
tempera tureexhib ts a noticeable conductivity at higher tempera
turas and may aiso be considered to be a semiconductor. Materials
with ever increasing forbidden band widths are ow being used as
semiconductors, gradually making the divisin into semiconductors
and dielectrics irrelevant.
The concept of a hole. Let us now discuss in more detail the be-
haviour of the electrons in the valence band in wliich as a result of
transition of some of the electrons to the conduction band some free
levels have appeared (Figure 5.14(b)).
Now the electrons of the valence band acted upon bv an external
eld can go over to the free levels and establish an electric current
in the crystal. Let us nd the instantaneous valu of this current.
The current -established by one electrn moving with a velocity
is
h = - <Fi
The total instantaneous current established by all the electrons of
the valence band is

where the sum is over all the States occupied by electrons.

For a band completely lled with electrons I t = 0, since there is
an electrn with the velocity v to correspond to everv electrn with
the velocity v
Xow let us imagine that all the States in the valence band except
one with the velocity y , are lled. The total current in such a band
will be
I = a 2
= rS
= g 2 v - r ? v>
i
'

Since the rst trra in the right-hand side is zero,

I = qy 3 (5.2S)
Henee the total current of all the electrons in a valence band with
one vacant State is equivalent to a current s e t up by the motion of
one particie with a positive charge q occupying the respective State.
Such a ctitious particie is called a h o le . If we attribute to the hole
a positive charge 4 - q numerically equal to the electrn charge, we
sho.uld also attribute to it a positiva eective mass m p numerically
equal to the negative enective mass of the electrn m n , which ini-
tially occupied that State cise-to the top of the valence band, since
156 Solid State Physics

oaly in this case will the current established by boles coincide botb
ia magnitude and in direction witb the current establisbed by th&
electrons of the almost completely filled valence band.
Table 5.2 presents the room temperature electrophysical proper-
ties and characteristics of the band- pattern of three typical intrinsic
semiconductorsSilicon, germanium, and indium antimonide.

T a b l e 5 .2

Eifective raass
Semiconductor ?eV) njlm"3) P i(o h m - m ) -
mn ITlp

Silicon 1 .1 2 ~ 1 0 18 2 X 103 i .08 m Q.3~m

Germanium 0 .6 6 3 x 1 0 1S 0 .4 8 0 .5 6 m 0 .5 9 m
Indium antimonide 0 .1 7 1 .4 x 1 0 " 6 x i 0 -5 0.0 1 5 /n 0.1S/7

We see that a reduction in the forbidden band width is followed by

a drastic rise in the concentration of fres charge carriers in the semi
conductor and a drop in its specic resistance. It mav be s.een from
the two last columns of the table that the enective mass of the charge
carriers may be much smaller than the electrn rest mass.

44 Impurity semiconductors
Semiconductors no matter how pur they are always contain some
impurity atoms, which create their own energy levels termed impuri
ty levels. Those levels may occupy positions both inside the allowed
and the forbidden bands of the semiconductor at various distances
from the top of the valence band and from the bottom of the conduc-
tion band. Frequently the impurities are introduced intentionally
to impart spcic properties to the semiconductor. Let us consider
the main types'of impurity levels.
Donor levels. Suppose that some germanium atoms in a germa
nium crystal are replaced by pentavalent arsenic atoms. Germanium
has a diamond tvpe lattice in which every atom is surrounded by
four nearest neighbours bound to it by valence forces (Figure 5.15(a)).
To establish bonds with those neighbours the arsenic atom uses four
valence electrons; the fth electrn takes no part in the bonding. It
contines to move in the field of the arsenic ion, where the field is
reduced in germanium by a factor of s = 16 (s is the relative permit-
tivity of germanium). Because of a weaker field the radius o the
electrons orbit increases 16-foid (as compared with that in an isoiat-
ed atom) and its bond energy with the arsenic atam ecreases about
e2 256 times, becoming equai to E 0.01 eV. When this energy
The Band Tneory of Soiids 157

is imparted to the electrn, the electrn leaves the atom and is now
free to move in the germanium lattice thereby becoming a conduction
electrn (Figure 5.15(b)).
In terms of band tbeory this process may be describen as foliows.
The energy levels of the fifth electrn of the arsenic atom occupy
positions between the valence band and the conduction band (Fig
ure 5.15(c)). Those positions are directly under the bottom of the con-
duction band at a distance of ?d 0.01 eV from it. When an elec
trn occupying such an impurity level receives additional energy
greater than E , it goes over to the conduction band (Figure 5.15(d))

Figure 5.15 Charge carrier excitation in an n-type semiconductor:

(a)at T = 0 K the atoms of pentavalent arsenic in the germanium
lattice are in a nonionized State; (b)ionization oi arsenic atoms
and genejration of conduction eiactrons at T > 0 K; (c)energy levels
of one of the ve electrons of every arsenic atom are honor ievels;
fd)electrn transition from a donor level to the conduction band
a t T > 0 K.

Fifth electrn
.158 Soiid Sara Physics

The rsmaining positive charge (a hola) is iocalized on the immobile

arsenic atora and does not takepartin elctrica! conductivity.
The impuxities which suppiy electrons are termed donors and the
energy levels of those impurities donar levels. The semiconductora
doped with donor impurities are termed n-type semiconductors.
Acceptor levels. Let us suppose now that some of tha germanium
atoras in the germanium lattice are replaced by trivalent indium
atoms (Figure 5.16(a)}. The indium atomlacks one electrn to estab-
lish bonds with all the four nearest neighbours. It can borrovv this

Figure 5.16 Charge carrier excitation in a p-type semiconductor:

(a)atoms of trivaient indium in the germanium lattice at T = 0 K
(the fourth bond of the indium atom is unpaired); (b)at T > 0 K
the electrons can go over to unpaired bonds^ of impurity atoms creating
an indium ion and a vacant level (hole) in the valence band o germanium;
(c) energy levels of unpaired bonds of indium atoms are acceptor levels;
(d) electrn transtion from the valence band to an acceptor level
at T > 0 K result in the generation of boles in this band.

Conduction band

A c c e p to r levels
G C O C -O -O -q -

(A

Conduction band

Acceptor levels

O o O O

V aience band

(b) (d)
The Band Thsory of Soiids 159

electrn from a germanium atona. Calculations show that the neces-

sary energy is of the order of E z 0.01 eV. The ruptured hond cor-
rcsponds to a hole (Figure 5.16(b)) since it results in the ormation
of a vacant state in the valence band of germanium.
Figure 5.16(c) shows the band pattern of gennanium doped with-.
indiuxn. Directly above the valence band at a distance of E &
0.01 eV away from it there are soma empty-levels of the indium
atonas. Those levels are so cise to the valence band that already at.
relatively low temperatures some electrons from the valence band go-
over to the impurity levels (Figure 5.16(d)). The y establish bonds-
with the indium atoms and loose their ability to move in the germa-
nium lattice playing no part in the' conductivity. Only the holes-
created in the valence band act as charge carriers.
The impurities that trap electrons from the valence band are termed
acceptors and the energy levels of such impurities acceptor levels.
The semiconductors doped with such impurities are termed p-type:
semiconductors.

45 Position o he Fermi Ievel

and free carrier concentration
in semiconductors
Dependence of freo carrier concentration on the position of the Fermif
Ievel. One of the main parameters o the gas of freo carriers in a.
semiconductor is its Chemical potential, u. As applid to the electrn
and hole gases the usual term ior it is the Fermi Ievel.
As we have ascertained in Chapter 3, the Fermi Ievel in metis is-
the last occupied Ievel in the conduction band (see Figure 3.4): at
absolute zero all levels below the Fermi leve! are occupied by elec
trons and all levels above the Fermi Ievel are empty. The concen
tration of the electrn gas in metis is comparable, as regards its-
order o magnitude, to the number of States in the conduction band;
because of this the gas is degenerate and the distrifaution of the elec
trons over the States is described by the Fermj-Dirac quantum sta-
tistics. The electrn concentration of such a gas is practicallv inde-
pendent of temperature.
In the intrinsic and low-doped impurity semiconductors the elec
trn (the hole) gas is nondegenerate and the distribution of electrons-
over the States is described by the Maxweil-Boltzmann classical
statistics. For such semiconductors the free carrier concentration is=
dependent on the position of the Fermi Ievel and on temperature.
Le us nd this dependence.
Figure 5.17 shows the band .pattern' of a nondegenerate semicon
ductor. At some temperature T other than absolute zero, there re
seme electrons in the conduction band of such a semiconductor and!
150 Solid State Physics

some boles in its vlenos band. Denote tbeir concentrations by n

and p respectively. Take as the zero energy level the bottom of the
conduction band. Choose a small energy interval dE lying between
E and E 4- dE cise to the bottom of the conduction band. Since
the electrn gas in a semiconductor is a nondegenerate gas, the num-
ber of electrons da in the energy interval dE (per unit volume of
semiconductor) may be calculated with_the aid of (3.28):

d a ^ ( 2 r a ^ /2e ^ Te - E/kBTE u-d E (5.29)

In nondegenerate semiconductors l is.negative [see (3.48)]. This
.raeans that the Fermi level .in such semiconductors is below the

F.i Figure 5.17 Band pattern of

:op a nondegenerate semiconductor:
I E z is the-bottom of conduction band,
3 J 1,15 -E -JF . y tbe top of valence band,
u the Ferini level. and Es
>****** i i- the forbidden band vidth.

'ulcnce O-md

bottom of the conduction band, as shown in Figure 5.17. Denote the

distance rom the bottom of the conduction band to the Fermi level
and from the Fermi level to the top of the valence band by u and
u' respectively. Evidently
u- u' = E s , or u' = (E u.) (5.30)
where is the width of the forbidden band of the semiconductor.
To obtain the total number of electrons in the conduction band at
temperature T we intgrate (5.29) over the energy vales correspond-
ing to the conduction band, that is, from 0 to F top:
Etop
n 4 jz \ e~s/kBT E l2dE
o
The function e~z/k3T decrsases very rapidly as E grows; therefore
it is permissible to substitute innity for the upper iimit to obtain

n = 4,1 j L j VV /A3 r f e~E!k-aT E in dE

The Band Theory of Solids 151

Evaluation o this integral yields

n _ 2 ^ ln m n ^ T j 3/2 ^ J1/* B r
(5.31)
A similar calcuiation carried out for holes generated in the valence
band yields the expression
xmpkBT \ 3/2 ^ _ ( + t i)/AB T _ 2 ^ 2rcmP 5 ^ ^ 3/2 &VL'k r
(5.32)
= 2 (- * V
where mp is the'effective mass of the hole.
It follows fro"m (5.31) and (5.32) that the concentration of free
charge carriers in a band is determined by the distance from the boun-
dary of this band to the Fermi level: the greater this distance the
smaller the carrier concentration (since p. < 0 and u' < 0).
According to (5.31) and (5.32) the product of n and p for any non-
degenera te semiconductor is

np = 4 ^ (mnmp) 3 / 2 g - V * B T (5.33)

Formula (5.33) shows that for a denite temperatura the product

of the electrn and hole concentrations is a constant for the respec
tive semiconductor. This is an expression of the law of mass action
as applied to the free carrier gas in semiconductora.
Let us now discuss separately the position of the Fermi level and
the free carrier concentration in intrinsic and impurity semiconduc
tora.
Position of Fermi level and carrier concentration in intrinsic
semiconductora. In intrinsic semiconductora the concentration of
electrons in the conduction band, n{, is equal to that of holes in the
valence band, pt:
= pj (5.34)
since everv electrn that goes over to the conduction band leaves
behind a hole in the valence band. Equating right-hand sides of
(5.31) and (5.32), we obtain
o ^ 2xmnkBT y t - j i / k ^ T ... o ^ 2a mpkBT ^ 3/2

Solving this equation for p, we obtain

3
kQT In (5.35)
4

This relation determines the position of the Fermi level in intrin

sic semiconductora. At absolute zero (T = 0)
(5.26)
10885
162 Solid State Physics

that is, the position o the Fermi level is exactly the'middle of the-
forbidden band (Figure 5.18). the temperature rises the Fermi
ievel shifts upwards towards the bo.ttom of the conduction band if
mp > mni and dowmvards towards the top of the valence band if
mp < mn. In many cases, however, the shift-is so small that the Fer-
mi level in intrinsic semiconductors can be considered to be ahvays
in the m iadle of the forbidden band.
Substituting p. from (5.35) into (5.31)- and (5.32), we obtain

rtt P, = 2 ( 2?t K y ) 3r- , : W (5.37)

It follows from (5.37) that the equilibrium carrier concentration

in an intrinsic semiconductor is determined by the width of the for-

Figure. 5.18 Position of Fermi level

in an intrinsic semiconductor at
various temperatures: at T = 0 K
the Fermi level is in the middle
of the forbidden bandeas temperature
risos the Fermi level does not change
its position if mn=m.D (straight
lino 1), shifts upwards if mn<m
(straight line 2), and shifts
downwards if mn > mp (straight
line 3).

bidden band and the temperature of the semiconductor, and the de-
pendence on T and E t is very strong. For instnce, at room tempera-
ture a decrease in E g from 1.12 eV (Silicon) to 0.08 eV (gray tin)
results in an increase in nine orders of magnitude of n. An increase
in the temperature of germanium from: .100 K to 600.Iv increases n
by 17- orders of magnitude.
Using (5.37), we may rawrite the law ;jof. mass action (5.33) as
np nf (5.38)
Position of Fermi level and carrier concentration in impurity
semiconductors. Figure 5.19 shows the .change'- in the Fermi level'
position with the increase in temperature for (a) n- and (b) p-type
semiconductors.
The low temperaturerange. At low temperatures the average energy
of lattica thermaL.vibrations, k3T, .is much less than the-width. of
the forbidden band, E g, and because'of that the vibrations are inca-
pable of providing sufncient excitation of the electrons of the valence
band to shift them to the conduction.band. But this energy is enougn
The Band Theory of Solids

Figure 5.19 Variation o Fermi Isvel position with temperature:

(a)in n-type semiconductora; (b)in p-type semiconductora (Ts is
the saturation temperatura o impurity levels and T i the temperature
of transition to intrinsip conductivity).

Figure 5.20 Excitation o Figure 5.21 Excitation of

the electrons occupying a donor the electrons occupying the valeuce
level and their transition to band and their transition to an
the conduction band. acceptor level.

l*
 164
- wiiu ?a:e Physic*
to excite and shift to the conduction band th* 4
the donor levels d (Figure 5.20) and tn th* eiec^ons occupying
occupying the acceptor levels E a (Figur* 5 2 1 ) 6DCe ?and ile oies
energy 100 times less than E g. Thereforp SUCe thls re<?uirs an
tically only the impurity charge carners temP.eraures prac-
semiconductors and holes in the n-tvDearo Q ,A S ,ln. the n-type
semiconductors. " excited in impurity
Calculations show that the position of the Fermi i. ,
tempera ture ra nge is rmi *evei mside this

M - - - ^ + 4 ia rT_ s i _ ^ j
L2(2 x m nkBT )3' * J (5.39)
for the n-type semiconductors, and

I l n f - ___
- L 2 (.2.-TmpkBT)3/- -]
J (3.40)

! fc - entrations

Substituting u and u' from (5.39)and (5. 40) into f

respectvely, tre obtain the following eipressions for the conreiifri

n=YW 3,2e~
(5.41)
of eloctrons in the n-tvpe semiconductors and

(5.42)
of holes in the p-type semiconductors.
Th i.'Tipui ity ezhausilon Tange (extrinsic mn cp) a a +~
rses th* electrn concentration in tte conducen band S ?
ana hat on the donor levels decreases-the donor levels helme el
S r UIOfaCCePt0r leTe!sn semicondultlm
1 exhaustiontie electrn concentration
conanctton band of an n-type semiconductor becomes nrac'ica lv
equal to the concentration of donor impurity, .Yd- practicaily
n
(5.43)
and the hole concentration in a p-type semiconductor, to that of ac-
ceptor impurity, A a: - 1 dC

(5.430
The Band Theory o Sones loo

The exhaustion, or saturation, temperature of the impurity levels,

Ts, is the higher the higher the impuritys activation energy, S
or ?ai and its concontration. For instance, for germanium with N d =
= 1022 m3 and E = 0.01 eV the saturation temperature is ap-
proximately 30 K.
The high temperature range. As the temperature is raised still higher
the excitation of intrinsic carriers becomes more intense, the semi
conductor increasingly approaching the State of an intrinsic semicon
ductor with the Fermi level approaching the position of that in an

Figure 521 Temperature dependence

of electrn concentraron in n-type
semiconductors:
1 impurity conductivity range,
2 impurity exhaustion range,
3 intrinsic conductivity range.

intrinsic semiconductor. Until the concentration of intrinsic car

riers remains much less than iYd (n{ -Yd), the total concntration
n -i- iYd remains practically constant and equal approximately
to :Yd-
However, at sufficiently high temperatures the intrinsic carrier
concentration may not only become equal to iYd but may substn-
tially exceed it (n{ > iVd). In this case n = n{ -- ,Yd is approximate
ly and marks the transition to intrinsic conductivity. The tem
perature of such transition is the higher the greater is the width
of the forbidden band and the impurity concentration. Forgermanium
with iYd 1022 m~3 this temperature is 450 K.
Above Fj the Fermi level in an impurity semiconductor coincides
with the Fermi level in an intrinsic semiconductor and is expressed
by formula (5.35), its carrier concentration being idntica! to that
of an'intrinsic semiconductor at that temperature, as described by
formula (5.37). Figure 5.22 shows schematically the dependence of
the natural logarithm of the electrn concentration in the conduction
band of an rc-type semiconductor on the reciprocal temperature.
Three sections may be marked on the curve: 1, corresponding to
impurity conduction; 2, corresponding to the impurity exhaustion
range; and 3, corresponding to intrinsic conductivity.
Finally, it shoula be pointed out that in contrast to intrinsic
semiconductors, in which both electrons and holes simultaneousiy
166 Solid State Physics

take part in electrical conductivity, in impurity semiconductors the

conductivity is mainly due to charge carriers o one sign: to electrpns
in ihe n-type semiconductors and to holes in the p-type. Sucia car
riers are termed mjority carriers.
Apart frcm them a semiconductor always contains minority car
riers as well: zz-type semiconductors contain holes and p-type semi
conductors, electrons. Equilibrium carrier concentrations may be
conveniently denoted as follows: nn0 and pn0 are the concentrations
of electrons (mjority carriers) and liles (minority carriers) in n-type
semiconductors, p p0 and np0 are the concentrations of holes (mjority
carriers) and electrons (minority carriers) in p-type semiconductors.
Using this notation, we may write the law of mass action (5.38)
in the folowing form: -
KnoPnO = 4 PpO^pO = (5.44)
It follows from (5.44) that doping a semiconductor by an electrically
active impurity, while increasing the mjority carrier concentra
tion. should inevitably decrease the minority carrier concentration
so as to keep the product of those concentrations constant.

46 Nonequilibrium carriers
As we aiready know at all temperatures other ihan absolute zero a
process of free carrier excitation, or generation, takes place in.the semi
conductor. Should this be the only process taking.place, the carrier
concentration would continuously grow with time. However, there
is a simultaneous process of free carrier recombination. The essence
of this process is that when a free electrn meets a hole" it may occupy
it, the result being annihilation of a pair of carriers.
At anv temperature an equilibrium is established betweed the
processes of thermal carrier generation and recombination character-
ized by appropriate equilibrium carrier concentrations. Such carriers
are termed equilibrium carriers. The law of mass action discussed in
the previous section is applicable only to them.
Besides thermal excitation there are other methods of free carrier
generation in semiconductors: by lignt, by ionizing partiles, by
injection through a contact, and others. Such factors result in the
appearance of additional free carriers, ezcess carriers, as compared
with the equilibrium carrier concentration. Another term.for them
is nonequilibrium carriers. Denote the concentrations of such carriers
by A/z and Ap respectiveiv. Then the total carrier concentration wiil
be
n n0 -j- An, p = p 0 -f- Ap (5.45)
where n0 and p 0 are the equilibrium carrier concentrations.
The Band Thaory of Sods 1S7

Every nonequilibrium carrier having been born in the semiconduc

tor lives a limited time before recombining, tbe time being dier^
ent for dierent carriers. For this reason an average carrier lifetime
x is introducen, with the notation xn for eiectrcns and xp for hoies.
The carrier generation procesa is characterized by the generation
rate g, which expresses the number of carriers (or carrier pairs) gene-
rated in a unit volume of the semiconductor per second.
The recombination process is characterized by the- recombinaiion
rate R, which is eq u alto the number of carriers (carrier pairs) recom
bining in a unit volume of the semiconductor per second. For elec-
trons
dn d(An)
R n = ~ ~~dt ~ dt (3.46)
and for holes
ap _ d (Ap)
R P= - dt dt (5.47)

where n and p are the total concentrations of electrons and holes,

respectively, at a given moment of time; An and Ap are the respec
tive excess concentrations at the same moment; and the minus signs
point to the fact that recombination results in a decrease in carrier
concentrations.
Suppose that light generales excess carriers in a semiconductor
whose concentrations are An0 = Ap 0. After the light is turned off
those carriers will recombine and their concentrations shall gradually
diminish. Since every excess carrier, for instarme,' an electrn, lives
on the average xn, their recombination rate will be An!x per second,
where An is the excess carrier concentration at the moment. There-
fore the recombination rate is
R = -
d (An) Un
(5.48)
"^71

A similar relation holds for holes:

d(Ap) Ap
>P
Integrating the two equations, we obtain
An = An0 t/77,.} Ap = Apoe_l/7r (5.50)
It follows from (5.50) that An = An je and Ap = Ap je for t = x.
Henee the average excess carrier lifetime is a time intsrval during
which the carrier concentration due to recombination decreases s
-= 2.73 times.
Free charge carriers diSuse in the volume of the semiconductor
and during their lifetime t cover an average distanca L termed car
rier ..difusin length. Calculations show that L depenas on x in the
ollowing manner:
L = V Dx ( 5 .-51)
 163 Sola Sfafe Physics

where D is the carrier difusin coeffient, related to their mobility

u. by the Enstein relation
D = k3Tulq (5.52)
where q is the electrn charge.,
The transition of the electrn from the conduction to the valence
band m ay take place directly across.the whole forbidden band E s,
as shown by arrow 1 in Figure 5.23, or indirectly, rst to the impnrity
level E im (arrow 2) and then from this level to the valence band
(arrow 3). Recombination of the rst type is termed direct recombina-
tion and of the second type recombination via an impurity level.

Figure 5.23 Excess charge carrier

recombination in semiconductora:
1direct recombination, 2 and
3recombination via impurity level.

In both types of recombination the same energy E is liberated.

The only dierence is that in the rst case this energy is liberated in
one act and in the second in two acts corresponaing to the transitions
2 and 3.
The energy may be liberated in. the form of a light quantum hv or
in the form of heat (phono'ns). In the rst instance the recombina
tion is termed radiative and in the second nonradiative. Calculation and
experiinent show that direct recombination plays an essential part in
semibondctors with a narrow forbidden band at relatively high
temperatures (room temperature and above). The principal recom-
bination mechanism in wide forbidden band semiconductors is non
radiative recombination via impurity, or defect, levels. However,
under appropriate conaitions a relatively high level of radiative
recombination. may be attained even in such semiconductors.
A favourable factor is, for instance, the increase in excess carrier con-
centration and in some cases higher doping. A remarkable material
in this respect is gallium arsenide (GaAs) in which, given favourable
conditions, radiative recombination may constitute as high as 50 per
cent or higher of the total. For this reason gallium arsenide is at
present the principal material for making iuminescent iodes and
' semiconductor lasers, which fina wide practical use.
6
Eiectrical Conducivity of Solids

47 Equiiibrium stafe o electrn gas

n a conductor in the absence
of an electric field
In the absence of an electric field the electrn gas in a conductor is
in an equiiibrium State aescribed by equiiibrium distribution unc-
tions. For a degenerate gas the appropriate function is the Fermi-
Dirac distribution function (Figure 6.i(a)) and for a nondegenerate
gas the Maxwell-Boltzmaan distribution function (Figure 6.1(b)).

Figure 6.1 The Fermi-Dirac (a) and Marwell-Boltzmann (b) distribution

functions.

It may be seen rom Figure 6.1 that the graphs of those functions are
symmetric about the axis of ordinales. This points to the fact that
the number of electrons in a conductor moving in the opposite di-
rections is always the same and their average velocity in any direc-
tion is zero.This explains the fact that there is no electric current in
a conductor (in the absence of a field) no matter how many fres elec-
trons it contains.
The equiiibrium of the electrn gas is established as a result of
the interaction of the electrons with the lattice deiects, this interac-
tion being accompaniea by energy and momentum exchanges. Such
defects inelude thermal vibrations of the lattice (phonons), lattice
imperfections, and' impurity atoms. The interaction results in elec
trn scattering and their random motion in the conductor.

169
170 Solid Safe Physics

48 Electron driff irfan electric ield

When an electric field <5 is applied to a conductor, an electric current
is established in it whose density according to Ohms law is
i = a (6.1)
The proportionality factor a is termed the specic conductance of the
conductor. Its dimensions are ohm-1cm l or ohm-1m _1. Good con-
ductors have a i0 7-108 ohm-1m _1; good dieiectrics 10_l2-10~u
ohni- l/m _1. Often it is more convenient- to use specic resistan-ce
p = l/c (6.2)
instead of specic conductance.
The specic resistance is measured in ohm-m. For metis p
^ 10~7-10-S ohm-m; for dieiectrics o 1012-10u ohm-m.

Figure 6.2 Forcss acting on a irse

electrn in a conductor in which
an electric field o has been
? F =-qS established.
Electron

vd

A current owiug in the conductor is an indication of the fact that

eiectrons acted upon bv the eld begin to move in a specic direc-
tion and that their distribution function experiences a chango. Such
directional motion is termed drift and the average velocity of this
m otion drift velocity v d. Let us calclate it.
The forc with which the eld e acts on an electrn is F = q&
(Figure 6.2). Acted upon by this eld the electrn shouid be ac'cel-
erated and its velocity shouid grow continuously. But in its motion
the electrn coliides with the lattice defects and as a result of scat-
tering looses the velocity it gained in the eld. The efect of the lat
tice may be form aily reduced to the action of a resistance forc Fr,
which hinders the electrn in its motion through the lattice. This
forc is proportional to v d and is directed against it:

Fr= ~m;nvd (6.3)

with / t a proportionality factor whose physical meaning w ill be

made clear subsequentiy, and mn the electrons eective mass.
Using (6.3), we may wrte the equation of the directional motion
o the electrn in the lattice in the following form:
Vd () co 1 (6.3')
$ ~ 7 m^ d (O
'""al = ~ a
ciecfrical Condudivity of Solids 171

. We see froin (6.3') that aftar the eld had been applied the veloci-
ty of the irectional jnotion of the electrons shall rise and the elec-
trons shall be accelerated until the resistance forc Fr, which is pro-
portional to v d (f), shall become equal to the forc F with which the
eld acts on the electrn, When those forces become equal, the re-
sultant forc acting on the electrn and, accordingly, its accelera-
tion, shall vanish.
From this moment the directional motion of the electrn shall
proceed at a constant velocity
q%t (6.4)
Vd =
mn
Since the electrn charge is negative, its drift is in the direction op-
posite to <f.
The ratio of the drift velocity to the eld intensity is termed car
riel' mobility:

For electrons un <C 0, and for holes uv > 0.

According to (6.4), the drift velocity in a eld of constant inten
sity remains constant. This is possible only if the forc F q<
with which the eld acts on the electrn is compensated by the forc
F r. Would the opposite be the case the drift velocity wouid grow
continuously and could become innitely high even in weak elds.
In this case electrical conductance would be innite and the elctri
ca!'resistance would vanish.
This would be the case if free electrons moved in an ideal regular
latt-ice with a strictly periodic potential. The electrn wave that
describes the behaviour of the electrn in such a lattice would pro
pgate in it practically without attenuation similar to a light wave
propagating in an optically transparent mdium.
The causes.of a fi'nite electrical resistance are various lattice imper-
fections, which result in the aeformation of the lattice periodic po
tential and which serve as scattering centres for the electrn waves
and attenuate the directional flux of the electrons in the same way
as light waves are scattered and a ray of light is attenuated in an
opaque mdium.

49 Relaxafion tima and mean free pah

Let us now find the physical meaning of the factor t .
Suppose that as soon as the velocity of the directional motion of
th electrons attains its stationary valu yd the eld % is turned o5.
Because of the coilisions of the electrons with the lattice defects
this velocity will start to diminish and the electrn gas wiii return
172 Solid State Physi'cs

to the State o equilibrium. Such processes leading to the establish-

ment o equilibrium in a system that was previously put out- of it
are termed, relaxation processes.
Setting = 0 in (6.3') yieids an equation describing the return
of the electrn gas to the equilibrium State, that is, the relaxation
process
(6.6)

Integrating (6.6), we obtain

Vi ( t) = vi e ~t/T (6.7)
where vd() is the velocity of the directional motion o the electrons,
(t is the time after the field had been turned ofi).
It follows rom (6.7) that x characterizes the rate at which the equi
librium State of the system is established: the less t is the sooner the
excited system will return to the State of equilibrium. The velocity
of directional motion of the electrons during the time t = x decreases
e ^ 2.7 times. The time x is termed relaxation time. For pur
metis x 10- u s.
The motion of the electrons in a crystal may conveniently be de-
scribed with the aid of the concept of mean free path. By analogy with
the kinetic theory of gases one may presume that an electrn in a
crystal moves along a straight Une until it meets a lat ice defect
and is scattered. The average distance /. that the electrn' traveis
between two consecutive scattering acts is taken as the mean free
path o the electrn.
Should the electrn loose its directional velocity completely al-
ready after a single scattering act returning to the former State of
random motion, its mean free path would simply be the product of
ts average velocity and the relaxation time x, which in this case
would simply be the free transit time of the electrn:

Howrever, often it is not one but on the average v collisions with

scattering centres that are required to nullify the directional velocity
completely. Oniy after v collisions do all traces of correiation
between the initial and the final velocities of the electrn disappear.
The time during which the directional motion of the electrn
becomes randomized will in ibis case too be termed relaxation
time. However, the mean path the electrn traveis during this
time is no longer a but
l av = yx (6.9)
The quantity l is termed transport mean free path.
It follows from (6.9) that
x va.Iv (6.10)
Electrical Conduciiviiy of Soiids 173

The appearance of the drift of fres charge carriers resulting in an

electric current is an indication of the fact that the eld o changas
the distribution of free electrons over the States, that is, the form' of
the distribution function /()> since the equilibriurn distribution
function f 0{3) can not be the cause of the current. The dotted lines
in Figures 6.1(a, b) show the graphs of the electrn distribution
functions after a constant driftjvelocity had beeu established. It
may be seen from Figure'6.1 that the eSect of the eld & on the elec
trn distribution function over the States is to shift the whole distri
bution by an amount v q0 x!m.n in the direction opposite to e.
Because of that shift the distribution functions are no longer sym-
metric about the axis of ordinates and the average velocity of the
electrons in the direction of the z axis is no longer zero (in the absence
of the eld this velocity was zero). It may be easily demonstrat-
ed that- the average velocity will in this case be equal to the drift
velocity i?d.

50 Specific conducfance of a conductor

Knowing the drift velocity of the electrons ud, we can easily calc
late the current density and the specic conductance of a conductor.
To this end imagine a cylinder with a unit base built inside a

Figure 6.3 Calculating current density in a conductor.

conductor with a side equal to v and directed along the direction

of drift (Figure 6.3). All the electrons inside this cylinder will in
one second pass through the base establisning a current with a density
i = qnvd anitu (6.)
Comparing (6.11) with (6.1), we obtain
o = qnii (6.12)
Substituting u, from (6.5) and t from (6.10), we obtain
na} kv
T= (6.13)
mn v
174- Sod Safe Physics

51 Elctrica! conducfivify of nondegenerafe

and degenerate gases
Up to now we did not distinguish between the nondegenerate and
degenerate electrn gases. Let us now try and find how the State of
an electrn gas affects its electrical conductivity. To tiiis end we
shall discuss in more detail the conductivity mechanism of the non
degenerate and the degenerate gases in more detail.
ongenerate gas. In the case of a nondegenerate gas the occupancy
of the conduction hand by electrons is so small that the}' practically
never come so cise together that their behaviour is limited by the
Pauli exclusin principie. The electrons are perfectly free in the sense
that the motion of anv one of thcm is not noticeably affected by
the others. Therefore all the conduction electrons of a nondegenerate
gas play an indcpendent part in the electric current- and in the elec-
trical conductivity of the conductor. For this reason formulae (6.13)
and (6.5) for the electrical conductivity of the nondegenerate gas and
for the electrn m obility should inelude the mean free path X, the
average number of collisions v, the average velocity of motion v,
and the average reiaxation time t of all the free electrons obtained by
averaging over the ensembie as a whole.
Taking this into account, we can write the expressions for the
electrn m obility and for the specic conductance of a nondegenerate
gas in the following orm:
qx _ q Kv
(6.5')
wn u
nq1' - nq1 ). v
(6.13')
mn "l71 v
Degenerate gas. The case of a degenerate gas is different. It may
be seen from Figure 6.1 (a) that for a degenerate gas all quantum
States to the let of vP are occupied by electrons. Beca use of that the
externai eld can act only on the electrons cise, to the-Fermi leve!,'
liting them to higher vacant levels by m oving them'frbm the let-
hand regin of the distribution to the right-hand regin, as shown by
the arrow 11'. This means that in a degenerate gas only the electrons
in the immediae vicinity of the Fermi level can take part in elec
trical conductivity. Therefore one should take or the reiaxation
time in expressions (6.5) and (6.13) the reiaxation time of the elec
trons whese energy is practically equal to the Fermi energy. Let us
denote it by xF.
Substituting tf for t in (6.5) and (6.13), we obtain the following
expressions for the electrn mobility and the specic conductance
Elctrica! Conductivify cf Sods it

of a degenerate gas;
qX-pVy
u = <rc? (6.5")
mn kF
azn q-n
0= Tj>- (6.13")
mn mn
where F is the mean free path of the electrons with Fermi energy,
t?F their velocity, and vF the number of coilisions after which the
directional velocity of such electrons becomes randomized.

Figure 6.4 Mechanical model of the behaviour in an electric eld

of (a) a nondegenerate and (b) a degenera te electrn gas; arrows denote
the drift velocity in the electric eld.

Here is a rough mechanical analogy to explain the diferent behav

iour of a nondegenerate and a degenerate electrn gas in an electric
eld.
Imagine small charged balls (particles) floating on the surface
of water in a Hat horizontal vessei A and moving at random with
diferent velocities in the absence of an external eld (Figure 6.4(a)).
Now let us place t-his- vessei in an external eld <*. The resulting
efect of the eld on the ensemble of the balls. as a whole, wilL sub-
stantially- aepend on how closely the balls are packed on the surface
of water. If the number of balls is: small, so that the distances be-
tween them are large, every one of them will move freely, practically
speaking, and will not interfere with the motion of its neighbours
(Figure. 6.4(a))... In t-his instanca the motion of the ensamble, as a
whole, shal be det-ermined by the average parameters of motion of
the individual particles ; by the average velocity v, by the average
relaxation time t , by the mean fres path k, etc. *
 17S Solid Sate Physics

If the balis are packed as closely as possible, so tbat there is no

place for any more balls on the waters surface, then the motion of
the ensemble as a whole under tne action of the fieid q will be deter-
mined fay the motion of the lower layer of the particles", CC,
which separates the occupied States from the vacant States
(Figure 6.4(b)), namelv, by^ the velocity of ihose particles vc, by
the relaxation tim c> h>yr the mean irse path LC) etc. In a degener-
ate electrn gas the part of this layer fs piayed by e le ctro s cise to
the Fermi leve!, which separates occupied States from the vacant ones.

52 Wiedemann-Franz-Lorsnz aw
The transport of electric charge in an electric fieid is not the oniy
result of the presence of the electrn gas in a solidanother is heat
transport in the presence of a temperature gradient. For this reason
it would be natural to expect that the electric and the heat conducti-
vities of a solid are interreiated. This interrelation was first experi-
mentallv established by G. Wiedemann and P. Franz and theoreti-
callv explained by L. Lorenz for the case of metis. They showed the
ratio of the heat conductivity $ of a metal to its specic conductance
a to be proportional to the absolute temperature f :
ZCfo = LT (6.14)
Expression (6.14) is the essence of the Wieemann-Fzanz-Lorenz
law; the proportionalitv factor L is called the Lorenz number.
The Wiedemann-Franz-Lorenz law can easily be obtained if one
makes use of the expressions for yC and cr deriven in the electrn theo-
rv of metis. Dividing expression (4.5S) for the heat conductivity of
a metal (wdiich is praciicallv equal to its electrn component) by

--?(})=r
(6.13*), we obtain

Comptmg::(6'.15) with (6.14), we find the theoreticai valu of the

Lorenz number
.i-
- = 2.45 x IC J W-ohm-K-2
5 v ff - - <5
6-16)
Table 8.1 shows the experimental vales of L for some pur metis
at 0 C. We see that the theoreticai valu o L agrees well with ex-
periment.
Tabte ,1

Ag Au Ca uu Ir Mo ?b

(10* W-ohm-K'2) 2-31 2.23 ; . ol .i i

Elctrica! Conducfiviy of Solids ^7

In semiconductor with a nondegenerate electrn gas the heat

conductivity is not entirely due to the electrons. A suhstantial part
of it is usuaily due to lattice conductivity. However, in this case too
the electrn component of the semiconductors heat conductivity
obeys the Wiedemann-Franz-Lorenz law, the only diference being
that its Lorenz number is not determinad by (6.16) but is

L = 2 (-)J (6.17)

53 Temperafure dependence
of carrier mobify
Let us discuss now one of
electrical conductivity of
carrier mobility. We shall
temperature rnge.
High temperature range.
minant part is played by
the principal problems of the t-heory of
solids the temperature dependence of
discuss separately the high and the lo v
In the high temperature ranga the do-
electrn scattering on lattice vibrations

Figure 6.5 Thermal vibrations oE

lattice atois (a is the amplitude
of vibrations, and the shaed circle
is eective scattering cross .
section).

(phonons). Every lattice atom vibrates at random around its equi-

iibrium position (Figure 6.5) remaining inside a sphere with a radius
equal to the vibration amplitude a. The cross section of this sphere
S xa2 may be taken as the scattering cross section of a vibrating
atom: an electrn moving in a conductor can run into one of such
disks and be scattered.
Other conditions being equal, the probability for an electrn to run
into such a disk will, evidently, be propcrtional to its cross section,
.and the mean free path of the electrn will be inversely propcrtional
to that cross section:
X CC 7- T
o 3-

The energy of a vibrating atom is proportional to the-square o

the amplitude: E <X a-. On the other hand, the average energy of
1--03S3
178 Solid Sfafa Physics

atoras vibrating in a solid is proportioaal to the absolute tempera-

ture of tiie solid, T, that is, E CC T. Therefore in the higa temp'era-
ture range the mean free path of the electrons dne to the thermal
lattice vbratioas .should be inversely proportioaal to the absolute
temoerature of the body:
i CC 1IT- _ (6.18).
Tais result could nave beea obtaiaed immediately with the aid
of formula (4.37). Accordiag to his formula the phonon concentra-
tioa in a conductor in the higa temperature range is proportioaal to
T, that is, nFa CC T. For electron-phonon scattering the electrn
mean free path should evidently be proportioaal to the phonon con
centraron and, consequently, inversely proportional to the absolute
temperature T, that is, X CC l/n ph CC 1IT. On the other hand, the
average momentum of a phonon at high temperatures is so great that
a single collision of the electrn with a phonon (that is, at v 1)
already results in a practically total loss of the electrons initial ve-
locity.
Substituting (6.18) into (6.5') and (6.5") and setting v = 1, we
obtain the following enpressions for the electrn m obility:
nondegenerate gas

u O C a - ^ - = r " 3/2 (6.19)

y Tu -
degenerate gas
u'cc i n a t -1 (6.20)
Henee, in the high temperature range, where the dominant eect
is scattering by phonons(bv the lattice vibrations),|the carrier mobil
ity (of electrons or boles) in a nondegenerate gas is"inversely pro
portional to Tz- and in a degenerate gas to T. We see that in this
instance too the dinerence in the behaviour of the nondegenerate and
the degenerate gases maces itself felt.
Low temperature range. In the low temperature range the domi
nant eSect is scattering by ionized impurity atoms. The mechanism
of the scattering process is such that the impurity ions deflect elec
trons passing cise to theni and thus reduce their velocities in the
original directions. As is shown in Figure 6.6, the velocity o the
electrn in the direction of the eld was v 0 before it was deected by
the positively charged ion. After the deection it feil to y '.
The problem of charged centres aeflecting charged partiles was
first solved by E. Rutherford who investigated the scattering of
cx-particles by the nuciei o Chemical elements. Applied to our case the
formula for v obtained by Rutherford assumes the following form:
Elctrica! Conductivity of Solids 179'

where v is the electrn velocity, s the dielectric constat of the crys--'

tal, and Zq the charge of the scattering ion.
This result is quite understandable from qalitative considera-'
ti-ons. The higher the electrn velocity, their efiective mass mn and
the eld intensity rednction factor in the crystal (the greater e is},-
the less the .electrons will be deflected from their original path and

Figure 6.6 Electron q scattered

by an ionized impurity atora
(Vg is electrn velocity before
scattering, and v aiter scattering)..

the greater will be the number of collisions needed to randomize

electrn motion. Evidently, v should decrease with the increase in
the charge of the scattering ion.
On the other hand, the mean free path of electrons being scattered
by ionized impurity atoms is inversely proportional to their concen-
tration and independent of temperature.
Taking this into account and substituting v from (6.21) into (6.5')
and (6.5"), \ve obtain the following:
for a nondegenerate gas

u a ^ a r 3a f /2 (5.22)
V

for a degenerate gas

u CC CC y| = constant (6.23)
v?
Henee in the low temperature range the carrier mobility aue to
scattering by ionized impurity atoms is proportional to P /2 for
conductors with a nondegenerate gas and independent of T for con-
ductors with a degenerate gas.
Figure 6.7(a) shows the temperature dependence of u for a nonde
generate gas, and Figure fi.7(b) an experimental u(T) curve for Si
licon. It follows from Figure 6.7 that experiment, on the whole,
supports the conclusions of the theory as to the natnre of the tempe
rature dependence of carrier mobility in nondegenerate conductors.
We have aiscussed the case when in the low temperature range
the main efect is due to scattering by ionized impurity atoms. How-
l(q Cy

ever, for .very pur and very perfect metis, containing negiigibl
amounts o impurities and lattice imperfections, phonon scatterin
12*
1S0 Solid Sraia Physics

mav turn out to be the principal charge-carrier scatering mechanism

in the low temperatura range. Let us nd the temperatura ependence
of u for ths case.
For electron-phonon scattering the electrn mean free path X is
inversely proportional to the phonon concentration nDh. Since in the
low temperatura range nph CC T3 according to (4.36),

' (6.24)

Now let us determine v the average numher of collisions the

electrn should take part in to loose its original directional velocity
At high temperaturas at which the average phonon momentum
yph sgual, ocaer of magnitude, to the electrn momentum
-v 1. At iow tempera tures pph pe and as a result v can be much
.greatcr than uaitj-, being substantially dependent on tenraerature
.since p ph rises with the rise in T.

.riaure 6.7 Temperatura dependence of carrier mobility in semiconductora:

;altheoretical curves; (b)experimental curves for Silicon doped
"Nvith didarent amount3 of phosphorus.

0.001 aoo2 0.004aoos 0.01 ao

Elctrica! Conduciviy of Sods 181

Figure 6,8 shows ths variaion of the momentum of an electrn

that took part in an.elastic collision with.a phonon. The collisioa
took place at point A. Before the collision the electrons momentum
was Pe and after the collision it becam9 pe. Since the collision vv'as
an elastic one, the absolute valu of the momentum has not changed:
Pe = Fe- Only the direction has changed so' that

Ps = Pe-f"Pp!i
The change in the direction of the electrons momentum brougkt
about by the collision entails a reduction in its valu in the original

3 Figure 6.3 Caiculating the nuraber

of collisions needed to nulliy tha
electrons momentum in a giyen
direction.

direction by Apa (Figure 6.8). It ollows from ABCD that

Ape = PphSn-|-

where is the electrn scattering angle. From &AEC it follows that

sin (cp/2) = PPii/(2pe). Therefore
Pl h
2p 3

It is by this quantity that the electrn momentum in the original

direction is reducen as a result of a single collision with a phonon.
To elimnate the electrn momentum in the original direction alto-
gether the following number of collisions is needed:
Pe \
a
pe Pph i pipa

In the iow temperatura range the energy of thermal lattice vibra-

tions (the phonon gas energy) is-, according to (4.30) and (4.36),
batuca CC T4 and the phonon gas concentration is CC T'1. There
fore the average phonon energy
lattice
eph : <X7
^ph
182 Solid Sfars Physics

increases in proportion to T.. Since the phonon mornentum is

i'

_ ePh
P>h ~ ~

(v is the velocity of sound in the crystal), the phonoa momentum in

this temperature range is also proportional to T :

Pph OC i (6.25)
Thereore
vCCpphCC T -n
- (6.26)

Substituting X from (6.24) and v from. (6.26) into (6.5*), we obtain

the following expression for free carrier mobility in pur metis in
the low temperature range:

uCX l i . ce r - 5 (6.27)
vy

Figure 6.9 shows the qualitative curve of the temperature depend-

ence of u. for pur metis. In the high temperature range (above the

F ig u r e 6.9 Temperature aependenca

of free electrn m obility in pur
metis.

Debye temperature 0 ) the carrier mobility a CC T~l, in the low tem

perature range (much below 0 ) u CC T~'. In the intermedate tem
perature range a gradual transition rom the r Lto the T~adependence
t-akes place. Finally, cise to absoluta zero the thermal vibrations
become so wreak that carrier scattering by impuritv atoms and lat-
tice defeets, which are ahvays present in a metal no matter how pur
and perfect it is, becomes of primary importance. In this case the
carrier m obility ceases to depend on temperature [see (6.23)] and the
i versus T curve follows a iine parallel to the temperature axis.
Eiectricai Conducfivify of Sods 133

54 Elacfricai conductivify o pur metis

Electrical conductivity of pur metis is due to the drift of irse
charge carriers of one sign. In the absoluta majority of pur metis
the charge carriers are free electrons. However, in somc metis, such
as beryllium and zinc, the charge is carried by boles.
The conductivity, or specic conductance, of. electrn metis is
escribed by formula (6.12):
a = qnu
Since the metis are degenerate conductors, the electrn concentra-
tion in them is practically independeht of temperature. Because o

Figure 6.10 Schematic piot oi the

temperature dependence o specic
resistance of pur metis.

that temperature dependence of specic conductance is determinen

entirely by the temperature dependence of the mobility of the elec
trons in a degenerate electrn gas, as discussed in the previous sec-
tion.
Substituting a from (6.20) and (6.27) into (6.12), we obtain the
following expression for a and the specic resistance p of pur me
tis:
in the high temperature range
cr == A/T, p = aT (6.23)
n the low temperature range
a = BIT\ p = &r5 (6.29)
Here A, B; a, and b are proportionality factors.
Figure 6.10 shows schematically the dependeuce of specic ress
tanos of pur metis on temperature. In the. high temperature range
tnis dependence is representeddoy a straight iine and in the low tem
perature range by a parabola of the th negree and in the vicinity
of absoluta zero by a straight line parallel to the temperature axis.
A.more rigorous quantum mechanical calcuiation enables the coef-
184 Solid Sfafe Physics

ficients A, B , a ,b in formulae (6.28) and (6.29) to be found. Table 6.2

shows the specific conductance of some pur metis at room tempera-
ture, calculated (atheory) and measured experimentally (<Jexperiment) (in
units ol O6 ohm^m-1).

Table 6_2

Na K Rb Cu Ag Au '

iheory 22 19 20 100 90 107 <

^experiment 23 15 8 64 67 68

It ollows that for Na and K, in which the conducting electrons

are almost in a free State, the agreement between theory and experi-
ment is satisfactorv. As the atomic mass increases so does the lattice
potential and the interaction of the conducting electrons with the
lattice. This means that the free electrn approximation becomes
less valid. The result is a discrepancy between athcory and
^eiperimeat W h iC h gTOWS.
Table 6.3 shows the ratios of the specific conductivity of gold
0O at 273 K to o at low temperatures, calculated and measured.
T a b l e 6.3

273 K 87.4 K 57.8 K 20.4 K 11.1 K 4.2 K

(cr3. o)theory i 0.2645 1 0.1356 0.0060 0.0003 3 X 10~5

(a3/c)experiment 1 0,2551 - 1314 -058 -0003 3X0-*
The agreement between theory and experiment seems to be quite
satisfactory-

55 Electrical conductivity
o meta! aiioys
In metal alloys too. the carrier concentration is independen! of tem
peratura. Therefore the temperature dependence of specific conduc
tance in alloys is determined entiiely by the temperatura dependence
of the carrier mobility. Let us discuss this probiem in more detail.
Suppose that some sites of an ideal metal lattice, for instance, of
a copper lattice, with a strictly periodic potential (Figure 6.11(a))
are at random replaced by atoras of some other element, say, gold.
Since the potential of the field of an imp.urity atom is not the same
as that of the matrix-atom, the.lattice potential will cease to be stric.t-
Eiecirical Conduciiviy of Solids ' 185-

ly periodic(Figure 6.11(b)). It will be distorted by tbe disoraered iru-

purity atoms. Naturally, such distortion w ill lead to carrier scatter-
ing and to the appearance of adaitional elctrica! resistance.
It was L. Nordheim who demonstrated that in the simplest case-
of binary alloys of the solid solution type the carrier mobility due to-
scattering on lattice imperfections is described by the folowing-
approximate relation:
ual CC c(1 co) (5.30)-

where co and (1 c) are the fractional pars of the metis consti-

tuting the alloy.

Figure 6.11 Violation of lattice potential s periodicity by impurity

atoms: (a)strictly periodic potential of ideal lattice built of atoms
of one kind; (b)violation of potential's periodicity by impurity atoms
ubstituting mtriz atoms at random.

wvwvwvw
----0 9----o------------- ----- O -----9----- - - ---- i
r

Substituting ual from (6.30) into (6.12) and keeping in mina that
p ij, y.e.obtain the following expression for the specic resistance-
of a binary alloy: :-j
Pai = p:[o>;(l c)] (6.31)-

where (3 is a proportionality factor.

The unction co (1 a) has a mximum at o> = 1/2, that is wheit
the concentrations of both components are equal. Figure 6.12(a)
shows the dependence of the specic resistance of copper-goia alloys-
on the gold contents. The curve passes through a mximum corre-
sponding to a 50 percent contents of copper (or gold) in the alloy.
It also foilows from Figure 6.12(a) that pal is much greater thart-
that of the pur components. For instance, at room temperaturas-
pCu = 1.7 X 10"3 ohm-m and pAu = 1.56 x 10~a ohm-m.. where as
Pso%Cu+ 50%au = .15 X 10"3 ohm-m. This is quita natural since-
1 35 Solid Sfate Physics

the disorder in the lattics has a much more detrimental eect on

the lattice periodicity than the thermal vibrations. If, however, the
alloyed metis taken in appropriate proportions form ordered alloys,
or metaliic compounds, with an ordered structure, the lattice pe
riodicity is recovered (Figure 6.12(b)) and the resistance due to
impurity scattering vanishes practically altogether. For the copper-
gold alloys the appropriate concentrations are those~ which corre-
spond to the stoichiometric composition of CusAu and CuAu (Fig
ure 6.12(a), solid curves). This may serve as a proof of the validity

Figure 6.12 (a) dependence of specific resistance of solid Solutions

of gold and copper on composition, (b) lattics potential s periodicity
recovered in the ordering o the structure.

of the quantum theory of electrical. conductivity, which maintains

that- the cause of the electrical resistance of solids is mot the colli-
-sion of the free electrons with the lattice atonas but their scattering
by the lattice defects wThich distort the periodic lattice potentiai.
An ideal regular imperfection-fres lattice with a strictly periodic
potentiai is incapable of scattering free charge carriers and must
therefore have cero resistance. This conclusin is supported by nu-
merous experiments carried out with extremely pur metis in the
low temperature range, the relevant data being presentad inTabie 6.3:
as the degres of purity of a metal is increased its specic resistance
near absolute zero diminishes, continuously tending to zero.. W e
would like to stress that this is not the phenomenon of superconduc-
tivity which we shail discuss later but the natural behaviour of abso-
lutalv ail pur metis in the extreme low temperature range, which
is a consequence of the quantum mechanical natura- of electrical re
sstanos.
For small impurity contents one may set in (6.31) (1 ) 1.
The a pai OC co. This specic resistance is independent of temperature
Elctrica! Conductivity of Soiids io<

and aoes not vanish at absoluta zero. It is termed residual resislivity

Pra* (sea Figura 6.16). .
At temperaturas other than absoluta zero a resistivity pr due to
electrn scattering by the lattice vibrations is added to the residual
resistivity and the total resistivity becoraes

p = Pres -f Pr (6.32)
This relation expresses Matthiessen's rule, whicn speaks of the adai-
tivity of specic resistance.
Let us discuss now the temperature coeff.cient of resistivity a. As
is well known, it expresses the relative variation of the specic re
sistance of a conductor whose temperature is raisea by K. For
pur metis p = pT and therefore
i Pf
a = (6.33)
pt dT

Experiment shows a roughly to be

i
a K -1 0.004 K -!
273

(see Tabie (6.4)1. For alloys p = pras -f- pr! therefore

1 do 1 dpr
a'^ ~ ~ p ~ d T ~ Presd-Pr dT
since pres is independent of temperature. This expression may be
transformed into
i 1. a
a l = (6.34)
t-rPres/Pr Pr dT 1-rPres/Pr

where a is the temperature coefcient. of resistivity of pura metis.

It follows from (6.34) that ccal should be less. than a of a pur
metal, the less the. greater pres is in comparison with pr . Usually
pres is an order o magnitude or more greater than pT. Therefore
a ai may be an order of magnitude or less than a. o a pur metal, and
this is on the whole supported by experiment (Tabie 6.4; the data
are for room temperature).
Tabie 6.4

Bronze Constantan
Nichrome
(83% Cu, ' (30% Ni, (54% Cu,
Cu Sn Ni 13% Sn, 20% Cr) 45% Ni)
1% Pb)

a (103 K-1) 4.1 8.2 0.5 0.13 -0.004

188 Sod State Physics

However, in many cases the temperature dependence of an alloys

resistance is much more complex than that which follows from the
simple additive rule (6.32), and the temperature coefficient f resis-
tivity of some alloys may be much less than one could expect from
(6.34). More than that, it oes not remain constant in a wide tempera-
ture interval but may in some cases even become negative as is, for
instance, the case with constantan (Table 6.4) and with some other
alloys.
A high specific resistance together with a low temperature coeffi
cient of resistivity made alloys valuable-materials for the produGtion
of various wire and film resistors and variable resistors (rheostats)
widely used in diferent fields of Science and technology.

5 6 Infrinsic conducfiviiy
of semiccnducfors
The electrical conductivity of very pur and perfect single crystal
semiconductors in the. not very low temperature range is due to in-
trinsic charge carriers, that is, to electrons and boles. Suca conductiv
ity is termed intrinsic.
Snce there are two types of carriers in the intrinsic semiconductor,
electrons and holes, its specific conductance is the sum of the conduc
tivity c n q(u.n due- to free electrons, with the concentration
and the mobility un, and of the conductivity ap = qp{up due to
the presence of holes, with the concentration p{ an the m obility up.
Since n = the total specific conductance of an intrinsic semicon
ductor is
<7 = o* + C7p = qn^ (un + Up) (6.35)

According to (5.37) the hole (or electrn) concentration in an in

trinsic semiconductor is

- - S S - ni = 2 ( i l ^ ! . ) 3/27 - v (2AB^

The carrier mobility in the intrinsic conductivity range is given

by (6.19). Substituting (5.37) and (6.19) into (6.35) we obtain

cri = a0e " V (2Ab^ (6.36)

w'here a0 denotes the preexponential expression.

It follows from (6.36) that cr, a as T oo. We thus conclude
that, ir the rule (6.36) rsmains valid for infinitely high temperaturas,
cra would be the specific conductance of the semiconductor as T oo.
The temperature dependence of can be conveniently representad
in the semilogarithmic coordinates. Taking the logarithm of (6.36),
Elctrica! Conductivity o Soiids 189

we ofaain

ln<x, = l n o , ~ (6. 37)

If we plot 1/T along tb.e x axis and ln a along the y axis, we will
obtaia a straight line that cuts o2 a section ln cr0 (Figura 6.13(a))
on the y axis. Th.8 tangent of a o -tais straight line to the z axis is

Figure 6.13 Temperatura dependence of intrinsic conductivity

of a semiconductor: (a) theoretical curve; (b)experimental piots
for germanium and Silicon. -

1000 500 333 250 K

equaito Eg!(2 ka). Plotting this dependence, we can nd the constant

<r3 and the width of the forbidden bansd-'-f?r Figure 6.l3(b) shows the
experimental ln cq versus 1IT dependence for pur germanium and
Silicon. The forbidden baas as determined frena the inclination
anglas of the curves turned out to be 0.72 eV and 1.2 eV wide.
Comparing the results of this section with those of the previous one,
we see that there is the foilowing principal dirence between metis
and semiconductora.. In metis, where the electrn gas is in a degen-
erate State, the carrier concentration is practicaily indepenaent of
temperatura and the temperatura dependence of conductivity is
determined entireiy by the temperatura dependence of carrier mobi-
lity. In the semiconductora, on the other hand, the carrier gas is
nondegenerate and its concentration aepends strongiy on temperatura
[see (5.37)]. Because of that their conductivity is entireiy determined
by the temperatura dependemos of carrier concentration [see (6.36)].
190 Sola Sfare Physics

For a specic temperature the carrier concentration and the conduc-

tivity of a semiconductor are determinad by the width of its forbid-
den band. This may be seen quite clearly from the data of Table 6.5
which contains the widths of the forbidden bands and the specic
resistances of the elements of Group IV of the Mendeleev periodic
table, the elements having tbe diamond-type lattice. As the width
of the forbidden band decreases from 5.2 eV (diamond) to 0.08 eV
(gray tin), the room temperature specic resistance diminishes by
16 orders of magnitude.
Table 6.5

Diamond Silicon Germanium Gray tin

Es (eV) O.C 1.12 0.66 0.08

p (ohm-m) 1010 3 X i O3 2 X 10-

57 Impurity (extrinsic) conducfivify

of semiconductors
The temperature dependence o specic conductance of nondegenerate
impurity semiconductors, as that. of intrinsic semiconductors, is
for the most part deterruined by the temperature dependence of
carrier concentration. Because of this, the curve representing the
temperature dependence of o must at least qualitatively be analogous
to the n versus T curve, where the latter is shown in Figure 5.22.
The temperature dependence of 1-n a for an impurity semiconductor
is represented qualitatively in Figure 6.14(a). There are three dis-
tinct regions on this curve ab, be, and cd.
The regin ab lies between absolute cero and the impurity satu-
ration temperature Ts. The carrier concentration in this regin is
described by formula (5.41):]]
Z Z 'Y 2 l^ p L )312

The m obility is determined mainly by impurity and imperfection

scattering and accoraing to (6.22) is proporional to T3/-. Substituting
(5.41) and (6.22) into (6.12) we obtain
Eb/(2AbD1 (6.38)
m ^*11
where a?m is a factor that depends weakly on temperature (as. com
pared with the exponential).
Taking the logarithm of (6.38), we obtain

ln <7.3, = In a?m Ed (6.39)

'! I. 'n
-AB1
Elctrica! ConcfeicHviy o Solios 19*

la tas in crIn versas 1T coordnate systena we get a straight lina-

which makes an angleN<z,m with the 1IT axis such that tan a Im =
= EJ{2k^) is| proportional to tlia imparity ionization energy Ed.
Henee regin ab corresponds to impurity, or eztrinsic, conductivity
which is due to imparity carriers freed as the result o the ioniza
tion o impurity atoras.
The regin be lies between the impurity saturation temperatura Ts.
and the temperatura o intrinsic conductivity T{. In this ranga all

Figure 6.14 Tempera ture dependence o specic conducanos o impurity-

semiconductora: (a) theoreical curve;, (b)experimental plots
or 3iiicon containing diSerent amounts o phosporus.
 -192 Solid State Physics

the impurity atoms are ionized but no noticeable excitation of in-

trinsic carriers takes place. Because of that the carrier concentration
remains approximately constant and equal to the impurity concen
tration n & N. Therefore in this regin the temperature dependence
of the conductivity is determinad by the temperature dependence
o f carrier mobility. If the principal carrier scattering mechanism
inside this regin is the scattering on thermal lattice vbrations,
which causes the mobility to fall with temperature, then specic
oonductance will also diminish with the rise in temperature. This
is just the case shown in Figure 6.14(a). But if the principal mechanism
is "impurity or imperection scattering,- then specic conductance in
the regin be will increase -with temperature.
The regin cd corresponds to the transition to intrinsic conducti
vity. Inside this regin the carrier concentration is equal to the
intrinsic carrier concentration. Therefore the conductivity of the
semiconductor in this regin is

a a, = a 0e~E*{2kBT)

In semilogarithmic coordnales In a versus 1T this dependence is

represented by a straight line cd making an angle a , with the 1IT
axis, its tangent being proportional to the width of the forbidden
ban d: tan a, = EJ(2k3).
Figure 6.4(b) shows the temperature dependence of the conduc
tivity of phosphorus-doped Silicon. A comparison with Figure 6.14(a)
shows that in the simplest cases the theory ensures a qualitative agree-
ment with experiment.
Thermistors. The strong dependence of the resistance of semicon-
ductors on the temperature is utilized in a wide class of semiconduc
tor devices, the thermistors. They are bulk semiconductor resistors
vith a large temperature coefcient of resistivity and a nonlinear cur-
rent-voltage characteristic.
Thermistors are used in measuring temperature and power of.
utraniflifrequenev radiation, for temperature compensation in va-
rious electric circuits, for timing relays, etc. Microthermistors, which
have smail dimensions and low thermal inertia, are being used in
the studv of heat exchange processes in plants and living organisms
iacluding early diagnosis of human ilinesses. The use of a thin semi-
conducting film in a bolometer made it possible to increase its sen-
sitivity to 0 10 W. Such bolometers placed in the focus of a para-
bolic mirror are capable of detecting aircraft, tanks, ships and other
bodies that radate heat at a aistance of the order of several kilome-
ters. A highly sensitive semiconductor bolometer detectad mira
ren radiation reflectad by the moons surface.
Elctrica! Conducfivity of Solids 1S3

53 Deviation from Ohm's law.

The eteci of a sfrong field
The proportionality between the current density i and the sla
intensity % demandad by tbe Ohms law (6.1) remains as long as cr,
which enters this law as a proportionality factor, remains indepen-
dent of g.
Let ns nd what are tbe conditions in whicb ibis reqnirement is
fullled.
According to (6.5') tbe carrier mobility in nondegenerate semi-
conductors u CC ilv, where v is tbe resultant velocity of carrier mo-
tion. It is tbe sum of tbe tbermal u0 and drift v velocities:

v = v0 -r vd
For weak fields
(5.40)

the resultant carrier velocity is v & v 0 and is independent o g.

Therefore botb tbe carrier mobility u and tbeir concentration and,
consequently, tbe specic conductance a ana are independent of
. Sucb fields are term-ed weak.
Henee Ohms law, whicb requires a linear dependence of on g,
is valid onlv in the case of weak fields comolying witb tbe coadition
(6.40).
As the field g increases tbe drift velocity u rises and in fields of
high intensity may become comparable in order of magnitude witb v0.
In this case tbe resultant velocity begins to be depeadent on g and
because of tbat mobility u and specinc conductance a too become
dependent on g. Naturally, the result is a distortion of tbe linear
dependence of i on g, tbat is, a deviation from Ohms law. Fields
in whicb sucb phenomena take place are termed strong.
Calpulations show tbat when scatteping by tbermal lattice vibra-
tions is the principal scattering mechanism in strong fields

u < X e~ i/Z, o = qmiCCe~u\ = aeC C 1/2 (5.41)

In still stronger fields the drift velocity yd ceases to be dependent

on e: the so-called drift velocity saturation efieci sets in (Figure 6.15).
Since i CC ud, sucb fields are also cbaracterized by current saturation.
The current-voltage characteristic of a semiconductor becomes , is-
tinctly nonlinear.
The rise in the resultant electrn velocity in an externa! field is
equivaient to the rise in tbe temperatura of tbe electrn gas. There-
. fore this effect is known as electrn gas heating and the electro ns whose
130885
 uIJi dlJl.l.
194 Solid Sfae Physics

>4
average kinetic energy exceeds that of the lattice atoms are termed ]
hot electrons.
Strong elds can bring about not only changes in carrier mobility i
but in their concentration as well. There are several mechanisms
leading to tbat result.
Thermoelectron ionization. In strong elds not only free electrons
become heated but to a lesser extent bound electrons too. Therefore
the probability of their transition to the conduction band increases
in the same way as it would increase if the temperature of the semi-

Figure 6.15 Nonlinear carrier drift velocity in semiconductors: i CC % (Ohm s

law) in regin OA , i ex y % in regin A S , and i
is independen! of % in regin BC (saturation offdriftjvelocity).

0 V m

conductor as a whole. would be raised by an appropriate amount.

This results in an increase in free carrier concentration and in the
specic conductance of the semiconductor, a. Such phenomenon be-
came known by the ame of thermoelectron ionization. Its theory was
developed by Ya. I. Frenkel.
Impact ionization. If conduction electrons of a heated electrn
gas receiv* sncugh energy to ioniza neutral atoms lifting their elec
trons to the conduction band and themselves remaining in the conduc
tion band, then there will be an avalanche-type- increase in the free
carrier concentration, until the process is counterbalanced by recom-
bination. This mechanism of free carrier breeding is termed impact
ionization.
Electrosfcatic ionization. In high-intensity elds the transition
o the electrons from the valence band to the conduction band by
means of tunnelling through the forbidden band becomes possible.
This enect is known as the Zener etfect, or electrostatic ionization.
The probability of tunnelling and, consequently, the tunnel current
aensity increase drastically with the increase in the eld intensity
and decrease with the increase in the wiath of the forbidden
band.
Eiedncal Condudivify of Sotidi 195

Figure 6.16 shows a qualitaiive curve of the variation of specic

conductance of germaium with fieid intensity (in semilogarithmic
coordinates). Aiso shown are approximate lfmits within which

ln a Figura 6.16 Qualitative dependaacs

of specific conductance
of gexmanium on aectric eld
intensity:
I ohmio regin, 2 Frenkel regin,
3 eiectrostatic. ionization regin,
4 breakdown regin.

those mechanisms of carrier generation resulting in the increase in

electric conductivity oprate (1 , 2 the ohraic and the Frenkel re-
gions; 3, 4 the regions of eiectrostatic ionization and breakdown).

59 The Gunn effecf

It was demonsrated in the previous section that in strong celas there
is a phenomenon of nonlinear drit velocity: the drit velocity changas
not in direct proportion to the eld intensity e, the result heing a
deviation from Ohm s law.
. An interesting efect o nonlinear drit velocity in gallium arseniae
was discovered by J. B. Gunn. It became known as the Gunnelject.
Figure 6.17(a) shows the pattern of the conduction bando gallium
arsenie. It has two mnimums in the [100] direction: one at k = 0
and the othr at k = 0.8 k0 (k0 is the wave vector corresponaingto
the Brillouin zone boundary). The second minimum is = 0.36 eV
above the first. In normal condit-ions the electrons of the conduction
baa occupy the rst minimum, where their eifective mass is rrx
0.072 m and the mobilitv = 0.5 V/(m2s). .'When an externa!
eld is applied to the crystal,' the electrons drit velocity becomes-
= U jf, which increases in proportion to I (the straight line OA,
Fig. 6.17(b)). This goes on until the heated electrons accumuiate
sufcient energy to go over to the upper minimum. where their e3ec-
tive mass is much greater (m'n = 1.2 m) and the mobilit 3r much lower
(2 = 0.01 V/(m2s)). Such a transition results in a drastic reduction
in the drit velocity (because o a lower electrn mobility) and the
currant density, that is, in. the a-ppearance o the regin AB with a
15**
 196 Soiid State Physics

negative dijjereniial conduciiuity (crdl{= dit'V). After most of tile

electrons move over o the upper mnimum any further increase in %
will be accompanied by aa increase in drift velocity vi = u, and
by a proportional increase in the curren: density i (regin BC). -
The presence of a negative dinerentiai condnctivity regin on the
current-voitage characteristic of a gaiiium arsenide crystal makes it
possible to devise on the basis of the Gunn efiect ultra-high frequency
UHF) oscillators known as Gunn diodes.

Figure 6.17 The Gunn enect: (a)structure of conduction band

in allium arsenide in [100] direction; (b)variation of drift velocity
and current density with the increase in electric eld intensity.

(a) (b)

The Gunn enect was nrst discovered in 1963. In 1966 a rst com-
mercial type of an UHF generator working at a frequency of 2-3 GHz
with a power output of approximately se iOO W in pulsed operation
was produced. At an electronic instrumentation and automatics
exhibition which took place in the United States in 1968, radars
using Gunn generators to measure the speed of moving objeets vvere
disnlayed. Those radars vrere so smail that they could be carried by
"rZ A " *&&*<*
nana.

60 Phofocondudivify o semicondudors
Let us tura a ray of light of intensity 70 on the semiconductor (Fig
ure 6.18(a)). Passing through the semiconductor the light is gradually
absorbed and its intensity is diminished, Cut out.an innitely thin
layer z a t a aistance x from the semiconductors surace. The amount
of luminous energy dJ absorbed in the layer dz is proportional to the
intensity J of ne light passing through this layer and to its thick-
ness dz\
d J = - k J x (6.42)
Elacfrical Conducivity of Soiids /

The minus sign shows that the energy diminishss; the tarn for he
proportionality factox k is absorption coejcieni. For z = l ,
= dJIJ. Thus the absorption coefficient is numericaily equal to
the relative variation of the intensity of iight passing thxough an
ahsorbing mdium of uait thickness. Its dimensions are reciproca i.
to lengh (m"1).
Integrating (6.42), we obtain
j = j 0 e-k s ~ (6.43)
The light absorbed in a semiconductor may be the cause of generation
of excess caxriers, which increase the total free carrier concentration.
The arrows 1 in Figure 6.18(b) show the process o excitation of the
conduction electrons and holes in the course of intrinsic absorptioa
of light by a semiconductor. A photon with an energy hv equal cr
greater than the forbidden band width E transports an electrn
from the valence band into the conduction band. The generated elec-

Figure 6.1S Generation of free charge carriers by light (interna!

photoeSect): (a)absorption of light by a semiconducting specimen;
(b) excitation of free charge carriers from valence band, 1 (intrinsic
absorption), and from impurity levels, 2 and 3 (impurity absorption);
(c) dependenca of absorption coefficient on wavelength (2intrinsic
absorption band, 2 and 3 impurity absorption bands, threshold
of pnotoeaect, Xlm : and , thresholds of impurity
photoconductivity).

im fnn i
X, *
r J r
i

(a)

ib) (c)
1S8 Solid Sfafe Physics

tron-hole paixs are free and can take part in the semiconductors
conductivity.
To excite electrons from the levels of impurity atoms the pho.ton
energy should be hy ^ Elm, where Em is the ionization energy of
those atoras. Such. impurity levels in Figure 6.1S(b) are Ex and is,;
the process of electrn excitation from these levels is shown by ar-
rows 2 and 3.
Thus if
hv ^ E g in case of intrinsic semiconductors and
hv ^ Elm in case of impurity semiconductors (6.44)
then excess charge carriers are generated in the semiconductor and
its conductivity increases.
The process of internal liberation of electrons due to the action of
light is terrned the internal photoeffect. The additional conductivity
of a semiconductor irradiated with light is terrned photoconuctivity.
The ame for initial conductivity due to the thermal carrier excita
tion is terrned ark conductivity, for it is the conductivity of the semi
conductor kept in darkness. Light can excite excess carriers both
from the intrinsic and from the impurity levels and accordingly two
types of conductivity can be distinguished: the intrinsic and the
impurity. Using (6.44), \ve can nd the threshold of this process, that
is, the mximum wavelength of light that is still photoelectrically
active:
X0 = ch!Eg for the intrinsic semiconductors
Xlm = chlEim for the impurity semiconductors (6.45)
where c is the velocity of light.
The ionization. energy for photoconductivity in pur semiconduc
tors E g lies in the range of 0.1-5 eV, the majority having E g 1-3 eV.
The threshold for the latter lies in the visible' part of the spectrum.
Many impurity semiconductors have Em of the order of decimal frac-
tions of an electrn volt and even lower. Forthem the threshold lies
in the infra-red part of the spectrum.
Figure 6.18(c) shows a schematic dependence of the absorption
coefficient k on the wavelength X for a semiconductor with two impu
rity levels E 1 and E% (Figure 6.18(b)). The absorption spectrum of
such a semiconductor consists of three absorption bands: the intrin
sic absorption band 1 corresponding to the electrn transition from
the valence to the conduction band, and two impurity bands (2
and 3). They correspond to electrn transitions from the impurity
levels Ex and Ez to the. conduction band (Figure 6.18(b)). Light with
a < a,0 = hcEz is practicaily com-pleteiy absorbed near the surface
in a laver z 10-6 m thick; its absorption coefficient is k 106 m~l.
The impurity absorption coefficient depenas on the concentration
Elctrica! Condudivity of Sods 199

o impurit-ies but seldom exceeds k = 1Q3 m-1. The less the impurity
ionization energy E s th greatar the mximum wavelength of pho
toconductivity is according to (6.45).
The impurity photoefect is only possibie if the impurity levis
Ex and , are occupied by eiectrons, that is, if the semiconductors
temperatura is below the temperatura of impurity exhaustion, Ts.
For this reason one usually has to cool_the semiconductor to be able
to observe photoconductivity, the necessary cooliag temperatures
being the lower the greater the mximum wavelength. For instance,
gold-doped germanium has km = 9 rm and must have iiquid nitro-
gen cooling (T = 78 K); germanium doped with the elements of
Groups III or V of the Mendeleev periodic table has la = 100 um
and needs Iiquid helium cooling (T = 4.2 K).
If the intensity of light entering the semiconductor is / , the amount
of luminous energy (the number of photons) absorbed in a unit vol
unte of the semiconductor per unit time will be kJ and the rate of
carrier generation will be
g= (6-46)
where j3 is the quantum yiel, which shows the number of free carriers
generated by an absorbed photon.
In the absence of recombination the number of excess carriers
would grow continuously with time. The eect of recombination,
whose rate rises with the concentration of excess carriers is to estab-
lish a stationary State in the semiconductor when the.generation rate
is equal to the recombination rate [see (5.49)]:
g = R = A/Iq/ t (6.47)
This State is characterized by a constant (stationary) excess carrier
concentration A n 0 equal to
An0 = gxn = Jtyxn (6.48)
Since the excess carriers have practicallv the same m obility as the
equilibrium carriers, the stationary (sieady-State) photoconuctiviiy
will be
(6.49)
It follows from (6.49) that the stationary photoconductivity of
a semiconductor and, consequently, the photosensitivity of semi
conductor radiation detectors is proportional to the excess carrier
lifetxme xn. From this poin of view it is advantageous to use mafe
riis with the highest possibie xn. However, this may substantiay
increase the time lag of the photodetector.
Indeed, consider the pattern of photoconductivity decay altar
the light source had been turned or (Figure 6.19). The recombination
procsss reduces the number of excess carriers in compliance with
 *.* >i it4' n wlilj
200 Soiid Sare Physics

the law [see (5.50)]

A n = &.n0e ~l/zn-
The same will be true for the semiconductor^ photoconductivity
decay (curve BC),:

^ph = phog ^ 71 (6 .5 0 )

It follows rom (6.50) that the greater the excess carrier lifetime xn
the siower the photoconductivity decay rate and, consequently,
the greater will be the raaiation detectors time lag.

Figure 6.19 Ris8 in photoconductivity of a semiconductor illuminated

by light and photoconductivity decay ater illumination has ceased.

It may easily be demonstrated that the tangent drawn to the pho

toconductivity decay curve aph() at point t0 cuts of a section numer-
ically. equal to xn, the excess carrier lifetime. This method is often
used for determining xn.
Figure 6.19 also shows the pattern of the rise in photoconductivity
after the semiconductor had been illuminated by a light pulse (curve
OB). The photoconductivity rises gradually and reaches the pla-
teau only after a lapse of some time. In this case too a tangent to
the curve aptl () drawn at 'the origin cuts o5 a section of the straight
line AB equal to xn.
Excitons. In the act of photoconductivity the electrons rom the
valence band are transportad to the conduction band and become
free electrons. However, the process may take another course when
the excited electrn does not tear its connections with its counter-
part hole in the valence band but forms an integral system with it.
Ya. I. Frenkei proposed the tem exciton for such a system. The exci
ten is similar to an excited hydrogen atomin both cases there is
an electrn moving about a unit positive charge and the energy spec-
trum is a discrete one (Fig. 6.20). The exciton leveis are near the
bottom of the conduction band. Since the excitons are electrically
Elctrica! Conducivity of Solas 201

neutral, their appearance does not result in the generation of addi-

tional cliarge carriers. Because of that, the absorption of light is not
accompanied by photoconductivity. The present point of view is tbat
tbe formation o excitons can in some cases result in photoconduc
tivity. The generated excitons for some time wander through the
crystal. Colliding with phonons, impurity centres, or other lattica
imperfections, the excitons may either recombine or decompose .
In the rst case the ground State is restored, the excitatiou energy
being transmitted to the lattice or emitted in the forra of light guan
ta (luminescence). In the second case a pair of free carriers, an eiec-

Conduction band Figure 6.20 Exciton States in

semiconductor.

Vaience band

tron and a hole, is created. They are responsible for the phtoconduc-
tivity.
Temperature greatly affects the photoconductivity of semiconduc-
tors. As the temperature decreases the number of dark carriers drops.
The result is, firstly, an increase in the ratio of photoconductivity to
the total conductivity and, secondly, an absolute increase in photo
conductivity due to the decrease in the photocarrier recombination
rate brought about by the decrease in the dark carrier concentra-
tion (the latter efect is observed only in semiconductors witk.pre-
vailing direct recombination).
Photoresistors. The photoconductivity efect in some semiconduc
tors is widely utilized in photoresistors. Figure 6.21 shows schemat-
ically one of the types of photoresistors. It consists of a thin semi-
conducting film 2 depositen on an insulating substrate 1, of metal
electroaes 3, by means of which the photoresistor is connected into
a Circuit, and of a protective organic lm 4. The most sensitivo pho
toresistors are maae of cadmium sulfide (CdS) the photoconductivity
of which is 10M08 times higher than the dark conductivity. Also
in-wide use are photoresistors made of lead sulfide (PbS), which are
sensitiva to the far inra-red radiation. Other semiconducting mata
rais are also being used.
202 Sola State Physics

The main advantage of the photoresistors over the vacuum photo-

cells is their high light sensitivity. For instance, the sensitivity of
seienium-cadmium photoresistors is 105 times higher than that- of
vacuum photocells. A disadvantage of the photoresistors is their
time lag.
Electrophotography. The internal photoefect in semiconductors
is widely used in so-called electrophotography: or zerograpky. . The
essence of this process is as follows.
A thin film of high resistivitv semiconductor (usually ZnO) is
deposited on a sheet of paper. Before the photographic process the

Figure 5.21 Schematic representaron of a paotoresistor:

1insulating substrate, 2semiconducting film, 3metal electrodes,
i protective coating.

film is negatively charged by a gas aischarge. When an image to be

photographed is projected onto such paper, the surface charge from
the illuminated parts leaks through the film much.more readily
than from the nonilluminated parts and accordingly an electric
image of the object remains on paper after the exposition. To devel-
op the electrical image the paper is sprayed by a weak spray of spe-
cial dry paint, or toner. The particles of toner are deposited on the
negatively charged parts of the paper thus developing the image.
The image is fixed by heating the paper to the temperature at which
the toner particles melt and adhere firmly to the paper.
The main advantage of electrophotography over normal photo-
graphy is the exclusin of Chemical development and fixation pro-
cesses. This makes it possibie to increase the speed of the photograph
ic process drastically, reducing the necessary time down to about
ten seconds. However, as yet electrography is inferior to normal pho-
tography in accuracy and fineness of reproduction and because of
that its appiication is limited to cases when great accuracy is not
needed (for instance, for multiplying printed texts, caras, etc.).
The well known Soviet-made duplicator Era14 operates on this prin
cipie.
Eiactricai Conduciivity o Sods 203

Semiconductor counters. Apart from light tiie internal photoefect

can be excited by irradiating the semiconductor with particles
with electrons, ions, cc-particles, etc. Sucb particles passing througa
tire semiconductor generate free charge carriers and thus increase
its conductivity or the current in a closed constant-voltage Circuit.
Since the number of generated carriers is proportional to the num-
ber of particles which enter the semiconductor, that number can
be determined from the changes in the current. This fact enables
semiconductor counters to be devised. Such counters are usually
graduated not in units of current but directly in the number of par
ticles. To enhance the sensitivity of the counter the variations of
current owing through the semiconductor are amplified with the
aid of special electronic devices.
Semiconductor counters are now n a State of high perfection.
They are widely used in nuclear research, in space technology, in
medicine, in dosimetry, etc. They will probably play the leadng
role in radiation detection and spectrometry.

61 Luminescence
A heated body radiates energy the power and the spectral composi-
tion of which depend on the temperature of the body. This radiation
is termed thermal. Its main feature is that it is an equilibrium process.
If we place a heated body into a cavity with walis of ideal reilectivity,
a dynamical equilibrium is established between the atoms radiating
energy and the radiation lling the cavity such that the number of
atoms radiating energy and returning to the nonexcited State per
unit time would be equal to the number of atoms absorbing radia
tion and going over to the excited State. This equilibrium can be
maintained any length of time. Practically the same equilibrium
radiation will be radiated by a heated body which is not surrounded
by reecting walis of a cavity if its temperature is held constant at
the expense of energy supplie to it.
Bodies can be made to emit light not only by means of heating.
Some materiais emit light after they have been irradiated with visible
or ultraviolet light, with X rays, y-rays, electrons, or other parti
les, when placed in an electric eld, etc. The emitted light may
be in the visible part of the spectrum, although the temperature of
the emitting body is low (room temperature and below). Such coid
emission of light is termed luminescence and the bodies exhibiting it
are termed luminophors; the luminescence excited by light is termed
photoluminescence. In contrast to thermal radiation luminescent
radiation is a nonequilibrium process. Should a luminescent body
be placed in a cavity with reecting walis it would loose energy by
radiation because the radiated energy reflecten by the walis would
204 Sciid Sae Physics

Figure 6.22 Illustration of Stoke s law.

Figure 6.23 Illustration

of Vavilovs law.

be absorbed by tbe body and entirely transfonned into the energy of

them al vibrations of its atoms. Therefore luxninescence would even-
tually cease and tbe entire energy accumulated in tbe excited lumi-
nopbor would be transformad into beat.
Tbe second important feature of luminescence is its long duration
in comparison with the period of optic osciilations equal to
10_13-10 1S s. The eraission of ligbt in tbe process .of luminescence
contines at least 10~10 s after tbe excitation has ceasea. In some
instances tbe emissio.n of ligbt may continu for seconds, minutes, _
bours and even montbs after tbe excitation has ceased. In accordance
witb tbe duration of ligbt eraission, photoluminescence is divided
into pbospborescence and luoreseence. Luminescence with a dura
tion of under 10~s s is usuaily termea jluorescence and tbat witb
duration of over 10_5-10~8 s is termed phosphorescence.
Tbe rst quantitative investigation of luminescence was under-
taken some 100 years ago by Sir George Gabriel Stokes. He succeeded
in formulating tbe ollowing rule which bears bis ame: tbe wave-
length of ligbt emitted in luminescence is longer tban tbe wavelength
of tbe absorbed ligbt (Figure 6.22). Subsequent experiments bave
pro ved tbat anti-Stokes luminescence is also possible vhen tbe wave-
length of luminescence is sborter tban tbat of the excitation.
An important cbaracteristic o luminescence is its efficiency (p),
brst introduced by S. I. Vavilov. Efciency is tbe ratio of tbe total
fcscncai Conductivijy o Soiids 205

energy emitted by a boy in the process of luminescence to the energy

absorbed by the boy in the process of excitation. Figure 6.23 shows
the dependencs of t]- on the wavelength of excitation. Insde soma
wavelength interval the efciency of luminescence rises in propor-
tion to the wavelength but then drops drastically to zero. This rule
was established by 3. I. Vavilovand is known as Vavilovs law. The
absolute valu of the efciency may be as high as 80 percent or more.

Figure 6.24 Energy diagram

of uorescent Iuminophor.

Let us now discuss the mechanism of luminescence of solid crystals.

Experiment shows that crystals with perfect lattice are practically
iacapable of luminescence. To make them exhibit luminescent pro-
perties defeets should be created in their structure.'The most eective
defeets are impurity atoms. Such impurities are termed activators.
Their contents in the matrix material hardly exceeds 10~4. Materials
widelv used at present are the so-cailed phosphor crystals compiex
synthetic crystals with a defect structure possessing high lumines-
ceut properties. A phosphor crystal usually contains three compo-
nents: the matrix,.. the activator and, the solvent. The matarais fre-
quently used as matrix matarais are ZnS, CdS, CaS, etc.; as activa
tors, the heavy metis Ag, Cu, Bi, Mn, etc., and as solventa, the
iowT-meiiing salts. The spectral composition and the efciency of
luminescence depend both on the matrix material and the activator.
Figure 6.24 shows the energy band pattern o a uorescent lumi-
nophor. There are impurity levels of the activator, A, between the
filed band I and the vacant band II. When the activator atom absorbs
a photon hv, an electrn from the impurity leve! A is transportad to
the conduction band II. As a free electrn it wanders reely in the
volume of the crystal until it meets an activator ion and recombin.es
with it returning to the impurity leve! A. The recombination is ac-
companied by the emission of a quantum of uorescent light. The
decay. time o luminescence of a Iuminophor is determined by the
lifetime of the excited State of the activator atoms, which seldom
206 Solid State Physics

exceeds 10~9 s. Therefore fuorescence is a short-lived process and

tenninates almost immediatelv after the irradiation of the body has
caased. . *
For durable luminescence characteristic of phosphorescence the
luminophor must contain not only the activator atonas A but also
electrn traps L near the bottom of the conduction band (Figure 6.25).,
Such traps may be formed by impurity atoms, interstitial atonas,
vacancies, etc. The light absorbed by the luminophor excited the
activator atoms: the electrons from the impurity level A go over to

Figure 6.25 Diagrara of electronic

transitions in the act
of phosphorescence.

the band I I and become free electrons. Trapped by traps they loose
their m obility together with the ability to recombine with the acti
vator ions. To librate an electrn from the trap the energy EL should
be expended. This energy the electrn can obtain from the lattice
vibrations. The time, t spent bv the electrn in a trapped State is
proportional to ; ft may be quite large if E L is great enough.
The electrn that left the trap wanders through the crystal. until
it is again trapped or recombines with an activator ion. In the latter
case a quantum of lumiaescent light is emitted. Henee the traps.
serve as centres where the energy of absorbed photons is accumulated
so as to be subsequentlv emitted in the form of luminescent light.
The duration of this emission is determined by the time the electrons
spend in the traps.
Experiments show that not in all cases is the transition of the
electrn from an excited State to the ground State accompanied by
the emission cf a light quantum. A much more frequent resuit is the
generation of a phonon. For this reason the pairity of the phosphor
crystals must satisfy the most severe requirements. Citen a negli-
gible impurity concentration (less than lO-4 percent) completely
extinguishes the luminescence.
The quantum theory presents a simple explanation of the funda-
meental iaws of-luminescence including Stokes. and Vavilovs laws.
ziecrrica Condud-iviiy o Sqids 207

Stokes Iaw. When a luminophor is irradiated by light quanta,

tho energy of the quanta is partly spent on the excitation o the acti-
vator atonas and partly is transfonned into energy of otber lypes
(mainly heat). Denote the fraction of the quantums energy spent on
exciting the activator atom by z. When the atom returns from an ex
citad to the ground State, a quantum of luminescent light will be
emitted with its energy equal, evidently, to e. The corresponding
frequency is v = elk and the wavelength is X ch/z. Since the
energy of the incident quantum s0 > s, the wavelength X of the
luminescent light should be longer than that of the light which
initiates luminescence (X > X0) and this is what Stokes law States.
When the incident quantum collides with an excited atom, its
energy e0 = kv0 may be added to the excitation energy e causing
the generation of a quantum with energy exceeding the energy of
the one which initiates luminescence. This is the origin of anti-
Siokes luminescence.
Vavilovs 'law. Gonsider the simplest case of every incident pho-
ton e0 = hv0 generating a luminescent photon e = h\ (quantum e-
ficiency unity). Then the efficiency of the luminescence will evi
dently be equal to the ratio of the energies of those photons: r =
= e/e0. Since s = hv hcIX, it follows that
q = v /v 0 = XJX (6.51)
From (6.51) we see that the luminescence efficiency should grow
in proportion to the wavelength of the excitation. as required by
Vavilovs law. When X0 attains a valu for which the energy of the
incident quanta is not enough to initiate luminescence, the e-
ciency drops abruptly to zero.

62 Fundamentis of superconducfiviiy
Phenomenon of superconductivity. Investigating the role piayed by
impurities in residual- resistance, H. Kameriingh Onnes in 1911
carried out experimenis with uitrapure mercnry. The resulta ot those
experiments were startling: at a temperature Tcr = 4.2 K the spe-
cic resistance p of mercury fell to zero (Figure 8.26). This phenome
non became known as supercondxictiviiy. The temperatura Fcr at
vrhich the transition to the superconducting State takes place is temed
critical, or transition, temperature. For thallium, tin and leaa it
is equal to 2.35 K,- 3.73 K and 7.19 K, respectively (Figure 8.26).
Since according to Ohms law p = /i, the condition p = 0
means that for a finite current d e n s i t - y i the intensity of the eiectric
fieid at any point of the conductor is zero: = 0.
Experiments carried out at M.I.T. showed that a current ..of
several hundred amperes once inducen in a superconducting ring
continen to flow without attenuation for a whole year.
 208 Soiid State Physics

Up to now over 20 pur Chemical elements and several hundred

alloys and Chemical compounds have heen found to be superconduc-
tive. They have transition temperaturas ranging from 0.01 to 20 K.
In 1933 W . Meiss'ner and R. Ochsenfeld round that the phenomenon
of superconductivity consista n'ot only in ideal conductivity, that.
is, zero specic resistance. The magnetic eld is pushed out of the
bulk f a superconductor no matter how this eld vas established
by an externai magnet or by a current ovring in the superconductor
itself. This means that the magnetic incuction B inside the super
conductor is always zero as long as it is in the superconducting State.

Figure 6.26 Abrupt changa in specio resstanos of mercury, tin, lead

and thallium in the course o transiticn to superconducting State.

P, Kr onm-m

In other words, the superconductor is an ideal iamasmtic whose

magnetic susceptibiiity x = 1- It "will be shown in the following
chapter that normal diamagnetics have jxj <C
Henee superconductivity is a eombination of two' simultaneous
phenomena that of ideal conductivity and of ideal diamagnetism.
The supereonductive state can be destroyed by a magnetic eld.
The necessary magnetic eld Hc. is termed critical. The valu of HCT
depen.ds .on the temperature: at 7 Tor the criticalheld intensity
is zero. W ith the decrease in temperature H tT rises and is maximai
at absolute zero. The temperature dependence o i? or for lead and
tin is shown in Figure 6.27.
Fundamentis of theory of superconductivity. Despite the fact
that over 60 years have gone by since superconductivity was rst
discovered, tha microscopio theory of this phenomenon is a quite
recent development due mainly to Bardeen, Cooper, Schriefer. The
ahbreviation for it is the BCS theory. Let us discuss this theory in
general temas.
Gap in the energv spectrum of conduction electro ns in a supercon
ductor. We again recall the causes of a nite electrical resistance of
normal conductora, for instanes. of metis in a normal su te. If we
gaegiect the periodic nature of the metaTs la-ttice potential (the Som-
elctrica! Conductivity o Sods 209

merf'ld moel), we can regar it as a potential trough or the electrons,

havin0, a flat bottom and filled by electrons up to the Fermi level p
(see Figure 3.4). The kinetic energy of such electrons is given by (5.11)
F - J L - !*!
Cj~ 2 m ~ Im
Figure 6 28(a) shows once again the E versus k dependence corres-
oondin0- to (5.11): thin horizontal lines denote the occupied levels,
the solid line denotes the Fermi level E-p, and K? and kp are \va\e
vectors corresponding to this level.

Figure 6.27Temperature dependence o critical field iatensity / / cr

in superconductor.

Ha , Uf* A/m

6.4

4.8

3.2

2 4 6 8 T, K

Figure 6.23 Dependence of iree electrn energy in conduction band

o a metal on wave vector: (a)in the absence o esternal field;
(blesternal field <. imparts additional momentum to electrons,
increasing their wave vector by Sk.

140SS5
210 Solid State Physics

The application of an electric field e causes a change in the elec

trn distribution over the States (see Figure 6.1 (a)); the electrons are
transported from the-let-hand side to the right-hand side. In Fig
ure 6.28(b) this corresponds to the transition of the electrons from the
regin of negative k's to the regin of positive k's. Such transitions
are possible s-ince. there is. a practically unlimited number of unoc-
cupied States above the Fenni level which the electrons can occupy.
Should there be no limiting factors, the momentum of the conduc-
tion electrn would in time Ai grow-8 under the influence of the field o
by an amount = HAk ae Ai and a current of a density i =
= qnv< = qnl{qoim)At = (cfnlm)oAt would be established in the
conductor, its magnitude growing infinitely with time. This would
correspond to infinite specific conductance o the conductor, since

lim a = lim Ai-*- co (6.52)

03 co 771
However, should it even be possible to realiza condition (6.52),
this w-ould still not amount to ideal conductivity, which is charac-
terized, as we have seen, by the condition that the current density
for % = O is finit-e: i 0. But the condition (6.52) cannot be real
izad in any case. The factors that prevent this are the processes of
electrn scattering by latt-ice defects and, primarily, by thermai 1at-
tice vibrationsphononswhich are present down to absolute zero.
riere the main part is played by elastic scattering processes which
change the electron's momentum to one directly opposite so that
they move over from the right-hand side of the distribution curve
to the left-hand side. The corresponding transitions in Figure 6.2S(b)
are those from the regin of positive k's to the regin of negative k's.
The rate of the scattering processes is the greater the greater the field
disturbs the equilibrium distribution of the electrons over the States,
that is. the greater the displacement to the right of the distribution
curve shown in Figure 6.1(a) by a dotted line. Those processes bring
the electrn drift velocity down to the valu ud = qex?m, the current
density to i qznoxF/m, and the specific conductance to a =
= q2nx?!m , where tf is the relaxation time of the electrons occupying
levels cise to the Fermi level.
Let us make the following point important for the future. At least
two conditions should be fulfilled to make elastic transitions, w:hich
are responsible for finite electrical resistance of a normal metal, pos
sible: (a) there should be States the scattered- electrons can occupy (in
other words, the corresponding energy levels should lie in the allow?ed
energy band); (b) the States the scattered electrons are to occupy must
not already be occupied.

3 The product' q% = F is the forc with which eld % acts on the electrn.
If d i is a time interval, F d is the impulse of forc.
Elctrica! ConducHvity of Soifds 211

For a normal metal with an energy spectrum of conduction elec-

trons as shown in Figure 6.28 both those conditions are fulfilled
making the scattering' processes possible.
Can a model of the energy spectrum of conduction electrons be
built which would make scattering processes (at least under certain
conditions) impossible even in the presence of scattering centres
phonons, impurity atoms, etc.?

Figure 6.29 Energy spectrum of conduction 9lectrons in metal with

a mobile energy gap (explanation in test).

(b)

r.

*, 0 k
(d)

Apparently,. yes. Such a spectrum is shown in Figure 6.29(a).

The dierence between it and the spectrum-shown in Figure 8.28
is that in it there is an energy gap Et_s with the Fermi level E?
in the middle. The iowrer part of the conduction band is completely
occupied by electrons: the upper part above the gap is completely
free. The. band pattern looks like that of an intrinsic semiconductor
at T = 0 K whose specic conductance in case of such occupation of
the bands is zero. Since the metal retains its high specic conductance
at T 0 K, it should he presumed that in contrast to a semiconduc
tor whose energy gap (the forbidden band) does not changa its position
in an externa! field, the ESmg in the conduction band of the metal
moves in the electric eld together with the electrn distribuion, as
shown in Figure 6.29(b). During a nite time interval A i the eld &
14*
212 Solid Sfate Physics

exists in the superconductor the electrn Y wave vector increases by

an amount Ak = p^ln, = qSEtih and the energy gap Etmg shifts to
the right together with the electrn distributinn by a distance Ak.
Now let us consider the possibility for the electrn q occupying
the upper leyel of the right-Iiand subband to be scattered. The arrows
1 , 2, S in Figure.6.29(b) show the possible scattering processes: 1 is
elastic scattering resuiting in the change from k to A; 2 are tran-
sitions to the levels of the lower left-hand subband; 3 are transitions
to the levels of the upper left-hand subband. It may easily be seen
that transitions 1 are forbidden since they terminate in the forbid-
den part of the energy spectrum. namely Ee_g. Transitions 2 are for
bidden by the Pa'uli exclusin principie since the corresponding lev
els are already occupiedby electrons. Transitions 3 , although allowed,
reqnire an ionization energy equal to Etng. If the metals temperature
is low enough so that the mean phonon energy /1cop < ' e.s, those
transitions are impossible.9
Henee there are conditons in vhicli even in the presence of such
scattering centres as phonons the scattering processes limiting
the conductivity in a metal whose electrn energy spectrum has
a mobile gap (Figure 6.29) cannot take place. Accordingly, such
a metal may become an ideal conductor, just like a superconductor.
Let us look again at Figure 6.29(a). The tangent to the curve E (k)
near the top of the lower. filled. part of the conduction baud runs
horizontally (dEldk = 0) and accordingly the translational velocitv
of the electrons occupying these levels v ~l (dE!dk) = 0 although
their momentum pt and wave vector Ay = p^h are quite large.
W e encountered a similar model.when we discussed semiconductors
(see Figure 5.12); this propertv of the electrons will prove to be es-
sential in constructing the model of superconductivity.

To be exact, there are phonons with an en:pgy,-'i.ph > e.g capable of

exciting electrons from the lower subband to the States of the upper subband
even at the lowest temperature. This should cause the appearance of vacant
levels in the lower subband and of normal electrons in the upper subband.
One would think that the appearance of vacant levels in the lower subband
would bring about the scattering of electrons responsibie for ideal conductivity;
in other words, that it would in eect destroy superconductivity. Actually,
as a more detailed cons.ideration shows, a rise in temperature is accompanied
by a narrowing of the energy gap (Figure 6.29(c)) so that no vacant level capable
of scattering the electrons of the lower subband remains in it. In other words,
phonons with an energy /zcoph > not only transform a supercoaducting
electrn A into a normal one 3 ' but destroy the superconducting State of this
electrn and of the electrn A ', which was paired with .4, by exciting it to the
normal State (Figure 6.29(d)). As temperature rises the number of energetic
phonons decreases, the width of the energy gap decreases together with the
number of superconducting electrons. On the contrary, the number o normal
electrons rises. At T = Tc- the width of the gap vanishes (Figure 6.34), ail
electrons go over to che normal State, and superconductivity is destroyed.
Eiectrical Conducfivify of Solids 213

Figure 6.30(a) shows.the dependence of the density of States in the

eonduction band of a normal metal on energy for = 0 K and Fig
ure 6.30(b) the pattern of this dependence in the presence of a gap
E. in the eonduction band. Near the edges of the gap the density
of the States is higher and because of that the band made shorter bv
Ee J 2 still has enough States to accept all the eectrons of the conduc-
tion band.
Henee if we were to prove that metis can actually have an electrn
energy spectrum with a lTaP and if the causes of its appearance could

Figura 6.30 Variation of density of States of fres eectrons in metal

with the appearance of an energy gap in the eonduction band:
(a) density o States versus energy plot for a normal metal;
(b) ditto for a metal with a gap in its eonduction band.

be established. the miracle of the ideal conductivity of superconduc-

tors would generallv be unveiled. For this reason the efforts of inves-
tigators in superconductivity were concentrated on the experimen
tal verieation of the presence of such a gap in the energy spectrum
-of superconducting metis. -
At present a number of methods have been devised capable not
only of detecting the gap but also of measuring its width. One of
them is based on the study cf far infrared radiation absorption by
metis. The idea of the method is as follows. Should a superconduc
tor be irradiated with electromagnetic radiation ox continuously vary-
ing frequeney a , it would not be absorbed as long as the energy of
its quantum remained less than the width of the gap V - (of course,
if there is such a gap). Intense absorption should start at a frequeney
(or for which (CT = Ee.s increasing with the frequeney to vales
common to a normal metal. Measuring caCr, one can determine <>.*.
The experimenta convincingly proved that there is a gap in the
electrn energy spectrum of superconductor. Tahle 6.6 shows the
vales o the gap width at T. Q K for some metis together with
214 Solid State Physics

the transition temperature. We sea that the gap Et,g is very narrow,
approximately 10 3-1 0 '2 eV v i de, and that there is a direct connec-
tion between the gap's width and the transition temperature T 0r;
the higher the Tcr the greater the E6mg is.

Table 6.6

Al Sn Hg V Pb Nb

e. g (0) (1 0 3 e V ) 3 .26 li.O 1 6 .4 14.3 2 1 .4 2 2 .4

T e r (K) .2 3 .7 3 4 .1 5 4 .9 7 .1 9 9 .2 2
e.g = 3 . 5 ^ B ^ cr fiO 3 e V ) 3 .6 1 1 .2 1 2 .5 1 4.3 2 1 .7 2 7.7

After the presence of a gap in the energy spectrum of conduction

electrons in superconductors vas proved experimentally, attention
turned to the problem of the origin of this gap.

Figure 6.31 Moving electrn

polarizas the lattice and puls
positive ions a littie away from
their equilibrium sites, thereby
creating an excess positive 'charge
which attracts another electrn so
that it forms an electrn pair
with the rst electrn.

Electron pair formation. As v e already know, the formacin of

forbidden bands in the energy spectrum of semiconductors is due to
the interaction of the electrons with the periodic eld of the crystal
lattice.
Tt vou ld be natural to suppose that the energy gap in the conduction
band of a metal in the superconducting State is also due to some adai-
tional electronic interaction that appears v nen the metal enters that
state. Its origin is as follows.
A free electrn moving through the lattice interacts with the ions,
pulling them from their equilibrium sites (Figure 6.31) and creating
an excess positive charge that may attract another electrn. For
this reason apart from the usual Coulomb repulsin a forc of attrac-
tion can arise between the electrons owing to the presence of the
positive ion lattice. If.this forc of attraotiou exceeds the forc of
214 Sold State Physics

the transition temperatura. We see that the gap Et_g is very narrow,
approximately 10_3-10-2 eV wide, and that there is a direct connec-
t-ion between the gap's width and the transition temperature T cr;
the higher the Tcr the greater the E %s is.

Table 6.5

Al Sn Hg V Pb Nb

e.g (0) (10* eV) 3.26 li.O 16.4 14.3 21.4 22.4
(K) 1.2 3.73 4.15 4.9 7.19 9.22
Ee.g = 3-b B r cr(lU3 eV) 3.6 11.2 12.5 14.3 21.7 27.7

After the presence of a gap in the energy spectrum of conduction

electrons in superconductor was proved e x p e r im e n ta l!y , attention
turned to the problem of the origin of this gap.

Figure 6.31 Moving electrn

polarizes the lattice and puls
positive ious a littie away from
their equilibrium sites, thereby
creating an excess positive 'charge
which attracts another electrn so
that it forras an electrn pair
with the first electrn.

Electron pair formation. As we already know, tne formacin o

forbidden bands in the energy spectrum of semiconductor is due to
the interaction of the electrons with the periodic eld of the crystal
lattice.
It would be natural to suppose that the energy gap in the conduction
band of a metal in the superconducting state is also due to some adai-
tional electronic interaction that appears when the metal enters that
state. Its origin is as follows.
A free electrn moving through the lattice interacts with the ions,
pulling them from their equilibrium sites (Figure 6.31) and creating
an excess positive charge that may attract another electrn. For
this reason apart from the usual Coulomb repulsin a forc of attrac-
tion can arise between the electrons owing to the presence of the
positive ion lattice. If.this forc of attraction exeeeds the forc of
Elctrica! Conduciivity of Soiids 215

repulsin, it wili be advantageous frota the viewpoint of energy for

the electrons to associate into couples. These became known as Cooper
pairs.
The ormation of a Cooper pair results in the reduction in the energy
of the two electrons by the amount equal to the binding energy of
the electrons in the pair, E b. This means that a conduction electrn,
which in a normal metal had a mximum energy ET at T = 0 K
(see Figure 6.28(a)), in the superconducting State has an energy E J 2

Figure 6.32 Schematic model

o f a Cooper pair.

less (the energy per pair'being Eh less) since this is the energy that
must be spent to break up the pair and move the electrons to the
normal State. Theretore in the one-electron spectrum there must be
a gap of Et'%= 2Eh between the upper level of a coupled electrn
and the lower level corresponding to the normal state. which is re-
quired for superconductivity. It may easily be seen that this gap is
mobile, that is, it can shift in an externa! hela together with the
electrn distribution curve over the States.
Figure 6.32 is a schematic representation of a Cooper pair. It con-
sists of two electrons oscillating about the induced pos.itive charge,
which in some ways resembles a helium atom. Each electrn of the
pair may have a large momentum p? and a large wave vector k?;
the pair as a whole (its centre of masses), on the other hand, can re-
main stationary having zero translational velocity. This explains a
paradoxical property of the electrons occupying the upper levels of
the filled part of the conduction band in the presence o a gap (see
Figure 6.29(a)). Such electrons have very large p's and k's (p ~ p?
and k ^ k?) and a translational velocity v CC E/ak = 0. Since the
central positive charge is induced by the moving electrons themselves,
the Cooper pair acted upon by an external field can freely drift in the
crystal, the energy gap moving with the electrn distribution, as
shown in Figure 6.29(h)..Hence the conditions for superconductivity
are fuilled from this point of view as. well.
 216
Soiid State Physics

However, not all the conduction electrons can fonn Cooper pairs
Since this process involves a change in energy, only electrons tha
can change their energy can form pairs. Ouly the electrn* in a

Figure 6.33 Estimatin? the number

of electrons capable of orming
Cooper pairs.

the ends of whose momentum vectors p? lie within the area of a given
ring make up a group every member of which has the same absoluta
valu of momentum p?. The number of electrons in each group is pro-
portional to the area of the respective ring. Since the area of the
ring.rises with cpso does the number of electrons in the band. Electrons
of anv group ntay form pairs. but the mximum number of pairs
will be formed b'y the electrons of the more numerous group. The
latter is made up of electrons whose momenta are equal in magnitude
and opposite in direction. The ends of the vectors pF of such eiec-
trons are not limitad to a narrow band'^t- spread over the entire
Fermi surface. Those electrons are so numerous in comparison with
any other electrons that practically only one group of Cooper pairs
is formedthat made up of electrons whose momenta are equal in
magnitude and opposite in direction.
A remarkable pecuiiarity of such pairs is the ordering of their
momenta: the centres of ma.sses of all the pairs have identical mo
menta, being zero when the pairs are at rest and nonzero when the
pairs move in the erystal. The result is a rather rigid crrela tion be-
tween the motion of every single electrn and the motion of all the
other electrons bouna into pairs.
The electrons move like mountain-climbers tied together by a
rope: siiould one of them leave the ranks due to the irregularities
Electrical Conducrivity of Solids 217

of the terrain (causea by the thermal vibrations of the lattice atoms)

bis neighbours would pul him back .10 This property makes the-
ensemble of Cooper pairs little susceptible to scattering. Accordingly,
should the pairs acted upon by some external forc be set in motion,
the current established by them would continu to ow in the super
conductor indenitely even after the factor that brought it to lile-
ceased to oprate. Since only the electric eld % can play the role
of such a factor, this means that in a metal in wdiich Fermi electrons'
are bonded into Cooper pairs a once excitad electric current i can
remain unaltered even after the field has vanished: i = constant at
o = 0 . This proves that the metal is actually in the superconducting
State and that its conductivity is ideal. Such State of the.electrons
may be roughly compared with the State of a body moving without
friction: the body having received a momentum can move incle-
finitelv and its momentum remains constant.
In the above we compared a Cooper pair to a helium atom. H ov-
ever, such a .comparison should be treated very cautiously. As vas
alreaay stated. the positive charge is not exactly constant and sta-
tionary. as in the case of a helium atom, but is induced by the moving-
electrons themselves and moves with them. Moreover. the binding
energv of the electrons in a pair is many orders of magnitude less
than their binding energv in the helium atom. According to Ta-
ble 6.6 the binding energv of Cooper pairs E = 10_3-102 eV. the-
corresponding valu for the helium atom being E = 24.6 eV. Be-
cause of that the aimensions of the Cooper pair are many orders of
magnitude larger than that of the helium atom. Calculations show
the eective diameter of a pair to be L zz ( 10~- i 0 -8) m; another
term for it is coherertce lensth. There are about 106 centres of masses
of Cooper pairs inside the effective volume [J of one such pai-r. For this
reason such pairs cannot be regarded as separately existing qnasi-
molecules. n the other hand, the accompanying colossal overlap-
ping of the wave functions of numerous pairs enhances the electrn
pairing enect so that it manifests itself in macroscopic proportions.11

10 Ya. I. Frenkc: Introduction to the Theonj of Metals, G I7TL, Moscov

(1950) (in Russian).
11 There is another analogy, a very proound one at that. betveen a Cooper
pair and a helium atom. The essence of it is that an electrn pair constitutes-
a system with integral spin, the same as the helium atom He does. It is a kiiovn
fact that the superfluidity of helium may be considerad as the result of a peculiar
eiect of bosons condensing on the lovest energv level. From this point o view
superconductivity may he regarded as superfluidity of the Cooper electrn
pairs. The analogy is a still more far-reaching one. Another helium isotope He,.
whose nucleus has a half-ntegral spin, does not e.vhibit superfluidity. But a most
striking new discoverv is that in the lowest temperatura range the atoms o f
3He can form pairs quite like the Cooper pairs and the liquid becomes superfluid-
One is justied in saving that the superfluidity of 5He is a sort o .superconduc
tivity of its atomic pairs.
218 Solicf Srata Physics

Henee electrn pairing is a typical coilective enect. There are no

Torces of attraction actlng between two isolated electrons to make
their coupling possible. In efect, the entire ensamble of tbe Fermi
electrons together with the lattice atoms takes part in the formation
of a pair. Because of that the binding energy (the gap width e.?)
too depends on the State of the .electron-adom ensemble as a whoie.
At absolute zero wheh'all the Fermi electrons are in pairs the width
of the gap is at its mximum, 2?e.g(0). The rise in temperature is ac-
companied bv the generation of phonons capable in the act of scat-
tering of transmitting energy to the electrons suicient to break up

Figure .34 Variation of the energy gap Ez.g(T).

the pair. At low temperatures the concentration of such phonons

is not large and the breaking up of a pair is a rare event. The disap-
pearance of some pairs cannot, naturally, lead to the disappearance
of the gap for the remaining pairs but makes it somewhat narrower,
with the edges of the gap drawing' closer to the Fermi level (see
Figure. 6.29(c)). W ith a further rise in temperature the phonon
concentration grows very rapidly, their mean energy growing as
well. The result is a steep rise in the bear-up rate of the pairs and,
accordingly, a drastic decrease in the gap width for the. remaining
pairs. At some temperature Tcr the gap disappears altogether (Fig
ure 6.34), its edges merging with the Fermi level and the metal return-
ing to the normal state. The temperature For is just the critica!
transition temperature that was mentionea at the beginning of the
-section.
It follows then that the critical temperature for the transition of
-a metal to the superconductive state shoula be the greater the great-
er the gap width at absolute zero, e.?(0), is. The 3CS theory gives
the foilowing approximate dependence of Ee,s(0) on Tar:
E&.s (0) = 3.5 sTcr (6-53)
which is in good agreement with experiment (see the last line in
Table 6.6).
Siecrica! Conducfivity of Soiis 219

Behaviour in an externa! electxic eld. Connect a Ion" cylindrical

superconducting specimen into an electric Circuit, as snown in Fig
ure 6.35(a). As the Circuit is closed, a homogeneous electric eld is
established in the specimen, % Vil, where V is the voltage across
the specimen and l is its length. Acted upon by the eld e all Gooper

Figure 6.35 Behaviour of superconductor connected into an electric

Circuit (explanation in test).

pairs contained in the specimen will start to move against the eld
with the same acceieration
2q%[t) q% (t)
^ 2m 2/71
where 2q is the pairs charge, and 2m its mass.
The current density in the superconductor will start to grow:
di
t 2 ( f ) 4 r = 5n- = '
where v is the drift velocity of the pairs, n j 2 their number, and
n3 is the concentration of superconducting electrous.
The current i generares a solenoidal magnetic eld H in the super
conductor (Figure 6.35(b)). Since i grows with time so does the
magnetic eld H. This results in the appearance of an incuced elec-
tric eld e Ia directed against e and of an induced current ln directed
220 Sod Sfae Physics

against i (Figure 6.35(c)). The current iln generales a magnetic

eld H la airected against H (Figure 6.35(a)). The resuit is the com-
pensation of the eid e inside the superconductor by the eld e ln
and of the eld H by the fieid H la so that the resulting electric eld
in the specimen % O (Figure 6.35(f)) together with the resulting
magnetic eld jTlnt = 0. For su'ch compensation to continu it is
necessary, rstly, that the current of Cooper pairs, i, be maintained
in the specimen indenitely after the end of transient processes.
To this end the specimen s resistance should be zero and this is so
if the specimen is in the superconducting State. Secondly, this
current should be localized in a thin surface layer X of the supercon
ductor (Figures 6.35(e, f)), for in this case it does not genera te a mag
netic eid inside the specimen but generates an externa! eld H
in the surrounding space just as a normal current does.
Henee after transient processes come to an end the following sta-
tionary state is established in the specimen:
6 0, = constant, i7lnt = 0
The rst two conditions correspond to ideal conductivity and the
third to ideal diamagnetism.
In the stationary State Cooper pairs move without acceeration
(ree motion) with the same drift velocity vd = p<g!m, where
is the momentum accumulated by the pair during the time the
Circuit was closed. The current set up by them is

As has already been stated before, this current is localized in a thin

surface layer X of the sarnpie. the magnetic eld of the current being
concentrated in this layer (Figures 6.35(e, f)). The parameter X is
termed the penetration deph. Theory gives the following expression
for this parameter:
* /"
K= (6.54)
V ^~ns.io
where u0 is the permeability of free space. For dierent superconduc-
tors X lies in the range from 4.1-S to 10-7 m.
It follows from (6.54) that at T Tcr when the concentration of
superconducting electrons vanishes becomes infinite. Physically
this means that as the metal returns to the normal state the layer
X in which the magnetic eid is localized spreads across the entirs
cross section of the sample and ideal diamagnetism vanishes.
Behaviour o superconductor in magnetic eld. Now let us suppose
that a magnetic fieid is set up in space containing a cylindrical
superconducting sample (Figure 6.36(a)). The eld induces a soienoi-
dal electric eld in the sample which sets up a solenoidal electric
Slectricai Conductivify of Solids 221

current in. The current ila crea tes a magnetic fred H la (Figure
6.36(b)) directed against the external eld and compnsalas it.
The eld Hln in its. tura generales the current % which compn
sales the current im (Figure 6.36(c)). The overall eect is the compensa-
tion of the externa! eld f 6xt by the eld fln and of the current
hn by n (Figure 6.36(d)). The total induced current flows in a thin
surface layer a in which its magnetic eld that compensates the

Figure 6.36 Behaviour of superconductor in magnetic eld (esplanation

in test).

external eld is localized. After the termination of transient proc

esaos the same steady State is established in the sample as was the
case when an em was applied to it:
= 0 , = constant, F/lnt = 0

Naturally, this State can be established only if the current m in

duced during the time the magnetic eld was switched on contines
indefinitely, that is, if the sample is in the superconducting State.
Destruction of superconducting State by eids. Before the eld
is switched on, the momenta of the electrons making up a pair are
equal in magnitude and opposite in directiou, the momentum of
the centre of masses of the pair is zero. Acted upon by the eld e
every pair as a whole attains some. drit velocity vd and increases
its energy by the amount
nii
mv
(

If this energy exceeds thebinding energy of the pair Eh = ES.J2,

the paixs start to hreak up and the superconducting State would
start to vanish. For this reason the condition for the transition ot a
metal from the superconducting to the normal State may be written
as follows:
2 22 Solid State physics

From here we can easily nd the drift velocity

and tlie current density

i / " -^e.g
cr = qnsvd = qn3 \ (6.5o)

for the case when superconductivity in a metal begins to vanish.

Setting ES'g = 5 X 104 eV. q = 1.6 X 10-19 C, and ns = 10~lsc m '3,
we obtain vd = 1.8 X 1Q4 m/s and iCT = 2.5 X 105 A/'cm2.
With. account taken of the fact that the current i sets up a mag-
netic field on the surface of the sample of intensity
H Ai (6.56)
(/. is the penetration depth of the magnetic field into the supercon
ductor) condtion (6.55) may be formulated as follows: the super-
conducting State of the sample w ill be destroyed when the intensity
of the magnetic field on its surface w ill attain the following critical
valu:
H /-cr (6-57)
Su'ostituting a from (6.54) and cr from (6.55) we obtain

f f cr = (/
^
]6.58)
--L0
For E* = 5 X 10-4 eV and ns = 1018 cm3, H zr m 104 A/m (or
100 Oe),
Thus when the magnetic field on a supereonductors surface attains
the critical valu H zr determined by condition (6.58) the supercon
ductivity is destroyed. Since the gap width substantially depends
en temperature [see (6.53)], icr and / f cr sliould also depend on tem-
perature, their vales decreasing with the nse in temperature. The
BCS theory gives the following dependence of Ffor and / cr on absolute
temperature:
iFor = i7Cr ( 0 ) [ l - ( F / r cr)2] (6.59)
I CT= r:dH cr for d^A (6.60)
where Tcr (0) is the critical magnetic field intensity at T = 0 K,
and d the diameter of the specimen. There is a satisactory agree-
ment betweon those relations and esperiment.
Practical uses of superconductivity. The field of practical appiica-
tion of superconductivity widens from year to year. First, it serves
as a basis for superconducting magnets. Such magnets are solenoids
or electromagnets wh a ferromagnetic core with the winding nade
Elctrica! Conductiviiy o? Solids

of a superconducting material. Calculations show that to establish

a magnetic field intensity of 8.105 A/m ( 105' Oe) in a solenoid
of a diameter of one metre a superconducting magnet requires 104
times less power than an ordinary electromagnet. Recently super
conducting magnets using N bsSn have been fabricated which enable
magnetic elds up to 6.108 A/m ( 7 X 104 Oe) to be produced.
Superconductivity is also being utilized to design modulators
(converters of weak constant current into an audio-frequency current),
rectifiers for the detection of modulated high frequency oscillations
in which the use is made of the nonlinearity of the superconductors
resistance in the transitional regin, commutators (noncontact
switches utilizing the phenomenon of superconductivity), cryotrons
(superconducting four-poles in which the magnetic eld at the input
Controls the output resistance), persistors and persistrons (super
conducting memory eiements for memory devices), etc.
Of highest practical importance is the problem of high temperatura
superconductivity. Of all the known materials the highest transition
temperature 'is that of the alloy (Nb3A l)4 -f- N b3Ge whose T0P
20 K. To obtain such a temperature liquid helium is needed.
What are the prospects for developing materials with higher critical
tempera tures?
The BCS theory demonstrates that Tor is directly related to the
forc of attraction bctween the electrons in the superconductor
and is determined from the followng approximate expression:
r cr = -/* (6.6i)
where 0 is the Debye temperature, and g is a constant (not exceeding
1/2 and usuallv less) dependent on the attraction forc between the
electrons. For g 1/3 the mximum critical temperature obtainable
with a material with 0 = 5 0 0 K would be Tcr = e~3 =!0.05 =
= 25 Iv. Naturally, this estimate is a verv rough one but still
it makes it clear that it is impossible to obtain high temperatu-
superconductivity (Tcr > 100 K) with the electrn pairing mecha-
nism discussed above.
Simultaneously with the development of superconducting materials
with increased Tac utilizing the electrn pairing effect via positiveiy
charged lattice ions, an intensive search for other mechanisms of
electronic interaction capable of more efficient attraction and, cou-
sequently, of providing superconductive materials with substantiai-
ly greater transition temperatures Tar goes on in the laboratorios
throughout the world. Should this search be successful and should
such materials be produced, the importance of this discovery would
prove to be comparable to the development of controlled therme
nudear fusin.
7
Magnetic Properties of Solids

63 Magnetic field in magnetic materials

Let us place a homogeneous body of volume V into a uriiform mag-
netic- eld of intensity H and induction B0 = u0H. Acted upon by
'the eld the body becomes magnetized obtaining a magnetic mo-
ment M. The ratio of the magnetic moment to the volume of the
body is termed magnetization J m:
Jm = MIV (7.1)
and when magnetization is not uniform it is equal to
J m = dMIdV (7.2)
Magnetization is a vector; in uniform magnetic bodies J m is
either directed parallei or antiparallel to R. The unit of magnetic
moment in the SI system is A-m2 and that of magnetization A/m.
The ratio of magnetization J m to the magnetic eld intensity
H is termed the magnetic suscsptibility yj.
X= SJH ' (7-3)
It may easily be seen that y is a dimensionless quantity. From
(7.3) \ve get ;#-
(7.4)
A magnetized body placed in an esternal eld establishes its
own eld which in isotropic magnetic materials away from its
esternal boundaries is directed either parallei or antiparallel to
the esternal eld. Denote the esternal eld induction by 5 0, the
proper eld induction by and the resultant induction by B. For
uniform magnetic materials B is an algbrale sum of B 0 and Bg.
3 = 5 0 -r Bl (7.5)
Esperiments show that
= V-oJm == 1 B0 (7.6)
"224
Magnetic Properties of Solids 225

wherefore
B (1 -f x) B0 (7.7)
The quantity
u = 1+ x _ (7-3)
is termed magnetic permeability.- It follows from (7.8) that
X= ^ - l (7-9)
Substituting (7.8) into (7.7), we obtain
B u, B0 = ,uuif - (7.10)
The unit of eld intensity H in the SI system is A/m and that of
induction B the tesia (T).

64 Magnetic properties of solids

All materials may be divided into three large groups according to
the absolute valu and the sign of their magnetic susceptibility
(Tabla 7.1): diamagnetics, paramagnetics and ferromagnetics.
Table 7.1

Diamagnetics x = u 1 Paramagnetics 7 = O 1 Ferromagnetics' 7. = u 1

Bi 18 X 10~5 CaO 580 X in- .Fe 1000

Cu 0.9 X 10~5 FeCL 360 X ID-3 Co 240
Ge 0.8 X 10-5 ISS0 4 120 X 10-5 Ni 150
Si - 0 .3 x to- Pt 26 X iO-5

Diamagnetics and paramagnetics. For diamagnetics (| % | <71)

% is negative and independenj| ,pf... the intensity of the externa!
magnetic eld and of temperature. Such materials are magnetizad
in the direction opposite t'o the direction of the externa! eld and
beca use of that they are pushed out of the regions of the highest
eld intensity.
Paramagnetics aiso nave |%j -< 1, but contrary to diamagnetics
X is positive. Such bodies are magnetizad in the direction o the
external eld and are arawn into the regions of mximum E.
Figure 7.1 (a) shows the dependence of / m on the eld intensity
for diamagnetics, 1, and for paramagnetics, 2. In both cases / :I1 is
pcoportional to H, this being an indication of the independence
of x of H. However, for paramagnetics this is observed only in reia-
tively weak elas at high temperaturas; in stiong elds and at low
temperaturas the plot J m [E) asymptotically approaches the lira it
i
2 26 Solid Srae Physics

valu / s, which corresponds to magnetic saturation of the para

magnetic (Figure 7.1(b)). Besides, y of paramagnetic bodiesisdepen
den! on temperature. This dependence was first studied by Pierre
Curie. He demonstrated that
X = C IT (7.11)
where T is the absolute temperature of the paramagnetic, and C is
a constant dependent on its natura. The term for it is the Curie
consiant and for expression (7.11) the Curie lavo.

Figure 7.1 Magnetization 7 m versus the magnetic ficld intensity H: (a)dia-

magnetics (2) and paramagnetics (2) in Weak and mdium fields-
at normal and high temperatures; (b)paramagnetics at low temperaturas
(or in very strong fields).

J,m J,m

H
J

(a )

Ferromagnetics. y o ferromagnetic materials. a typical represen-

tative of which is iron. is also positive but immeasurably greater
than that of paramagnetics. Besides, y depends on H. Apartirom
iron this group ineludes also nickel, cobalt, gadolinium, dysprosium,
holmium, erbium and some alloys.
The rules governing the magnetization were first investigated by
the Russian physicist A. G. Stoletov. Figure 7.2 shows the depen
dence of B (a), of magnetization / m (b), and of susceptibiliiy
y (c) on H for soft iron. B and Jm at first rise quickly with the magne-
tizing field but then the rise slows down and at some E a cise
to the mximum valu of / 5 is attained; any further slow- increase
in induction is due solely to the increase in H. This State correspond
to technical saturation of the ferromagnetic: as this State is ap-
proached, y 0.
A carefulstudy of the magnetization curve shows that as E in
creases J m rises not coninuously but stepwise. This is espepially ap-
uarent in the regin of the steep rise o the magnetization curve. Fig
ure 7.2(b) shows a magnified section of the magnetization.curve (en-
Magnetic Properfies of Solids 127

Figure 7.2 Magnetiza tica of erromagnetics: (a)induction B versus

eld inteasity E] (b) magnetization J m versus eld intensity H
(the right-hand sida shows a magnifi^d section of the magnetization curve)-r
(c)-magnetic susceptibility versus fied intensity E.

closed in a circle). This section consists of a iarge number of steps cor-

cesponaing to individual junaos accompanying the variation of / rt)
with a continuo us rise in H. The stepwise na ture of the magnetiza-
tion process was discovered bv Heinrich Barkhausen and became-
known as the Barkhausen effect.
Figure 7.3 shows the plot of a complete remagnetizing cycle of a-
ferromagnetic. It may be seen from Figure 7.3 that uring the remag-
netization the variation of B lags behind the variation of H and
vvhen H = 0 is not equal to zero but to ?rss. This lagging of B'
behind H has been named magnetic hysteresis and the induction-
S res residual magnetic induction, or remanence. To remove it a demag-
netizing eld Ha termed coersive forc should be applied. The closed:
curve A B resf f ZA'BrSSII'c A, wnich describes the remagnetizing cycle,
is termed the hysteresis loop. The area of this loop is proportional to-
the work that should be expended to remagnetize a ferromagnetic of
5*
228 Sod Sraia Physits

uait volume. In the course of remagnetization this work is completely

transformed into hea. Therefore when the ferromagnetic is remag-
netized many times in succession its temperatura risas, the eect
being the greater the greater the area of the hysteresis loop.
Ferromagnetic materiais are ciassied as soft an d * hard",
or having a high coersive forc. Soft magnetic materiais used for
manufacturing cores of electric motors and instrumenta _have a
low coersive forc and high permeability. The best alloys of this type
(Supermalloy, for instance) have p,s as high as 105, saturation induc-
tion B s 1 T and a coersive forc H a of only 0.32 A/m. Their hyster-
esis-loop area is so small that their remagnetization losses are some
500 times less than those of soft iron. Hard magntic materiais are
characterized by a high coersive forc and by high residual magneti-
zation. For instance, Magnico used for manufacturing permanent
magnets has Hz 5 x 103 A/m and ?re3 = 1.35 T.
YVhen ferromagnetic materiais are heated their magnetic proper-
ties become less pronounced: there is a drop in the vales o x. ti,.
/ m, etc. There is a temperature 0 Cfor every ferromagnetic at which
it looses its ferromagnetic properties. This is known as the ferro
magnetic Curie point. By way of an example we shall show the Curie
points of some ferromagnetics.
Cobalt Iron Nickel 30% Permalloy
0 C (C) . . . 1150 770 360 70

Above 0 Cferromagnetics turn into paramagnetics with their charac-

teristic linear dependence of /% on T (Figure 7.4), which is quite
well represented by the following relation known as the Curie-
Weiss laur.
X = C/(T - 0) _ (7.12)
with C the Curie constan! and the paramagneiic Curie point (it
is somewhat higner than 0 c).
Figure 7.5 shows the temn^^uEe dependence of mximum magneti
za t ion. of iron, nickel, and cobalt. The ratio Z76c is plotted along
the x axis and the ratio J 5{T)/JS (0) along the y axis. In such rela-
tive coordiuates the dependence of magnetization on temperatura
is aescribed by the-same curve for all ferromagnetics. As tempera-
ture rises magnetization drops becoming practically zero at the
Curie point.
Ferromagnetic single crystals are characterized by anisotropic
magnetization. Figure 7.6 shows the magnetization curves of iron
(a) and nickel (b) crystals in the [111], [110], and [100] directions.
It follows from Figure 7.6 that there are directions in the single
crystai in which it is easier to magnetiza the crystal and obtain
magnetic saturation at relatively small vales of magnetic Seid
intensitv. Those directions are termed directions of easy magneti-
Magneiic Properfies of Soiids 229

Figure 7.4 Temperatura dependence

173: o magneiic susce ptihility
of erromagnetics:
0C the ferromagnetic Curie point;
and 0 the paramagnetic Curie point.

J,(r)/J,(u) Figure 7.5 Temperatura dependence

of mximum magnetization of iron,
nickel and cobalt.

1 ! ' \ t,gc
0.2 0A 0.0 (LS ii)

Figure 7.6 Magnetizaron plots of iron (a) and nickel (b) single
crystals in directions [100], [110] and [111].

a) (M
230 Solid State Physics

zation. For iron this direction is the [100] and for nickel .the
[ l] direction, It is much more dihcult to magnetize iron in'the
[100] and [111] directions and nickel in the [110] and [100] direc
tions, substantially greater vales of magnetic fieid intensity being
needed to attain magnetic saturation. Those directions are termed
difficu.lt magnetization directions. The integral

Js
t'F'm \ Mo-nT j m (/ .13)
o
taken along the magnetization curve expresses the work spent on
magnetizing the crystal in the given direction. This work is trans

figure 7.7 Variation oi length

o erromagnetic samles with
magnetization (magnetostrictioo).

formed into free energy of the magnetized crystal. It may be seen

from Figure 7.6 that the least free energy is that of the crystal magne
tized in the easy direction and the greaiest is that of tha crystal
magnetized in the difcult direction.
Magnetization of ferromagnetics is accompanied bv a change in
their dimensions and shape. This phenomenon became known as
magnetostriction. Figure 7.7 shows the relative change in the length
of rods made of nickel, of annealed and of cast cobalt, of iron, and
of Steel magnetized in helas of gradually increasing intensity. The
greatest relative contrction is that of nickel (aimost 0.004 per-
eent); iron and Steel rods increase their length a little in weak helds
and contract in strong fieids. On the contrary, cast cobalt rods
contract in weak elds and increase their length in strong helas.
In compliance wth the Le Cnatelier principie to the enect that
a svstem resists the inluence of external factors striving to change
its State, the mechanical deformation of erromagnetic hodies result-
ng in the change in their shape an d dimensions should inluence
Magnetic Properties o Solas 231

the magnetization o such bodies. Specincally, i the body being

magnetized contrais in the given direction, then application o a
compressive stress in this direction should favour magnetization and
application o an extending stress should make magnetization more
difcult. The variation o magnetic properties o strained ferromagnet-
ic bodies is termed the magnetoelastic efect. Some erromagnetic
materials are so sensitive to internal stresses caused by deormations
that this property is utilized for the purposes o strain measurements.
When a erromagnetic is magnetized in an alternating magnetic
eld, its dimensions change with a frequency double that o the
eld. This property is used in magnetostrictive oscillator capable o
generating powerful ultrasonic vibrations with a frequency up to

Figure 7.3 Dependence of linear expansin coeEcient o iron-aickel (a)

and o iron-platinum (b) alloys on composition.

a * 10t
12
T

] ;
xT
M U
4 ! i
M
O i I j i |
i'j11 i '
-4 1 1 1
1 1 1
.O XI) 100
r, n Pt
(a) hl
several megahertz. Sucli oscillators are employed in ultrasonic
devices for the machining and cleahing o solid objects, in sonars
used to measure depth of waterways, and in numerous other devices
and instruments.
An interesting problem is that o thermal expansin of ferromagnet-
ic bodies. Thermal expansin of solids is, as we know. due to the
anharmonicity o vibrations of particles around their equilibrium
sites. For diamagnetic and paramagnetic solids anharmonicity is
the only cause of the change in their dimensions upoa heating. By
forc of this such bodies always expand with the rise in temperatura.
Let us denote the linear expansin coecient due to anharmonicity
of atomic vibrations by a x. The situation in erromagnetic materials
is not so simple. A change in their temperatura is accompanied by
a change in their magnetization and, consequently, in dimensions.
Ni S. Akulov has proponed the term thermostriciion for this pheno-
232 Soiid Staie Physics

menon. Denote the linear expansin coefcient dus to thermostric-

tion by a 2. The total therma! expansin coefcient of a ferromagnet-
ic will be a = a x -f cu* The coefcient a, is ahvays positiva while
a 2 may be either positive of negativa. Thereore the total thermal
expansin coefcient of a ferromagnetic material may be positive,
zero, or negative. For instance,' the group of ferromagnetic materials
with a negative ferromagnetic component of the thermal linear
expansin coefcient inciudes Invar alloys. Figure 7.8 shows the
dependence of the thermal expansin coefcient of iron-nickel (a)
and iron-platinum (b) alloys on their composition. The a of alloys
containing about 36 % nickel is about 10 times less than that of pur
nickel or iron: a of alloy containing 58 % platinum is negative
such an .alloy does not expand upon heating but, on the contrary,
contracts.
Invar alloys are widely used in instrument manufacture, metro-
logy, aviation, and manufacture of electric lamps and radio valves.
Depending on practical purposes, alloys with very small, zero, or
even negative thermal expansin coefcients can be used.

65 Msgnefic properfies of aoms

The orbital magnetic moment of an atom. The atom of every element
is made up of a positively charged nucleus and an electrn shell.
Many magnetic phenomena can be adequately explained with the
aid of Bohrs theory in which it is assumed that the electrons of the
shell move in denite orbits. Each such electrn will establish a
closecl current equal to I = qv (v is the frequency of rotation of the
electrn in the orbit, and q its charge). The magnetic moment of
the current is M IS qyS (S is the area of the orbit). Since
S = rcr2 and v = y/(2xr) {v is the linear velocity of the electrn
in the orbit), it follows that .
M = i 1 vqr/2 (7.14)
The magnetic moment of the electrn which is due to its motion
around the nucleus is termed orbital magnetic moment. We shall
denote it by u. This moment is perpendicular to the plae of the
orbit as is required by the right-hand screw rule (Figure 7.9).
The orbital angular momentum of the electrn is
Pi mvr (7.15)
wherc m is the electrn mass. It is oppcsite to Uj. Comparing (7.14)
and (7.15), we nd -
Magnetic Properies o Sods 233

The ratio
yi- 2m (7.17)
pi

is termed the gyromagneiic ratio.

As required by quantum mechanics, p and its projection plH on

Figure 7.9 Orbital magnetic

moment j, and orbital angular
momea tu m p oi an electrn.

the direction of the magnetic eid H can assume only discrate

vales, namely
p, = Ti V 1 ( 1 + i) ' (7.18)
PiH = mlh (7.19)
where l is the orbital quantum numher, wnich can assume only the
following vales:
l = 0, 1, 2, ..., (7 .1 8 ')
n vales in all (rc is the principal quantum number);
is the magnetic quantum.-nu-mber, whieh'can assume only..the
following vales:
mi = l, - (l - 1), 0, ..., -f l (7.19')
21 -f 1 vales in all.
Because of that the magnetic moment .t and its projection
on the direction of H may assume only tuQ following discreta vales:
(7.20)
Pi
plH mlPB (7.21)
where
}IB: lL 9.27 x 1024A-m: (7.22)
2m
234 Sod State Physics

is the Bohr magneton. It is ths quantum of the magnetic .moment

and is accepted as a unit for measuring magnetic moments of atomic
systems.
In a complex atom whose electrn shell is made up of many elect-
rons, the total orbital magnetic moment is found by adding up the
moments. of individual electrons in compliance with the rules of
space quantization. The moment of closed electrn shells is zero
Bv forc, of tais only the atoms with partially filled shells can have
a nonzero orbital magnetic moment. But even in the latter case
should the partially filled shell lie cise to the external shell and
should the interaction of the atoms in the solid State be strong, the
magnetic moments of the partially filled shell would be frozen in":
their orientation in an external field would be so impaired that they
would take practically no part in the magnetizaron of the body.
For instance, such is the behaviour of orbital magnetic moments of
the electrons of the partially filled 3d shell of the elements beionging
to the iron gro'up.
The spin magnetic moment of an atom. Apart rom the orbital
angular momentum the electrn has an intrinsic angular momentum
p s termed spin. It is known from quantum mechanics that
Ps = y'3/2 (7.23)
and that the projection of the spin on the direction of the field II may
assume only the following vales:
P;H= JU2 (7 24)
There is an intrinsic magnetic moment us connected with the
intrinsic electrn angular momentum whose valu was first experi-
mentaily determined by Otto Stern and Walther Gerlach. Their
experiments demonstrated that the projection u sH is numerically
equal to the Bohr magneton:
usH (7.25)

(the minus sign reects the negative nature of the electrn charge).
The gvromagnetic ratio for the intrinsic moments of the electrn is
Usff _ Q (7.26)
PsH ' m
It is twice as large as for the orbital moments.
In atoms containing a large number of electrons p s should be added
up like vectors with account taken of the rules of space quantization.
The total spin moment of closed shells is zero,. the same as the orbit
al moment. Table 7.2 shows by way of an example the data on the
spin configuration of the 3d shell of free atoms of the elements of
tile iron group.
Magnetic Properies o Sods 235

Tabie 7.2

Se Ti V Cr Ma Fe ' Co Ni

Total 1 2 3 5 5 4 o 2
spta i u m u m w u n m u u h u itm tu
The spins are least compensated in the chromium and raanganese
atoms and, correspondingly, they nave the mximum total spin
raoment. However, suen orientation of the spins is not usualiy re-
tained in the solid State and because of that the total atomic spin
monient in the solid is difierent. For instance, in the iron lattics
the average number of Bohr magnetons per atom is not 4 but only
.2.3; in chromium it is 0.4, and in a-manganese it is 0.5.
Magnetic moments of nucleus. Atomic nuciei too have a spin and
a magnetic moment connected with it. The order of magnitude of
the nuclear spin is the same as that of the electrn. Since the nuclear
mass is soma 103 times greater than the electrn mass the nuclear
magnetic moment, in compliance with (7.25), is three orders of mag
nitude less than the electrn magnetic moment. Therefore, as a rst
approximation, the efect [of nuclear magnetic moments on the
magnetic properties of bodies can be neglected. This dees not
mean that those moments do not play any role at all. In some
phenomena not discussed in this book that role, may be quite
important.
The total magnetic moment of an atom. The total magnetic mo-
ment of the electrn shell of the atom is determinen as ollows.
Using the rules of space quantization, one nds the total orbital'
angular momentum: PL = 5 where p lt is the orbital angular
momentum of the ith electrn. The numerical valu of P is deter
minad by the quantum number L
PL = hyrJ lT T T ) (7.27)
The number L may be any integer between the mximum ana the
mnimum vales of the algbrale sum T;: o* ^-e orbital quantum
numbers of individual electrons. Next one nds the total atomic
spin: Ps = y\ip Si, where p si is the spin o the ith electrn. The
numerical valu of P3 is determinad by the quantum number S :
P3 = R V S (S - t 1) (7.23)
The number S may assume vales lying in the interval between the
mximum and the mnimum vales of the algbrale sum T\ Sf of
spin quantum numbers of the individual electrons, the dierence
between successive vales of S being unity.
233 Solid State Physics

Finally, one finds the total atomic momentum equal to the

vector sum of Pr. and P s, that is P/ = P -f Ps- The numerical
valu of Pj is determined by the intrinsic quantum number 7.
Pj = Y 7(7 t 1) (7 .29)
which may assume the following set of vales:
7 = L + S, L + S - 1, L S, if L > 5
7 = 5 + 7, 5 + 7 - 1 , ..., 5 - 7 , if 5 > L (7.30)
The only allowed orientations of Py in an external field are such
that its proiections on the direction of the field are mltiples of h:
Pjh = mjh (7 .31)
whcre vn: is the magnetic quantum number equal to
m j = - J , - ( / - 1)...... 0, 1, 2, J (7.32)
2/ -f 1 vales in ali.
The atomic magnetic moment corresponding to the total momen
tum P j is
M j = - g u By / ( / - l ) (7.33)
with projections on the direction of an external field H equal to
3 (7.34)
\vhere
, / ( / S ($'-!-1)-( 4-1) (7.35)
is the Lande factor, or magnetic spitting factor which takes account
of the difference in gyromagnetic ratios of the orbital and the spin
moments making up the total atomic magnetic moment. For L 0,.
that is in the case of a pureiy spin magnetism, g = 2; for 5 = 0,
that is, in tile case of a .pureiy-orbital magnetism, g 1.
Often the term atomic magnetic moment is taken to mean not
(7.33) but the mximum valu of the projection M JH. For instance,
the magnetic moment of a hydrogen atoro, in the ground State (L =
= 0, S = 1/2, g = 2) characterized by / = 1/2 is taken to be equal
to uB; for a free iron atom with a frozen in orbital magnetic moment
7 = 5 = 2 , g = 2, and ilfa = 4uB.
AU atoms and ions with closed shells have 5 = 0, L = 0, and 7 = 0.
Therefore the magnetic moments of such atoms and ions are zero.
Paramagnetism owes its existence to the presence in an atom of
partially fiiled shells. According to the Pauli exclusin principie
there may not be more than two electrons with opposite spins in one
stae. The total spin moment of those electrons is zero. Such electrons
are tsrmed pairsd. If an atom or an ion contains an odd mumber
Magnetic Properties o Sclids 237

of electroas, one of them will 9 unpaired and the atom will have
a permanent magnetic moment. If the atom contains an evan num-
ber of electrons, two cases are possible: either all electrons are
paired and the total spin moment is zero or two or more electrons are
unpaired and the atom has a permanent magnetic moment. For
instance, H, K, Na, Ag have odd numbers of electrons, one of them

Figure 7.10 Schematic representation

ot atomic magnetic moments in
paramagnetic (a), 'lerromagnetic (h),
antiferromagnetic (c),
and ferrimagnetic (d) matarais.

(c).

(d)

unpaired; Be, C, He, Mg contain even numbers of electrons, all of

them paired. Oxygen aiso contains an even number of electrons, but
two of them are unpaired.
Magnetic moments of many molecuies are zero because only sorne
of them contain unpaired electrons. First of all these are tile free
radicis, which play an excepfcionally important part in many
Chemical reactions. As examples of such radicis are free hydroxyl
(OH), free methyl (CHa), and free ethyl (C..rl5). The presence of
unpaired electrons in molecuies and in free radicis makes them
magnetic.
Classication of magnetic materials. When the orbital and the
spin moments are added up, a complete compensation may take
place and then the total atomic moment will be zero. I such a com
pensation does not take place, the atom will have a permanent mag
netic moment. Accordingly3 the magnetic properties of bodies will
be different.
Materials whose atoms have no permanent magnetic moments are
aiamagnetic. Materials whose atoms have a permanent magnetic
moment may be either paramagnetic, ferromagnetic. antiferromag-
netic, or ferrimagnetic. Namely, if the ineraction between the at
omic magnetic moments is zero or very weak, the material will be
2 38 Solid Sfste Pnysics

paramagnetic (Figure 7.10(a)); if the neighbouring magnetic moments

tend to align themselves parallel to one another, the materiai will
be ferromagnetic (Figure 7.10(b)); if the neighbouring magnetic
moments tend to align themselves antiparallel to one another,-the
material will be antiferromagnetic (Figure 7.10(c)); finally, if the
neighbouring magnetic moments tend to align themselves antiparal
lel to one another but their magnitude is not the same, then the
material will be ferrimagnetic (Figure 7.l0(d)).

66 Origin of diamagnefism :
The cause of diamagnetism is a change in the orbital motion of the
dectrons acted upon by an external magnetic eld. It is common to
all materials but is often overshadowed by strong para- and ferro-
inagnetism. In its pur form diamagnetism is displayed by materials
whose total atomic magnetic moment is zero.

Figure 7.11 EGect of magnetic field on orbital motion of an electrn:

(a)eld H is perpendicular to orbit plae; (b)orbit precession
in magnetic eld.

(b)

Precession of electrn orbits in a magnetic field. Consider the

motion of an electrn in an orbit of radius r (Figure 7.Tl(a)). In the
absence of field H the centripetal forc applied to the electrn is
FcP = mv2 J r = m),r (y0 is the linear and 0 the angular velocity
of the electroas motion). When external eld H perpendicular to
tile plae of the orbit is applied^ the electrn is acted upon by the
Lorentz forc qv0B 0 directed along the radius of the orbit
Magnetic Properfias o Solios 2 39

(.B 0 is the fields induction). The resultant centripetal forc will be

F = Fcv + FLorentz or mc2r = m.(x>\r qa>0rB0
Jt oiiows hat
mr (a2o) = mr (co c) (co-f c0) 2mrco0coL <7co0rS o (7.36)
where'"
0L = cco0= - - 5 0= u0if (7.37)

is called the Larmor angular frequency.

Thus, a magnetic eld changes the angular frequency of an orbil-
ing electrn. It may be seen from (7.37) that this change is the same
for all electrons no matter what the radius of their orbits and the
linear velocity of their motion are. The airection of col coincides
with that of B 0.
Generally, when H is. not perpendicular to the plae of the orbit,
its eiect is to excite precession of the orbit around the direction of
the eld (Figure 7.Tl(b)): the perpendicular p; to the plae of the
orbit describes a cone around H. Calculations show that the angular
velocity of such a precession is expressed by formula (7.37).
Induced magnetic moment of an atona. Magnetic susceptibility of
diamagnetics. The precession of the electrn orbit resulta in an
additional motion of tile electrn around eld H. This motion is
superimposed on its orbital motion. The magnetic action of this
additional motion is equivalent to that of a closed current

A /= gvL = L Bn (7.38)
7 2
oT
where vL is the precession frequency related to the angular frequency
by the expression col = 2rcvL. The minus appears because of the
negative charge of the electrn.
The magnetic moment of the elementan' current A/ i.-

where S is the area bounded by the patk of the electrn precessing

around eld H. Calculations show that T = 2jrr73. where r- is
the mean square of the electrons distance from the nucleus. There-
fore
q2r
Au =
tjm (7.40)

It follows from formula (7.40) that in a magnetic eld every elect

rn acquires an additional so-called inducsd magnetic moment direct-
ed against
3 H. The annearance
X i. of this moment is the causo of the
240 Sciid Sfafa Physics

magnetization of the body in the ireciion opposite o that of the

magnetic eld, which is characteristic of diamagntica.
The magnetic moment of an atona containing Z electrons is found
by adding up the' moments of individual electrons:

(7-41)
i
where r\ is the mean square distance of the ith electrn from the
nucleus. The sum of r\ may be replaced by the product Zar, where
a~- is the mean square distance of all the electrons from the nucleus.
Then

<7-42>
Multiplying (7.42) by the number of atoms per unit volunte., rc-,
we obtain the magnetization J m:
Za-na
- Zq-na-
J,n nAM u0H (7.43)
6 m 3o = 6m

The magnetic susceptibility is

y J rn _ . UpZq-na- (7.44)
H 6m \ /

Assuming that a IO-10 m and n 5 X 1023 m-n. we obtain

y ^ 10-SZ. Tnis is in good agreement with the data of Table 7.1.
Moreover, from (7.44) it follows- that magnetic susceptibility of
diamagnetics is independent both of temperatura and of magnetic
eld intensity 7/ and rises in proportion to the atomic number of
the element, Z, which is in full agreement with experiment.

67 Origin o paramagnefism
Langevins classical theory of paramagnefcism. The classical theory
of paramagnetism developed by Paul Langevin is based on the
idea that the atoms of paramagnetic materials have a permanent
magnetic moment M, that is, they constitua permanent magnetic
Jiples and that the interaccin betweea these dipoies is negiigible.
The energy of such a dipole in a magnetic eld II is
Un Mu0JT eos 9 (7.45)
where 9 is the angle between M and H (Figure 7.12(a)).
The mnimum of 57m corrssponds to 9 = 0. Therefore all th
dipoies tena to orient themselves in the .direccin, p the externa!
Magnetic Propertres of Soiids 241

eid, this being bampered by thermal motion. The total magaetic

raoment of the material is made up of the projections of tbe magaetic
moments of the individual atoms on the direction of H. Since the
magnitude of those projections is MH = M eos 9, the problem of
the quantitative calculation of the magnetization of the material
is reduced to the calculation of the average valu of M H that corre-
sponds to the State of equilibrium between the orientational enect
of the eld and the aisorientational enect of thermal motion. Just

Figure 7.12 Explaining classical theory of paramagnetism: (a)magnetic

rnomeat M and its projection on magnetic field E; (b)caiculating
total magnetic moment of a paramagnetic; (c)plot of the Langevia
function.

H
H

this problem was solved by Laagevin with the aid of methods of clas
sical statistics. He supposed that the orientation of M with respect
to H can be arbitrary and that accordingly the angle 0 can assume
all vales.
The probability for a dipole to align itself at an angle in the
interval (9, 9 d6) to H (that is inside the solid angle dQ; see Fig
ure 7.12(b)) is determined by the Boltzmann distribution function:
llnM H eos 9
W = Cle~u^ k*TdQ = Cl ex p( ) dQ
ks?
where Cx is a normalization constant.
It may be seen from Figure 7.12(b) that dQ. = 4n sin 9 6; there-
fore

where C is a new constant.

t S 0 8 8 5
242 Solid 5ae Physics

The average valu of M H is

Jt
] eos 0 exp (]10MH eos Q/k3 T) sin 0 0
MH = M eos 0 = M ............................................ (7-46)
J'exp (y.0M H eos Q/k3 T) sin 0 d0
o

l those integris are evaluated, the result is

(7.47)
) - J ( 4 f p)
where
Q h/u0i/ (7.4S)
P 7. <T\
*3 T
The magnetization is
J m = nNIu nM ^ coth 3 3- j (7.49)

where n is the number of atoras per unit volume, and the magnetic
susceptibility is

Since tlie atomic dipoles acted upon by a eld aliga themselves

in its direction, such will be the direction of the magnetizaron of
the body as a whole, which is characteristic of paramagnetics.
Let us expand coth 3 in a power series: coth (3 = j3_1 (3/3
(3-/45 ... . For 3 1 we can liraiL ourselves. with the first
tvvo ternas of the expansin. Then
iLgnM2 (7.51)
7"ra 1 3k%T

In full agreement with experiment J m is airectly proportional

to H and inversely proportional to T. The second of the formulae
(7.51) expresses the Curie law: x = CiT. The Curie constant C
= nPii0/(3ka).
For atonas M uB: then for II iO8 A/m we see that MH\i0 ~
10~23 J and for T = 300 K we see that k3T 3 X 10-21 J.
Henee the condition 1 is alnaost always satised. Only in very
strong elds and at very low temperatures is (3 1 and the direct
proportionality between I m and II is no longer naaintained. In
the limiting process, as (3*- co, coth (3 and the magnetization
beconaes saturated, the corresponding mximum valu being
J , = nM (7.52)
Magnetic Prcpertias o Soiids 243

Magnetic saturation involves the alignment o magnetic moments

o all atonas in the direction o the field.
The unction L (P) = coth 3 1/p' is termed the Langevin func-
tion. ts plot is shown in Figure 7.12(c). For small Ss a good approx-
imation or the plot is the segment o the straight lin e a l; as p co
the unction ()->-1.
Fundamentis o quantum theory o paramagnetism. The classical-
theory is incapable of providing a consistent -explanation o the
magnetic phenomena as the result of the raotion o electric charges.
The existence of molecular currents necessarily involves the acknowi-
edgment o the fact of the stability of electronic motion in atoms,
a fact unacceptable for classical physics. The assumption that all
orientations of magnetic moments wiih respect to H are possible,
which is the basis o Langevins classical theory, is also wrong. Those
difficulties have, on the whole, been overeme by the quantum theory
o naramagnetism. Lets consider brielv the essence of this theory.
There are 2/'-f- 1 ways in which the atomic magnetic moment
.1/ may align itsei in a magnetic eld ( / is the intrinsic quantum
number). The probability of each such orientation is determined
by the Boltzmann distribution W = Ce^0^ jh^ iK3j {Mjh is
the projection of M j on H). The average valu o M JK vill be
+J
yp I .uLlph/
o-VjjHH
2 xVjsesp ( 3T
ir _ (7.53)
JH -
v / ,uniVj
2 j es? \ ksT
- j

The diSerence betveen (7.53) and the classical expression (7.46)

is that integra tion is replaced by summation over the discrete direc-
tions in which the vector MT may be aligned. Evaluation o the sums
in (7.53) vields the ollowing result:
M jk ?TlibS j (P) (7.o4)
"vhere
P= k-aT (7.55)
2/-M 9rn_i
Bs($) 2J
c o t h ^ p p- coth 2J
(7.56)
Function B j { 6) is termed the BriUouin funcior,.
The magnetization and the magnetic susceptihility are equal to-
-m = M JHn = n g j uB Bj (p) (7.57)
x= ^ 5 y(p) . (7.58)
244 Solid State Physics

For p < 1, B/ f f i P(J + l)/(3J) and

r _ at V | / ( / + l ) F lF .,_ /(/+ !) . _ _
m 3ks T X -------------3c%f-------- (7 - 9 )

.It follows from (7.59) that for p < i the quantum theory results
in a linear dependence o J m on H and in an inverse dependence of
Jrr, and x on T, which agrees with experiment. In sfcrong elds and
at very low temperatures p oo,
coi11 coth ^-1, B j (P) -*-l

and the magnetizaron attains the saturation valu

/ s = ng/uB (7.60)
Materials used for experimental tests of the theory of oaramagne-
tism are the Solutions and crystalline hvdrates of sals, which con-
tain ions with nonzero magnetic moment. 3uch are, for instance, the

Figure 7.13 Calcuiating paramagnetism of free eiectrons.

ions of the elementa of the groups of iron and rare earths.

In Solutions' and in cryjm]Jine hvdrates the ions are so far-apart that
their interaction may be neglected, which is a necessary condition
for paramagnetism. Experimental investigations of such compounds
have produced results in good agreement with the theory.
Paramagnetism of electrn gas. Accoraing to (7.51) and (7.59),
paramagnetic susceptihility is inversely proportional to temperature.
However, some metis have been discovered to exhibit paramag
netism independent of temperature. It was Wolfgang Pauli who
demonstrated that this is due to the paramagnetism of free eiectrons
that constitute the electrn gas.
Figure 7.l3(a) shows the conduction band o a metal. It is schemat-
icaily representad in the form of two hal-bands containing eiectrons
with opposite spin moments us = pB. When H 0, the number of
eiectrons in both hal-bands is equal and the total magnetic moment
Magnetic Progenies of Sods 245

of tile electrn gas is zero. When the eid H is applied, evsry electrn
of the left half-band acquires an additional energy Um tu 0u.3f
and every electrn of the right half-band an energy U"m = }a0.(.b//.
The result is the appearance of a difference between the quasi-Fermi
levels E]'= Ep = 2u0lu Bif (Figure 7.13(b)) for the electrons of the
right and the left half-bands which is equalized by means of spin-
ip of some of the electrons of the right half-band and their transi-
tion to the left half-band (Figure 7.13(c)). Since al! the interna!
levels in the half-bands are occupied, the only electrons whose spins
can be reversed are those occupying levels in the zone where the
Fermi distribution is fuzzy (see Figure 5.6(b)) and where there are
vacant levels. The number of suctt electrons, according o (3.43)), is
An ^ ~Ej-pn (7.61)
where n is the electrn gas concentration.
Of this number Ar C exp (.i0pBif/A-Br) electrons vi 11 have
their magnetic moments orientad in the direction of H and An" =
= C exp (u.0iu3iT/A-Br) against H (C is a constan!). The magnetic
moment per unit volume of a metal due to spin-ip is
Jme = (An' An") p3 = Cu3 (gP e~*)
where
(3= u.oj.s.Bf7/BT
Since An = Ar -j- An" C (e^ ~r s~^), it follows that C '=
An (es -r Substituting this into the expression for / ra, we
nd
e3 _ ,-S
, J me = A.uuBrt= Anu3 tanh p
-j-e 3

For p <C 1, tanh {3 3 and J me = S.n\i%ii0H'/k3T. - Substituting

An rom (7.61), we bctain
u 3
J me UnH
n ~g~ Uo (7.62)

The paramagnetic susceptibility of the electrn gas is

UoUB 7
Xe ~ n ~T~ (7
A more accurate caiculation yielas
3 UoUB
Xe 9n (7.64)

It mav be seenrrom (7.64) that the magnetic susceptibility of the

eiec
iectron gas should be independent of temperatura, which is what
is observed in practice.
246 Sod State Physics

Production of low fcemperatures using the method of adiabatic

deinagnetization of paramagnetic samples. The atoms of paramagnet-
io materiais possess a permanent magnetic moment. In the absence
of an esternal eld as a result of thermal motion of the atoms the
orientation of their magnetic moments is almost completely random.
The quantitative measure of this disorder is the entropy S , which
in this case is terraed magnetic entropy S m. In compliance with the
Boltzmann principie
Su = k3ln W& (7.65)
where FM is the thermodynamic probabilitv, which -in this case
is equal to the number of ways the n atoms of the paramagnetic
sample can be distributed among the 2J 1 sublevis into which
every atomic level splits in a magnetic eld. Its valu may be obtai-
ned from the expression
Wu = (2J + l)n (7.66)
SubstiLuting (7.66) into (7.65) \ve obtain
S u = kBn In (2J -f i) (7.67)
When the magnetic eld is applied and its intensity increased,
an ever increasing number of magnetic moments is oriented in the
direction of the eld, the result being a reduction in the magnetic
entropy. When the State of magnetic saturation is reached, the grea-
test possible order in the arrangement of magnetic moments is
established and Su vanishes.
Henee the process of magnetization of a paramagnetic sample up
to saturation is accompanied by the decrease in its entropy by the
amount
AS = Su = kBn ln (2/ -J- 1) ' (7.68)
If the magnetization is performed at a constant temperatura T,
the decroase in S by the amount AS results in the generation of an
amount of heat equal to AQ TAS. 'filis heat is transmitted from
the sample to the surroundings,. usually to liquid heiium. After
equiiibrium has been established the heiium is removed and the
sample is leffc thermally insulated. In such conaitions it is slowly
adlabatically demagnetized with the result that its entropy again
rises by AS. The rise in entropy requires heat, which can be supplied
only by the thermal vibrations of the lattice since the sample is
thermally insulated from the surroundings. Because of that its tem-
perature drops. Using this method it was possible to obtain tempera-
tures below 0.001 K. The possibility. of obtaining still lower tempera-
tures is limited mainiy by the fact that alreadv at H 0 the atomic
energy levels are to soma extent split into sublevis because of the
interaction of the magnetic moments with each other and with the
nuclous.
Magnetic ProperHes o Solids 247

B Origin cf ferromagnefism
Elementary carriers of ferromagnetlsm. A magnetized body acquires
a magnetic moment M mad8 up of regularly orientad atoraic magnet-
ic moments and an angular momentum P made up of regularly
oriented atomic angular momenta. According to (7.17) and (7.26)
the ratio MIP must be equal to ql2m if the magnetization is due to
orbital magnetic moments of the atoms and to q/m if it is due to
spin moments.
The aDpearance of a magnetic moment in the course of magnetiza
tion was rst established in experiments of A. Einstein and W. J.

Figure 7.14 Experiments on the nature o erromagnetism:

(a)the experiment of Einstein and de Haas; (b)experiment of Barnett;
(cjexperiment of Ioe and Kapitza.

\\\N\\\V

Before After
heating heating

de Haas and became known as the Einstein-de Haas efect. In those

experiments a small iron rod 1 suspended on a thin elastic thread
2 was placed inside a solenoid 3 (Figure 7.14(a)). In the course of
magnetization the rod turned and twisted the thread. The direction
of rotation of the rod changed with the change in the direction of
magnetization. The angle of rotation was measured witli the ala of
mirror 4 xed on the rod which reflecten a beam of light on scale
5. The experiment made it possible to determine M and P and to
nd the gyromagnetic ratio y MIP.
S. J. Barnett made an experiment which was the reverse of the
Einstein-de Haas experiment: he observen the magnetization-of a
quickly rotating iron roa. Such magnetization is caused by the
tendenc-y of electronsthought of as tops possessing angular me-
248 Soild Sfafs Physics

mentato arrange their axes of rotation (spins) in the direction of

tile rotation axis of the body (Figure 7.14(b)). In another experiment
the Soviet physicists A .F . Ioffe and P. L. Kapitza quickly heated
a magnetized rod to a temperatura above the Curie point". Before
heating the orientation of the electrn tops was ordered~(Fig. 7.14(c))
and their total angular momentum was nonzero. TVhen heated
to a temperature above the Curie point the tops" chancad their
orientation to chaotic (Figure 7.14(d)) and their total angular momen
tum became zero. Because of that the demagnetized rod as a whole
acquirea a rotational momentum that could be measured in the
experiment. Measuring in addition the magnetic moment of the
magnetized rod one could find the gyromagnetic ratio y M/P.
The experiments demonstrated that the gyromagnetic ratio for
ferromagnetic materials is M/P = q/m, that is, equai to the
gyromagnetic ratio for the intrinsic moments of the electrn. This
proved that ferromagnetism is due not to orbital but to spin magnetic
moments of electrons, which is consistent with the electronic structu-
re of the atoms of elements that exhibit ferromagnetism. Since the
magnetic moments of closed shelis are zero and since the outer valen-
ce electrons are collectivized in the process of formation of the metal-
lie state, ferromagnetism must be a property solely of the transitio-
nal elements, which have incomplete inner shelis. Thosc elements
inelude the transitional metis of the iron group, which have an imcom
plete 3d shel, and rare earth elements with an incomplete 4/ shell.
Since, on the other hand, the orbital magnetic moments of the
electrons of the 3d shell are frozen in and their contribution tothe
magnetic properties is negligible, ferromagnetism of the elements
belonging to those groups can be due only to the atomic spin moments,
which in this case are quite large-(Table 7.2). This hypothesis was
first expressed by the Russian scientist B. Rozing in- 1892 and was
expioited later by the French physicist Fierre Weiss. The iatter
assumed that there is an intense molecular field H in a ferromagnetic
proportional to Ier s.alnration magnetization / s:' ,
H = Ut (7.69)
where a, is termed the molecular field. constant. This field is responsible
for the spontaneous magnetization of a ferromagnetic.
The introduction of a molecular eld made it possible to explain
a wide range of phenomena observed in ferromagnetism. However,
the natura of the eld itself remained a mystery for a long time.
At rst it was supposed that the origin of forces which orient the
spin moments is purely magnetic, that they appear as a result of an
ordinary interaction of spin magnetic moments (spin-spin interaction).
The energy of this interaction is of the order of Um u|/fl3, where
a is the interatomic distance in the iattice o the ferromagnetic.
Substituting uE = 9.27 x 10"M A-m2 ,apd,fa..a$ lO-10 m, we obtain
Magnefic Properies o Solids ;249

Um 10~23 J. Tliis is about two orders of magnitude less than

the room temperature thermal energy of a latice atom wbich disturbs
tbe orderly spin arrangement. It follows then tbat the magnetic
spin interaction is incapable of effecting their parallel orientation
characteristic of the ferromagnetics at tempera tures beiow the Curie
point and that the origin of the molecular field which enects such
parallel spin orientation should be nonmagneticu Subsequently this_
conclusin was proved by direct experimenta of Ya. Dorfman.
The role of exchange interaction in ferromagnetism. In 1S28 Frenkel
mane the assumption that the origin of the orces wbich are respon-
sible for the denite mutual orientation of atomic magnetic moments
is electrostatic. Tney are the result o exchange interaction of the
electrons of inner incomplete atomic skells. We have already discussed
this type of interaction in dealing with the nature of the covalent
bond (see Section 3). The exchange interaction involves a change
in the energy o the system. This is easily seen from the example
of the simpest system of two hydrogen atoms (see Figure 1.5).
Accoraing to (1.11) and (1.12) the energy of such a system is
(7.70)
where E0 is the energy of two noninteracting hydrogen atoms, K is
the energy of Coulomb interaction of electric charges making up
the atoms, S is the overlap integral whose valu lies in the range
O ^C F^l, and A is the exchange energy (in Chapter 5 we called
it tbe exchange integral).
Calculation shows that A can be expressed by the following relation
- 4 = - /( S ,.S j) (7.71)
where S and S are the total spins of the interacting atoms. and
J is the exchange integral (it is a measure of the probability of
electrn 1 going over to atom B and of electrn 2 going over to
atom .4). In the- case of two interacting hydrogen atoms

(7.72)
Here qzir and q2!rl2 are the interaction energas of the nuclei between
themselves and of the electrons betw-een themselves, respectively;
q2/rbl and q2ira are the energies o attractiou of electrn 1 to
nucleus b and of electrn 2 to nucleus a; apa(1) and \(:&(2) are the
wave functions that describe the motion of electrons 1 and 2 around
nuclei a and b, respectively; ijja(2) and ij;b(l) are the wave functions
that describe the probabilities for electrons 2 and 1 to be cise to
nuclei a and b, respectively, that is, the probabilities of the atoms
A. and B exchanging electrons; and dV1 and V are vQlurne eie-
ments.
250 Sod State Physics

It foilovvs from (7.72) that both positive and negative terms enter
th.9 exchange integral. Therefore the sign of the exchange integral
may be either positive or negative. This is determined by the part
played by the positive and negative ternas of the exchange integral,
which in its tura depends on the relation of the dimensions of the
electrn slaells taking part in the formation of the exchange bond
and on the interatomic distance.
The sign of the exchange integral determines what orientation
of the spins o electrons taking part in the exchange bond is advanta-

Fioure 7.15 Spontaneous

magnetization of a ferromagnetic.
Exchange [orces cause paraliel
orientation of the spins of electrons
belonging to inner partiallv ftlled
sn eils.

geousthe paraliel or the antiparalel. t follows from (7.71) that

when the sign of the exchange integral is negative (/ < 0), the
exchange energy .4 wiil be negative and, consequenthr, the svstems
energy U vil! be iess than the energy 2E0 of the individual atoms
[see (7.70)] if the spins S; and S;- o the electrons taking part in, the.
exchange bond are antiparalel: As has been mentioned
in Chapter 1 this case corresponda, to the formation of a Chemical
bond between the atoms and the creation of a molecule [the sym-
metrical State described by (1.11)]; below we shall see that this is
also a necessary condition for antiferromagnetism.
When the sign of the exchange integral is positive ( / > 0), the
exchange energy A will be negative and the energy of the system as
a whole will be less than the energy of the individual atoms if the
spins Sf and S;- of the electrons taking part in the exchange bond are
paraliel: S S;-
Henee the paraliel orientation of the spins of neighbouring atoms
may too be advantageous from the energy point of view if their
exchange integral is negative. This is the necessary condition for
ferromagnetism since the paraliel arrangement of spins and, conse-
quentlv, of spin magnetic moments resuits in spontaneous magneti-
Magneric Prcperlies o? Solids 251

Figure 7.18 shows the depeadence of the exchange integral-/

on the ratio of the lattice constant a to the diameter d-of theSd
shell of atoms of the'iron group transition metis. It may be seen
from Figure 7.16 that at ai > 1 .5 the exchange integral is positiva;
at ald < 1 .5 it turns negative, its ahsolute valu increasing with
the decrease in ald. It follows then that of all the transition metis
only iron, cobalt, and nickel should be ferromagnetic, which is
indeed the case. Manganese and other-elements of the group, for
which ald < 1 .5 , are not ferromagnetics. Should it, however, be
possible to increase somewhat the lattice constant of manganese

Co Figure 7.16 Dependence. of exchange

integral on the ratio of the lattice
parameter'to the diameter of the
a/d inner parially filie 3d shell
in transition elemcnts of the iron
group.

so that the ratio ald would approach 1.5. one could expect manganese
to become a ferromagnetic.
Experiment support this view. For instance, inclusin of small
amounts of nitrogen into the manganese lattice increases its lattice
parameter and results in the appearance of ferromagnetism. Ferro
magnetic properties are also exhibited by the Mn-Cu-Al alloys
(Heusler alloys) and by the compounds like MnSb and MnBi in
which the distances between the manganese atoms are greater than
in pur manganese crystals.
Henee the necessary and adequate conditions for ferromagnetism
are the existence of. incomplete- internal atomic shelis and the posi-
tive sign of the exchange integral which cause the parallel orienta-
tion of the spins.
Domain structure of ferromagnetic substances. Let us isolate
a regin A insicie a ferromagnetic crystal (Figure 7. 7(a)). Suppose
that exchange forces establish a parallel orientaton of spins of all
the electrons o incomplete atomic shelis, as shown in Figure 7.15.
Regin A will be magnetized to saturation. What will be the equi-
librium spin orientaton in the regin B helow regin .4? If the spins
in regin B are oriented as in A , there are two magnets with like
poles in contact with eacii other (Figure 7.17(b))- Suclx a state is
unstabie since it is characterized by the mximum energv of magnet-
ic i-nteraction. The stable state will be that in which the magnetic
felas cf the eontacting regions are joined together, that i, a state
in which the magnetizaron of the neighbouring regions of the crys-
252 Sod State Physics

Figure 7.17 Ferromagnetic divided into domains (regions o spontaneous

magnetization).

A A S
=S 8 Xy

(>)

Figure 7.18 Structure o boundary

la ver separating two domains
(Bloch walls); N denotes poles
on the sample's surace.

tal is opposite (Figure 7.17(c)). Calculations show that as long as

the width of regin .4 does not exceed several interatomic distances,
the dominant factor is the rstthe orientatioa action of the exchange
forceswhose effect is to magnetize the layers of regin 3 in con-
tact with regin A in the same airection as that of regin .4. As the
area of A widens, the importance of the second factor (the increase
in the energy of magnetic interaction) grows and ftnally it becomes
predominant: the width of regin A reaches a critical valu and
the magnetization of the neighbouring regin 3 rom now on pro-
ceeds in the opppsite irection. The critica! width of the regin of
spontaneous MgS'dfeation is dependent on many Motors, but usual-
ly it does not exceed sevefal micrometers.
Thus in the ahsence of an esternal eld a ferromagnetic crystal
should consist of a great number of seprate and rather small regions
magnetized to saturation. Those regions have received the ame of
regions of spontaneous magnetization, or domains. Domains are sepa-
rated by layers in which the orientation of the spins changes from
that of one domain to that of the other (Figure 7.18). Such transition-
al layers between domains became known as Bloch walls. In iron
their thickness is about 300 lattice constants (some 1000 A). Figure
7.19 shows the domain pattern of a ferromagnetic predicted thecret-
ically (a) and a photograph of the domain st'ructure of an edge of
a ferrosilicon crystal (b); arrows indcate the directions of spontaneous
magnetization in the neighbouring domains..;
Magnetic Proparfies o Sods 253

Qualifcative analysis oE the magnetization curve. Spontaneous

magnetization takes place ia directious o easy magaetization. In the
abseace o an esternal eld the mutual orientation of the domains
is such that the total magnetic moment of the ferromagnetic as a
whole is zero (Figure 7.20(a)) since this corresponds to the minimum
of the sysems free energy. When an esternal eld H is appiied, the
ferromagnetic is magnetized acquiring a nonzero magnetic moment.
The nature of the physical phenomena which take place during the'
magnetization of a ferromagnetic is such that the process may he
subdivided into three stages.
(1) Displacsment of domain boundaries: Place the crystal shown
in Figure 7.20(a) in a magnetic eld H. The orientation of the magneti
zation vector Jn of diereut domains with respect to H is not the
same: J m of the first domain malees the smallest angie with H and
that of the third domain the largest. When H is increased it becomes
aavantageous from the viewpoint of energy for the most favourably
oriented domain 1 to grow at the expense of domains 2, 3, and 4
(Figure 7.20(b)). The mechanism of this grwth is the displacement
of the domain boundaries. For this reason the first magnetization
stage became known as the displacement process.
The displacement of the boundaries contines until the first domain
spreads over the entire crystal. Figure 7.21 shows the magnetization
curve of a single crystal. The displacement process is represented
on this curve by the section OA. In small H's magnetization procesas
smoothly and is reversible; in strong fields it is a jumpy and an ir
reversible process leading to the Barkhausen effect.
(2) Rotation. When H is further increased, the spontaneous magne
tization Jm begins to rotate towards the eld (Figure 7.20(c)). The
magnetization proceeds now at a much slower rate than in the first

Figura 7.19 Domain structure of ferromagnetics: (a)theoreticaliy

predictet pattern oi a ierromagnetics divisin into domains;
(b)photograph. of an edge of errosiiicon with decorated domain
boundanes..- . ; ~
254 Soiid-State Physics

Figure 7.20 Processes involved in magnetization of a crystai:

{a)boundaries of iour dcmains into which tbe crystai has been divided
(the arrows indcate the direction of vector Jm ); (b)displcemete
of boundaries and growth of the most favourably orientad doraain 1 with
the increase in the magnetizing field H; (c)rotation of magnetization
vector Jra in the direction of H.

Jm Figure 7.21 Magnetization plot of

a ferromagnetic: OA section
corresponding to the process
of dispiacetnent of domain boundaries,
A B section corresponding to
rotation o magnetization vector,
and B C section corresponding
to the paraprocess.

stage and ends when vector concides with H. At this stage the
magnetization reaches technical saturati'on (Figure 7.21; section
A B).
(3) Paraprocess. After technical saturation is reached the magneti
zation contines to grow with the increase in H although at a drasti-
cally reduced rate. The explanation is that at any temperature other
than absolute zero not all the spins in the regions of spontaneous
magnetization are oriented parallel to each other. Because of the
thermal motion of atoms the orientation of some o the spins is anti-
parallel. The application of a strong magnetic eid can efect the
reorientation of these spins. The spin reorientation corresponding
tG tlie paraprocess is representad by the section BC.

69 Aniferromagneiisrn
As was established in the preceriing section, when the exchange
integral is negativa, the preferential orientation of the spins of
neighbouring lattice sites is the antiparallel 'one. In this case the
spin arrangement can be also an ordered one, but there wl be no
Mdcnefic Propertie; of So'.ics 255'

spontaneous magnetization because tire spin momcnts o the neigh-

bouring iattice sites are antiparallel and compnsate one another.
Figure 7.22(a) shows the magnetic structure o MnO dctermined
with the aid o neutrn spectroscopy (oniy the magnetically active-
Mn atoms are shown in the figure). The structure may be regarded
as a complex one consisting o two sublattices magnetized in oppo-

Figure 722 Magnetic structure o antierromagnetic (a)

and the temperatura dependence of its magnetic susceptibility (b);
the iattice o an antierromagnetic (MnO) can be considered as consisting
of two sublattices wnose magnetic momento ara antiparallel.

site directions. Such structure can exist only below a certain tempera-
ture termed the antierromagnetic Curie point, or the Nel poini.
At ahsolute zero the magnetic moments o the sublattices are mutu-
allv comoensated and the total magnetic moment of the antiferro-
magnetic is zero. As the temperature is raised the antiparallel:
arrangement of the spins is graduaily disturbed and the magnetiza
ron of the antierromagnetic rises; it reacnes its mximum at the-
Nel poxnc at which the orderiy spin arrangement vanishes altogether
and the antierromafgnstic turas luto a paramagnetic. As the tempera-
ture is raised' stiil higher the magnetization decreases in the same-
wav as that of every paramagnetic. Figure 7.22(b) shows the temper
ature dependence o the magnetic susceptibility o MnO whose eL
point is Ty m 120 K in a fteld E 4 X 1Q4 A/m.

70 Ferrimagneism. Ferrifes
The magnetic moments oLthe sublattices in antiferromagnetics are-
equal in magnitude and opposite in direction with the regult that
they completely compnsate one another. However, the re are cases
256 Soiid State Physics

when the magnitude of the magnetic moraents o tlie sublattices is

not the same owing, for instance, to the difference in the number or
in the nature of atoms that make up the sublattices (Figure 7.23(a)).
This leads to the a.ppearance o a ffnite difference in magnetic moments
of the sublattices and to an appropriate spontaneous magnetization
of the crystal. Such an uncompensated antiferromagnetism is termed'
ferrimagnetism.
The external behaviour of a ferrimagnetic is similar to thatof afer-
romagnetic, but because of the difference in tlieir internal structure
the temperature dependence of their spontaneous magnetization may
be quite dirrerent. For instance. the magnetization of a ferrimagnetic
does not necessarily decrease monotonously with the rise in tempera-

Figure 7.23 Schematic diagram of magnetic moments in a ferrimagnetic

lattice in general (a) and specically in magnetita FeO-Fe2Oj (b).
One o the sublattices is mane up of the half of trivalent iron ions,
the second subattice being made up of the other half of trivalent iron
ions and o bivalent ions of iron or of a substituto metal.

ture but can pass through zero even before the Nel point is reached.
Magnetite FeO-Fe20 3 can serve as an example of a ferrimagnetic.
The negativa oxygen ions forra a face-centered cubic lattice in which
there are one bivalent (Fe2+) and two trivalent (Fe3+) iron ions to
every FeO -Fe20 3 molecuie. The bivalent iron ions may be replacsd
by bivalent ions of other metis, for instance, Mg, Ni, Co, Mn, Cu,
etc., so that the general formula o materials of this class known as
ferrites assumes the forra MeO -Fe30 3, where Me stands'or a bivalent
metal. One of the sublattices of the complex ferrite lattice is made
up o one half of the trivalent iron ions and the other of the other
half of trivalent iron ions and of bivalent ions of iron or the substitute
metal. The magnetic moments of the sublattices are antiparallel.
Thereors the magnetic moments of the trivalent. iron ions are mutual-
Magnetic Properties o Sods 257

ly compensated and the piagnetization is dua to the magnetic momeis

o the bivalent metal ions (Figure 7.23(b)).
A remarkable property o ferrites is the combinatioa o excellent
magnetic parameters (high magnetic perineahility, small coersive
forc, high saturation magnetization, etc.) with a high elctrica!
resistance (o the order o 103 ohm-m). This particular property enabled
ferrites to revolutionize the neld o high and uitra-high requency
elecfronics. It is well known that ordinarv low res.istivitv
( 10-3 ohm-m) ferromagnetic ma-terials caunot be used at such
frequencies because o the extremely high eddy current losses. This was
the reason why ferrites have occupied a unique position in this eld.
Lately ferrites with a high coersive forc have been rieveloped.
They are used to construct permaueat magnets capadle o competing
with eiectromagnets. Ferrites with a rectangular hysteresis loop
are now widely used as digital storage elementa in computers.

71 Magnetic resonance
The magnetic moment o atonas, ions, and radicis with unpaired
electrons is determined by relatiou (7.33). There are 21 4- 1 ways
in which this moment may be aligned in a magnetic eld f0, there

Figure 7.2i Splitting o a ievei

oi aa atoin with. J=Z/2 in a magnetic
cf
eld.

Initial leve!

being correspondingly 2 / -f 1 dinerent projections o the moment

on the direction o the eld. The energy corresponding to each such
projection is
Um = o= Mjg'oPbH q (7.73)
Therefore an atomic energy level spiits in a magnetic eld into 2J -f-
-r 1 sublevis (Figure 7.24) the separations between which are
tdTja = g U g j.l3 .rf0 (/.Z^)

In the State oi thermal equilibrium the atoms are distributed

over those sublevis in accordance with the Boltzmann law:
nt = C exp (J 11
kBr } *B T
.n oq
 258 Solid Stste Physics

where ^ is the number of atoms occupying tho level with m} J ,

and Tin the level with m: = J 1.
To effect transitions of the atoms rom the lower to the higher
sublevis an external eiectromagnetic eld can be used. Spectro-
scopic selectioa rules ailow only of such transitions which result
in a unit change in the magnetic quantum number:
Anij = (7.75)
that is, only transitions between adjacent sublevis the energy di'er-
ence between which is g u0u.f0. Such transitions can be excited by
an eiectromagnetic field whose energy quanta are
n a = g p o p s ifo (7.76)
Sinca transitions from the lower to the higher ievels require a sup-
ply of energy, an intense absorption of eiectromagnetic energy will
set in if condition (7.76) is ulfilled.
Condition (7.76) is the condition for electrn paramagnetic reso-
nance (EPR). The resonance requency, as impiied by (7.76), is
a function of the constant magnetic eld intensity H 0. At II0
5.6 X 105 A/m, vres 2 X 104 MKz, which corresponda to the
wavelength ?. 0.016 m.
A similar phenomenon is the nuclear magnetic resonance (NMR)
due to the nuclear magnetic moment. For instance, in the case of
prctons the nuclear resonance for II0 = 5.6 X 104 A/m occurs at
a frequeney of vres ~ 30 MHz corresponding to a wavelength of
eiectromagnetic radiation k 10 m.
The rst successful experiments on electrn paramagnetic reso
nance were carried out by E. K. Zavoisky in 1944. He measured the
losses of eiectromagnetic energy in an electrical circuit caused by
paramagnetic absorption. In 1945 H. C. Torrey and R. V. Pound
used Zavoiskys method for the rst successful experiments on the
nuclear resonance of protons in solid parafn. That moment marked
the start of a rapid development of microwave spectroscopya formi
dable branch of physics dealingwith the interaction of radiowaves
with matter.
The magnetic resonance is extremely widely used in dierent
felas of science and technologv.
The nuclear magnetic resonance is the main and the most accurate
method for measuring the magnetic moments o atomic nuclei.
NMR has been helpful in collecting the data on the structure of
liquida, dielectric crystals. metis, semiconductors, and polymers.
The rst investigations of population inversin of energy Ievels
utilized in lasers were carried out with its aid.
The electrn paramagnetic resonance makes it possible to study
partiles possessing upnaired electrons and processes in which such
partiles take part. Those partiles inelude the conduction-electrons.
Magnefic Properiies of Soiids 259

the free and bondad radicis, many atonas and ions. EPB is success-
fully applied in the study of the naechanisms of Chemical reactions,
the radiation efiects in matter and in live tissues, the electronic
State of solids (metis, dielectrics, and semiconductors), and in many
other important fields of Science and technology.

72 Fundamentis of quantum electronics

Stimulated radiation. Negative ahsolute temperatuxes. The develop-
ment of microwave spectroscopy in recent .years led to ene of the
most momentous technical discoveriesthe discovery of the eid
quantum electronics based on the ideas first put forward by Soviet

Figure 7.25 Two-level quantum system.

*
. -O<X><X;-CH0^0<K><>vO

physicists V. A. Fabrikant, N. G. Basov, and A. M. Prokhorov.

Let us take a look at those ideas.
An external radiation'airected at a quantum system causes not
only the transitions from the lower to the upper leveis in which
the energy is absorbed but also the transitions from the higher to
the lower levis in. which energy is liberated. Such radiation is
termed stimulated.
Consider the simplest quantum system with only two leveis (Fig
ure 7.25). When N radiation guanta pass through a system, the diner-
ence A in the number of transitions from the lower leveis to the
higher leveis and from the higher leveis to the lower leveis will be
proportionai to the transition probability w identical for both the
direct and the reverse processes, to the number of quanta N, and
to the difference in the population of the leveis (n2 mj:
A = icN(n2 nx)
In conditions of thermal equilibrium the distribution of the
partiles over the leveis is described by the Boitzmann law:

(7.78)
17*
250 Solid Siafe Physics

Smce E z xtx, it rollows that nz <' nx and because of that reso-

nance absorption exceeds stimulated radiation and the system absorbs
incident eiectromagnetic energy eventually transorning it into
heat.
For stimulated radiation tp exceed resonance absorption the ther-
mal equilibrium of the system must be disrupled by raising the
population of the higher levels above that of the lower levels. that
is, to make n2 > Such population is termed population inversin.

Figure 7.26 Level population at various tempera tures.

oooooocooo
T= o K T >o K r <o

Quantum States with population inversin may conveniently be

described with the aid of the concept of negative absolute temperature.
From (7.7S) we llave
T = - .- (7.79)
In equilibrium nz Ez E1 E> 0, and T > 0. In case of
population inversin > nx, E x > 0, and consequeatly
T < 0. Figure 7.26 shows the occupancv of States at various tempera
turas. At T = 0 K all the partiles occupy the lowest level11 and
the systems energy is at its mnimum (i?mln = ?x). As the tempera-
ture rises some of the partiles go over to the higher level and the
energy of the system increases. At T oo the population of the
levels is equilized and the systems energy reaches the mximum
valu it can ha.ye in equilibrium [2?mai n {El 4- E z)l2], At
T < 0 K the p^puiation of the higher level exceeds that of the
lower level and because of that the energy of the systems turns out
to be greater than E mar. Henee the energy regin corresponding to
negative temperaturas lies not below the absolute zero, as would
appear at the flrst glance, but above infinite temperature. Should
such a system be brought in contact with a body whose temperature
is positiva, the heat would be transported from the system to the
body until a State of thermal equilibrium would be reached.
It should be pointed out that negative temperatura is a purely
quantum effect and may be observed onlv in systems with a limiten
set of levels.
12 To avoid contradiction with the Pauli exclusin principie Et and .
must be taken to mean narrow energy bands: from Ex to Ex -- E "and troca
En tO En -f- E.
Magnec Properties of Solids 261

One may realiza the population inversin in practice, for instance,

by quickly reversing the direction o a constant magnetic field
so that the time of reversal would be less than the relaxation time.
With the eld f0 ^ the initial direction the population of the lower
level is greater than that of the upper one; when the eld is reversen

Figure 7.27 Population inversin in

2 O^KXXXXKKXHO n2 a two-levei quantum system.

quickly the initial population of' the levels remains unchanged but
with respect to the new field H 0 direction it will De populated in-
versely (Figure 7.27).
Principies, of operation of masers. Suppose that an externa! signal
with the resonant frequency co ,= (E* Ex)/h is applied to a two-
levei system with a population inversin. This signal will induce

Figure 7.23 Excitation o paramagaetic amplifier (a) and radiation

o ampiied signal fb).

(a)

the transitions of the partiles: E.,* Ex and Ex-~ E.x. Since in the
case of population inversin the number of partiles on the higher
level exceeds that on the lower leve! and since the transition pro-
hahilities wX2 and w21 are equal, the stimulated radiation Ex
will exceed resonance absorption Ex-*- E and the signa! will be
ampiied. Thereore such a system will do the Job of an ampian^1
of eiectromagnetic radiation. The term for it is paramagnetic ampu-
ftsr (usuallv masar). ..
262 Scid Safa Physics

In practice not two-level but three-level quantum, systems ara

use for pararaagnetic ampliers. The active nateriais in thera are
diamagneicaiiy diiuted crystals of paramagnetic salts; in-wide
use are ruby crystals (A1,03) doped with chromium and germanium
ions. Figure 7.28(a) shows a quantum system with three levels:
Ex, E n_, 3. The dotted line represents the dependence of the number
o? partiles n on the energy E in thermal equilibrium. Sucia a'System
is placed in a magnetic eld H 0 and irradiated with high-frequency
clectromagnetic radiation with a frequency )13 = {Ez Ex)n.
This process is termed pumping. In pumping fields of high intensity
(Figure 7.28(b)) saturation is achieved, when the numbers of partiles
on the levels Ex and E 3 are equal (n3 = nx) and less than that on the
level Ex: nx < ? i,, n3 <.n. Should a signal with a frequency co21 =
= (E Ex)/ be applied to such a system, it would be amplied
by the stimulated transitions of the partiles from the level Ez
to the level Ex. The system will work as an amplier.
For a high amplification factor the dierence in the population
of the levels must be as great as possible. The way to do it is to cool
the system to liquid helium temperaturas. The concentration of the
magnetic atoras should be. low to exelude interactioa of magnetic
moments, which results in the widening of the absorption line and
in the decrease in the amplification factor. One solution is to use
diamagnetically diiuted crystals of paramagnetic salts, of which
the ruby crystal doped with chromium or germanium ions may
serve as an example.
The main advantage of the paramagnetic microwave frequency
ampliers is their ability to work at very low temperatures and,
consequently, at low noise levels. This makes it possible to receive
signis too weak for ampliers of conventional types. The frequency
tuning of the amplifier is done by changing the intensity of the
eld H0 which in its turn changas the resonance absorption frequea-
cy.
Principies of operation of quantum ----- - '.....r
on the basis of the nogative tsmperaiure quanm asiera a positiva
feedback should be provided to make sustained osciilations possible.
To this end the system is placed inside a cavity with reecting walls.
In conditions in which each spontaneously generated quantum /ico
st imilla tes the generation of, on the average, more than one quantum,
the amplitude of the electromagnetic osciilations o the appropriate
frequency <3 will grow continuously. The system will become seif-
excited. The raiative energy contained in the cavity is removed via
a wave guide. Such cavities are termed cavity resonaiors. They are
constructed from highly conductiva materials and their dimensions
are cise to the wavelength radiated by the lser. The feedback signa!
in provided by radiation reflectad by the cavity walls, a carsful
choice of the dimensions and shape of the resonator being necessary
Magnafic Prooeriies of Soiids 263

to obtain the right phase reiationship (the refiected radiation must be

in phase with the generated one at the point o generation).
The first quantum generators were designed bv N. G. Basov and
A. M. Prokhorov. They used two-ievel oscillators and the generator
operated oa beams of ammonia molecules. The ammonia molcula
NH, is made up of three hydrogen atoms arranged n the base of a
triangular pyramid and a nitrogen atom in the pyramids vertex

Figure 7.29 Ammonia molecule (a) and its energy levels (b).

(Figure 7.29(a)). This is the ground State o the molecule Ev The

molecule has an excited state E (Figure 7.29(b)) in which the nitro-
gen atom is forced into the base plae. The normal NH3 molecule
is asymmetric and by forc of this has a nonzero dipoie moment. The
dipole moment of an excited molecule because of its symmetry is
zero. This fact makes it possible to seprate the normal and excited
molecules by passing the molecular beam through a nonuniform
electric eld. Such a neld deects normal polar molecules; the excited
nonpolar molecules are nct delected.
Figure 7.30(a) represents a schematic view of a molecular generator.
It is made up of - three parts: the beam source A , the separation
system B. and the cavity resonator C.
The source of the molecular beam is a smail space 1 closed on one
side by a fine mesh 2. A gas pressure of one mm Hg is maintained
inside the space. The molecules forming the beam pass through the
mesh into a vacuum chambee praciicaiiy without collisions. A quad-
rupole condensen B (Figure 7.30(b}) is placed in the way of the
beam setting up a nonuniform eld (Figure 7.30(c)) that sorts out
the molecules. The excited molecules (on tho higher level) are con-
centrated cise to the condenser axis and the normal molecules aro
deected to the walls. In this way the beam cise to the condenser
264 Soltd Sfate Physics

axis is made to contain mainly the excitad molecules and thus a

quantum system with population inversin, or with negative tempe-
rature, is created.
From the separator B the molecular beam enters the resonator C
turned to the frequency of the radiative transitions from the upper
to the lower level (to the frequency o> = (E Ex)h] and induces
in it electroniagnetic oscillations with this frequency. The electro-
magnetic energy is transported via the wave guide ?.The molecules

Figure 7.30 Schematic representaron of a quantum generator operating

on a beam of ammonia molecules: (a)lay-out of amplier (4 is the source of
beam of ammonia molecules,
B the separation system,
C the cavity resonator); (b)quarupole condenser of separating system;
(c) electric field in condenser.

Output signa!

in the beam practically do not internet and becaus of that the spec-
tral line of the oscillations is very'narrow (i kHz at v = 24 MHz).
Another advantage of the molecular generator is its long-term fre
quency stability. The oscillation frequency is determined solely by
the structure of the ammonia molecule (that is, by E* Ex) and
is therefore independent of other generator Circuit parameters. This
made it possible ter-usb- the ammonia oscillator as a fbquncy stan
dard. The error of a dock using suen a frequency standard for its
pendulum does not exceed one second in 300 years of continuous
operation.
In recent years quantum generators fox the infrared and the optical
spectral intervals (lasers) have been developed. Their use opens up
wide possihilities for'the development of communication, in loca-
tors, etc. For instance, an optical communication channel would be
capable of carrying up to 10 000 TV progrmeles. Quantum gen-
erators nowada37s are widely used for the machining of tough mate-
rials, in medicine, and in other eids of Science and technoiogy.
Contact Phenom en a

73 Work funciion
The work function eoncept. The positive ions that make up the meta)
lattice establishin it an electric eld with a positive potential that
changes periodically along a straight line passing through the lattice
sites (Figure S.l(a)). As a rough approximation, this variation may
be negected and the potential be considerad constant and equal te

Figura 8.1 Metal as a potential

through: (a) intsrnal potential
of metal; (b) potential energ;/
oi electrn in metal

V0 at every point of the metal. A free electrn has a negativa poten

tial energy L\ = qV0 in this fela (q is the electrn charge).
Figure 8.1(b) depicts the change in the potential energy of the
electrn as it passes from the vacuum into the metal: in vacuum U =
= 0, in the metal U U0 = qV0. Although such a change
has the nature of a jump, it takes place over a distance 5 equal approx-
imateiy to the lattice parameter. It may be seen from Figure 8 .l(b)
that the metal is a potential trough for the electrn and that work
should be perfonnea to get the electrn out of it. This work is ierra ed
the work junction.
 <tJilute,jila,
266 Soiid Srare Physics

Should the electrons liave no kinetic energy, the work needed to

librate them. woold be equal to the depth of the potential trough Ua.
However, electrons possess kinetic energy oi translational motion
-even at absolute zero since they occupv all the lower energy leveis
oi the potential trough up to the Fermi level u. Therefore the energy
needed to make them leave the metal is less than U0. The least work
must be performed to librate the electrons occupving leveis cise to
the Fermi level. It is equal to the separation %of the Fermi level from
the zero level and the term for it is thermoynamic work function.
The problem of estimating the electrn work function of a semicon
ductor is somewhat more complicated. As may be seen from Figure 8.2

Figure 3.2 Electron work function

of semiconductor.

the electrons may leave the semiconductor from the leveis of the con-
duction band at the expense of work %0, from che impurity leveis at
the expense of work %i and from the leveis of the vaience band at the
expense of work y., and The least work %g is required to librate
the electrons from the conduction band. However, emission o f only
the conduction electrons woud, upset the equilibrium of the electrn
gas to reestablish which the electrons should go over to the conduction
band from the impurity leveis and from the vaience band. Such tran- .
sitions require work to be performed and in adiabatic conditions this
work is performed at the expense of the interna! energy of the crys-
tal, that is, as the State of thermal equilibrium is restored the crys-
tal is cooled. If the electrons leave the semiconductor from the va
ience band, to restore equilibrium soma electrons must go over from
the conduction band to the vaience band, which results in liberation
of energy and the crystal being heated. The equilibrium will be main-
tained and the temperature will remain constant only if the electrons k
leave the semiconductor from the leveis both beiow and above the
Fermi level in appropriate numbers. Theory shows that to maintain
equilibrium the average energy of the electrons leaving the semicon-
Confaci Pher.omena 0/

ductor should be equal to the Fermi energy, and this is the work func-
tion although there may be no electrons on the Fermi le'vel itself.
The work function is measured in electrn volts. The ratio of the
work function to the electrn charge is the voltage equivalent of the
work function. The work function measured in electrn volts is nu-
xnerically equal to its voltage equivalent.
Eect of adsorbed layers on work function. Molecular layers ad
sorbed by the surface of the solid, in particular monomolecular layers,
greatly affect the work function. Figure 8.3(a) shows a monatomic
caesium layer on the surface of tungsten. Caesium is an alkali metal.
Its outer valence electrn is bonded to the nucleus much more weakiy

Figure 3.3 Eormation of electric

double layer in the course
of adsorption of caesium (a) and
of oxygen (b) on tungsten surfaca.

(a)
^Oxygen ons

(b )

than the valence electrons of the tungsten atom. Thereiore in the

process of adsorption caesium donates its valence electrn to tungsten
and turns into a positively charged ion inducing a neegative chage of
equal magnitude in the metals surface layer. When tungsten is cov-
ered by a monatomic, caesium layer, an electric double layer is
puilt up at the surface with. its outer side charged positively. The
potential difference in this double layer aids electrn emissron out
of tungsten. Thereiore the electrn work function, as eterauned
from experiment, drops in the presence of the caesium layer from
4.52 eV for pur tungsten to 1.36 eV. The same is the eect of mon
atomic layers of other electropositive metis: barium, cerium, tho-
rium, etc.
The reduction of the work function by the adsorption of electro
positive metis is wiaely used for manufacturing vacuum tube cath-
odes, photocathodes, etc.
Quite different is the effect of oxygen adsorbed on the metal'
p a.

surface. The. valence electrons in the oxygen atom are bonded muc
268 Soiid State Physics

more strongly than in metis. Tkerefore in the process,;of adsorption

the o x y gen atom instead of donating electrons accepts two electrons
from tlie metal and turns it into a negativeiy charged ion. As
a result the outer side of the electric double layer becomes negativeiy
charged (Figure S.3(b)) and the resulting electric field prevens the-
electrons from leaving the metal, thereby increasing the work unc-
tion.

74 Contact of two metis

Contact potential difrerencc. Consider the process which takes place

when two metis (Figure 8.4(a)) whose energy diagrams are shown in
Figure S.4(b) are brought together.
The electrn gas in the individual metis 1 and 2 is charactezed
by the respective Chemical potentials ux and u2, the thermodynamic

Figure 3.4 Origin of contact potential dierence between n-type

conductora.

(c)

work functions being and jr. Let us bring the metis closer to
gether to within suca a distance d that an eflective electrn exchange
by means of thermionic emission or by direct transition from one
metal to another is possible. At the initial moraent after the contact
has been established t'here will be no equilibrium between the elec
trn gas in the first and second metis since the Chemical potential
(the Fermi level) lu is above ut. The diere-nce in the Fermi le veis,
Contad Phenomea 269

m _ u.lt results in the prevailing transition of tlie electrous from the

second metal to the first in the course of whicn the rst metal is
charged negatively and the second positively. The appearance of the
charges causes a shift in the energyleyeis of the metis: all the levels
in the negatively charged metai 1 rise and in the positively charged
m etal^sink as compared with their positions in uncharged metis.
This may easiiy be understood from the following simple consider-
ations. To move an electrn from, for instance, the zero level of
an uncharged metal to the zero level of a metal charged negatively
to a potential V1 work should be performed numerieally equal to
qVx. This work is transformed into the potential energv of the elec
trn. Therefore the potential energy of an electrn occupying the
zero level of the negatively charged metal will be higher bv the
amount qVx than the potential energy of an electrn occupying the
zero level of the uncharged metal. The zero level of a positively
charged metal- sinks below the zero level of an uncharged metai the
Teason for this being the same. A similar shift occurs in the position
of ot'ner energy levels of the metis 1 and 2 including that of the
Fermi level.
As soon as the continuously rising Chemical potential of the metal
1 (ti,) and the continuously sinking Chemical potential of the metai
2 (jj.) level out (Figure 8.4(c)), the cause for the predominant low
of the eiectrons from the rst metal to the second disappears and a
dynamic equiiibrium is established between the metis resulting in
the corresponding constant potential difference between'the zero
levels of both metis (Figure S.4(c)) equal to
r, - (& - * .)/? (S.i)
This potential difference is termed the external contad potential
difference. It follows from (3.1) that it owes its existence to the differ
ence in electrn work functions of the contacting metis: the elec-
trons leave the metal with the smaller work unction and settle in
the metal with the g-reater work function. -?-
After the Chemical potentials have been equalized the kinetic ener
gy of eiectrons occupying levels cise to the Fermi levels of both met
is is not the same: that of eiectrons in metal 1 is equal to En
and that o eiectrons in metal 2 to E7z (EFi A En ). Whea a direct
contact is established between the two metis, a predominant difu-
sion process of the eiectrons from the second metal to the rst sets
in and contines until the so-calied internal contact potential difference
Vi = ( E ^ - E m)!q (S.2)
is established.
Thickness of the electric double layer at the contact o two metis.
A double layer (Figure 8.5(a)) is established at the contact of two
metis across which the potential changes. abruptly by the amount
270 Solid- State Physics

Vl (Figure 8.5(b)). Let us estmate the thickness of this layer. Sup-

pose that it is a plae condenser with the distance between the plates
equal to the double layer thickness d. Denote the charge on each pate
by 0 and the potential difierence by Ft. The capacitance of a plae
condenser with the pate area of 1 m3 and a dielectric with relative
permittivity e = 1 is C = s Q/d (e0 is the permittivity of ree space).
Using the reiation C = Q/Vx, we can rewrite this formula in the form
Q/Vi = s0/<. Henee we obtain
d = eoVJQ
The thickness of the double layer cannot be less than the lattice
parameter a 3 . A t F, 1 eV such a layer can be established if

Figure 8.5 Externa! ( 7 C) and internal (y ) contact potential

dierences appearing at the instant two dierent metis are brought
in contact.

id
(b)
% - <tlc
/
ii
. :t
'b'Mp
1

a charge Q ^ Fs0/a 3 X 10-2 C is transported from every square

metre of the contact surface of the rst metal to the second. This
corresponda to .n Qlq -^ 2 X 1017 m~2 electrons. There are ap-
proximately 1019 atoms on every square metre of a metal. If we as-
sume that each of them donatas to the electrn gas one valence elec
trn, we will obtain for the surface density of the electrn gas the
valu ns 1019 m~2. Comparing An with ns, we see that evemfor the
narrowest possible double layer (a 3 A) only two percent of free
electrons need be transported from the contact surface o one metal
to another.
Because of a small change in the electrn concentration in the con
tact layer and because of the small thickness of this layer in compa-
rison with the electrn mean free path the conductivity of the layer
cannot be much less than that of the bulk metal. The current passes
through the contact of two metis just as easiiy as through.the metis
themseives.
Contact Phenomena 27t

75 The mefal-semiconducior contact

Barrier laye?; Coasider a metal-semiconductor contact. Suppose-
that a metal M with its work function equal to Xm is brought in con
tact with an rc-type semiconductor S whose work function is x&
(Figure 8.6).
If Xm> Xs> the electrons will ow out of the semiconductor'into-
the metal until the Chemical potentials um and are equalized and
a State of equilibrium is established. A contact potential difference
Vc will be established between the metal and the semiconductor
whose order of magnitude will be the same as that in the metal-
metal contact (several volts). To establish such a contact potential

Figure S.6 Formation o f barrier layer in metal-semiconductor contact.

difference approximately the same number of electrons should be

transportad from the semiconductor to the metal as in the case of
the contacto! t-wo metis. For a lattice parameter of-the semiconduc
tor a 5 A (germanium) and an electrn gas concentration in it
n 1021 m~3 there will be ns 0U electrons on 1 m2 of the semi
conductors surface. Thereore the transport of 10lT electrons.
entails the depletion of about 103 atomic layers of the semiconduc
tor.
Henee the equalization of the Chemical potentials of a metal and
a semiconductor in contact with one another necessarily involves-
the transport of electrons to the metal surface from _a boundary*
layer of appreciahle Vvidth d of the semiconductor (Figure S.6).
The ionized impurity atoms remaining in this layer establish a
static positive space charge. Since there are practicaliy no free charge-
carriers inside this layer and since its width greatly exceeds the-
electrn mean free path, its specic resistance is very great and be-
cause of that it is termed barrier layer.
Effeet of eontact eld on semiconductor energy levels. The contact
potential difference Va between the metal and the semiconductor is-
built up over the entire width d of the barrier layer (Figure 8.6).
Assuming that a 5 , we obtain for the number of depleted atomi&
layers AN 103 and for the thickness of the barrier layer d &
5 X 103 5 X 10~7 m. For Va i V the eld intensity will
be o: m V jd s 2 X 106 V/m. This is at least three orders of mag-
272 Sod State Physics

-nitude less than the intensity of the internal crystal eld, which is
responsible for the energy band pattern of the semiconductor. -There-
fore the contact eld cannot appreciably affect the band spectrum
(the forbidden band width, the impurity ionization energy, etc.).
Its action is limited to the .d'eflection of the semiconductors energy
bands. Let us dwell on this.
Figure 8.7(a) shows the energy band aiagrams of the metal M and
the semiconductor S before thev have been brought in contact. When

Figure 3.7 Enect of contact eld on semiconductor s energy levels:

(a) band patterns of metal M and semiconductor S\ (b)deflection
o semiconductcr's energy bands by contact eld.

contact h i^ f^ h ""achieved and equilibrium has-been established, a

positive space charge throughout the barrier layer width d is built up
(Figure 8.7(b)). In the absence of the contact eld the energy levels in
the metal and in the semiconductor are represented by horizontal
straight lines. This expresses the fact that the energy of an electrn
occupying this level, for instance, the lower level of the conduction
band, is the same everywhere in the semiconductor and does not
depend on the position of the electrons. In the presence of a contact
potential difference thepicture is changed: inside the layer in which
the contact eld is concentrated a forc acts on the electrn pushing
it out of the layer. To overeme this forc work should be performed,
this work being equal to the potential energy of the electrn in the
contact eld. Therefore as the electrn moves inside the space charge
layer its potential energy <p (x) increases reachiug its mximum cp3 =
Conac phenomana 273

= qV0 at the semiconductors surface. Quantum mechanical calcu-

lations lead to the conclusin that the application of an external
eld to the semiconductor results in an inclination o its energy
bands in relation to the horizontal Fermi level. The contact eld
acts in the same way causing a deflection of the energy bands. The
quantity cp0 is termed the equilibrium potential barrier for electrons
going over from the semiconductor to the metal.
To estmate the unction cp (x) we apply the Poisson equation
known rom electrostatics. This equation relates the eld potential
V (x) to the density p (x) of the static space charge responsible for
this eld. The equation is of the form
1= p (*) (8.3)
dx~ e0e
where e is the relative permittivity of the semiconductor.
It is expedient to go over from the potential V (x) to the potential
energy of th electrn cp (x) = qV (x) and to write the Poisson
equation for cp (x) as
dzcp
(8.4)
~?
In calculating the space charge density p (x) we shall assume all
the donor atoms iVd to be ionized and their electrons to be transferred
to the metal. Then p (x) = qN. Substituting this into the Poisson
equation. we obtain
z<p
(8.5)
1=?
If we assume that there is no contact eld at a distance x ^ inside
the semiconductor, we will be able to write the boundary conditions
for this equation in the form

(8 .6)
* * > - 0' ( r L) , - i =
Solution of equation (8.5) with the boundary conditions (8.6)
yields:
(8.7)

It follows from (8.7) that the potential of the contact eld diminishes
parabolically with the increase in x in the semiconductor.
For x = 0 we find that (p = %s. Substituting this into (8.7),
we obtain the widtli of the barrier layen
Oo oV
Y c.
( 8 . 8)
r
where n^0 N d is the concentration of electrons (majority carriers)
in the n-fcype semiconductor.. 5tr:
13-0SS5
274 Solid Sfaie Physics

It follows from (8.8) that the thickness of the barrier layer d in

creases with the contact potential dinerence V0 determined. by the
difterence in work functions and decreases with the concentratfon of
majority carriers in the semiconductor. Table 8.1 shows the vales
of d calculated with the aid of (8.8) assuming that Vc = 1 V and
e = 10. *
Table 8.1

nnt (m-3) d (m)

I034 3 x 10-7
103- 3 X 10 a
1020 3 X 10-s

It follows from the table that for the electrn gas concentrations
typical of semiconductors (1020-1024 m-3) the thickness of the sur-
face layer containing practically no free carriers mav attain vales

Figure 8.3 Formation of antibarrier

layer in metal-semiconductor contact.

? (x )
Vn

^ < *s

from one to three orders of magnitude greater than the electrn mean
free path. This is the reason wny the resistance of the barrier layer
is enormous.
I the work function of an rc-type semiconductor exceeds that of
a metal, the electrons shall be transported from the metal to the
semiconductor setting up a negativa charge in its contact layer (Fig
ure 8.8). In this case the energy of the electrn cp (z) as it approaches
the surface does not increase but, on the contrary, decreases with the
result that the bands are deected in the opposite direction. This
leads to an increase in the free charge carrier concentration inside
the contact layer of the semiconductor and to a consequent increase
in its conductivity. For this reason such a layer is termed antibarrier.
Reetifieation at a metal-semiconductor contact. A remarkable
feature of the barrier layer is a drastic aependence .of its resistance
on the direction of external voltage applied' to the contact. This
 Contad Phenomena
275

Figura 8.9 RcctiCcation

in metai-semiconductor contact:
(a) equilibrium State o contact;
(b) external voltage applied
in reverse direction;
(c) external voltage applied
in forward direction.'-

dependence is so strong that it

resuits in practically a unidirec-
tional conductivity of the con-
tact: a cnrrent passes easily
through the contact in the formar
direction and much worse in the
reverse direction. This is the
essence of the rectifying property
of a metal-semiconductor con
tact. Let us discuss this point in
detail.
Figure 8.9(a) shows the energy
band pattern of an rc-type metal-
semiconductor contact in the
State of equilibrium. The poten-
tial barrier for the electrons
going over from the metal to the
semiconductor is equal to the
diSerence in work functions
7.m Xs'- fr the electrons pas-
sing from the semiconductor to
the metal it is cp0 = qVc. Denote
the electrn flux from the metal
to the semiconductor by n.s
and the flux from the semicon
ductor to the metal by n^.3. The
corresponding current densities
flowing through the contact are
s>m and z'm. s. Since in the State
of equilibrium the total current
flowing through the contact is
zero it follows thatt is.ra m.s* Denots the current density corre-
snonaing to the equilibrium currents tf.
s.m and 4 .3 hy eqi

0
*s. m
i
m. = lq (8.9)
18 *
276 Solid State Physics

Apply an external potential dierence V to the contact in the di-

rection of the contact potential diereuce Fc imparting a positiva
charge to the semiconductor with respect to the metal (Figure 8.9(b));
such direction of F is termed reverse. The resistance of the barrier
layer is usually some orders of magnitude greater thanthe resistance
of the other parts of the Circuit and because of that practically the
entire applied voltage is built up across the barrier layer. The energy
levels in the positively charged semiconductor are deflected down-
wards by qV from their initial positions. The same will be the dis-
placement of the Fermi level xn. The deection takes place along the
entire barrier layer thickness d across which the potential rises by
F. The bottom of the conduction band Ec and the top of the valence
band Ey in Figure 8.9(b) have been drawn so as to take account of
the new position of the Fermi level. It may be seen from this drawing
that the external voltage F applied in the reverse direction raises the
potential barrier for the electrons going over from the semiconductor
to the metal to
<p (0) = <p0 + qV (8.10)
According to (8.8) the thickness of this barrier layer will be

d = ']/ <!s(Ve + y) (8.11)

V 9nno
Henee the external eld applied to the contact in the reverse di
rection raises the potential barrier for the electrons flowing from the
semiconductor to the metal and increases the barrier layer width.
The picture will be different if a fonvard voltage is applied to the
contact (Figure 8.9(c)). In this case all the levels of the negativelv
charged semiconductor including the Fermi level un are deflected up-
wards by the amount'-gF lowering the potential barrier for the
electrons flowing from the semiconductor to the metal by the amount
qV. The barrier height becomes
^ ^ _ q,o _ qy (S.12)

The width of the space charge layer decreases accordingly and be-
comes equal to

d = ]/" 18.13)
Y Qnno
The change in the potential barrier height disturbs the equilibrium
between the electrn fluxes flowing through the contact in both di-
rections. When the external voltage F is applied to the contact in
the reverse direction, the current aensity m>s corresponding to the
electrn flux ns.m decreases e<r'**T times since according to the
Boltzmann law the number of electrons flowing from the semicon
ductor' to the. metal capable of surmounting the barrier <p0 -j- qV is
Conrac Phencmana 277

eqV/hs T times less than the number owing thrcugh. tbe equilib-
rium barrier qp0l that is, ns.m = ns.ms~?V/aBi . Therefore the cur-
rent im.s becomes equal to
i - f
f

The current density s.m corresponding to the electrn flux

will remain equal to since the external eld does not change the
height oi the barrier for electrons owing rom the metal to the semi
conductor: its height remains equal to the diSerence in work unc-
tions, 'Xm Xs-
The total current density in the reverse direction is (Figure 8.9(b))
r = i ^ e - ' v,V _ (e- "*r - 1 ) (8.14)
The current flows from the semiconductor to the metal. As the re
verse voltage Y is increased the exponential e~qy,h*T tends rapidlv
to zero and the reverse current .density to its limit valu eq. The
current density eq is termed saturation current density and 7eq =
= iSi'eq saturation current (S is the cross-sectional area of the metal-
semiconductor contact).
When a fonvard external voltage V is applied (Figure 8.9(c)),
the potential barrier for the electrons owing from the semiconduc
tor to the metal is lowered by the amount qV and because of that
the current density of those electrons increases e^ 'k%T times in
comparison with eq becoming

The current density s.m remains equal to teq. Therefore the densi
ty o the fonvard current (from the metal to the semiconductor)
will be

.... r= m. s h. m= eq ~ *) (8.15)
and 'it grows exponentially with V. Combining.-(&.14) with (8.15) ,
we obtain
= q ( = , s T - 1) (8.16)
(V = |F for fonvard bias and V = j V\ for the reverse).
Formula (8.16) is the equation of the current-voliage characieristic
(CVC) of a rectiying metal-semiconductor contact. Figure 8.10
depicts the CVC of such a contact.
The ratio of the fonvard current to the reverse current for the
same absolute valu of applied voltage is termed reciification razio.
Its -valu for good rectiying contacts may be as high as tens and
even hundreds of thousands. .
The potential barrier at the metal-semiconductor interface is
citen termed Schoiiky barrier. Presentiy Sckottky barrier dioaes
278 Soiid Siae Physics

with extremely short switching times are being developed. The

diodes have switching times ag short as 10~u s. This makes it pos-
sible to use them eectively in radioelectronic pulse circuits, in
Computer and automation circuits where there is a need for high

i Figure 8.10 Current-voltage"

characteristic of metal-serniconductor
contact.

operational speeas, that is, where extremely short and quickly

recurring electrical pulses have to be processed.
The nonrectifying (antibarrier) metal-semiconductor contact is
used to provide ohmic contacts bv means of which the semiconduc
tor device is connected into the electrical Circuit.

76 Conacf beween iwo semiconductors

of different types of condudivity
The progress in semiconductor electronics is based mainly on tie
use of contacts of two impurity semiconductors of different conduc-
tivity types. Such a contact is termed a p-n junction. Let us briey
discuss its properties.
Preparation of ^-/^junctions. It is impossible to obtain a trae
p-n junction by means of a mechanical contact of the n- and p-types
of semiconductors because the lattice discontinuity at the interface
contains more defects than there are impurity atoms on each.o the
contacting suraces. Therefore the p-n junction was successfully
prepared only when the art of making it in the form of an interna!
boundary in the bulk of a single crystal semiconductor was mastered.
One of the more wiaely used methods of preparing p-n junctions
is the method of alloying, an example of which is aiscussed below.
An rc-type germanium wafer witn a piece of indium placed on it.
(Figure 8 .ii(a )) is held in a furnace at a temperatura 500-600 C
in a hydrogen or argn atmosphere. The indium melts ana dissolves
some germanium (Figure 8.l(b)):. As the wafer is slowly cooled
germanium saturated with indium preciptates-- 'rom the melt
Contad Phenomena 279

It crystallizes as a continuation of the siDgle crystal of the wafer.

Since the germanium doped with indium has a p-type conductivity,
there will he a p-n junction at the boundary between the rc-type
single crystal that had not heen dissolved and the recrystallized
regin (Figure 8.11(c)). The indium drop alloyed to the germanium
surface is used as an ohmic contact.
The p-n junction can be prepared by difusing acceptor impurities
into an rc-type semiconductor or donor impurities into a p-type

Figure 3.11 Fabrication of alloyed p-n junction: (a)room temperature;

( b)tamperature T 500 C; (c)room temperature.

-In-Ge eutectic
Molten In p-type G e_
In saturated with Ge single crystal
vm ?m
w & m |

(b )
single crvstal
(c)

semiconductor. The difusin process may be carried out from the

gaseous, liquid, or solid phases. The penetration of the impurity
and the depth of the p-n junction is determined by the temperature
and the time of the difusin process. The p-n junction itseif is the
boundary that separates the regions of diferent conductivity types.
A widely used method is the epiiaxial method of preparing p-n
junctions which consista in the deposition via Chemical reactions
on. for instance, an n-tvpe Silicon wafer of a p-type single crystal-
line Silicon lm in the gaseous phase or recrystallization from the
liquid phase being used for the process.
A method growing in popularity in recent years is the ion implan-
tation method, in which energetic ions of specic impurities (energy
in the range of 50-300 keV) are directed at the semiconductor surface
and pentrate into the bulk of it (to a depth of the order of 0.1-0.5 una,
depending on the energy and the type of impurity).
Equilibrium State o a p - n junction. Let the plae M M be the
internal boundary between two semiconductor regions of diferent
conductivity type (Figure 8.12(a)): to the left is the p-type semi
conductor, for instance, p-germanium, with an acceptor concentra-
non iVa, and to the right an n.-type semiconductor (n-germanium)
with'a donor concentration /Yd. For the sake of simpiicity we shall
assume iYa == N d to be equal to 10" m-3. Figure 8.12(b). depicts
the change in the acceptor and donor concentrations along the x
280 Soid State Physics

axis perpendicular to the plae MM. At point 0 lying in this plae

the acceptor concentration abruptly vanishes and the donor.con -
centration increases abruptly from zero to N.
The majority carriers in the n-type regin are electrons, and
in the p-type regin holes. The majority carriers are due almost
entirely to ionized donor and acceptor impurity atoras. At temoer-
atures outside the extreme low temperatura ranga practically* all
of those impurity atoras are ionized and because of that the electrn
concentration in the n-region ( nn0) can be assumed to be equal to
the donor concentration N d {nn0 iVd), and the hole concentration

Figure 8.12 Equiiibrium State of p-n junction: (a)internal boundary M M

between p- and n-re_gions; (b)impurity istribution in p- and n-regions;
(c)distribution o immobile charges in p-n junction; (d)position
of Fermi leveis at the time of imaginary contact of p- and a-regions:
(e)deflection of energy bands in the course of p-n junction formation
and formation of a space charge layer.

Zero ievel
i
- 1
1

w m m m * . .
O 80G 0 9
a f
%l>i V-?
i \
^>Fermi ievei A n
i
i

eoe60000

' (d)
1
 i

Contacf Phenomena 28f

in the p-region (p p0) can be assumed to be equal to -the concentra-

tion of the acceptor atoras (pv0 N 3).
Besides majority carriers those regions contain also minority
carriers: the n-region contains holes (pn0) and the p-region elect-
rons (np0). Their concentrations may be found from the law of mass
action (5.44)
nnOpnO PpoKpO ~ n'-
where n is the concentraron of carriers of one sign in the intrinsic
semiconductor (germanium). At = p p0 1022 m-3 and n
= 1019 m~3 we get pn0 = np0 =_1016 n r 3.
We see that the hole concentration in the p-region is six orders-
of magnitude higher than in the n-region. Such a diference in the
concentrations of carriers of one type is the Cause of difusion flux es
of electrons from the n-region to the p-region (ftn_o) and of holes
from the p-regon to the n-region (pp- n). The difusion flux of the
electrons out of the n-region imparts a positive charge to this regin,
and the hole flux from the p-region imparts a negative charge to
the p-region. Such charges raise the position of all the energy leveis
including the Fermi level in the p-region and sink it.in the n-region.
The electrons continu to flow from the right to the left and the
holes from the left to the right until the gradually rising Fermi
level of the p-region (pp) reaches the level of the gradually sinking
Fermi level of the /i-region (uR). As soon as those leveis are equal-
ized a State of equilibrium is established between the n- and p-regions
when the electrn flux from the n- to the p-region (nn_ p) is compen-
sated by the electrn flux from the p- to the n-region (nP_ n) and
the hole flux from the p- to the n-region (pp_) is compensated by
the hole flux from the n- to the p-region (pn_,):
nn-P = ^-n> Pp^n^Pn-p (8.17)
As the electrons leave the contact layer of the n-region a static
positive spaee vcharge of ionizad donor atoms is left in it (Fig
ure 8.12(c)). Denote the width of this layer by dn. As the holes leave
the contact layer of the p-region a static negative space charge of
the ionized acceptor atoms is left there. Denote the width of this
layer by dp. A contact potential dierence Vc is built up across
those layers, which constitutes a potential barrier' <p0 localized
in the p-n junction; cp0 prevens the electrons from going over from
the n- to the p-region and the holes from the p- to the n-region.
Calcuiations show that

<p0= AbT ln -222. = k3T ln -222- (8.18)

npo . Pno
It follows from (8.18) that cp0 is the greater the greater the ratio
of the majority carrier concentration in one regin of the p-n june-
232 Soid .Sfaa Physics

tion to the concentraron of the carriers of the same type in another

regin where they are minority carriers. At nn0 = 102a m~3, np0 =
= 1016 m 3, and T = 300 K we see that cp0 0.45 eV.
Figure 8.12(d) shows the energy hand pattern of the p- and n-re-
gions at the instant of their nnaginary contact, that is, before equi-
iibrium between them has been estabiished. It may be seen rom
Figure 8.12(d) that un lies above pp.
Figure 8.12(e) shows the energy band pattern of those regions
after equilibrium has been estabiished. The Fermi levels Un and up
coincide and there is a space charge layer between the p- and n-re-
giocs that spreads into the n-region to a depth dn and into the p-region
to a depth dp forming a potentiai barrier with the height qp0 = qVc.
Comparing Figures 8.i2(d) and (e), one may easiiy see that

<Po m, |Ap (8.19)

The width of the space charge layer d = dn 4- dp, as in the case
of the metal-semiconductor contact, is determined by the height
of the potentiai barrier cp0 and by the concentrations of majority
carriers in both regions of the p-n junction and p p0:

:say
d = y /* 2es(p0 inno ~ Vpn) no' ' Pp) (8.20
r nnaPpo nnoPt)o

It may, however, be demonstrated tliat in the p-n junction the

barrier width d depends eventuaily only on the majority carrier
concentrations nn0 and p D0. Indeed, substituting (5.38) into (8.18)
and the result into (8.20), we obtain
2ss0 (nn pm)
k3T In | nno??o j
Q~. nno?p

It follows from (8.20) that the space chargeTayer width is the great-
er the less the majority carrier concentra tion in the n-and p-regions
of the semiconductor.
If One of the regions, for instance the ?i-region, is substantially
less doped than the p-region, so that n^a p p0, we will obtain
from (8.20) the following:
1/ 2ge0<Pa (8.21)
^ 2'no
In this case almost the entire space charge is concentrated in the
low-doped (high resistivity) ra-region, the same as in the case of
a metal-semiconductor contact.
Rectifying properties of p - n junctions. A remarkable property of
the p-n junction on which the operation of most semiconductor
devices is based is its ahility to rectifv alternating current. Let us
discuss this property in more detail.
Conacf Pnenomena 2S3

Currents flowing through p - n junction in equiiibrium. In the State

of equiiibrium fluxes of majority and minority carriers flow through
the p-n junction (Figure 8.12(c)). These fluxes are such that the
flux of eiectronsmajority carriersflowing from the n- to the
p-region (nn_ p) is, according to (8.17), equal-to the flux of elect-
ronsminority carriersowing from the p- to the n-region (nP^n).
In the same way the flux of holesmajority carriersflowing from
the p- to the n-region (pp_ n) is equal to the flux o holesminority

Figure 8.13 Calculating minority

ic -i. carrier currents owing through.
equiiibrium p-n junction.
p -regin -regin
Ln 'ti Lp/~p
_O
,IItr

-<7<C
/
d p i.

~T~ N
/ qoc
r s
df in

carriersflowing from the n- to the p-region (p,,_;,) Denote the

current densities corresponding to those fluxes as oilows: the flux
corresponda to in, n p^ n to ip and p _ P to ?_eq. In accordance
witi (8.17) we can write
^n-eqi tp
cq ( 8 . 22 )
Adding the right- and left-hand sides of those equations, we obtain

l n T* "p = ta-eq W Jp-eq

The left-hand: side o the relation expresses the component of the
full current due to the majority carriers, and the right hand side
the component due to the minority carriers. The full current flowing
through the p-n junction will evidently be zero:

i (ti T ip) (^n-eq "T P-eq) 0 (8.23)

Let us calclate n_eq and To this end cut out a unit area
S o the left boundary 1 of the p-n junction (Figure 8.13). Using
it as a base, buiid a cylinder with the side L J - n, whers Ln is the
diffusion length of the eiectrons in the p-region and their average
lifetime. Since the diffusion length is the average distarme the cer
ner passes during its lifetime, the ratio Lnh n evidently expresses
the-average speed of eiectrons diusing from the bulle of the p-region
where their concentration is np0 to th boundary 1 where thev are
drawn into the ccntact eld Sc and transportad to the n-region.
284 Sola Safe Physics

Figure 8.14 Rectifying action

o{ p-n junction: (a)equilibrium
siate of p-n juaction; (b ) orward
voltage applied to p-n junction;
(c)reverse voltage applied
to p-n junction.

The number of electrons con-

tained in the cylinder is equal to
its volunte Lnxn multiplied by
the electrn [concentration np0,
that is Lnnpojxn. A ll those elec
trons will pass through the unit
area S in one second aDd [will
be transported to the rc-region
establishing a current with a
density
tn-eq = ?T<n?ipo^n (8.2-i)

One may similarly calclate

by. building a cylinder o
unit base with the side equal to
LpJxp at the boundary 2 of the
p-n junction:
p-eq = q L PP n0/Xp (8 .2 5 )

Henee in the State o equilib-

riuxn in the p-n junction
n. n-eq = ?21n/tpo/Tn,

p = p-eq = ?-^pPn.O^? (>.2)

(c)
Forward current. Apply a or-
ward voltage V to a p-n junction
in a State o equilibrium (Figure (8.14(a)) by ccnnecting the positiva
terminal o the power suppiy to the p-region and the negative termi
nal to the n-region (Figure 8.l4(b)). This voltage brings the no ten-
tiai barrier or the majoriy carri.ers dowa to cp0 qV. Thereore
the electrn ux rom the n- to the p-region (nn_p) and the hole
flux from the p- to the n-regin will increase e?F/f!3- times with
the resuiting similar increase in the maiority-carriers
ties in and A:
; _n T n <?V/&3 T /
*p H^-fpsrnOs ' wp
Coniacf Phenomena 2S5

At the same time the current densities of minority carriers i

and p_eq whose magnitude is independent of the p-n junctioiFs
barrier height shall remain the same as expressed. by formulas (8.26).
Therefore the total current flowing through the p-n junction to which
a forward voltage V has been applied and termed forward current it
will now be not zero but

h = {"nJTp) (n-eq ~r p-eq) = ? [ np0 + Pao) (e5^ * 3 " 1)

(S.27)
Reverse current. Apply now a reverse voltage V to the p-n
junction by connecting the negativo terminal of the power supply
to the p-region and the positiva one to the n-region (Figure 8.14(c)).
This voltage raises the potential barrier of the p-n junction to <p9 -r
qV with the result that the uxes of the majority carriers nn_ ?
and pp-n decrease eqV,kzT times together with their currents in
and ip. The atter will be equal to:

. = e - qVIV h ^ h = oL pPm

The total current owing in the p-n junction termed reverse current
T will be
' . . L
h ~p) (n-eq-rp -eq )::=l?( T ~ N + 7 f 4 [ < ' " r' v - l ) (S.28)
' >71
Current-voltage characteristic (CVC). Combining (3.27) and (8.28),
we obtain the equation for the current-voltage characteristic of the
p-n junction:

i (8.29)
npo ~r ~ Pao - i )

where V > 0 is the forward and V < 0 the reverse voltages.

Let us analyse this formula. With. the increase in the reverse
voltage V the exponent e-?v/A3i Q and the expression
(e-?V/S3i _ l. Accordingly, the current density ir tenas
to its limit

e q = - 7 ( - r - 'l p o r ^ - P n o ) (3.30);

termed saturaiion current density. Practically this valu is reached

already at qV & 4 k%T, that is for V 0.1 V. It follows rom (8.30)
that gq is determined by the minority carrier uxes through the
p-n.junction. Since their concentrations are not large, is a smaii
quantitv. For germanium p-n junctions of the type discussed here
{nno pp0 1022 m~3) at room temperature it is of the order of
10-2 A/m2; for Silicon p-n-"junctions it is much less.
235 Sod State Physics-

When a forward voltage V is applied to the p-n junction, the

currsnt density through it increases exponentially and reaches big-
vales already at small voltages. Substituting (8.30) into (8.29)
we obtain:
= eq(e- 9V/ABr _ i ) (8.31)

Figure 8.15 shows the plot of a currentwvoltage characteristic

of a p-n junction which corresponds to equations (8.29) and (8.31).

Figure 8.15 Current-voltage

I characteristic o p-n junction.

It is drawn to diSerant scales for the forward and reverse branches

since shouid the same scale be used for the reverse current as for
the forward current the reverse branch would coincide with the
x axis. Indeed, for Vr = 0.5 V the reverse current density is '
i r sq and for Vt = 0.5 V the forward current density i
eq.s/.o25 i^e20 Sq 109, since at T = 300 K (room temper-
ature) kBT 0.025 eV. As we see the rectication coefficient at
such a voltage is itliT 109 and this proves thaf a p-n junction
exhibits a practically unidirectional conductivity.
Deerioraiion o rectifving properties at high temperatures. ccord-
ing to (5.44)
PnO ^v^nO* ^?0 = ni P p O
where
2rc Y n l7lr)
i-2( n-
It raay easily be seen that will rise rapidly with the increase in
temperatura while nn0 iYd and pp0 Arawill remain practically
constant. Therefore at some temperatura n5 may become as high
as nna
no or np Then the concentrations of the minority carriers wiih
O'

be as nigh as the concentrations of the majority carriers:

Pno TLiJn 1
j j i j nna and = nf/ PpO PpQ/Ppo Pp O -

The notential barrier in the p-n junction which is responsible for

Contad Phenomena 237

its rectifying properties will then cease to exist:

<p3 = kBT l n ^ - t t k 3T l n ^ Q
PjJO
togethsr with the abiiity of the junction to rectify alternating cur
ren!. It ollows from (5.37) that the corresponding temperature
will be the higher the higher the forbidden band width E g of the
semiconductor. For germanium p-n junctions (Eg = 0.62 eV) the

Figura 8.16 p-n Junction breakdown:

2 thermal, 2avalanche; 3 tunnei.

highest operational temperature is 75-90 C; for Silicon p-n junctions

(Eg = 1.12 eV) it can be as high as 150 C.
Breakdown in y > n junctions. If the reverse voltage is continuously .
increased, a voltage F.d will be reached at which the resstance
of the barrier layer drops drasticaliy and the reverse current jumos
up. This phenomenon became known as the p-n junction breakdown
(Figure 8.16).
There are di5erent types of breakdown: thermal, tunnei (or Zener),
and the avalanche breakdown in accordance with the nature of the
physical processes that cause the reverse current to grow abruptlv.
Thermal breakdown occurs when the heat generated hy the reverse
current flowing through the p-n junction is not completely removed
from it and raises its temperature. A rising temperature leads to an
increase in the reverse current and this, in its turn, raises the temper
ature still further, etc. This Progressive process results eventually
in thermal breakdown. The character of the increase in current
during such a breakdown is depicted in Figure 8.16, curve 1.
When the electric eld intensity in the p-n junction is high enough,
impact ionization of the- atoras o the semiconductor may take place.
This wll result in an avalanche-ype increase in the carrier con-
centration and in the avalanche breakdown; the character of the
increase in current is shown in Figure 8.16, curve 2.
In a narrow p-n junction already at comparativeiy low reverse
voltages an electric eld may be established high enough for the
tunnelling of the electrons through the p-n barrier to.take place.
The resuiting breakdown is termed tunnei, or- Zener, breakdown
-3 S Sod State Physics

and the character of current increase r (V) for this case is shown
in Figure 8.15, curve 3.
In most cases the p-n junction breakdown is a harmul eect
that limits the practical use of the junction. At the same time the
eect was utilized to develop a wide range of semiconductor devices.
known as Zener diode voltage regulators, which will be discussed
in the ollowing section.

77 Physlcal principies of semiconductor p-n

junction devices
As han already been stated before, the rapia progress in semiconduc
tor eiectronics %vas the direct result of the development of the p-n
junction technology on which. the design of various semiconductor
devices is based. Let us discuss the general principies of such devices.

Figure S.17 Schematic reprssentation of a rectier diode:

l p-7i junction, 2 and 3 passive regions, 4ohmic contacta.

Rectier diodes. The nonlinear current-voltage characteristic of

the p-n junction (Figure 8.15) enables it to be usen to rectify alter-
nating current. A two-terminal semiconductor device fulllng such
function is termed semiconductor rectifier diode. Figure 8.17 shows
the principal schematic diagram of such a diode. It consista of a p-n
juaction J , passive n- and p-regions ,2, and 3, and ohmic contacta 4.
The high-resistivity regin of the crystal is termed base of the diode.
In our case it is the n-region 3 of the width W.
Nowadays p-n junction rectier diodes are made almost exclusive-
ly of Silicon. The efciency of such diodes is almost 100 percent
and in combination with their low weight, small dimensions, ease
of servicing, etc. this mane them a very widely used device for such
applications. Various diode types are designed to rectify currents
from several mili i amperes to several hundred or thousand amperes.
For greater currents the diodes are connected in parallel. The mxi
mum reverse voltage for various types lies in the range from 50 to
600 V. It may be much higher for special diode types. For use in
hgh-voltage rectiers the diodes are assembled in-series in stacks.
The reverse currents of various rectier types lie in the range from
fracions of a microampere- to teas o milliamperes.
Conlacf pnenomena 289

Impulse and high-frequency diodes. The second very important

field o application o he semiconductor diodes is the eld o impulse
electronics, computar electronics, automation, VHF electronics, etc.
In such applications the diode is required to process pulses o mni
mum duration and o mximum repetition rate. Thereore one o the
main requirements o diodes designed or such circuits is the speed
o operation, that is, switching speed from the direct to the reverse

Figura 3.18 Transient processes in diode: (a)majority and minority

carrier distributions; (b)-^minority carrier injection by forward voltage
and their difusin into the bulk o semiconductor; (c, d}variation?
of forward and reverse voltages and currents in a diode switched
on in the forward direction and reswitched to reverse direction.

State. To nd what lies at the origin o this speed let us discuss the
physical processes that take place when a p-n junction is switched.
Figure S.iS(a) shows the distribution o the majority and minority
carriers in the p- and n-regions f an equilibrium p-n junction.
When a direct voltage V is applied to the diode, the potentiai barrier
o its p-n junction sinks by qV and the majority carrier flux through
the junction increases esvkbt times with the result that the hole
concentration at boundary 1 and the electrn concentration at boun-
dary 2 rise to the vales o pn (0) pna and nv (0) > np0, respec-
tively (Figure S.18(b)). For the direct current to flow it is necessary
or those carriers to be drawn into the bulk o the semiconductor:
the boles should be drawn into the ?i-region and the electrons into
the p-region. The recomhination o those minority carriers takss
place inside the regions, or on the contacta if the width o the regions
is small in comparison with their difusin length. The removai
o the minority carriers proceeds either by means o difusin whose
rate is the higher. the greater the concentration gradient o boles.
290 Sod Saa Physics

(dp/dx) at boundary 2 and thatjof electrons (dn/dx) at boundary 1

or in special diode types by meahs of drift in built-in electric fields.
At the initial moment after the forward voltage had been applied
(Figure S.iS(c)) this gradient is extremeiy bigh since the boles trans
poned to the n-region and the electrons transportad to the p-region
are concentrated in narrow layers cise to boundaries 2 and 1. There-'
fore the initial forward current in the diode is bigh, being limited
practicalty only by the resistance of its passive regions (plateau 1
in Figure 8.18(d)). As the holes enter the n-region and the electrons

Figure 3.19 Equivalen! circuit o diode: i? non-linear resistance

of p-n junction, C capactanos o p-n junction; R 0bm ohmic resistance
o passive regions and contacts.

the p-region their concentration gradient drops and so does the

forward current (Figure 8.18(e)). After a time equal to the minority
carrier lifetime x (or to the transit time of the minority carriers
from the boundaries 1 and 2 to the contacts 4, which is even shorter)
a steady-state (independent in time) distribution of the holes in
the n-region and of the electrons in the p-region is established
(Figure 8.1S(b)) and the forward current assumes its normal valu
(Figure S.lS(d)).
When the diode is switched from the forward to the reverse State
(Figure 8.18(c)), the initial reverse current is very high since the
minority carrier concentrations at boundaries 2 and 2, which are
responsihle for this current, are high: the magnitude of the current
is actually limited by the resistance of the passive regions of the
diode (plateau 2 in Figure 8.18(d)). In the course of time the excess
carriers at the boundaries graduallv dissolve, their concentrations
arop to equilibrium. vales (pn0 and np0), and the reverse current
assumes its normal valu (Figure 8.18(d)). This- process lasts about
the same time as the rst (lifetime or transit time).
Thus, when the diode is switched, transient processes (of carrier
accumulation in the forward biased diode and of carrier dissipation
in the reverse biased diode) take place in it, limitingits switching
speed. Since the duration of those processes is approximately equal
to the minority carrier lifetime x and since a reduction in x increases
the switching speed, the tendency is to make t as short as possible.
Another method is to reduce the transit time of carriers by making
the dioaes as thin as possible (severa! microns or iess).
Contad Phenomena 291

On the basis of the aforesaid we can draw the conclusin that

a p-n junction behaves with respect to a short alternating signa! as
a resistance R located in the barrier layer shunted by the capacit-
ance C of the p-n junction (Figure 8.19). When a forward bias is ap-
plied, the initial current through the diode is mainly the current
which charges the capacitor C. This~current can be quite high. When
the diode is switched to the reverse bias, the initial reverse current
is mainly the discharge current of the capacitor C and it too can be

Figura 8.20 Circuit diasrram o f the simplest voltage regulaor using

a Silicon Zener diode Z.

1
H, *'01:t

quite high. To improve the speed of the diode and its high-frequency
performance it is evidently necsssary to reduce the p-n junction
capacitance C. This is done by reducing the area of the p-n junction
to the minimum. This raeasure together with other measures enabled
the switching speeds of modern diodes to be reduced to approximate-
ly 1 0 '5 s and the operating frequencies to be raised to 103 He.
Voltage regulators. A small increase in the reverse voltage in the
prehreakdown range causes a substantiai increase in the reverse
current (see Figure 8.16). This eect is being used to stabilize voltage
and the device is called the Zener diode regulator.
Figure 8.20 depiets the simplest Circuit diagram of a de voltage
stabilizer utilizing. a Zener diode. When the input voltage 7a is
increased, the diode s resistance drops drastically and the current
in the Circuit of the voltage drop resistor R r increases causing the
voltage drop across it to increase; the voltage across the load resis
tance R\ (the output voltage 7 0Ut) may remain practically constant.
Power supplies using Zener diodes are now quite a match or normal
cells.
Tunnel diodes. There is another very interesting and practically
important t3rpe of semiconductor devices, the so-called tunnel diode
which utilizes the quantum mechanical effect of electrons tunnelling
through a narrow potential barrier. The diodes are constructed from
heavily doped degenerate semiconductor material in which the
Fermi level lies not in the forbiaden band but just libe in metis
in the coduction band of an n-type semiconductor or in the vaience
band of a p-type semiconductor. Figure 8.21 (a) shows the energv-

9*
292 Sod Sae Physics

tha there is a partial overapping of the valence band of the p-region

and the conduction band of the rc-region. This malees possible the
tunnelling of electrons from the n-region to the p-region (flux 1)
and from the p-region to the rc-region (flux 2). Flux 1 constitutes
the reverse tunnelling current and flux 2 the direct current. In the
absence of an externa! eld those currents' are equal and the total
current through the junction is zero.
When a direct voltage is applied to the junction, the overapping
of the bands becomes smaller (Figure 8.2(b)) and because of this
flux 2 exceeds flux 2-and a direct current passes through the junction

Figure 8.21 Principie of operation and current-voltage characteristic

of the tunnel dioe.

increasing with the direct voltage V until a mximum, is reached

corresponding to a voltage at which the bottom of the conduction
band of the n-region coincides with the Fermi level of the p-region
(Figure 8.21(c)). When V is increased still further, the direct current
diminishes because of the decrease in the number of occupied States
of the n-regin lying opposite the free States of the p-region (Figure
8.21(d)). When at a voltage V the bottom of the conduction band
of the n-region coincides with the ton of the valence band of- the
p-region, the overapping of the bands ceases (Figure 8.21(e)) and
the tunnel current turas zero, but a smail direct current appears
as in a usual dioae. It rises rapidly with a further increase in V in
accordance with (8.27) (Figure 8.21(f)).
A remarkable property of the tunnel diodes is the negative di-
ferential resistahe regin 1-2 on the current-voltage characteristic
similar to that of the Gunn diode. This.makes it possible to use
Contacf Phencmena 293

those diodes for the generation o VHF uscillations up to requencies

o about 10u Hz. The tnnnel diode was one of the rst devices with
a switching time o a fraction o a nanosecond (O-10 3) and this
enabled it to be used in impulse circuits of digital computers and
in various automation circuits. jO n ly the majority carriers work
in the tunnel diodes and this malees them much less sensitive to
ionizing radiation than the bipolar semiconductor devices, this
act being of special importance for space explorations.
The development of the tunnel diodes is an excellent illustration
of the act that the quantum mechanics, formerly an exotic Science,

Figure 3.22 Schematic representa tion o n-p-a junction transistor

and its connection into Circuit.

became for the modern engineer a powerful tool which he must

master in order to be able to take an active part in the progress
of modern technology.
Transistors. Rapid progress in semiconductor electronics became
possible only ater the invention in 1948-1949 by J. Eardeen,
W . H. Brattain, and W . Shockley of the semiconductor amplier
the transistorwhose characteristics and whose designacin were-
similar to those of the vacuum tube but which had some substantial
advantages over the latter.
Figure 8.22(a) shows the schematic representation of an n-a-n
junction transistor. The transistor is made of three regions: the leit
n-region E termed emitter, the' miadle p-region B termed base, ana
the right n-region C termed collector. Those regions are separaed
by two p-n junctions: the emitter and collector junctions. By means
of ohmic contacts the transistor is connected into the Circuit: one
o the possible connections (the common base connection) is shown
in-Figure 8.22(a), where R-in is the equivalen! input resistance o
the transistor and Ront its equivalen! output resistance. It may
be seen that the emitter p-n junction is biased in the forward and
the collector in the reverse direction. t ....
294 Sod Siaie Physics

To understand the physical principies of transistor operation let

us turn again to Figure S.lS(b). When a forward bias is applied to
the emitter p-n junction, the concentrations of the majority carriers
o eleclrons in the p-region and o boles in the rc-regionincrease
drastically as compared with their equilibrium concentrations.
This phenomenon is termed minority _carrier injection and serves
as the basis for transistor operation.
The electrons injected from the emitter E into the base B (Fig
ure 8.22(a)) diase to the collector C\ with a base that is narrow
in comparison with the minority carrier diusion length practicaily
all the injected electrons wili reach the collector junction and will
be drawn in by its eld into the collector Circuit of the transistor.
Therefore the collector current 7C will be approximateiy equal
to the emitter current Je: 7C = cz7e, where a i is the common
base current amplification factor.
Now imagine that an external signal rln small in comparison
with the bias voltage V is applied to the input resistance i?n. The
input currentthe emitter signal currentwill be 7e = Vln/Rln
and the output voltagethe signal voltage across the collector
junction (or the equivalent collector resistance 7?out)will be Fout =
= 7ci?out = aJcRout. Therefore the voltage amplication factor
a r of the transistor in the common base connection will be
CLy = V in^Tout == &Ro\iV Pin ~ -^ouk-^a
Since f?in is a small dierential resistance o a forward-biased
p-n junction and 7?out is an enormous resistance of a reverse-biased
junction (7?0ut 7?in), &v 1 and may be-'as high as 103 (for de
current). Since in this connection only the voltage is amplied,
the same will be the power amplication factor a P- = P0utP\n
ce.v-, The source of the additional signal power dissipated in the
collector Circuit is the collector power supply Fe-
Figure 8.22(b) shows the transistor connected into a common emit
ter circuit. In this case the signal from the source 5 is applied
between the emitter and the base, the output signal being taken o
the emitter and the collector. The input signal aeets the emitter
7e, the collector 7C, and the base I-a currents, the latter being the
dierence of the formar two (7b = 7e 7C). Since 7C = x l s, it ollows
that the common emitter current amplication factor
T
Q__ JC T1 C _ n ^
A
-b 7 T
i z 1z 1 *a
can be made very high ( ~ 1 0 4), but from considerations of stabilitv
and of requeney response it is usuallv held in modera transistors
in the ranga from 40 to 100.
Transistors have round universal application in electronics:
in low- and high-requeney amplier and osciliator circuits, in
Conac Phenomena 2S5

switching circu ts, in triggers and multivibrators, n low- and high-

requency detector ckcuits, etc. Almost the whole of modern com-
mercial and special-purpose electronics is based on semiconductor
devices the most important of which is the transistor.
Photoelectric devices; p - n junction photocells. When a p-n junction
is iliuminated an emf is established in it. This phenomenon is uti-
lized in barrier ayer .photocells which. may serve , as indicators of

Figura 3.23 Semiconductor barrier layer photocel: (a)schematic

representation; (b )energy band pattern o p-n junction.

(b f

radiative energy independent of external power sources and as con-

verters of radiative energy into electrical energy.
Figure S.23(a) shows the schematic representation of a photocel.
A narrow dinused yi-iayer is fabricated on the surface of a p-type
semiconductor wafer.so that a p-n junction is iormed. In the absence
of illumination the p-n junction is in the State of equiiibrium and
an equiiibrium potential barrier qVa (Figure 8.23(b)) is established
in it. When the junction is iliuminated, electron-hole pairs are
generated mostly in the p-region because light passes through the
narrow ?i-layer without absorption. The electrons generated in the
p-region aiffuse to the p-n junction and are drawn in by the contact
eid and transporten to the n-region. The holes are unable to sur-
mount the barrier qVc and reniain in the p-region. Because of that
the p-region acquires a positive charge and the .u-region a negativa
one and an additionai forward voltage Fptl is established across
the junction. The term for it is photo-emf, or photovoltage.
At present the most eicient converters o solar energy are Silicon
photocells (solar baUeres). They are usea as power supplies for
receivers - and transmitiera installea on sateilites and even on the
296 Soiid State Physics

ground. Calculations show the mximum efciency (theoretical)

o the Silicon energy converters to be as high as 22-23 percent. The
efciency of the best modera types is about 15 percent. Germanium,
copper oxide, seienium, silver sulde, suifurous thallium and other
semiconductor photodiodes ere widely used as indicators of radiative
energy. Their integral (in the entire spectrum) sensitivity is much
higher (lO'MO3 times) than that of the external photoefect cells.
Their main disadvantage is their great inertialiy.
Photodiodes. The photodiode is a photocell connected into a Circuit
in-series with an external power supply (Figure 8.24(a)). In the

Figure 8.24 Photodiode: Circuit. diagram (a) and current-voltage

characteristic (b).

absence of illumination I a negligible so-called dark current i

ows through the junction (Figure 8.24(b)). When the p-n junction
is illuminated, excess carriers are generated and the current rises
in propcrtion to I causing .voltage drop aeross the load resistor R\.
Substantial advantages of the photodiodes over the external photoef-
fect elements are smaller dimensions and lower weight, high integral
sensitivity and low operating voltage.
Luminescent diodes. The passage of a forward current through
the p-n junction involves, as ws know, minority carrier injection:
of electrons into the n-region and of boles into the n-region. The
injected carriers recombine with the majority carriers of the respec
tive regin, their intensity decreasing with the distance from the
p-n junction (Figure 8.18(b)). In many semiconductors the recom-
bination is nonradiative: the energy liberated in the recombination
process is absorbed by the crystal iattice, that is, turas eventually
into heat. However, in such semiconductors as SiC, GaAs, InAs,
GaP, and InSh the.recombination is radiative: the energy of reccm-
bination is liberated in the orm of radiation quanta, photons.
Coniacf Phenomana 2S7

Because of that a forward current lowing through the p-n junction-

made of such materials is accompanied by the emission of light
rom the junction regin.
This phenomenon is utilized in the luminesceni diodes. Such
diodes are used in dispiays, they may be used in computers for data,
input and output and-in other applications requiring reliadle lumi-
nous indicators. Low operating voltages, low power consumption
and a long Service lile are the advantages of the light emission diodes
over other electroluminescent light sources.
Semiconductor lasers. In recent years inensive work has been in
progress on the semiconductor sources of coherent radiationsemi
conductor laserswhich open up possibilities for the direct conver
sin of electric energy into the energy of coherent radiation..
In Figure 8.25(a) the solid line shows the electrn distribution
corresponding to the equilibrium State and the dotted line the distri-
bution corresponding to the nonequilibrium State, in which the-
concentrations of electrons in the conduction baa and of the hoies
in the valence band are above the equilibrium vales. Such band'
occupancv corresponding to population inversin is shown in Fig
ure S.25(b). The peculiar point about it is that the light quanta with
the energy Hu> E 3 ( Ez is the forbidden band vidth) cannot be
absorbed by the system. Indeed, such an absorption involves the
transfer of an electrn rom the top leve! of the valence band to the-
lowest level of the conduction band. Since there are practicallv
no electrons on the top leveis of the valence band and no vacant
States at the bottom o the conduction band the probability of such
a process is extremely small. This creates favourable conditions
for the stimulated emission and for an avalanche of photons. The
light quantum 1 (Figure S.25(b)) stimulates the recombination o
the electrn and the hole (a-transition), which resulta in the emis
sion of an identical quantum 2. Since the quanta are not absorbed

Figura 3.25 (a) electrn distribution piots for equilibrium (solid line)
and inverse (dotted line) States of semiconductor; (b)development
of photon avaianche cause by inducen radiation o a quantum system
with popuiation inversin; (c) in a quantum generaior only the radiation
which propagates aiong the 0 0 ' axis is amplied.
298 Sod State Physics

by the system, they subsequenfcly stimulate tile emission of two

new guanta, etc. To make ooe photon take part in many stimulated
emission acts, two strictiy parallel mirrors 1 and 2 (Figure 8.25(c))
are arranged on the opposite sides o the lser crystal to roflect the
incident photons- and return them. into the crystal. Only those pho-
tons are amplied which move strictiy along the 0 0 ' axis, for only
such photons are repeatedly relected by mirrors 2 and 2. A ll other
photons leave the active lser space immediately or after a limited
number of reections (in Figure 3.25(c) such photons are shown by

Figure 8.26 Schematic representaron

o semiconductor lser:
/ active regin o lser;
2relecting faces (mirrors)

dotted lines). The result is a highly directional and highly monochro-

matic beam of radiation along the 0 0 ' axis.
There are various methoas of creating a population inversin
-of the semiconductor energy bands. The best prospects ofiers the
minority carrier injection through a forward biased p-n junction
made in a aegenerate semiconductor. Figure S.26 shows the st-ructure
of a semiconductor lser in which such method of pumping is used.
The lser is a dioda with a p-n junction 1 made in the forra of a bar.
Highly polished faces 2 of this bar made strictiy parallel to each
other play the part of mirrors -that reflect the photons.
The interest for the semiconductor lasers is due to some of their
remarkable properties.
First of all, they have a high efciency which may in principie
reach 100 percent. This is due, on the one hand, to the quantum
mechanical nature of the lser as a system in which only the working
-energy levels are excitea, and on the other, to the fact that in a semi
conductor lser the electric energy is directly converted into coherent
radiation without any intermedate steps as in all the other lser
types.
Another remarkable properiy of the semiconductor lser is that
it is possible to modulate the coherent radiation directly by changing
the current through the p-n junction. This enables them to be used
in Communications and in televisin as well as in ultra-high-speed
computers for wdiich their miniature dimensions are of special impor-
ance.
Contact Phenomena 233

73 Fundamentis o integraied circuif

electronics [microelecfronics]
Tile p rog ress o f m odera. S c ie n c e a n d te c h n o io g y requires e le c tr o n ic s
t o provide fo r it e fic ie n t complex electronic eq u ip m en t. Such e q u ip -
m e n t o f ten c o n ta in s h u n d r e d s o f th ou sa n d s o f eiem en ts connected
in to a C ircu it by m ea n s o a similar number o f co n n ection s.
Electronic equipment is responsible for the main part of costs
of .production of modera miiitary equipment and aircrat. Accord-
ing to Western sources, tile cost of electronic equipment makes
up over 70 percent of the cost of a modera guided missile and over
50 percent of the cost of a modera bomber.
To produce such highly sophisticated equipment soma difncult
problems had to be solved, including the problems of drastic reauc-
tion in weight, aimensions, power consumption, and in price and
of increasing the reliability of electronic devices.
Should modera electronic apparatus be assembled from compo-
nents manufacturad by'the industry severa! decades aqo, its weight
would have been tons, its dimensions cubic mstres, and its power
consumption hundreds of kiiowatts. The latter fact would have
sufficed to malee it impractical for many elds of the
economy.
But those are not the only drawbacks. As the complexity of the
electronic equipment increases its reliabilitya factor of primary
importance especially in miiitary, computar and atomatic produc
tion line applications diminishes.
Finally, the problem of bringing down the costs of electronic
equipment is also not devoid of importance and this problem can
only be solved by far-reaching automtica of the technologica!
processes, which in tura requires the aevelopment of appropriate
technoiogy.
T h e effo rts t o s o lv e th o s e p r o b le m s le a d to the cre a tio n o f m in ia
tu re e le c t r o n ic e ie m e n ts a n d b lo c k s based on solid -sta te te c h n o io g y
an d t o th e m ic r o m in ia t u r iz a t io n o f e le ctro n ic e q u ip m en t r e s u itin g
fin a lly in th e b ir t h o f a n e w fie id o f m o d e ra e le ctro n ics m ic r o e le c -
t r o n i c s w h o se m a in o b je c t i v e is the p r o d u ctio n of h ig h ly r e lia b ie and
e c o n o m ic a l m ic r o m in ia t u r e e le c t r o n ic circ u its and a p p aratu s.
T h e m o d e r a w a y to s o lv e th is p r o b le m is to devise new p r in c ip ie s
o f c o n s t r u c t in g e le c t r o n ic c i r c u i t r y w h ic h w ou ld m ake p o s s ib le the
f o r m a t io n o f a C ircu it as a w h o ie on a m in iature s e m ic o n d u c to r
c r y s t a l in ste a d o f a s s e m b iin g it fr o m separata c o m p o n e n ts . S u ch
s o l i d e le c t r o n ic c ir c u it s d e s ig n e d fo r s p e c i c a p p lica tion s are term ed
integraied circuits (IG ). T h e in t e g r a ie d Circuit, iik e an o r d in a r y
e le c t r o n ic C ircu it, is m a d e u p 'o f a c tiv e eiem ents (tra n sistors, d io d e s )
a n d o f p a s s iv e e ie m e n ts (c a p a c it o r s and resistors).
3C0 Soiid State Physics

A semiconductor IC is fabricated on a singie-crystal wafer (usually

Silicon) with the aid of methods o local aoping with appropriate
impuriies to produce on it transistors, diodes, capacitors, and
resistors and to connect them into a Circuit. The dimensions of the
water are typicaily (10-2) X (5 X 10-3) X (2 x 10-4) m3, the area
of the active elementa, fo_r instance of a transistor, being under
10 " 9 m2.
The integrated circuits are usually characterized by packing
densi'ty and by negree of integration. The packing density is the
number o elements per unit volume of the IC, and the degree of
integration, the -number of elements making up the IC. Table 8.2
presenta data on the packing density and on failure rate of circuits
of dierent generations.

Tabie S.2

Number of elements per

Circuits using Failure rate ?* (hr_1)
cubic xnetre

Pre-1941 elements 3.5 X ID4 10 5

Miniature elements l .S x l O 5 5 x q -5
Semiconductor IC 3 .0 x 1 0 * Negligible

It follows then that the changeover from the circuits assembled

from pre-1941 components to modern IC increased the packing
density some 1Q5 times. There are reports of packing densities of IC
of up to 1015 m-3.
The degree of integration of an 'IC may vary in a wide range
from tens to hundreds or thousands of elements on each wafer. IC of
over 100 elements are termed big iniegrated circuits (BIC).
The powr consumption of IC, depending on the type, lies in the
range of hundreds of milliwatts to several microwatts.
Thus, the changeover to electronic equipment designed around IC
practically solved the problems of dimensions, weight, and power
consumption. Electronic computers are an impressive example of
this. The rst Soviet computers which were assembled from vacuum
tuhes and radio components (Minsk, Ural, etc.) occupiea whoie
buildings, weighed tons, and cosumed tens of kilowatts of power.
Electronic blocks assembled from IC have dimensions of the order
of 10-2 m3 and consume only hundreds of watts while special comput
ers used, for instance, for launching and conrolling missiles and
spacecraft have dimensions of the order of 10~3 m3, weigh tens of
lograras and consume power of the order o l tens of watts.
P r e s e n t ly BIC are b e in g u sed in sin g le w a fe r e le c tr o n ic c a l c u la t o r s .
T h e C om p u ter w a fe r ( c a lie d c h ip ) o f such a d e v ice is (5 x 5 )i0 "s m 2
and c o n t a in s a b o u t 5- th o u sa n d tra n sistors. BIC fo r e l e c t r o n ic t im e -
Contaci Phencmena 301

pieces including wrist watches have been developed as well. In suca

watches two wafers containing about 2 thousaad transistors ara used.
A substantial advantage of IC is that bccause mass-produced IC
ara much cbeaper than equivalent circuits assembled from compo-
nents. Modern technology makes it possible to arraage about a thou
saad IC on one singie-crystai wafer of 5 X 10"2- m in d O,Xi0 * b S i, ^

if a hundred such wafers are processed at a time, about a million

IC can be produced in one technological cycle.
The progress in microelectronics is a very rapid one. During the
last decade a distance was covered rom the simpiest IC to BIC.
In the nearest future it is expected that most electronic equipment
shall be based on integrated circuitry with the degree of integration
increasing 100 to 1000-foid and a much greater reliabiiity being
attained. However, there are obstacles on this road, namelv the
so-calied tyranny of numbers of microelements which aiready
today crowd c.omplex equipment in tens or hundreds of miliions.
To' overeme this obstacle it will, probably, be necessary to change
over from the conventional IC to functional circuits, that is, to devices
designed for specic functions and operating on some specic prin
cipie of solid State physics as a whole and not as a sum of individual
elementa (transistors, diodes, etc.).
A s the s im p ie s t e x a m p le of a fu n c t io n a l d e v ice one m a y c it e th e
a c -d c c o n y e r te r . T h e c o n v e n t io n a l C ircuit o f such a c o n v e rte r co n sista
o f a tra n sfo rm e r, r e c tifie r s (s e m ic o n d u c to r or v a cu u m d io d e s ), a n d
a lte r. T h e fu n c t io n a l c o n v e r t e r co n sists o f a resista n ce r e g i n in
w h ic h the a c e n e r g y is tr a n s fo r m e d in to h eat, o f the c e n t r a l l o w
e le c t r ic b u t h ig h h e a t c o n d u c t i v i t y re g i n , and o f a t h e r m o e le c t r ic
re g i n in w h ic h h e a t is c o n v e r t e d in to de p ow er. In s u ch a d e v ic e
it is im p o s s ib le to se p ra te r e g io n s e q u iv a le n t to th e c o m p o n e n te
o f a c o n v e n t io n a l C ircu it. H e re th e c r y s ta l as a w h o le fu ifills the
c o m p le x f u n c t io n o f an a c -d c c o n v e r te r .
T h e t r a n s it io n t o fu n c t io n a l c ir c u its sh ou ld resu lt in a d r a s tic
d e c r e a s e in th e n u m ber. o f c o m p o n e n ts and th e r e o r e in the
d ecre a se in th e c o s t an d in d im e n s io n s and in the im p r o v e m e n t o f
r e lia b iiit y .
T h e p ro ce ss o f c r e a t io n o f n e w s c ie n ti c and t e c h n o lo g ic a l tre n d s
in e le c tr o n ic s a n d o f d e v is in g d e v ic e s and e q u ip m en t b a sed o n n e w
p rin c ip ie s is a c o n t in u o u s on e, th e o u n d a tio n fo r it b e in g th e u t i l i -
z a tio n o f th e t o p -r a n k in g a c h ie v e m e n s in the fu n d a m e n ta l an d
a p p lie d S cien ces, first o f a li, in p h y s ic s . H ere the le a d in g r o le b e lo n g s
to s o lid State p h y s ic s w h ic h d e te r m in e s th e m a in strea m o f p rog ress
in m o d e rn e le c t r o n ic s .
9
Thermoeiecric and Galvanomagneic
Phenomena

The thermoelectric eSects inelude the Seebeck, the Peltier, and the
Thomson eneets, and the galvanomagnetic the Hall, the Ettingshau-
sen, and the Nernst eSects. Some of those phenomena have found
wide application in practice; therefore a look at them is not only
of .educational but of practical interest as well.
Let us discuss briefly the physical background of those phenomena.

79 The Seebeck effecf

In 1822 T. J. Seebeck discovered that an eiectromotiye forc VT
is established in a Circuit consisting of two conductors 1 and 2 made
cf different materials if the junctions of these conductors, .4 and B r
are kept at diSerent temperatures, and Tz (Figure 9.1(a)). This
emf is termed thermal emf. Experimenta show it to be in a narrow
temperatura intervalproportional to the diSerence in- the tem
peratura the junctions A and B :
7 r = a (T2 - (9.1)
The proportionality factor
a = dVt IT (9.2)

is callea diersniial, or specic, thermoeiecric poivsr. It is deter-

mine by the material o the conductors and the temperature.
There are three sources of thermal emf: the directional current
of the carriers in the conductor due to the presence of a temperature
gradient (the volumetric component Prv)> change in the position
of the Fermi leve! (the junction component Fj), and the drag of the
eiectrons by the phonons (the so-cailed phonon drag efect).
Let us discuss the physical natura of each of those sources.
Volumetric component o thermal emf. Suppose that a temperature
diSerence (T., Tx) is maintained at the terminis of a uniform
conductor AB (Figure 9.1(b)) so that there is a temperature gradient

302
Thermoeiecric and Gaivancmagneic Phenomena 305

dT/dx in the direction from B to A . The current carriers in the Hot

end have greater kinetic. energy and greater speeds of motion than
the carriers in the coid end. Therefore a current will low in the-
conductor rom the hot end to the coid; this current will charge
the conductor. In cases when the current is carried bv electrons the-
coid end will accumulate a negative charge and the hot end a posi-
tive charge, and a potential difference Vv will be established between

Figure 9,1 The Seebeck eect: (a)thermoelectric Circuit;

()origin of volumetric and junction components of thermal emf.

thera. This is the volumetric component of thermal emf. The difieren-

tial thermoelectric power corresponding to this component is

(9.3)

<xv may be estimated as follows. The pressure of the electrn gas

in the conductor is
(9,4>

where E is the average energy of .electrons in the conductor, and n.

their concentration.
The temperature gradient occasions a pressure gradient to com
pnsate which a eld e should be established in the conductor sucii
that
dp ap dT
q&n =
d- dT dx
From here cty may easily be found:
__ dVv o, / dT \ - 1 dp
(9.5>
^ dT S l dx ) qn oT
As a rale, in an n-type conductor ccv is directed from the hot end
to the coid. However, there are exceptions to'this rule which we will.
discuss below.
304 Soiid Sfafe Physics

The junction component o thermal emf. The change in temperatura

occasions a change in the position of the Fermi level. In nrtype
conductors the Fermi level sinks on the energy diagram as the temper-
ature is raised (see Figure 5.19(a)). By forc of this it should he
higher on the coid end of an ji-type conductor than on its hot end.
The difference in the Fermi level positions is equivalent to a poten-
tiai difference

dvl = - J W dT (9.6)

And this is just the junction component of the thermal emf. The dif-
erential thermoeiectric power corresponding to this component is

The resultant difierentiai thermoeiectric power

\ dp 1 da
(9.8)
nq. d T q T
We apply the relation (9.8) to conductors of various kinds.
Thermoeiectric power of metis. Substituting the average energy
of electrons of a degenerte electrn gas from (3.45) into (9.4),
we ohtain the following expression for the pressure of the electrn
gas in a metal:
9
-A nE? (*b T)2
o
Dierentiating this expression with respect to T and multiplying
it by 1nq, we ohtain
ccv ^3 EL (9.9)
q 3 E?
The temperature dependence of the Fermi level in metis is given
..by relation (3.44):

u = E* E l ( M i ) 2]
12 l E? ) _
Diferentiating it with respect to T and multiplying by 1Iq, we
ohtain
n*kn k3 T
(9.10)
oq

Substituting (9.9) and (9.10) into (9.8), we ohtain

_ ( a j__1 \ 3^ (9.11)
-m 3q \ 1 : E?
A more accurate calculation for metis with a quadratic depen-
-dence of the electrn energy on the wave vector produces the rol-
Thermoelecrric and Galvanomagneic Phenomena 3C5 I
i.
lowing result: !I
_ ,, , . k3 T
(9.12)
3 T (1^ r) " s F
where r is the exponent in the relation
(9.13)
which expresses the aependence of the electrn mean free path on
energy E.
Table 9.1 presents the vales of r for various mechanisms of elec
trn scattering on atomic and ionic lattices.
Table 9.1

Scattering on thennal vibrations

Ionic Iattice Scattering on
Atomic impurity atoms
iattice T> 0
T< 8

1/2

It follows from (9.12) that for metala a m a T, in full agreement

with experiment. Since kQ T E? the thermoeiectric power of
metis is quite small. For instance, for silver E ? = 5.5 eV and
k3 T = 0.025 eV at T 300 K; substituting this into (9.12) we
obtain am 8 X 10~8 V /K , which is quite cise to the experimen
tal valu a ; 5 X 10-8 V/'K.
It follows from (9.13) that when r < 0 the more energetic elec-
trons nave a shorter mean free path Since the diusion current
in this case is directed from the hot end to the coid end, the siga
of the volumetric componen!' of the thennal emf will be reversen.
This may cause the re versal of the sign of the metals thermal emf
as a whole. Such effects are observed, for example, in some transitioa
metis and in some alloys.
As was already stated before formula (9.12) is valia for metis
with a quadratic E {k) aependence. In metis and in alloys with
a complex Fermi surface the contribution of the. various regions
of this surface may difer not only in absoluta valu but in sign
as weil, with the result that the thermoeiectric power may be zero
or cise to zero. For instance, lead has a zero thermoeiectric power.
For this reason thermoeiectric power is usualLy measured in relation
to lead.
The current direction in the hot junction of a thermocouple made
of aa ra-type conductor and of lead will be determined by the polarity
of the conductors charge. For a normal conductor whose hot junction
acquires a positiwe charge the current in it will be-directed from
20-0833
306 Sod Sae Physics

coaductor to lead (Figure 9.2(a)). In this case the conductors ther-

raoelectric power is assumed to. be-negative.
In case of an rc-type conductor acquiring an anomalous charge
(Figure 9.2(b)) the current in tbe bot junction will flow from lead
to conductor and a will be positive; X will be positive for a normal

n-type conductor Figure 9.2 Magnitude and sign

i- i o tbermoelectric power determined
Hot junction
in relation to lead (explanation
in text).
(a) j
n-t ypg conductor
Pb Tf^Hot junction

777 .
p-tvpe conductor
'* -----11
Pb / !| Hot junction

77
p -type conductor too, whose bot end acquires a negative cbarge
(Figure 9.2(c)).
Tbermoelectric power of semiconductora. The pressure of electrn
gas in a nondegenerate semiconductor is

p= nE = nk3T

Diferentiating tbis expression witb respect to T and multiplying

by i Inq, we obtain
ccY= ^ - [ ' l - f T ] . _ (9.14).

A more rigorous calculation yields

a v = -^ [r -4 7 1 | A ] (9-15)
The Chemical potential in a nondegenerate u-type semiconductor
is given by relation (3.26):
^ jS
pn k s ? la 2
DiSerentiating p* witb respect to T and multiplying by 1/g. we
obtain
d !n n
os = T dT (9.16)
k3 T
Substituting (9.15) and (9.16) into (9.8), we obtain
/ , o l*w ) _ k"Q 2(2Kmnfc3n * /-l
CLn 1 , ii* .n
ir 1 - k3 T j 9 nri3
/p, Jry
jb .l i
Thermoalectric and Galvanomagnetic Phertomena 307

The minus sign in front o the right-hand side |is in accordance with
the conventional polarity o the thennoelectric power.
For a n-type semiconductor
^3 r 2 (Ixmpk-BT)*1-
a. = I r- 2- f i n (9.13)
1 L ph3
Let us estmate the valu o a o an extrinsic semiconductor, for
instance, or n-type germanium with n = 1023 m -3 at T 300 K.
Substituting those vales into (9.17), we ohtain a 10-3 V /K .
Henee the thennoelectric power o semiconductors is three orders
o magnitude greaterthan that o metis.
For semiconductors with bipolar conductivity, in which the
electric current is carried both by electrons and holes, the expression
for the thennoelectric power is:
a ppup a nnun
a n .p (9.19)
P p ~ n

It ollows from this relation that i the electrn and hole concentra-
tions and their mobilities turn out to be equal, the thennoelectric
power may be quite small 'or even zero.
Phonon drag o electrons. The phonon drag effect. discovered by
L. E. Gurevich in 1945, consists in the ollowing. W ith a temperature
gradient in the conductor the phonons drift from its hot end to the
coid end at an average velocity yPb- In the presence o such drift
the electrons scattered by the driting phonons are themselves in
volved in the directional motion from the hot end to the coid end, their
velocity being about equal to The accumulation o the electrn?
on. the coid end of the conductor and their depletion on the hot
end resulta in the appearance of a thermal emf Vp^.
G. E. Pikus in 1956 calculated the diferential thennoelectric
power due to the phonon drag and obtained the ollowing result:.

k3 TPh 9.20>
ph 3q k3T
Here yPj, is the phonon drift velocity. and tp^ and r e are the phonon
and electrn relaxation times.
In the low temperature range this component o thennoelectric
power can be tens or hundreds of times greater than the volumetric
and junction components.

SO The Pelfier effect

Let a current I flowr in a Circuit consistir g of two conductors I
and 2 (Figure 9.3) made o diferent materiais. The Joule heat Q
= PR i will be liberated in the junctions A and D (i? is the jnc
eo*
3 08 Soiid State Physics

tions resistance, and t is the time the current ows). At junctioas

of identical conductors oaiy tais heat will be Iiberated and from this
point of view there is no difference between the junction and- the
rest of the Circuit. At the same time at points of junction of different
materiais an additionai heat apart from the Joule heat will be liber-
ated or absorbed, heating the_c_ontact in the former case or cooling
it in the latter. This phenomenon was discovered in 1834 by

Figure 9.3 Diagram ol Circuit used

to observe the Peltier eSect.

J. G. A. Peltier and is termed the Peltier eijeci; the additionai heat

iiberated or absorbed in the junction is termed Peltier heat, Q?.
Experimenta show it to be proportionai to the current I and the
time the current passes through the contact t:
QlP = n It (9.21)

The proportionaiity factor n is termed the Peltier coeificient. Its

valu is determined by the materiais and by temperatura.
There is a direct connection between the Peltier and Seebeck
eects: the temperatura dinerence causes a current to ow in a Cir
cuit consisting of diSerent materiais and a current flowing through
such a Circuit sets up a temperature dinerence. The expression for
this relation is due to W. Thomson (Lord Ivelvin), who is the author
of the thermodynamic theory of thermoelectric phenomena. He
demonstrated that:
a - UIT (9.22)
The Peltier eSect is due to the dinerence in the average energies
of the conduction electroas. in unlike materiais. By way of an exam-
ple, let s consider the junction of a metal with a nondegenerate
n-type semiconductor (Figure 9.4). After equilibrium. had been estab-
lished their Fermi levels will coincide. Only the electrons cise to
the Fermi level whose average energy are practically equal to the
Fermi energy take part in the conductivity in the metal.
Denote the average energy of the conduction electrons in the
semiconductor by En. This energy is not equal to the thermai energy
of the electrons 3k3 772 since the relative prt played by fast elec-
trons in the electrc current is greater than that of slow electrons.
A calculation madehor the case of a nondegenerate electrn gas
Thermceleciric and Gaivanomagneic Phenomena 309

yields
En = (r + 2)kBT (9.23)

where r is tile exponent in (9.13).

Suppose that the electric current owing through the junction
is such that the eiectrons flow from the semiconductor to the metal.

n -tv p e Figure 9.4 Energy band pattern

M e tal se m ic o n d u c to r oi metal-semiconductor junction
illustrating mecbanism
En
o the Peltier eect.
T

l
t 1 -p i
(f,

- L'
---
> /// / / v 'y . ' / \

It may be seen from Figure 9.4 that each electrn t.hat goesover from
the semiconductor to the metal carries with it an additional energy
equal to

= -H -d (9.24)
This energy is the Peltier heat and it is liberated near the junction.
When the direction of the current is changed, the eiectrons going
over from the metal to the semiconductor absorb heat and ccoi the
junction.
Dividing LE by the electrn charge, we obtain the Peltier coe-
ficient:

(9.25)

iubstituting u from (3.26) and En from (9.23) into (9.25), we obtain

k3T (9.26)
1 7lh*
<Z L

A similar relation. may be obtained for the junction of a metal

with a p-type semiconductor
knT r 2 (2nmpkBTr/- - /O
n mp (r 2) -fin
310 Solid State Physics

For a junction o two metis the Peltier coecient may be founa

rom (9.22):
n 1>s (tXi cc.i) T ' (9.28)
Substituting a rom (9.12), we obtain
/A , / T
n1>2 Zq ^ ^ ( Ft ?2 )
(9.29)

81 The Thomson effect

Imagine a homogeneous conductor AB with a temperature gradient
cLTIx along its length and carrying a current / (see Figure 9.(b)).
W . Thomson predicted theoretically that in sucia a conductor apart
rom the Joule heat an aaditionai amount o heat Ox proportional
to the current I, the temperature difference {T T-), and the
time t should bo liberated or absorbed depending on the direction
of the current:
Qx = x l {Tt - r o t (9.30)
The heat Qx is termed the Thomson heat and the proportionality
factor x the Thomson coefcient. It is determined by the material
of the conductor and by temperature. According to Thomsons theory
the diference in Thomson coefhcients of two conductors is related
to their diSerential thermoelectric power by the expression;-
^C1.2 Ti /Q qx\
dT T ^
The Thomson effect is due to the fact that in a conductor in which
a temperature gradient exists the carrier flux carries not only the
electric charge but heat as well. Suppose the current in the conductor
AB (Figure 9.1(b)) flows in the direction corresponding to the elec
trn flow rom the hot end 3 to the cola end A. The hot electrons
as they arrive in the coid regions give up their extra energy and
heat the conductor. When the direction of the current is changad,
the conductor is cooled.
In quanitative calculations of the Thomson enect one should
take into account the thermal emf set up in the conductor, which
in the former case w ill retard the electrons and in the latter accele-
rate them. This thermal emf can changa not only the magnitude
of the Thomson coefcient but even its sign.

82 Gavanomagnefic phenomena
The Hall efect. Suppose a current of density i ows in a conducting
bar of width a and* thickness b (Figure 9.5). Choose points C and D
on the side faces of the bar such that' the potential diference between
nermoeiedric and Galvanomagnefic rhenomena 311

them is zero. Should this bar be placed, into a magnetic eld with
induction B, potential diSerence VH termed Hall emf would appear
between points C and-D. It follows from experiments that in mag-
netic fields not too strong
REBia (9.32)

The proportionality factor i?H is termed the Hall coefcient. Its

imensions are L3/Q (L is length and Q electric charge) and it is

Figure 9.5 Layout used to observe the Hall effect.

measured in cubic metres per coulomb (m3/C). Let us consider the

physical origin of the Hall effect.
The Lorentz forc FLorentz acting on an electrn moving from
right to let at a speed v (Figure 9.5) is
F Lorentz OY X B

If v B, the forc wiH be equal to

^ Lorentz =

The Lorentz forc deects the electrons to the outer face of the
bar (dotted line in Figure 9.5), and the bar receives a negative charge.
Uncompensated positive changas accumulate on the opposite side.
This results in an electric eld directed from C to D:

where 7 H is the potential diSerence between points C and D (the

Hall emf).
The eld e H exerts a forc F q oa the electrons, this forc
being directed against the Lorentz forc. When F = FLorer.tzi the
transverse electric eld compensates the Lorentz forc and no more
electric charges are accum.ulated. on the sida laces o the bar. From
312 Solid State Physic*

the conditions o equiiibrium

qvB = qoH (9.33)
we obtain
e g = vB
Multiplving this relation by the dista ace a between points C and D>
\ve obtain
Fg = ac- vBa
Taking into account that i = qnv and consequently u = {qn),
we obtain

Fh -Bia (9.34)
nq

Thus theory produces an expression for FH that coincides with

the relation (9.32) obtained from experiment. The Hall constant
turns out to be equai to
(9.35)

It foliows from (9.35) that knoving the absolute valu and the sign
of the Hall constant we can nd the concentration and siga of the
charge carriers in a conductor: RH of rc-type conductors is negative.
and of p-type conductors positive.
If we measure in addition the specic conductance o = qnu of the
conductor, we will be able to nd the carrier mobilitv u since
i?Hcr = . (9.36)
Mobility uH determined from (9.36) is the Hall. mobility and
it may not coincide with the drift mobility defined by (6.5).
In the derivation of (9.35) it was assumed that all carriers in-the-
conductor nave the same speea v. Such an assumption is vaiid in
..cgse of metis and degenerate semiconductors but it is totally unac-
ceptable for nondegenerate semiconductors, in which the carrier
veiocities are distributed in accordance with the Boltzmann law.
A- more rigorous derivation, which aecounts for this fact, yields
the foilowing expression for i?H: j
i?H = Al{an) (9.37)
where A is a constant dependent on the scattering mechanism of
carriers n the crystal. The typical vales of A are given in'Table 9.2
below.
In bipolar semiconductors the currant is carried simultaneously
by electrons and holes. Since their charges are opposite and they
move in opposite directions in an eiectric eid, the Lcrents forc
FLorentz = ?v. X B deects them in the same direction. Because
Thermoelecfric and Galvanomagnetic Phenomena 3 1?

Table 9.2

Scattering by thermal vibrations

Ionic lattice Scattering by im-
Atomic purity ions
lattice T< 0" T> 0

.4 1.17 0.99 1.11 1.93

of this, other conditions equal, -their Hall emf and Hall coefcients-
wiil be smaller than in unipolar semiconductors. Calculation yields-
the following expression or Ru of bipolar semiconductors:

A
i?H = (9.3S)

where n and pt are electrn and ho'ie concentrations, and un and up.
their mobilities. Depending on .which of the two terms in the nu-
merator is greater, the sign of the Hall coefficient may either be posi-
tive, or negative.
For intrinsic semiconductors, in which n = formula (9.3S)-
assumes the form:
.4
(9.39)
1? u p ~T a n

It follows from (9.39) that in the intrinsic range the sign of the Hail
coefficient is determined by that of the carriers with greater m obili-
tv. As a rule such carriers are electrons. Therefore when an extrin-
sic p-tvpe semiconductor goes over to the intrinsic range, the sign
of the Hall coefcients changes. Hall coefficient (at room tempera-
ture) for some metis and intrinsic semiconductors is presentad
below in Table 9.3.
Table 9.3

Cu Zn Bi Ge Si

i?H(10-u m3/C) 5.5 3.3 103 O10 1013

As indicatea, the Hall coefcient of semiconductors is many

orders of magnitude greater than that of metis. The explanacin
is that the carrier concentration in semiconductors is much less-
but the mohility, on the other hand, is much greater than in metis.
Ettingshausen eSect. The thermal velocities of electrons in non-
degenerate semiconductors lie in a wide range. In such conditions
equation (9.33) may he valid not for all the electrons but only tor
some of them whose average velocity is v0. F o r, electrons whose-
314 Soc Sfae Physics.

velocity is v > v0 we have qvB > qeH and. tney will be deected
to the right-hand face of the pate (Figure 9.6(a)). The electrons
whose velocity is v < v0, so that qvB <C gH , will be deected to
the left-hand face of the pate.
Fast electrons reaching the right-hand face give up their extra
-energy to it and thereby heat it. The slow electrons, which accumu-
late on the left-hand face, replenish their energy dficit at the ex
pense of the thermal energy of the crystal and thereby cool it. Thus

Figure 9.6 Diagram expLainirtg the origin of the Ettingshausen

and Nernst efiects (a) and o specic. resistance variations
in magnetic fieid (b).

-a transverse temperatura dierence T = TD Tc is established.

This phenomenon is termed Ettingshausen eiject. '
Nernst efect. The electrons entering a homogeneous magnetic
fieid B perpendicular to their velocity v start moving in a circie
vvi til a raius

r ~ ~qE~ (9-40)
It follows from (9.40) that the fast electrons are rotated by the mag
netic eld less than the slow electrons (Figure 9.6(a)). Therefore
the front face of the. pate will be richer in hot electrons and will
he heated while the back face will be richer in slow electrons and
will be cooled. A longitudinal temperature diference Ts TA
will be established. This is the Nernst effect
Yariation of conductora resistance in magnetic fields (magneto-
resistance). The trajectories of electrons moving in a magnetic fieid
with velocitiss other than v are curved (Figure 9.6(a)) and this
results in a rsduction of their effective mean fres path in the direc
cin o the eiectric current. If the mean fres path in the irection
Thermoelectric and Gaivancmagnaiic Phenomena 315

of th.8 current in the absenes of a magnetic eid is l0, in the magnet-

ic eid it is equal to the projection of the are AD on the direction
of the current I, that is X = X0 AX (Figure 9.6{b)).
Since the carrier m obility u is proportional to the mean free path,
the decrease in the mean free path by A? shouid bring about a
decrease in m obility by Au and in the semiconductor^ conductivity
by Acr so that
Aa/cr = Att/u = AX/>.0
The theory provides the following expression for the relative
increase in specic resistance-of extrinsic unipolar semiconductors:

( 9 . 41)

where B is the magnetic eid induction, and c a coefcient dependent

on the carrier scattering mechanism.
The ratio Ap/p is terme.d inagneoresistiuity. It follows from (9.41)
that by measuring magnetoresistivity one can directly nd the
carrier mobility.

83 Practica} applicafions of thermoeecric

and gaivanomagnetic phenomena
Thermoelectric phenomena. For a long time the only application of
the Seebeck enect was in measurements. Placing one junction. of
a thermocouple in a thermostat held at a known constant tempera-
ture and the other junction into the object the temperature of which
is to be measured, one can determine this temperature rom the
thermal emf established in the couple. Such measurements are
quite simple, reliable and sufficiently accurate and can be carried
out in a. wide temperature range.
However, after semiconductors were discovered it became pos-
sible to use the Seebeck eect for the direct conversin of thermal
energy into electric energy. The devices used to this end are termed
thermoelectric generators and the elementa of which they are assem-
bled thermoelements. The man mainly responsible for their develop-
ment and wide pubiieity was the Soviet physicist A. F. IoSe.
The rst thermoelectric generators were produced beore World
War II and were used in the war to power radio equipment. The
thermoelectric generators were mounted in the bottom of a special
kettle and heated in the process of boiling water.
In 1953 a commercial type of thermoelectric generator of 3-Tv
power for battery receivers was produced; later thermoelectric
generators of 1-kW power and more were produced. Presently gen
erators designed for hunareds of kilowatts are being developed.
316 Sciid Sfae Physics

The midseventies saw the appearance of thermoelectric generators

utilizing the heat o radioactiva decay of Chemical elements. An
example o such a generator is the generator Beta-1 with a power
o 150-200 W , which operates o the radioactive cerium isotope
Ce-144. It was designed to power electronic equipment of automatic
radiometeorologic stations, earth satellites, etc.
In 1964 an experimental atomic reactor-energy converter Romash-
ka (Camomile) with a power of 500 W for direct conversin of heat
energy into electric energy was built.
Work is in progress on thermoelectric generators that would
utilize the thermal energy of the suns radiation.
It is a regretful fac, but the efciency of even the best modera
experimental thermoelements does not rise above S percent.
The Peltier eect is beginning to be widely used in practice mainly
for various cooling devices: home refrigerators, devices for cooling
aircraft electronic equipment, microcoolers for biological applica-
tions, various thermoelectric thermostats, temperature-controlled
microscope supports, etc. Quite possible,. the Peltier eflect will
in the near future be used for heating dweliings in winter and for
cooling them in summer.
Galvanomagnetic phenomena. The most widely used galvano-
magnetic phenomenon is the Hall eect. Apart from applications
in the study of electric properties of materials it served as a basis
for the design of a wide class of instruments: magnetometers, dc-ac
and ac-dc converters, signal generators, phase meters, microphones,,
etc.
In recent years attempts are mad to use the Ettingshausen eect
in cooling devices. Writh the right choice of materials "and with the
optimal geometrv o the cooling crystal it is possible to obtain tem
peraturas of the cola face of the crystal of over 100 'C below that
o the surroundings.
Appendices

Appendix I
Derivation of the Maxwell-Boltzmann distribution unction. To ob-
tain expressioa (3.25) consider a collision of two partiles one of
which is iu a State with the energy Ex and the other in a State with
the energy E v After the collision the partiles will go over to States
with energies E 3 and E 4 respectively. Let us dene the term reverse
collision as a collision that returns the partiles to the initial States
with the energies Ex and Ez. Thus we shall consider collisions of
two types:
(Eu E z) - * ( E 3l EJ (direct)
{E3> E ,\-+ {Ex, Ez) (reverse)
The rate of direct collisions Q is proportional to the average
nurnber oi partiles in the initial state, that is f(Ex) and /(.,),
and is independent of the nurnber of partiles in the final state be-
cause the gas is nondegenerate:
<?d = cf(Ex) f(Ez) (.l)
where c is a proportionality factor.
The nurnber of reverse collisions is proportional to / (E3) f {E4):
Q, = cf[Ez)/(,)
In the state of thermodynamic equilibrium shouid be equai to Qr:
f(Ed KE) = ; ( 3) /( ;) (1-3)
Making use of the energy conservation law, Ex - f Ez = Ez E *,
we mav rewrite the expression in the form:
/( ,) f(E.) = f(E.) /( , + E, - E,) (1.4)
TsTote that Ex, Ez, E3 must be regarded as independent quantities.
Taking the logarthm of both siaes of (1.4), we cbtain
- in f(Ex) -r ln f(Ez} = in f(Es) -f- ln f{Ex - f Ez E3) (1-5)
317
318 Salid Sfafe Physics

Dierentiate this sum with respect to Ex. Since E and E z are

independent of Ex they may be assnmed to be constant. Then
1 i f ( E x) 1 dftE^ + E . - E s ) d[Ez+ E , - E 3)
(1.6)
<t, f ( E i + E . E3) d(
(F-
i-r -A ', 3) d1

Since d (E1 -f- E z E 3)/dE1 = 1, it follows tbat

t d f(E z) 1 d f ( E , E ~ - E 3)
(1.7)
f [ E x) aEx j(Ei~j-E* 3) d{ Ex-\-E* E 3)
Dierentiate (1.5) with respect to Es:
1 df(E2) _ 1 df (E, E. E3)
(I.S)
() dEn {E-y-f-E2 3 ) [\~T-.n E 3J
Comparing (1.7) with (1.8), we obtain
1 df (Ej) _ 1 df (E.)
(1 .9 )
{E 1 ) dEl / (*) <

The left-hand side of (1.9) is independent of Ez, the right-hand

side is independent of Ex\ therefore each of them is equal to some
constant independent of the partiles energy. Denote it bv 3. Then
we may rewrite (1.9) as follows: ------------
i e(E )
= B (LIO)
t [) dE J
Integrating (LIO), we obtain
/ (E) = Ae& (I.H )

where A is the integration constant. Experiment shows that

A= (1. 12)

Substitnting (1.12) into (1.11), we obtain

f^{E) = e'xlh^Te~ElkBr (1.13)

ppendix I!
Derivation of the Fermi-Dirac distribution nnction. Consider, as we
did in Appendix I, the direct and reverse partile coilisions. Recail
that in the case of a nondegenerate gas the rate of coilisions was
independent of the mimber of particles in the nal stages and was
entirely aetermined b y the number of particles in the initial stages.
The situation in the case of a degenerate ermion gas is a difieren t
one: if a State is already occnpied, it cannot accept another ermion
and the collision w ill not take place. For this reason in the case
of a degenerate fermion gas the rate of coilisions is proportional not
only to the average number of particles in the initial States but
Appendix II 319

to the average number of vacant States with the energies. E 3 and E k

as well.
Since / F (E) expresses the probahility for the State with the ener-
gy E to be occupied, 1 / F (E) expresses the probahility for it
to be vacant.. Thereore the average numbers of vacant States with
the energies Ez and 4 are 1 f ? { E 3) and 1 ; F (f?4), respective-
ly. Accordingly,' the rates of direct and reverse collisions are:-
Q i = < * (E J um a - h it - h (
Q, = cfr ( ,) h ( ,) [1 - n ( ^ 1 a - h { E S (II.2>
In the State of ttlermal eqnibrium
f? (i) fr (E ti h (^s)l ^ ~ h (OI
= h (E3) t ( 4) [1 - h (ES' [1 - 7 f (EJ1 (II.3>
Diviaing both sides of (II 3) by / F ( x) fv (En) / F (Es) / F (*), we-
obtain

r 1
- h [Eo
\ r 1
L h (E.)
1n
J
r / 1- r J rL /
l f (Ez) f
1
(4)
1]
.
(H.4>

Comparing this equation with (1.4), one may easilv see that the-
function iff(E) 1 for a degenerate fermion gas satises the
same condition as is satised by the function f (E) in the case of
a nondegenerate gas. This maces it possible to use the result (I.iO)r
which in this case takes the form
d Mi (II.5>
[ v (E) dE
where y is a constant. Integrating (II.5), we obtain

- J - - { = B e y s (II.6>
J p ( ) v
where B is the integration constant.
The following considerations may be of use to estmate the cons-
t-ants y and B. When the condition /F (E) C 'I is satised, the fer
mion gas becomes nondegenerate. For such a gas we can neglect
unity in the left-hand side o (II.6) and rewrite the expression in-
the form
f?(E) = B~le-ys (II.7V
Comparing (II.7) with (I.II) and keeping in miad (L12), we obtain:-

3 = A-1 = s-'llhT, y = - p = iJksT (I I -5 )

Substituting into (II.6), we obtain

f?(E) (II-9)
320 Soiid State Physics

Appendix II!
Derivation of the Bose-Einstein distribution function. In contrast to
ermions bosons ca occupy both vacant States and States already
occupied by other bosons and they do it the more readily the greater
is the occupancy of the States. Therefore the rate of direct colli-
sions Ex E3, E -+ Ei will be the greater, the greater the numbers
-of partiles in the initial States /Bose(i) and / Bose( 3) and the
higher the occupancy of the nal States / Bose ( 3) and boscE)'-

Qi. c/Bose () / Bose ( 2) [1 f /Bose ( 3)] [1 ~T/Bose ('4)] (III.1)

The units in the brackets take account of the bosons ability to go
over not onlv to occupied States but to vacant States, for which
/Bose^a) = BosEi) = , as well. For / Bose() < 1 (the condi-
tion of nondegeneracy) the expression in the brackets in (I I I .1)
becomes unity and Qd becdmes equal to the rate of direct collisions
for the partiles of a nondegenerate gas. For the rate of reverse colli
sions 3 - Ex, Ei ^ E.: we obtain
(?r == c/Bose ( 3) /Bose ( 4) [I "T /Bose (l)l [1 T/3osc ( 2)] (III.2)
In the State of equilibrium Qd = Qr:
/Bose (Ei) /Bose (") [1 T* / Bose ( 3)! [ 1 ~ /Bose ( 4)]
/Bose ( 3) /Bose ( 4) [I t / b os( i)1 [1 ~r f Bose (")] (III.3)
Dividing this expression by / Bose(i) /Bo*e(*) / bosc( 3) / boscO^O.
v e obtain
r r. 1
.1 - /Bose*(3) 11 -/Bose(-i)
J r-/Bose1() i1-
(III.4)
A comparison between (III.4) and (1.3) shows that the function
i 7Boie()- ~ 1* for bosons satishes the same equation as the func
tion /'( )- for- a nondegenerate gas. Therefore we may malee use of
the resuit (I.iO) writing it for bosons in the forra
1
1 -Y (III.5)
/Bose 1 d \ /Bose ()
Integrating (III.5), we obtain
, i ~ + = BevE (III.6)
/Bose (&)
The vales of the constants y and B that enter this expression are
the same as in the case of the degenerate fermiongas:
y = IkBT, B = e ' m Br (IH-7)
Therefore
f _ 1 (III.S)
3se--jTrT7^r
.<
Appendix IV 321

Appendix !Y
Table 1

VI NA

VIII C

Vi l II
VII A
III A

VI A
V AI

V II

V A
>
> > > 0

I H He

11 Li Be B C N 0 F Ne

III Xa Mg transition elements Al Si P s C1 Ar

IV K Ca Se Ti V Cr Mu Fe Co Ni Cu Zn Ga Ge As Se Br Kr

V Rb Sr Y Zr Xb Mo Te Ru Rh Pd A g Cd In ?l Sb Te I Xe

VI Cs Ba La1 llf Ta IV Re Os Ir Pt Au H? TI Pb Bi Po At Rn

VII Fr Ra Ae-
1

I cas; II class III class IV

class

i Lauthanicies
- Actinides

Table .2-

21-0885
 Glossary of Symbois and Noaiions

A Iattice basis; Madelung coastant; energy o exchanga interaction

a lattice constant, Bohr radius
B magnetic field induction
Cv constant volume heat capacity
Cs heat capacity of electrn gas
c capacitance; Curie constant; Coulomb interaction energy
velocity o light
diSusion coecient
electrn diSusion coefcient
hole diSusion coecient
barrier width; space charge layer tvidth
energy; Young s modulus
acceptor excitation energy
born energy
bottom o conduction band
donor excitation energy
electrn gas energy
excitation energy
energy gap
? Fermi energy
orbdden band -width
e

"'lattice lattice energy
Bry.m energy of normal mode
Ev top o valence band
% electric eld intansity
contac-t eld intensity
F free energy; orce
interaction orce
/ (E) distribution unction
Bose (E) Bose-Einstein distribution unction
te (E) Fermi-Dirac aistribution unction
tei (E) Maxwell-Boltzmann distribution unction
G number o States; shear modulus
oa generation rate; anharmonicity coecient; Land factor
g (E) density of States
g (u ) requency distribution unction o normal modes
H magnetic eld intensity
I light flux; current
te base current
te collector current

322
Giossary o Syrr.faols and Nofafions 3'23

/e cmitter current
i~ current density
i forward current density
ir reverse current density
i3 saturation current density
J lignt intensity; srtchange integral; intrinsic quantum number
/m magnetizaron
Is saturation magnetization " " ...
k wave vector of electrn -
k absorption coeicient
k-Q Boltzmann constant
i" heat conductivity
7Clattics lattice heat conductivity
electrn gas heat conductivity
L ~ linear dimensin; diSusion length; Lorentz number
L diSusion length of electrons
Lv diSusion length o boles
V orbital quantum number; transport mean fres patb
M mass; eiectric dipoie mcment, magnetic moment
Mj total magnetic moment of atom
ni electrn rest mass; partile mass
m.Sff eSective mass
mj ' magnetic quantum number of atom
mi magnetic quantum number of electrn
m. n electrn eSective mass
hole eSective mass
A" number of partiles
VA Avogadro s number
N'{E) total distribution unction
A a acceptor concentration
A'd aonor concentration
Arlm concentration of impurities
n ' concentration of particles; electrn concentration in conduction
band
n. , equilibrium electrn concentration in intrinsic semiconductor
nn0 ' equilibrium majority carrier concentration in n-type semiconductor
nph phonon concentration
7i u0 quiiibrium minority carrier concentration in p-tvpe semiconductor
Pj atomic angular momentum
p momentum; pressure; bote concentration
electrn momentum
pp Fermi energy electrn momentum
Pj hole concentration in intrinsic semiconductor
p orbital angular momentum
Pna equilibrium minority carrier concentration in n-type semiconductor
Pr)-a phonon momentum
p,j0 equilibrium majority carrier concentration in p-type semiconductor
pj, electrn spin
Q quantiy of heat;, eiectric charge
aestruction energy
Qp Peitier heat
sublimation energy
qJ electrn charge; phonon wave number
R recombination rate; gas constant
- Hall coeicient
Rn electrn recombination rate
Rp hole recombination rata
r ' distancs between particles
 324 Soiid State Physics

equilibriura interparticle distance

S entropy; siip plae
T temperature
Te Curie temperature
Ts r critical or transition temperature
TF Fermi temperature
intrinsic conductivity transition temperature
h impurity exhaustion temperature
P potentiai energy
bond energy
rb magDetic energy
mobility
electrn mobility
n hele mobility
F potentiai; voltaje; tensin; volume
V- contact potentiai dference; collector voltage
F,ph photo-emf
Ff thermal emf
velocitv
<o thermal velocitv
Ld drift velocitv
l'e electrn velocitv
iP Fermi energy electrn velocitv
IF work
particle s displacement from equilibrium position
a polarizability of molecules; free surface energy; linear thermal ex
pansin coecient; temperature ccefcient of resistance; diierential,
or specific, thermoelectric power
bond rigiditv cocfhcieut; quantum eficiency (vield)
r phase volume
relative shear deormation, gyromagnetic ratio
reiative permittivity; quantum energy; relative extensin deorma
tion
Spb phonon energy
0 Dbye temperature; paramagnetic Curie point
pC ferromagnetic Curie point
wavelength; mean free path; magnetic fteid penetration depth in
conductor
^ph phonon mean free path
u .-vp.-Chemical potentiai (Fermi level); molecular mass; magnetic suscepti-
biiity; magnetic moment
PB Bohr magneton
a electrn orbital magnetic moment
Fermi leve! in n-type semiconductor
r Fermi level in p-type semiconductor
Ms electrn intrinsic magnetic moment
V frequeney; number of collisions; Poissons coecient
n Peitier coecient
p specic resistance; space charge density
a specic conductance; normal stress
a theoreticai strength
dif diSerential conductivity
al intrinsic specic conductance
Fin impurity (extrinsic) specic conductance
^ph photoconductivity, phonon eective cross section
a oho stationary photoconductivity/
ay real (techaical) strength
lifetime; relaxation time; tangential'(shear) stress; durability
Glcssary of Symbois and. Nciaticr.s 325

TCr critica! saear stress

xn electrn lifetime
x~ hole lifetime
<p' electrn potential energy; angle
cp0 equilibrium potentialbarrier in p-n junction
X thermodynamc work function; magnetic ' ,
susceptibiiity
\jj wave unction
a angular trequencry
cq Debye frequency
c e L a r m c r frequency
 Bibliography

General
0.1 V. L. Bonch-Bruevich and S. G. Kalashnikov: Semiconductor Physics,
Nauka , Moscow (1977) (in Russian).
0.2 G. 1. Epifanov: Physical Principies of Microelectronics, Mir Publishers,
Moscow (1974).
0.3 P. T. Qreshkin: Physics of Semiconductors and Dielectrics, Vysshava
Shkola , Moscow (1977) (in Russian).
0.4 L. S. Stil bans: Semiconductor Physics, Sovetskoe radio , Moscow (1967)
(in Russian).
0.5 K. V. Shaiimova: Semiconductor Physics, Energiya", Moscow (1976)
(in Russian).
0.6 V. "V. Novikov: Theoretical Foundation of Microelectronics, Vysshaya
shkola", Moscow (1972) (in Russian).
0.7 J. M. Ziman: Principies of the Theory of Solide (2nd edition), Cambridge
U.P., Lonaon (1970).
0.8 R. L. Sproull: Modern Physics (2nd edition), W iley, New York (1964).
0.9 K. Seeger: Semiconductor Physics, Springer, Wien (1973).

Chaper 1
1.1 G. C. Pimental and R. D. Spratly: Chemical 3 onding Clarified Through
Quantum Meckanics, Hoiden-Day, San Francisco (1969).
1.2 W. Haberditzl: Bausieine der Materie und chemische Binding, Deutscher
Verlag der Wissenschaften, Berln (1972).
1.3 L. Pauling: General Chemistry (3rd edition), W. H. Freeman, San Fran
cisco (1970).
.4 J. A. Campbell: Chemical Systems, W. H. Freeman, San Francisco (1970).
1.5 I . Ficini, N. Lumbroso-Bader, and J.-C. Depezay: Elemente de ckimie-
physique, Hermann, Paris (1968-69).
1.6 C. A. Wert and R. M. Thomson: Physics of Solids (2nd edition), McGraw-
Hill, New York (1970).

Cnapfer 2
2.1 t a. I. Frenkel: Iniroduction to the Theory of Metals (4th edition), Nau
ka , Moscow (1972) (in Russian).
2.2 M. Kh. Rabinovich: Strengih and Supersirengik of Metals, Izd-vo Akad.
Nauk ciSSR, Moscow (1963) (in Russian); Strengih, Temperature, Time,
Nauka , Moscow (1968) (in Russian).

326
Sibiiography 327

2.3 A. V. Stepanov: Fundamentis of the Theory of Practical Strength of

Crysials, Nauka", Moscow (974) (in Russian).
2.4 G. M. Bartenev'and Yu. S. Zuev: Strength. and Ruplare of Hyperelastic
Materials, Khimiya , Moscow-Leningrad (1364) (n Russian).
2.5 Physical Metallurgy, R. W. Cann (Ed.), North-Holland, Amstsrdam
(1965). i
2.5 Fracture, H. Liebowits (Ed.), Academic Press, New York (1972).
2.7 F. A. McCerntrock and A. S. Argn: Mechanical Behaviour oj Materials,
Addison-Wesley, Reading, Mass. (1966).
2.8 See 1.6.

Chapfer 3
3.1 Y. G. Levich: Introduction to Siatistical Physics (2nd edicin), GITTL,
Moscow (1954) (in Russian).
3.2 F. Rei: Statistcal Physics (Berkeiey Physics Course, Yol. 5), McGraw-
Kill, New York (1972).
3.3 H. Bethe and A. Sommerfed: Electronentheorie der Metalle, Springer,
Beriin (1933).
3.4 C. Kitte: Introduction to Solid State Physics (4th edicin), Wiiey, New
York (1971); Elemenary Solid State Physics: A Short Course, Wiiey,
New York (1962).
3.5 See 0.1-0.7.

Chapfer 4
4.1 A. F. lorie: Physics of Semiconductors, Inosearch, London (1960).
4.2 A. J. Dekker: Electrical Engineering Materials, Preutice-Hali, Engle-
wood CLifls. N. J. (1959).
4.3 G. A. Siack: Heat conduction in solids in Encuclopaedic Dictionary
of Physics, J. Thewlis (Ed.) Vol. 3, Pergamon Press. Orciord (1361),
601-610.
4.4 See 0.1-0.8.

Chapfer 5
5-1 J. Callaway: Energy Band Theory, Academic Press, New York (1964).
5.2 Semiconductors, N. B. Kannay (Ed.), Reinaoid, New York (1353).
5.3 R. A. Smith: Semiconductors, Cambridge U.P., London (1959).
5.4 J. N. Shive: The Properties, Physics, and Design of Semiconductor Devices,
Van Nostrand, Princeton, N .i. (1959).
5.5 See 0.1-0.9, 1.6, 3.3, 4.1.

Chapfer 6
6.1 G. Mirdel: Elektrophysik, Beriin (1970).
6.2 A. C. Rose-Innes and E. H. Rhoderick: Introduction to Superconductivity,
Pergamon Press,, Ordord (1969).
6.3 I. Bardeen and J. R. Schrieer: Recent developments in superconducti-
vity, in Progr. Low Temo. Phus.,. Yol. 3, Chapter 4, North-Holland,
Amserdam (1961).
6.4 D. K. Douglass and L. M. Falikov: The superconductivity energy gap,
in Progr. Low Temp. Phys., Yol. 4, Wiiey, New York. (1964), 97.
323 Soiid State Physics

5.5 D. N. Langenberg, D. J. Scalapino, and B. N. Taylor: The Josephsoa

eBect". Sci. Amsr. 214, No. 5 (1965), 30-9.
5.5 J. Berdeen: Physics Today 26, No. 7 (1973).
6.7 J. Schrieer: Physics Today 26, No. 7 (1973).
5.8 L. Coooer: Physics Today 26, No. 7 (1973).
6.9 J. Giaver: Science 183, No. 4131 (1974).
6.10 B. D. Josephson: Science 184, No. 4136 (1974).
6.11 W. P. Jollv: Cryoelectronics, The English U.P., London (197)
6.12 See 0.1-0.91 1.6, 2.1, 3.3, 4.1, 4.2,. '.2-5.4.

Chapier 7
.l S. V. Vonsovskii: The Coniemporary Science of Magnetism, Gostekhizdat,
Moscow-Leningrad (1953) (in Russian).
7.2 Ya. G. Dorman: Magnetic Properties and Structure of Matter, Gostekh
izdat, Moscow (1955) (in Russ'ian).
7.3 L. V. Kirenskii: Magnetism, Izd-vo Akad. Nauk SSSR, Moscow (1963)
(in Russian).
7.4 See G.1-0.S, 1.6, 2.1, 3.3, 4.2.

Chapier 8
8.1 Ya. I. Fedotov: Fundamentis of Physics of Semiconductor Devices, So-
vetskoe radio , Moscow (1969) (in Russian).
8.2 I. p. Zherebtsov: Electronics (reprint of the 2nd edition), Mir Publishers,
Moscow (1975).
3.3 V. V. Pasvnkov, L. K. Chirkin, and A. D. Shinkov: Semiconductor
Devices. Vysshaya shkola , Moscow (1966) (in Russian).
8.4 A. F. Gorodetskii and A. F. Kravchenko: Semiconductor Devices, Vys-
shaya shkola , Moscow (1967) (in Russian).
8.5 S. N. Levine: Principies oj Solid-State Microelectronics, Holt, Rinchart
and Winston, New York (1963).
8.6 J. H. Kalish: Microminiature Electronics, Bobbs-Merrill, Indianapolis
(1967).
3.7 G. F. Alfrev: Physical Electronics. Van Nostrand, Princeton, N.J. (1964).
S.S See 0.1-0.9, 1.6, 4.1, 5.2-5.4, 6 .1.

Chapier 9
9.1 See 0.1-0.7 0.9, 3.3, 4.1, 5.3, 5.4, 6.1.
 Index
Acceptors, 159
Activators, 205
Akulov XV S., 231
Amplier, pararaetric, 261
Anisotropy, crystal, 50
Antibarrier .ayer, 274 Brillouin zone, 145
Approximation, harmonio, 4S
Band, allowed eaergy, 140
couduction, 142
energy, 139
orbidden energy, 140
hybrid, 141
valonee, 142
Bareen J., 20S, 293
Bartenev G. M., 78
Barkhausen H., 227
Barrier ayer, 271
Basov X. G., 253
Boltzmann L. E., 89
Bond, covalent, 16
hydrogen, 22
ionic, 15
metallic, 21
van der Waals, 14 Curie point, antierromagnetic,
Borazon, 41
Bose S. N., 90
Boson, 88
Brattain W. H., 293
Breakdown, p-rt junction, 237
avalanche, 287
thermal, 287
tunnel, 2S7
. Zener, 2S7
Burgers vector, 59
Carriers, equilibrium, 166
majority, 166, 280
- mmority, 163, 283
nonequilibrium, 136
Chemical poteatial, 86.
Coefcient, absorption, 137
linear expansin, 125
Coherence length, 217
Contact potential dierence,
nal, 269
internal. 289
Cooper L., 208
Cooper pairs, 215
Coordination number, 33
Counters, semiconductor, 203
Curie P., 226
erromagnetic, 228
paramagnetic, 22S
Current, orward, 2S4
reverse, 285
330 Index

Curreat (cont.) Einstcin, A., 90

saturation, 285 * Electrophotography, 202
Current-voltage characteristic, 285 Electron pairs, 214
Energy, internal, 85
Dark conductivity, 198 Entropy, 246
Debve frequency, 111 magnetic, 246
temperatura, 111 Epitaxial xnethod, 279
Defects, Frenkel, 42 Equation, current-voltage characte
Schottkv, 43 ristic, 277
Destruction of solios, brittle, 71 van der Waals, 11
plstic, 71 Exchange forc, 19
viscous, 71 interaction, 19
Diamagnetism, 223 Excitons, 200
Dielctrica, 153
Diiusion length, carrir, 167 Fabrikant V. A., 259
coefficient, carrier, 168 Fermi E., 89
Diodc-s, Gunn, 195 Fermi level, 96
high frequency, 28S energy, 97
impulse, 289 Fennions, 88
rectifier, 288 Ferrites, 256
tunnel, 291 Ferromagnctics, 226
Zener, 231 Fields, strong, 193
Dirac P. A. M., 89 weak, 193
Dislocations, edge, 59 Fluorescence, 204
screw, 61 Frank, F. C., 67
Dispersin curves, 109 Frank-Read source, 67
Domains, ferromagnetic, 251 Frenkel Ya. I., 249
Donors, 158 Function, Bloch, 144
Dorfnian Ya., 248 Bose-Einstein, 320
Drit, carrier, 170 Brillouin, 243
velocity, 170 distribution, 90
velccity satnration, 193 Fermi-Dirac, 318
Durability, 77 Langevin, 243
Maxwell-Boltzmann, 317
ESect, Barkbausen, 227
Einstein-de Haas, 247 Gap, energy, 146
Ettingshausen, 313 in superconductors, 211
Gunn, 195 Gas, degenerate, 174
Hall, 310 electrn, 138
' magnetoeiastic, 231 nonaegenerate, 174
Nernst, 314 Generation, carrier, 167, 197
Peltier, 307 rate, 167
Seebeck, 3C2 Grneisen constant, 126
Thomson, 310 Gurevich L. E., 307
Zener, 194
Gyromagnetic ratio, 233
 ndex

Heat conductivity, 127 Hooke s, 46

o electrn ga3, 130 Stokes , 207
lattice, 123 ~ " Vavilov's, 205
o metis, 130 Leipunskii O, I., 40
Heisenberg W., 45 Lie-time,_ average carrier, 137
Heitier W. H., 19 London._F.,-'12, 19
Holes, 155 Lorentz number, 176
Hysteresis, 227 Luminescence, 203, 296
anti-Stokes, 204 .....
- Luminophors, 203
mpurit-ies, 44
Indices, site, 29
odirection, 30 Madelung constant, 13
plae, 30 Magnetizaron, 224
Injection, minority carrier, 294 Magneton, Bohr, 234
Integrated circuits, 299 Magnetostriction, 230
Interaction, deiormation, 14 Masers, 261
dispersin, 11 Mass, eSective o electrn, 148
induction, 14 negative eective-, 150
orienta tional, 13 Matthiessen s rule, 187
Interstitals, 42 Maxwell J. C., 89
lo ge A. F., 243 Meissner V.'., 208
Ionization, electrostatic, 194 Mobility, carrier, 17
impact, 194 Hall, 312
thermoelectron, 194 Modes, acoustical, 109
normal, 110
optical, 110
Kapitza P. L., 24S Modulus, e'iasticity, 48
Keesom W. H., 13 shear, 50, 52
Youngs, 48
Moment, inducen magnetic, 239
Langevin P., 240 magnetic, o nucleus, 235
Larmor requency, 239 orbital angular, 133
Lasers, semiconductor, 297 orbital magnetic, 232
Lattice, body-centered cubic, 35 Mosaic structure, 42
Bravais, 26
coordination, 32
< ace-centered cubic, 35 Negative digerential conductivi
hexagonal close-packed, 35 ty, 196
hydrogen. 32 Nondegeneracy, 89, 93
ionic, 32
metallic, 32
molecular, 32 Ochseneld R., 208
with a basis, 27 Orosvan E., 73
Law, dispersin, 142 Oscillators, linear harmonio, 112
Dulong and Petit, 117 normal, 113
332
Paramagnetics, 225
Paramagnetism o electrn gas, 244
Paraprocess, 254
Path, mean ree, 172
transport mean ree, 172
Pauli exclusin principie, 88
Peltier J. C. A., 308
Permeability, magnetic, 225
Phonons, 2
Phonon drag, 307
Phosphor crystais, 205
Phospborescence, 204
Photocells, 295
Photoconductivity, 199
PhotoeSect, internal, 198
Photoluminescence, 203
Photoresistors, 201
Photovoltage, 295
Pikus G. E., 307
Planck M., 105
p-n junction, 278
Polanyi M., 71
Polarizability, 13
Polymorphism, 3o
Polymorphic modifications, 3S
Popula tion inversin, 210
Potential barrier, 273
Pound R. V., 258
Precession, electrn, 238
Prokhorov A. M., 259
Quantization o phase space, 92
Quantum number, magnetic, 134
principal, 133
orbital, 133, 233
spin, 134
Quantum yieid, 199
Radiation, stimulated, 259
thermal, 203
Read W. T., 87
Recombination, ree carrier, 166
radiative, 168,
rate, 167
Index
Recombination (cont.)
nonraiative, 163
Recrystallization, 55
Rectification, 274
ratio, 277
Relaxation, 54
processes, 172
time, .172___
Remanence, 227
Residual magnetization, 277
Resistance, tbermai, 126
Resistivity, residual, 187
temperatura coeficient o, 137
Resonance', electrn paramagnetic, 258
nuclear paramagnetic, 258
Resonators, cavity, 262
Rozing B., 248
Saturation, tecbnical, 226
Scattering, pnonon-phonon, 127
Schottky barrier, 277
Schrieer J. R., 208
Seeheck T. J., 302
Semiconductors, impurity, 156
inrinsic, 102, 153
nondegenerate. 94
rc-tvpe, 15S
p-tvpe, 159
Shokley W ., 293
Slipping, 51
Sip planes, 52 .
system, 52
Sommereid modei, 209
Spin, 234
Stepanov A. V., 83
Stoletov A. G., 225
. Strength o soids, real (tecbnical), 71
theoretical, 71
Stress, critical shearing, 53
shearing, 51
yieid, 50
Superconductors, 207
Superconductivity, destruction o, 221
practica! uses o, 222
Suseeptibility, magnetic, 224
Temperatura, critical, 207 Tunnelling, 138
ezauscion, 155
Nei. 255
negative, 259
Vacancy, 43
saturaticn, 165
Vector, wave, 143
transition, 207
Tnermoeiectric power, diSeren-
tial, 302
of metis, 3C4 Weiss P., 248
: : semiconductora, 306 Whiskers, 8i

en
specic 302 Work function, 26
7 n i ~ astriccin, 231 thermodynamical, 268
Torray H. C., 258
Traasistors, 293
Twinning, 55 Zhurkov S. N., 78
plae, 55 Zavoisky E. K., 258