Вы находитесь на странице: 1из 142

THE HEIDELBERG SCIENCE LIBRARY I

Volume 3
Order
and Disorder
in the
World of Atoms
A. I. Kitaigorodskiy
Translated by Scripta Technica, Inc.
Edited by S. Chomet, King's College, London

SPRINGER-VERLAG NEW YORK INC. 1967


Originally published as "Poriadok i Besporiadok Y Mire AtomoY" by Nauka
Press, Moscow, 1966.

All rights reserved, especially that of translation into foreign languages. It is


also forbidden to reproduce this book, either whole or in part, by photomechan
ical means (photostat, microfilm and/or microcard or any other means) with
out written permission from the Publishers.

ISBN 978-0-387-90004-9 ISBN 978-1-4615-7559-7 (eBook)


001 10.1007/978-1-4615-7559-7

1967 by SpringerVerlag New York Inc. Libra,ry of Congress Catalog Card


Number 6721458
Softcover reprint of the hardcover 1st edition 1967

Title No. 3913


CONTENTS

Chapter I-DISORDER 1

1. What constitutes a disordered arrangement? 1


2. The gaseous state of matter 3
3. Thermal motion in gases . 5

Chapter 2-0RDER 9

1. The symmetry of wallpaper patterns. 9


2. Crystals. 14
3. Invisible lattices 15
4. Crystals and the close packing of spheres 17
.'5. Crystals that are not close-packed assemblies of spheres 24
6. Same atoms but different crvstals 31
7. Long-range order 32
8. Order in microcrystalline bodies 33

Chapter .3-ELEMENTS OF ORDER IN DISORDER 36

I. Short-range order and the structure of liquids 36


2. The amorphous solid state 39
3. Liquid crystals 42
4. Thermal motio]] of particles in liquids 44

Chapter 4-ELEMENTS OF DISORDER IN ORDER 46

1. Thermal motion of atoms and molecules in crystals 46


2. The gas-crystalline state of matter 47
3. Block structure 49

v
4. Dislocations 51
5. The dislocations move 55
6. Ideal crystals 58
7. Defects within the blocks 60
8. Crystals with errors . 62
9. Order and disorder in binary alloys 64
10. Magnetic order 70

Chapter 5-0RDER AND DISORDER IN THE


WORLD OF LARGE MOLECULES 78

1. Long and branched molecules 78


2. Bundles of long molecules 80
3. Behavior of bundle polymers 82
4. Alignment of polymer molecules to form single crystals . 84
5. The structure of polymers 86
6. The living cell 88

Chapter 6-TRANSITIONS BETWEEN ORDER


AND DISORDER 92

1. Iron vapor and solid air 92


2. Water-an exception to the rule. 97
3. The growth of crystals 98
4. Spiral growth 102
5. Conversions between crystal structures 107
6. Delayed transitions 115
7. Particles do change place in cry~tals 117

Chapter 7-0RDER OR DISORDER? 121

1. Probability and disorder 121


2. The tendency toward disorder 122
3. The tendency toward order 125
4. The struggle between order and disorder 128

vi
INTRODUCTION

Our main aim is to examine whether the atoms and molecules


constituting the world around us are distributed in space in a
random and disordered fashion, like pebbles on the beach, or in
an ordered pattern like the cells of a honeycomb. However, it is
often impossible to make such a clear-cut distinction, and it is
better not to use "order" and "disorder" as absolute terms but to
speak instead of a "degree of order" and a "degree of disorder."
These concepts are fairly new in science. Up to about 20-30 years
ago it was still believed (and in fact this belief can still be en-
countered today) that certain states of matter - such as gases,
liquids, and amorphous solids - were characterized by a totally
disordered distribution of the constituent particles, whilst crys-
tals, by contrast, exhibited perfectly ordered lattices. According
to the present view, on the other hand, order and disorder often
coexist inseparably from each other, though there are admittedly
many cases in which "order" or "disorder" does describe quite
accurately tbe actual state of affairs.
Symptoms of disorder have recently been found in seemingly
perfectly regular molecular structures, and symptoms of order in
seemingly perfectly chaotic aggregations of particles. These dis-
coveries led to the formulation of new and important laws cor-
relating the structure of substances with their properties, and to
tIlt' explanation of many phenomena in terms of changes in the
degree of order. Furthermore, transitions between order and dis-
order are very important, partly because they underlie various
technological processes.
One example of the practical importance of order-disorder
phenomena is represented by polymers, in which the spatial ar-
rangement of the various atomic groups shows features of order
side by side with features of disorder. These features collectively
define the stereo-regularity of the molecular chains in polymers,
which in turn determines the properties of these materials. If we
vii
had not discovered the laws of order and disorder, our life would
hardly have become inundated as it has by synthetic polymers in
the form of fabrics and plastics, to mention only two examples.
Order-disorder phenomena have thus ceased to be merely a
topic of theoretical discussion: they have come to affect everyday
life. The author feels that this is a sufficient raison d' etre for the
present book.

viii
CHAPTER 1

DISORDER

1. What constitutes adisordered arrangement?


By a rough and ready definition, any collection of items in which
we can see no pattern or design is characterized by disorder.
These items may be particles or objects of any kind, and will
sometimes be referred to as subsystems, their collections being
called systems, though they may show no "system" in the every-
day sense of the word. When the subsystems form a recognizable
sequence or any other spatial pattern, we can no longer speak
of disorder. However, to avoid such loose definitions in physical
sciences, let us consider what the physicist understands by
"disorder".
Imagine several thousand grains on a chessboard. If it is found
on closer inspection that, for example, the grains occupy only
the white squares, or that they occupy both the white and the
black squares but in such a way that each square carries the
same number of grains, or that each square carries an even num-
ber of grains, or that there is some regularity in the number of
grains along the knight's move, we can be sure of the absence
of ideal disorder.
Let us now shake out a bagful of grains onto a clean chess-
board, making sure only that the grains form a single layer. They
are then distributed literally "as chance would have it." But
what does this really mean? To examine this point more closely,
we put a strip of paper on the chessboard and count the grains
along it. Suppose this first count gives an answer of 167. We
then place the strip in other positions and record the number of
grains in each case. If we thus obtain a set of rather similar
figures (e.g., 159, 172, 165, 167, 169, and so on), we can say that
2 Order and Disorder in the World of Atoms

there are no sites or lines that are particularly favored and others
that are particularly unfavored; this is one of the criteria of ideal
disorder. When such a situation prevails, the greater the number
of grains, the smaller will be the relative deviation in the density
of the grains along various lines on the board. In addition, the
quasi-equality of the counts should be observed irrespective of
the position of the strip.
The above characteristic of ideal disorder is known as iso-
tropy, a term derived from the Greek and meaning "equivalence
of all directions." Conversely, bodies in which the various direc-
tions are not equivalent in this sense are called anisotropic.
Returning to the above example, if we find that the strip of
paper encounters consistently greater numbers of particles along
one direction than along the direction at right angles to the first,
the distribution is clearly anisotropic and is not characterized by
ideal disorder.
Ideal disorder must satisfy the requirement of a uniform
density distribution. If we replace the strip of paper with a wire
mesh of the size of one square of the chessboard, and thus sub-
divide the square into small squares (bearing only in mind that
the latter must be sufficiently large to accommodate a reasonable
number of grains), we find that the small squares will behave
exactly like the large ones, each containing roughly the same
number of grains, say, 10, 9, II, 12, 8, 10, etc. This further con-
firms the equal density of distribution. It must be added, how-
ever, that uniform distribution is a necessary but not a sufficient
condition for disorder. It does guarantee, though, that the grains
do not agglomerate.
Speaking of order and disorder, it is necessary to specify
exactly what kind of order one has in mind. The grains in Fig. 1
are drawn accurately, but they nevertheless illustrate the con-
cepts of ordered and disordered distributions. The order and
disorder in question concern here the numbers of the grains. It
would be just as reasonable to arrange spheres of equal size but
of two different colors in perfect rows, and to speak of a dis-
ordered arrangement in the sense that the colors form a random
sequence.
In this book we shall be dealing mainly with order and dis-
order in the geometrical arrangements of particles.
Disorder 3

0) b) c)

Fig. 1. Grains on a chessboard: a) Simple order: the white squares carry


two and the black carry three grains b) More complex order: all white
squares carry even numbers and all black squares the same number of
grains c) Complete disorder

2. The gaseous state of matter


Having obtained a better picture of disorder, we can now
examine various aggregates of atoms and molecules searching
for ideal disorder, and observe that the gaseous state is char-
acterized by perfect disorder as regards the disposition and the
movement of the constituent particles.
Although there is no microscope on earth that could directly
reveal the constantly moving particles of gases, physicists can
describe in fair detail the phenomena that occur in this invisible
world.
Under normal conditions (at room temperature and atmos-
pheric pressure) one cubic centimeter of air contains 25 x 10" 8
(25 billion billion) molecules, so that each molecule has a space
of about 0.4 x 1O-19 cm 3 , corresponding to a cube with a side of
0.34 x 10-6 cm. The molecules are of course minute. For example,
the molecules of oxygen and nitrogen-the major constitutents
of air-have a maximum diameter of 0.4 x 10- 7 cm. (Examples
of the models of some simple molecules are shown in Fig. 2.)
Given the number of molecules in 1 cm 3 of air, we find that the
average distance between the molecules is about ten times as
long as the maximum molecular diameter, so a cube capable of
accommodating 10 x 10 x 10 = 1000 such molecules in actual
fact contains only one.
Consider a set of dime-size pebbles arranged randomly on a
smooth surface in such a way that on average each square meter
4 Order and Disorder in the World of Atoms

0) b)

o <> <t,

o co
0
c o <> ()

c c
()

CO 0
<> 0
00. 0 000 0. C Q

0
Q <> <> <> 0
<>
0 0
0

0 o
C>
0

"" C>
o <>

0 0
0

oj

00 0 ~ 0 ------ -~-- ~

o o 0 0 0 0 0 :
0 0 I
o 0 0 0 1
o 0 0 0 0 0 0 I
o I
00 o 0 0 CI
I
I
I
o 0 0 CI o o
I
,
C
I
I
c: 0 0 0 o 0 0
I
o 0
I o
I
o 0

,
I o 0

I
o 0 0

o
o 0
o
0
o
,LI __________,,--,-_-,,-_-,0,---,

b)

Fig. 2. Molecular models: a) oxygen b) carbon dioxide c) benzene.


The outer contours (which are in fact far from sharp, because the dens
ity of molecular substance tails off only gradually) are found from the
packing of molecules and atoms in crystals
Fig. 3. Dimes distributed in disordered fashion over an area of 1 m2 :
a) perspective view b) plan

of the surface contains one hundred pebbles. This gives an idea


of the density of molecules in a gas.
Let us now isolate 1 cm 3 of air under normal conditions and
imagine that we can determine the density of the air separately
in each half of this volume at every moment of time without
Disorder 5

introducing a physical partition. We shall then find that the


density-and therefore the number of molecules-in the two
halves is practically the same. We can continue such measure-
ments in the hope that a detectable difference will eventually
be brought about by the disordered motion of the molecules, but
we shall have little chance of success. Theoretical calculations
show that no such difference would appear even in a billion
years.

3. Thermal motion in gases


Figure 3 may be regarded as an instantaneous snapshot of a gas,
showing the positions of the molecules frozen in an instant of
time. In fact, such a photograph would require an extremely
short exposure. The shutter of this nonexistent camera would
have to remain open only during the time in which the molecule
moved by no more than some tenth of its diameter, as otherwise
the photograph would resemble a time exposure of a ballroom
filled with dancing couples. One-tenth of a molecular diameter
amounts to about 1O~8 mm, and the molecular velocity is of the
order of 1000 m (1,000,000 mm) per second. Our exposure
would thus have to be something like 10-14 sec.
Is the molecular motion also completely disordered?
Each molecule in a gas is in a state of constant movement. Let
us now consider a gas at a low pressure, such that the molecular
interactions reduce to collisions, and concentrate on the behavior
of a single molecule. Now it is flying to the right. If it encoun-
tered no obstacles it would continue moving in the same direc-
tion at constant velocity, but its path intersects countless other
molecules and collision is unavoidable. Following the collision,
the molecules fly apart like two colliding billiard balls. Our
molecule may bounce away in any direction, and at a slower or
a faster speed, depending on the individual case. Clearly, being
subject to such random collisions, the molecule will shoot back
and forth inside the vessel in which the gas is contained.
Sometimes the collisions will follow one another in quick suc-
cession. In other cases, the molecule may be left in peace for
a relatively long time. The molecule may build up a high
velocity as a result of a series of fortuitous "pushes," or it may
6 Order and Disorder in the World of Atoms

be prevented from building up an appreciable velocity by


opposing collisions.
If the gas contained in the vessel is not heated, compressed,
or subjected to any other influence, its energy (composed of the
kinetic energies of its molecules) will remain constant. This may
seem paradoxical since the individual energies of the molecules
are in a state of continuous flux, but the answer is simply that
as one molecule is slowed down at some instant another is
simultaneously speeded up, so that the relative proportions of
particles moving at certain velocities remain constant. All that
happens is that individual molecules constantly change their
allegiance from one velocity group to another.
Let us now take a look at the numbers of molecules moving
at low, medium, and high velocities. At first sight it seems that
the proportion of molecules moving at, say, between 1 and 2
meters per second, should be equal to the proportion of those
moving between 101 and 102 meters per second. In the case of
ideal disorder there should be no preference for any particular
velocity. However, as soon as we examine the problem of the
number of very fast molecules we see that this is not the case.
In the extreme case, the whole energy of the gas might be con-
centrated in the very fast molecules. This is obviously nonsensi-
cal, so that the proportion of such molecules cannot be very
high and the numbers of molecules moving at different speeds
are not the same. In fact, most molecules exhibit a medium
velocity.
In everyday life, the "average" is often the most frequent
occurrence. In the case of streetcars scheduled to arrive at a
stop every four minutes, the usual (average), most frequent
period of waiting will be four minutes, more rarely 1-2 minutes,
and only infrequently 10 or 15 minutes. The greater the de-
parture from the average, the more rarely will it occur.
The same reasoning can be extended to people crossing a
busy street at a certain point within a period of, say, one minute.
Generally, their number will be composed of equal proportions
of men and women, and deviations in one or other direction will
become rarer in proportion to their magnitude. It would be rare
indeed to observe 100 men and no women. However, no matter
how high the probability, it is never synonymous with certainty.
The average case is not always what we encounter most often.
Disorder 7

Such coincidence of the two quantities will only occur if the


deviations on either side of the norm are equally probable.
Suppose we are standing by the side of a motor-racing track,
and we have established that half of the motorcycles go past at
a speed greater than 130 miles per hour, while the other half are
slower. We call 130 miles per hour the most probable (most
often observed) velocity. If any deviations on either side of this
average are due to the skill of the drivers and the service me-
chanics, we shall observe such small deviations with equal
frequency. The most probable velocity will then coincide with
the average velocity. But now suppose that 10% of the machines
get into difficulties and move off the track, passing us at only 20
miles per hour. The most probable velocity is the same as before,
but the average velocity immediately becomes much lower,
being calculated as the total speed of the machines divided by
their number.
It is easy enough to calculate the average and the most
probable velocity of racing motorcycles and the percentage of
cases in which the velocity will be greater or smaller than this
average or the most probable velocity by a given value. How-
ever, such calculations are of little value when they are applied
to individual cases. Given sufficient information, we can draw a
number of conclusions about the character of the velocity dis-
tribution, but we cannot predict the velocity of a particular
machine. All probability calculations are most reliable when
they deal with a large number of objects or events.
The number of molecules in a cubic centimeter of gas has
already been given. Under normal conditions this number ex-
ceeds the total population of the Earth by a factor of ten billion.
This means that probability calculations for molecules will be
very exact.
Using the theory of probability, physicists have calculated the
numbers of molecules (in a unit volume of a gas) moving with
particular velocities, and found that the average velocities occur
most often. In view of the great number of molecules, this
average velocity remains the same from instant to instant, as do
the relative numbers of molecules possessing definite velocities.
Consider a numerical example which is illustrated in Fig. 4.
At about 15C the most probable velocity of nitrogen molecules
is something like 500 m per second: 59% of the molecules move
8 Order and Disorder in the World of Atoms

at 300-700 m per second. Only 0.6% of the molecules have the


low velocities of 0-100 m per second, and only 5.4% are faster

5.4
o 100 JOO .500 100 1000 fJOOm/sec

Fig. 4. The velocity distribution for nitrogen molecules at


room temperature. The numbers (proportional to the areas
of the columns) indicate the percentages of molecules
moving with given velocities

than 1000 m per second. The mo~ecular velocity distribution was


first calculated theoretically by Maxwell.
The disorder in the motion of gas molecules appears in that
the same number of molecules moves in any given direction at
any given time, and that the majority of the molecules have a
velocity close to the average.
How often do molecules collide, and how far can they travel
without undergoing collision? These questions too should be
answered in terms of averages, because the time between colli-
sions will vary over a wide range. There is of course no regular
connection between any two successive collision-free periods.
The average time and the distance between collisions (the
mean-free path) depend on the size of the molecules and on
the density of the gas. The larger the molecules, and the higher
their overall density, the more often will collisions take place.
The mean-free path under normal conditions is 0.1 micron for
hydrogen and 0.05 micron for the much larger molecules of
oxygen. 0.1 micron-one ten-thousandth part of a millimeter-is
not a very short distance on the molecular scale: it is in about
the same relation to a molecule as 25 m to a billiard ball.
Knowing the average velocity and the mean-free path we can
easily calculate the time between collisions. For hydrogen, it is
5 x 10-0 seconds.
CHAPTER 2

ORDER

l. The symmetry of wallpaper patterns


While ordered arrangements are encountered fairly frequently
in nature, ideal order is perhaps best discussed on the familiar
example of wallpapers.
Figure 5 shows a large number of identical flowers. This is a
simple pattern, made by repetition of the same drawing. The
smallest unit of this pattern can be found by constructing two
lines starting from any point on a flower to the corresponding
points on the neighboring two flowers, and it is easy to extend
this procedure to form a parallelogram. The entire pattern can
obviously be reproduced by moving this parallelogram along
the two starting lines of the construction by distances equal to
the sides of the parallelogram. Such a parallelogram is called an
elementary or unit cell. Depending on the symmetry of the
pattern, the elementary cell may be a square, a rectangle, and
so on.
It will be evident that this smallest repeating unit can be
chosen in different ways, obtaining different parallelograms each
containing the same parts of the drawing. It is immaterial
whether one and the same flower is wholly or only partly con-
tained within the repeating unit.
The task of the wallpaper designer does not end with the
preparation of the drawing. This would be the case only if the
pattern could be made up in only one way, by repeating the
basic drawing in parallel. However, apart from this simplest
method, there are 16 other ways of covering the wallpaper by a
regularly repeated unit drawing, making a total of 17 types
(plane groups) of the disposition of drawings on a plane
9
10 Order and Disorder in the World of Atoms

Fig. 5. A simple wallpaper pattern illustrating some aspects of the lattice


structure of crystals

(Fig. 6) ". The existence of only 17 plane groups can be demon-


strated by geometrical methods.
The basic unit is here simple, but as in Fig. 5 it itself is devoid
of symmetry, being not unlike a comma. The first illustration of
Fig. 6 shows a set of such commas making up the simplest pat-
tern. The other patterns are all based on the same fundamental
unit, but joined to make symmetrical subpatterns; the differences
between them are determined by differences in the symmetry
of the disposition of the commas.
"This figure is taken from A. V. Shubnikov's "Symmetry," a somewhat more
advanced work.
Order 11

For example, the first three cases of Fig. 6 have no mirror


plane of symmetry-no vertical mirror could be placed so that
one part of the drawing could be a "refleCtion" of the other part.
This is however possible in cases 4 and 5, and in cases 8 and 9

Fig. 6. The seventeen plane groups: the unit cells are indicated
12 Order and Disorder in the World of Atoms

we could have two mutually perpendicular mirrors. Case 10


contains fourth-order (fourfold) axes of symmetry'" perpen-
dicular to the plane of the paper, case 11 third-order (three-
fold) axes, and cases 13 and 15 sixth-order (sixfold) axes.
The planes and axes of symmetry do not occur in these illus-
trations as isolated entities, but run in parallel families. Once we
find a point through which we can pass an axis (or a plane) of
symmetry, we can easily locate the others .
.Let us now choose in these patterns the smallest fragments
( elementary cells) which, when displaced parallel to themselves
by distances equal to the length of their sides, will reproduce the
entire pattern. This will raise several interesting points.
In the first place, this elementary cell may be a parallelogram
(e.g., case 1 in Fig. 6; the smallest fragment isolated on the de-
sign is fairly small; a better idea of the form of the cell may be
obtained by mentally extending the sides of the cell). The
elementary cell may also be a rectangle (cases 3 and 4), a
rhombus with an angle of 60 0 , or a square.
Secondly, the elementary cell may contain various numbers of
basic units (commas): one in case 1, four in case 8, six in case
17, and so on.
In the description of symmetrical patterns it is not always the
best plan to choose the cell as small as possible. The main con-
sideration in choosing the cell is to give the clearest possible
picture of the pattern as a whole-and the smallest, i.e., elemen-
tary, cell may not do this.
Consider case 9, showing a pattern that contains mutually
perpendicular planes of symmetry. The rhombic elementary
cell would not reveal this high symmetry. Therefore, in this and
similar cases, a better cell for the description of the pattern is a
rectangle, containing not four but eight basic units.
It is completely immaterial whether the corners of the ele-
mentary cell are placed at the "heads" or the "tails" of the
commas, or at some other position in the white space between
them. In cases 14 and 15 the chosen cell shows the overall sym-
metry somewhat better than, say, that in case 8. However, this

*This means that the drawing rotating around this axis would come into coinci-
dence with itself four times.
Order 13

does not make much difference, and we can, if we wish, move


around the corners of the cell in case 8, keeping of course the
dimensions of the cell constant and its sides mutually parallel to
themselves.
The ways of filling the elementary cells with the pattern units
are different in all cases, and this is the primary difference
between the 17 cases illustrated in Fig. 6. Thus, having designed
the basic unit, the designer must also say in which of the 17
possible ways his wallpaper is to be printed. For example, in
case 8 the drawing must be positioned in the shaded area (one-
quarter) of the elementary cell, and must then be reflected in
two "mirrors" (Fig. 1).

~f~ f~ fi f~l(
~~~ ~3 ~3 ~ ~ " 3~
~f~ f~ f~ f~f~f
3~ ~t J~ ~~ J~ J~
~f ~f ~f ~f ~f ~f
J~ ~~ J~ ~~ J~ ~~
Mirror planes Mirror planes

Fig. 7. Two different arrangements of the basic drawing, with the same
plane group (case 8 in Fig. 6)

The above 17 types of symmetry do not, of course, exhaust


the variety of patterns based on one and the same unit. The
designer must further indicate how the basic drawing is to be
placed with respect to the cell boundaries. Figure 7 shows two
wallpaper patterns obtained from the same starting unit, the
only difference being that this unit is oriented differently with
respect to the mirror lines. Both these patterns correspond to
case 8 of Fig. 6 . .
We shall not give the rules for constructing the patterns in
the remaining cases.
Let us now see what is the connection between wallpaper
patterns and crystals.
14 Order and Disorder in the World of Atoms

2. Crystals
The first scientific opinions about the nature of crystals appeared
only in the 17th and the 18th centuries. Figure 8, giving an idea
of these early views, is taken from an 18th century book, whose
author (the French mineralogist Pere Haiiy) regarded crystals
as close-packed structures built up of minute blocks. Actually,
this is a very feasible idea. A strong blow will cleave a crystal of
calcite (calcium carbonate) into fragments which may have
different sizes but which have the correct form, similar to the
form of the "parent" crystal. "Surely," argued Pere Haiiy, "fur-

oj

Fig. 8. A crystal made up of unit blocks (a), and an idealized form (b)

ther fragmentation of the crystal will proceed in the same way,


until we arrive at the smallest fragment, invisible to the naked
eye, representing the basic crystal of the given substance."
These basic blocks are so small that the stepped crystal faces
appear smooth to the eye. But what exactly is this basic build-
ing block?
The secret was discovered very much later in investigations on
the diffraction of x rays by crystals. The building block-now
called the unit or elementary cell of the crystal-turned out to
consist of a number of atoms, in the same way as the elementary
cell of a wallpaper pattern consists of a number of basic draw-
ings.
Order 15

The crystal planes are fully analogous to wallpapers; in fact,


it may be said that the crystal represents a three-dimensional
lattice that can be built up by successive parallel displacements
of the three-dimensional unit cell.
How many ways are there of constructing three-dimensional
lattices from elementary units? This complex mathematical
problem was tackled by F. S. Fedorov, the founder of structural
crystallography, who showed that there are 230 ways (space
groups) of accomplishing this task. Fedorov's theory was fully
confirmed experimentally about 20 years after its publication,
and no crystal has as yet been found which does not belong to
one of the 230 space groups.
All our modern ideas about the internal structure of crystals
have been obtained by x-ray structural analysis, based on an
optical phenomenon-diffraction of x rays-discovered in 1912
by the German physicist Max Laue. In this method a narrow
beam of x rays is aimed at a small 0.5-1 mm crystal situated
in front of a sheet of photographic film. The diffraction phe-
nomenon gives rise to the appearance of a series of deflected
rays in addition to the central ray passing straight through the
crystal. When the film is developed, it shows a number of spots
marking the positions of the various rays, and the crystal struc-
ture can be deduced from the positions and the intensities ot
these spots. Needless to say, the interpretation of such patterns
is far from being easy.
The originators of x-ray structure analysis, which has now
been used to study the structure of many thousands of crystals,
were W. H. and W. L. Bragg.

3. Invisible lattices
Some simple crystals consist entirely of atoms of one kind. An
example of such a simple crystal is diamond, a pure form of
carbon. The crystals of common salt are built up of sodium and
chlorine ions (electrically charged atoms). More complex crys-
tals are formed by molecules, themselves made up of various
atoms. However, we can always pick out a certain smallest
repeating group of atoms consisting, in the simplest case, of
a single atom.
16 Order and Disorder in the World of Atoms

Like wallpapers, crystals are based on an elementary cell, i.e.,


a parallelepiped whose successive parallel displacements by dis-
tances equal to the length of its edges will eventually reproduce
the entire crystal. Such displacements are called primitive
translations.
The repeating groups of atoms (or individual atoms) are
disposed within the elementary cell according to one of the
230 space groups.
The comers of the unit cells are called nodes or points. These
points usually coincide with the centers of atoms making up
the crystal. Not all atoms will of course be found at these
points. In very complex crystals the elementary cell is a "tilted"
parallelepiped, in more symmetrical crystals a rectangular paral-
lelepiped, and in the most symmetrical (cubic) crystals a cube.
Considering only the points of the crystal structure and
imaginary lines connecting these junctions, we arrive at a regu-
lar, three-dimensional arrangement forming as it were a "skele-
ton" of the crystal, this skeleton being called the space lattice.

Fig. 9. Model of a crystal lattice. It would be more correct to omit the


connecting wires and to replace the spheres by points

A particular characteristic of the crystal structure is the dis-


position of the "layers" at equal intervals. Figure 9 shows a
model of a crystal lattice constructed from spheres joined by
Order 17

rigid wires. Suppose we shrink to atomic size and take a stroll


along one of these wires. Leaving one sphere and continuing
along the wire, we encounter ever new vistas-at least until
we arrive at the next sphere in line. From now on, the view will
repeat itself exactly. Becoming bored with this, we turn in
another direction, but again, no matter which direction we
choose, the surroundings will repeat themselves with monot-
onous regularity.
The size of the unit cell may vary over a wide range from
crystal to crystal. The smallest distances, of 2-3 angstroms (an
angstrom is lO~8 cm), between adjacent nodes are encountered
in the simplest crystals made up of atoms of only one kind. The
largest spacings, running to hundreds of angstroms, are found
in the complex crystals of proteins, which are already visible
under an electron microscope.
Although crystal lattices exhibit a great variety, certain prop-
erties are common to them all. It is not difficult to see that
their ideally flat faces are planes passing through nodes occupied
by atoms. Such planes can of course have any orientation. Which
of them will then actually bound the growing crystal? Consider
in the first place that not all of these planes would be occupied
equally densely by atoms. Experiment shows that crystals are
bounded by the most densely populated planes.
We shall now clothe the crystal skeletons in flesh and turn
to the packing of particles within the crystals.

4. Crystals and the close packing of spheres


Atoms, the structural material of crystals, have a complex struc-
ture: their positively charged nuclei, themselves made up of
smaller particles, are surrounded by negatively charged electrons
orbiting at various distances. Nevertheless, in certain cases it is
convenient for the study of the atomic arrangements in crystals
to. regard the atoms simply as spheres. This assumption neglects
their complex structure, but it does reflect the very important
point that crystals are built up on the principle of the closest
packing of spheres, as has been shown by x-ray analysis.
To appreciate this principle more fully, let us take some
billiard balls and arrange them in the densest possible packing.
First of all we obtain a dense layer such as that shown in Fig.
18 Order and Disorder in the World of Atoms

10, in which the central atom is in contact with six of its neigh-
bors. Clearly, we cannot do better than that. Now we can put
other layers on top. The arrangement in
which the spheres of the second layer
are directly above those of the first is
evidently not very space filling. The
closest packing is achieved by placing
the spheres of the top layer over the
triangular gaps left in the first. Note
that, if the spheres are all of the same
size, then only every second gap can be
Fig. 10. Billiard balls occupied. The hollows remaining unoc-
"racked" before the
game are closely packed
cupied are shown in black in Fig. 11.
in a single layer There is again only one way to achieve
close packing with two layers of spheres.
We could of course occupy the black gaps and leave the white
ones empty, but this would not affect the final result.
The situation changes when we come to put on the third
layer. To obtain the greatest degree of space filling, we should
put the spheres into the gaps left in the second layer, but now
we find that there are two alternatives: the centers of the
spheres can lie above the centers of the
first layer or over the black gaps. The de-
gree of space filling is the same in both
cases, but the two arrangements are
quite different.
The fourth layer further increases the
possible number of packings to four. In
the same way, there are eight possibil-
Fig. 11. Twolayer pack ities for five-layer arrangements, and so
ing of spheres. The
spheres of the second on. Clearly, the number of different but
layer lie over the white equally space-filling arrangements be-
hollows. The black hoI comes very great as we increase the
lows form small spaces
surrounded by six
number of layers.
spheres We are now in a position to trace the
connection between crystal lattices and
the spacing of spheres. We know that the lattice is based on an
elementary cell which can reproduce the entire crystal by primi-
tive translations. No matter what direction we follow in a crystal,
Order 19
the structural pattern will repeat itself at regular intervals.
Hence the crystal is an arrangement of atoms ( ~pheres) in
which the position of the layers repeats exactly after a certain
number of layers. If this repetition begins from the 14th layer,
the height of the cell consists of 13 layers. The 14th layer is
then directly above the first, the 15th above the second, etc.
The simplest kind of packing is a two-layer one, in which the
third layer repeats the first, the fourth repeats the second, and
so on. This is the so-called hexagonal close packing, shown on
the right in Fig. 12, together with its lattice (skeleton). In this

A
c
fj

A
A

c ~ A

<><:> B

<>0<>
f3

A
A

Fig. 12. The commonest two close-packed arrangements of spheres. The


figure shows both the packings and their "skeletons." Centers of spheres
lying in the same repeating layers are indicated by the same symbols

figure, the atomic centers are marked by circles, crosses, and


squares. The atoms marked by crosses fit into the gaps of both
the upper and the lower layer. This kind of structure is found,
for example, in crystals of magnesium.
20 Order and Disorder in the Wurld of Atoms

A very common arrangement is a three-layer one, in which


the repetition begins from the fourth layer. Figure 12 (bottom
left) shows that in this kind of packing we can choose a cubic
elementary cell. The dense layers are disposed perpendicularly
to the cube diagonal joining the two atoms whose centers are
marked by circles.
Imagine that the cell. is tilted so that this diagonal assumes
a vertical position. The atom marked with a circle will be at
the bottom, and will be the only one belonging to the first layer.
The second layer lies close-packed on the first, and the centers
of six of its atoms appearing in the illustration are marked by
"/
squares. These are situated at the corners and mid-points of the
sides of a triangle. The third dense layer, indicated by crosses,
is also disposed over a tria,ngle. Finally, the fourth layer, which
repeats the first, consists of only a single atom, marked by
a circle.
The unit cell is thus a cube with atoms occupying the corners
and centers of the faces. This structure, called face-centered
cubic, is found in a number of metals, for example, aluminum,
copper, nickel, and at high temperatures, iron.
A real crystal is therefore a system of densely packed particles
whose arrangement is repeated periodically in space. The
junctions and the lines connecting them are imaginary, serving
only to aid us in visualizing the geometry of the crystal and
selecting the unit cell.
As regards the actual size of atoms, accurate x-ray measure-
ments have shown that the atomic radii of various elements lie
within the relatively narrow range between 0.5 and 2 angstroms.
So far we have considered the packing of identical spheres,
i.e., crystals made up of only one kind of atoms. Let us now turn
to spheres of different sizes, i.e., to crystals containing more than
one kind of atoms. Here again experiments show that the crys-
tals can often be represented as close-packed assemblies of
spheres. The crystals are often made up of ions, i.e., electrically
charged atoms.
The next problem is therefore how to pack most efficiently
equal numbers of large and small spheres. We recall that in
the packing of spheres of uniform size, some spaces are left
empty. It can be calculated that these empty spaces amount to
Order 21

about 7~ of the total volume. The spaces are of two types: the
first are surrounded by four spheres disposed at the corners of
a regular tetrahedron; and the second by six spheres lying at
the corners of an octahedron (Fig. 13). The tetrahedrally sur-
rounded spaces are smaller, and there are twice as many
of them.

Fig. 13. Structure of crystals made up of atoms having different sizes-


a) The small atom is situated in the smaller (tetrahedral) space. For the
sake of clarity, the surrounding spheres are shown only as sectors-
b) The small atom is situated in the larger (octahedra l) space- c)
Method of cutting out sectors from six neighboring spheres (the front
sphere is removed)

It is now clear that, in the case of spheres of two different


sizes, all we have to do is to pack the larger spheres as closely
as possible and put the smaller ones into the gaps. Not all the
gaps need be occupied.
It can be calculated that any close packing of identical
spheres contains one large and two small gaps for every sphere.
Very small spheres will fit into these spaces-if they are too
large they will push the larger spheres apart and destroy the
22 Order and Disorder in the World of Atoms

close packing. This concept, too, can be extended to many crystal


structures.
The crystals of common salt (sodium chloride ) form a close-
packed three-layer structure of larger chlorine ions (white
spheres in Fig. 14a ) with the smaller
sodium ions (black ) filling all larger
gaps. Each sodium ion is thus
rounded by six chlorine ions.

oj bJ
Fig. 14. Packing of ions in common salt (a) and in cadmium chloride (b).
The chlorine ions are larger than the metal ions

The form of iron sulfide known as magnetic pyrite Qr pyrrho-


tite consists of a two-layer packing of large sulfur ions with the
smaller iron ions occupying all the larger spaces. In lithium
oxide, containing two lithium atoms for every oxygen, the close
packing consists of the large oxygen ions; the tiny lithium atoms
fill all the smaller spaces, so that each lithium is surrounded by
four oxygens. In cadmium chloride (two atoms of chlorine for
every atom of cadmium), the close packing is built up of chlo-
rine ions (white spheres in Fig. 14b). The smaller cadmium ions
occupy the large spaces in every third layer of chlorine ions (Fig.
14b). For the sake of clarity, two layers of large spheres not
containing small spheres have been left out from this figure;
the elementary cell is also shown.
The above are the simplest examples of the space filling in
close-packed structures. Very detailed and valuable work in
this field by Linus Pauling, W. L. Bragg, and N. V. Belov dis-
Order 23

closed a large number of regularities which explain the mechani-


cal, optical, and electrical properties of many minerals.
It should be stressed again that the representation of atoms
as rigid spheres may be convenient, but it does not indicate the
complex nature of the atoms. Our simplified model is obtained
by imagining a spherical shell around each atom, bounding the
electron orbits with the nucleus at the center.
To complete the picture of crystal structure, we should now
indicate some features of the behavior ~f atoms within crystals.
The point is that when atoms are combined in a crystal part of
their electrons is pooled, and belongs more to the crystal as a
whole than to their individual atoms. These so-called free elec-
trons are no longer confined to the field of a single atom but
migrate over the whole crystal attaching themselves temporarily
to positive centers and leaving them again. Unlike the atoms of
liquids and gases, which move chaotically, the ions making up
the lattice oscillate only slightly about their equilibrium posi-
tions. On the other hand, the free electrons in a crystal behave
under certain conditions very much like the atoms in a gas,
and are in fact sometimes known collectively as electron gas.
In most cases only a very small proportion of the electrons in
a crystal is made up of free electrons. This is the situation in
crystals composed of oppositely charged ions, such as sodium
chloride. Here practically all the ten electrons of the positive
sodium ion, and all the 18 electrons of the negative chlorine ion
move around their own nuclei. (It is necessary to distinguish
between free electrons and electrons which pass from sodium
atoms to chlorine atoms--one per atom-to form the corre-
sponding ions. The latter electrons are firmly held by the
chlorines.) .
The picture changes when we come to consider metal crystals.
Here the atoms can yield, but not take up, excess electrons. As
a result of this, all the ions are positive and an appreciable part
of the electrons is shared. Each atom gives up 1-3 of its outer,
less strongly bound, electrons.
In monovalent metals such as lithium or sodium one electron
is bound to the nucleus much more weakly than the others, and
is given up and pooled in the crystals of these metals. In divalent
metals such as calcium or barium, the number of the more
24 Order and Disorder in the World of Atoms

weakly bound electrons is two, and in the corresponding crystals


each atom gives up on average two electrons.
Atoms assemble into crystals because they are attracted to one
another. Attraction occurs between oppositely charged ions, for
example, between negative chlorine ions and positive sodium
ions. Each chlorine ion attracts six sodium ions and is therefore
surrounded on all sides by oppositely charged particles. In tum,
each sodium ion is surrounded by six chlorines. This gives rise
to the observed close packing in the resulting crystal.
The attraction is opposed by forces or repulsion, which begin
to operate as the atoms closely approach one another. This
repulsion is due to the interaction between electrons of the con-
verging atoms. Thus, the distance between ions in a crystal
represents a compromise between attraction and repulsion
forces.
But how is it that, though all the ions are positive in a metal
crystal, they are still strongly bound into a closely packed
assembly? When the atoms come close enough to one another,
they begin to share part of their electrons, and these shared
electrons "cement" the ions into a strong lattice. Further ap-
proach of the atoms is prevented by the electrostatic repulsion
between like charges.

5. Crystals that are not close-packed assemblies of spheres


As has just been shown, many crystal structures can be repre-
sented as close-packed assemblies of spheres, in which each
atom or ion is attracted equally from all sides by neighboring
particles. It may be said that each atom or ion behaves as a
sphere uniformly covered with glue. On the other hand, there
are ju~t as many cases in which this simple model is inapplicable,
and in which the atoms behave not as spheres but as more
complicated units. Moreover, they may be covered nonuniformly
with the hypothetical glue. Suppose, for instance, that the glue
has only been applied to eight spots. Each sphere will then
be surrounded by only eight neighbors, and not by twelve as in
the most closely packed structures.
These deviations from the rule of closest packing should not
come as a surprise. In fact, the surprising thing is that so many
structures do conform to this simple rule.
Order 25

Atoms are complex electrical systems, and the strength of the


attraction exerted by them may be different in different direc-
tions. As a result of this the surrounding electrons may be dis-
torted into nonspherical envelopes. All this leads to the fact
that, apart from the closest packing, we also find other struc-
tures in the world of atoms.
An example of a structure based on
a sphere with eight spots of glue is
shown in Fig. 15. This is the structure
found, for example, in iron. The iron
lattice is cubic, with the atoms situated
at the corners and in the centers of
cubes. The same structure is observed
in lithium, potassium, cesium, and many
other substances. Fig. 15. Body-centered
cubic structure, encoun-
Mercury atoms behave as flattened tered in a number of
spheres. In Fig. 16 the structure of metals
crystalline mercury is compared with
the structure of crystalline copper, an ideal close-packed struc-
ture of the type ... ABCABC ... The character of the disposi-

Cu Hg
A
A

c
c
B

A A

Fig. 16. Copper (Cu) exhibits an ideal close-packed struc-


ture _ The unit cell is a cube whose corners and face centers
are occupied by atoms. The structure of mercury (Hg) is
similar, but the distances between the atoms within a layer
and in different layers are not the same

tion of the atomic centers is clearly the same, but in the mercury
crystal the distances between the layers are smaller and the
distances between the atoms of one and the same layer are
larger.
26 Order and Disorder in the World of Atoms

The "glue" on uranium atoms is clearly distributed non-


uniformly. Figure 17 shows two layers of the close-packed struc-
ture ... ABAB ... on the left, and two layers of uranium atoms
on the right. The atoms of the upper layer no longer fall into
the spaces in the lower layer. It is as if the atoms in the lower

Fig. 17. Structure of uranium (right) compared with the hexagonal close
packing

layer were "sticky" on the left, and those in the upper layer on
the right. Thus the structure of uranium cannot be represented
as a closest packing of spheres.
Many examples of such "faulty" close packing could be given,
ranging from minor discrepancies to cases in which the resem-
blance to the packing of spheres is completely lost.

0)
b)
Fig. 18. a) The structure of diamond (also germanium and white tin)
b) The structure of graphite
Order 27

Figure 18a shows the structure of diamond. Here every carbon


atom has only four nearest neighbors, and it is clear that in this
case we could no longer represent the atoms by spheres. If we
did draw a sphere around each atom in such a way that the
spheres were touching, then 75% of the space occupied by the
crystal would remain empty. Experimentally, it is known that
the carbon atoms in diamond are bound exceptionally strongly
(giving rise to its great hardness), and their electron shells
overlap. The attraction forces are directed toward the corners
of a regular tetrahedron, making angles of 109 30' with one
another. The diamond structure is also found in silicon, gernia-
nium, and white tin.
Spheres with four "sticky" spots
can be used to construct a model
of the structure of ice. Each pair of
oxygens is joined here through a
hydrogen atom. In these four
bonds each hydrogen is joined to
two atoms of oxygen. There is of
course no contradiction between
the chemical formula of water and
the structure illustrated in Fig. 19,
in which the hydrogen atoms are
shown in black. The structure of
ice is very loose, as can be seen Fig. 19. The crystal structure
from the large empty spaces in the of ice
diagram. If we extend the structure
above. the plane of the paper, these spaces will become broad
tunnels penetrating the lattice.
Apart from diamond, Fig. 18 shows the structure of graphite.
The carbon atoms in graphite are arranged in layers, but these
are not the layers of a close-packed structure. We could not
construct the layers of graphite from contacting spheres. In the
first place, such an attempt would result in a packing with holes,
and, secondly, the neighboring layers would hang in the air
without touching each other. As in diamond, the carbon atoms
in graphite are held by very strong bonds. However, in this
case the bonds are separated by angles of 120. The atoms in
adjacent layers are 2.5 times as far apart as atoms in the same
28 Order and Disorder in the World of Atoms

layer. Obviously, the layers in graphite are bound much more


weakly than atoms within the layers themselves.
In graphite the layers are flat. Arsenic and phosphorus also
give layer structures of this kind, but in these the atoms in each
layer are not arranged in the same plane. In arsenic (Fig. 20)

~
, ' :;',l' : {/:
: "..
/...
... ...

.... .. I /'

'~

Fig. 20. The atoms of arsenic form layers which


are shown isolated on the right of the figure

each atom is again bound strongly with three neighbors, but in


contrast to graphite each atom is at the apex of a pyramid.
Gray selenium is an example of a structure composed of
chains of strongly bound atoms. Here each atom is strongly
joined to only two neighbors, forming an infinite spiral wound
around a straight axis. The distances between neighboring
spirals are much larger than the distances betwee;n nearest
atoms of one and the same spiral.
Now consider Fig. 21, showing the
structure of crystalline iodine; anal-
ogous structures are found in crys-
talline bromine and chlorine. This
time the atoms are represented as
intersecting spheres to underline the
fact that each iodine atom has only
one nearest neighbor. Each such pair
Fig. 21. An example of a of atoms constitutes an iodine mole-
simple molecular structure,
found in iodine, bromine,
cule. The bonds holding atoms in
and chlorine molecules are considerably stronger
than intermolecular forces. The dis-
tances of the closest approach of different molecules are dictated
by the radii of the spheres. The flattening shown in the figure
is an expression of the strength of the intramolecular forces.
Order 29

Despite all this variety, so far we have discussed only a few


examples of the structures of simple substances. These examples
have included crystals made up of the closest packing of spheres,
in which every atom is bound identically with all 12 nearest
neighbors. In ideal and nonideal close-packed assemblies of
spheres there are no groups of strongly bound atoms. At the
same time, strongly bound layers could be distinguished in
graphite, and chains in crystals of gray selenium. Finally, some
crystals are built up from molecules and from groups of strongly
bound atoms.
The same variety of structures is found in substances com-
posed of various kinds of atoms. We shall not deal with the
crystals of chemical compounds composed of layers and chains
of strongly bound atoms, since these are fairly rare.
Crystals exhibiting close packing of atoms of one kind fitting
into the interstices left by atoms of the other kind are not com-
posed of molecules. Let us return to the now familiar example
of sodium chloride in Fig. 14a, made up of alternating ions of
sodium and chlorine. Each sodium ion has six identically dis-
posed neighboring chlorine ions, so that we cannot say that a
given sodium ion is bound preferen-
tially to anyone of them. The crystal
does not contain any molecules of
common salt, consisting of one so-
dium and one chlorine atom.
This situation is by no means the
rule. Consider the crystals of carbon
dioxide (dry ice), which exist only
at low temperatures and are made
by intense cooling of the gas at a
reduced pressure. Here we find that
the crystals are built up from mole-
cules (one such molecule is shown in
Fig. 2b). Fig. 22. Close packing of
the molecules of carbon
This structure is further clarified
dioxide. Carbons are shown
by Fig. 22. Each carbon has only in black
two nearest neighbors-two oxygen
atoms. Each oxygen has only one carbon as the neares tneighbor.
Closely bound three-atom groups-molecules of carbon dioxide
-stand out clearly in the crystal.
30 Order and Disorder in the World of Atoms

The flattening of the spheres in Fig. 2b indicates that the at-


traction between atoms in each carbon dioxide molecule is far
stronger than the attraction between individual molecules. As a
result of chemical forces, the electron shells of the two carbons
and one oxygen in the molecule overlap and interpenetrate to a
large extent. The same flattening was used in Fig. 21 to represent
the molecular nature of iodine crystals.
Countless investigations have now established that the model
of crystals built up of spherical atoms or ions is valid for metals,
alloys, and many inorganic salts. Molecular crystals are en-
countered in organic chemistry and in some inorganic com-
pounds, such as mercuric chloride. How are these, often oddly
shaped, molecules arranged in the crystals?

Fig. 23. Principle of the structure of organic


crystals. Independently of how they are arranged,
the molecules tend to occupy the smallest possible
volume.

Just as forces of attraction acting between the ions give rise


to close-packed assemblies in ionic crystals, the mutual attraction
between molecules results in the formation of closely packed
molecular arrangements. The general principle is simple: the
protrusions of one molecule fit into the recesses of another
(Fig. 23).
Order 31

This principle leads to the conclusion that there can be no


structures in which the planes of symmetry pass between mole-
cules, since the protrusions of one molecule would then en-
counter the protrusions of another. Only certain forms of sym-
metry (8-10 space groups) are therefore possible for molecular
crystals. As has been shown by the present author, the principle
of close packing allows a prediction of the nature of the mutual
disposition of molecules, the symmetry of the crystal, and some
other properties.

6. Same atoms but different crystals


The soft matt black graphite and the brilliant, transparent, ex-
tremely hard diamond are made of the same type of atoms. Both
are forms of carbon. The reason for their complete lack of
similarity is that their structures are completely different.
Figure 18 shows the two lattices in question, with every atom
having three nearest neighbors in graphite and. four nearest
neighbors in diamond. It also demonstrates very clearly just
how strongly the atomic arrangement can influence the proper-
ties of the resulting crystals. Consider some of the differences.
Graphite is made into refractory crucibles capable of withstand-
ing temperatures of 2000-3000C, while diamond bums above
700C; the specific gravity of diamond is 3.5, that of graphite
only 2.1; graphite is a conductor, while diamond is not.
Not only carbon can form different crystals. Practically every
substance existing in the crystalline state exhibits several modi-
fications-six in the case of ice, nine in the case of sulfur, four
in the case of iron, and so on.
At room temperature iron exists in the body-centered cubic
structure illustrated in Fig. 15. Each atom is then surrounded by
eight neighbors. A closer packing occurs at higher temperatures,
with every atom having 12 nearest neighbors. The former modi-
fication of iron is considerably softer than the latter. The hard
modification can, however, exist at room temperature; it is
brought about by heating the iron to a high temperature and
rapid cooling. Thus the existence of several modifications of the
same substance can be extremely important from the technologi-
cal point of view.
32 Order and Disorder in the World of Atoms

The above two examples show clearly that different modifica-


tions of one and the same substance may have completely
different structures. Countless other examples could be given
to illustrate this point.
Thus, yellow sulfur for.ms in the crystals puckered eight-
membered rings, i.e., molecules made up of eight atoms. The
same rings are found in red sulfur, but they are disposed differ-
ently with respect to one another.
Yellow phosphorus exhibits the cubic body-centered structure,
while black phosphorus is modeled on graphite.
Gray tin has the same structure as diamond, while the struc-
ture of white tin can be mentally derived by strongly compress-
ing the diamond structure along the axis of the cube. As a result
of this, the number of nearest neighbors becomes six instead of
four.
The occurrence of more than one crystal modification is also
often encountered in organic compounds. In this case the modi-
fications differ in the arrangement of their constituent molecules.

7. long-range order
This term can be used to describe the arrangement of atoms in
crystals; thus a crystallite 1 mm across may easily contain the
same atomic pattern repeated regularly millions of times. Most
solid bodies are basically crystalline, consisting of assemblies of
crystal grains measuring perhaps 0.001 mm. The long-range
order in these grains is clearly expressed by the periodic repeti-
tion of analogous atoms at equal distances along the cell axes.
In most cases the periodicity of long-range order is equal to
2-4 layers of atoms or molecules, but in some very interesting
substances the structure begins to repeat only after dozens or
even hundreds of layers. The second situation is exemplified by
silicon carbode, which occurs in nature in several modifications.
In a rare form (to which we shall return later), structural repeti-
tion begins only after 243 layers of atoms.
The width of the repeat pattern is very small, being 3-7
angstroms in metals and not more than 20-30 in most crystals.
Large periods approaching 1000 angstroms (0.1 micron) are
encountered in proteins. Such spacings have been found in the
Order 33

Fig. 24. Print of an electron microscope photograph of bovine tendon.


The periodicity in the arrangement of the large protein molecules appears
as the regular alternation of dark and light bands.

molecular structure of bovine tendons ( Fig. 24), porcupine


quills, sea gull feathers, and mammoth tusk. This periodicity
undoubtedly plays an important part in the life processes of
animal tissues. We may expect that the biological properties of
tissues will eventually be correlated with their structure.

8. Order in microcrystalline bodies


As was already mentioned in the preceding section, the over-
whelming majority of solids have a crystalline structure, being
composed of crystallites (grains) generally visible under the
microscope. Thus, intergrown crystals can be found in individual
grains of sand and even in clay. The crystals in clay are rather
remarkable, since they allow water to penetrate between the
atomic layers-this is the explanation of the familiar swelling of
clays. In fact, we can say that solid noncrystalline substances
are the exception rather than the rule.
The properties of solid substances are determined by the size,
structure, and arrangement of the crystallites, as well as by the
forces binding these into a coherent mass.
Any treatment carried out on a metal will affect its grains.
34 Order and Disorder in the World of Atoms

Consider a specimen of cast metal. Its grains are fairly large


and randomly oriented. How will these grains behave when the
metal is drawn out into a wire? Investigations show that drawing
-or any other form of mechanical working-leads to the ap-
pearance of a certain degree of order in the arrangement of the
grains, and to a decrease in the size of the grains. This statement
may seem rather puzzling: what kind of order can we have when
the grains have a completely irregular shape? The answer is
that despite the irregularity of their outer boundaries, the grains
are still crystals; the ions in their lattice are packed just as
regularly as in well-formed crystals, and we can tell how the
unit cell is oriented in each individual fragment. Before the
mechanical treatment, the cells are strictly ordered only within
the grains. After the treatment, the cells of different grains tend
to assume a certain common orientation. This is known as the
appearance of texture.
In Fig. 25 texture is shown on the
example of the ordering of some
planes, say, the planes populated most
densely by ions, marked by series of
dots. The phenomenon of texture was
first observed by N. E. Uspenskii and
S. T. Konobeevskii.
The actual texture produced depends
on the type of working (rolling, forg-
ing, drawing, etc.) to which the metal
is subjected; the grains may orient
themselves so that the diagonals of
their unit cells lie either along or
oj across the direction of working. The
bJ
extent of texturing depends of course
on the degree of working. The pres-
Fig. 25. Diagrammatic rep- ence of texture strongly affects the
resentation of a metal speci-
men, a) with and b) without
mechanical properties of the finished
texture article. Prolonged studies of the grain
size and distribution in metallic ar-
ticles have now thrown some light on the treatment of metals
and have indicated how such treatment should be carried out.
Another very important technological process connected with
Order 35

the reorganization of crystal grains is annealing. If rolled or


drawn metal is heated at a sufficiently high temperature, new
crystals begin to grow in place of old. This secondary crystalliza-
tion has a random orientation and gradually obliterates the
texture. As the temperature (or time) of annealing is increased,
the old grains are progressively replaced by new ones. In fact,
the new grains may be grown to a size visible to the naked eye.
As a result of annealing, the metal becomes less hard and more
plastic because of grain growth and disappearance of the texture.
CHAPTER 3

ELEMENTS OF ORDER IN DISORDER

1. Short-range order and the structure of liquids


As was shown in the previous chapter, many solids are composed
of pseudo spherical particles forming close-packed assemblies.
The free space in such an arrangement amounts to about 26% of
the total volume. Let us now see what happens when we melt
such a solid, e.g., copper. Experiment shows that its volume
increases by 3%. More specifically, it is the free space that in-
creases from 26 to 29%. The atoms will now have some room to
budge from their proper places, and so the ideal order of the
parent crystal is destroyed.
The change that accompanies the melting of copper can be
illustrated with the aid of the following model: A rack such as
that used in pool is first packed with many billiard balls as
tightly as possible, and then its area is increased so much that
the density of the packing (i.e., the number of spheres per unit
surface) decreases to the same extent as in the melting of
crystalline copper. When the rack is now shaken, the balls begin
to oscillate about their equilibrium position, and when a ball-size
empty space is occasionally formed somewhere, adjacent balls
can change place.
Can the resulting arrangement be regarded as complete dis-
order? It certainly shows some features of disorder: however a
strip of paper is placed on this set, the number of balls under it
will be the same. But do the various distances between them
occur with equal frequency? In the case of liquids, the answer is
no, not quite.
No sphere can ever come so close to another that the distance
between their centers is smaller than the sum of two sphere
radii. This deviation from ideal disorder is also found in gases,
but there it makes little difference since the nearest gas mole-
36
Elements of Order in Disorder 37

cules are separated by a gap of some ten molecular diameters.


So, despite the finite size of atoms and molecules, we may speak
of complete disorder in gases.
To show that there is a fundamental qualitative difference
between the arrangement of particles in gases and in liquids,
one need only count the particles in ever-increasing volumes
drawn around a given particle. Although this can be done in
space, we shall again use a two-dimensional model. We select a
molecule and draw round it concentric circles with radii equal
to two, three, four, and more radii of the circle representing the
central molecule. Now, how m~my nearest neighbors are there?
By nearest neighbors we mean those molecules whose centers
lie on the innermost circle. Combining this model with the actual
melting of copper, we can readily answer this question, since we
know the volume of the innermost field and the packing density.
The volume available to each atom in molten copper is only 3%
greater than the volume of the atom itself. Calculations show
that the innermost region can accommodate on average 11.6
atoms. Each copper atom thus has only about twelve nearest
neighbors, whose centers are one atomic diameter away.
The situation is completely different' when we pass on to
gases: the innermost region here contains an average of not 11.6
but only 0.026 particles, i.e., interatomic (or intermolecular)
contact will be found only in 26 cases out of 1000. Owing to the
finite size of these particles, the latter cannot approach one
another more closely. While a change from 0.026 closest neigh-
bors to zero is small and unobservable, the change from 12 to
zero in the case of liquids is sharp and can be detected experi-
mentally.
This does not exhaust the differences between gases and
liquids; another fundamental difference associated with the
persistence of some order in liquids lies in the fact that some
interatomic distances in liquids will be found more often than
others. The interatomic distances that acquire this predomi-
nance are those which best approximate the close-packed ar-
rangement. Figure 26 shows that in ideal two-dimensional order
the central particle singled out has six closest neighbors whose
centers lie on the circle drawn at a distance of one radius from
the boundary of the central particle, i.e., the six closest neighbors
38 Order and Disorder in the World of Atoms

are in contact with the central atom. These neighbors are marked
by 1 in the figure. As we proceed outward, these are followed by
particles in the second region (circles marked by 2), third region
(marked by 3), and the fourth region (marked by 4). Regions 2,
3, and 4 contain, respectively, six, six, and twelve spheres.

0)

b)
Fig. 26. a) Ideal close packing. The short-range order
is perfect, being characterized by the fact that the cen-
ters of spheres lie on definite boundaries b) The
type of packing found in liquids (the structural loosen-
ing is exaggerated for the sake of clarity)

As mentioned above, the space placed at the disposal of each


sphere in a liquid is somewhat larger. This leads to an increase
in the total volume of the assembly. If ideal order were neverthe-
less maintained, the number of closest neighbors would remain
the same, and the interatomic distance would only increase by
about 1%. Needless to say, this is a very small change.
Elements of Order in Disorder 39

The model of packing in a liquid (Fig. 26b) shows that the


central sphere is still surrounded by six closest neighbors whose
centers are near the innermost boundary. However, as we pro-
ceed outward disorder becomes more and more obvious: the
short-range order that prevails around our reference point gets
blurred at the edges, and gradually changes into disorder (this
is why it is called short-range order). We could choose any
atom as our central reference point and we would still arrive at
the same result, because the short-range order is the same
'around every atom in a liquid. Furthermore, this is the only
kind of order that solids and liquids have in common.
We have seen in the last chapter that atoms in crystals often
do not behave as spheres. This also applies to liquids. Thus, in
the case of ideal short-range order in atomic liquids, each atom
should have about twelve nearest neighbors. How does this
agree with experiment? It has been found that when metals with
close-packed structures are melted, the retained short-range
order is in fact characterized by an average of about twelve
closest neighbors around each atom. When each atom in the
crystal has only eight closest neighbors (e.g., lithium, sodium,
and potassium), this state of affairs will also persist in the
molten state, but the average number of closest atoms is now
somewhat more than eight. In the case of simple substances in
the crystals of which the atoms are firmly bound to few neigh-
bors, these bonds are destroyed on melting and the atoms ac-
quire more neighbors in the liquid state.

2. The amorphous solid state


The word "amorphous" means "devoid of form." Amorphous
substances are thus the opposite of crystals, which possess a
regular polyhedral shape. On the other hand, a piece of iron can
have a grossly irregular shape, and yet it is not amorphous. This
is because crystallinity is not necessarily reflected in the outward
appearance. It is quite sufficient if the innumerable small grains
that make up the substance are crystalline. But if the outward
appearance is not always a good indication of the internal struc-
ture, what are the reliable signs showing that we are dealing
with a crystalline or microcrystalline material? Well, first of all,
40 Order and Disorder in the World of Atoms

a sharp melting point. When heat is supplied to such a material,


the latter absorbs it up to a certain point and then suddenly
melts without further increasing its temperature. The tempera-
ture corresponding to this event is called the melting point. It
is strictly constant, and is characteristic of the substance in
question. In the case of amorphous materials, on the other hand,
there is no such sharp melting point. Thus, as we heat ordinary
glass-a well-known amorphous substance-it softens and grad-
ually liquefies, giving first a viscous and then, at higher tempera-
tures, a free-flowing liquid that can be poured from one crucible
into another as easily as water. This behavior is due to the struc-
tural properties of amorphous materials; which are such as to
make us regard these substances to be more like liquids than
bodies-just like liquids-have only short-range order.

oj bJ
Fig. 27. Structure of quartz (the silicon atoms are shown in black, oxygen
in white). This twodimensional representation naturally simplifies the
actual state of affairs a) Crystalline quartz b) Amorphous quartz:
basically similar to a), but imperfect

The structures shown in Fig. 27 belong to the same chemical


substance, quartz or silicon dioxide, in two forms: crystalline and
amorphous. The difference between them is clearly visible; the
glass can be regarded as a crystal gone wrong. Nevertheless, the
number of nearest neighbors is the same in both cases.
Long-range order is closely connected with a sharp melting
Elements of Order in Disorder 41
point, and the lack of a sharp melting point in amorphous sub-
stances is a reflection of the absence of long-range order. A
sharp melting point marks the temperature at which long-range
order is destroyed, leaving behind only short-range order. No
such drastic change occurs when amorphous materials are
heated. In this case, the atomic arrangement remains the same,
only the atomic mobility is increased. As the temperature is in-
creased, more and more atoms slip away from their group,
seeking new company. Beyond a certain point, this fickle be-
havior becomes a habit with all the atoms, the number of posi-
tional changes per second becomes the same as in water, and
the substance becomes as fluid as water.
The amorphous state is freakish, and we shall see later that
crystallinity is the natural state at low temperatures. However,
the formation of long-range order is not always easy. Thus, the
molecules may have an awkward shape highly unsuitable for
an arrangement at once regular and tightly packed. For example,
if we take tadpole-shaped molecules, we can never marshal
them into regular close packing. We must forego either the
long-range order or the close packing. Thus, if we crumple the
long tails of the molecule, we lose the long-range order, but we
do achieve close packing. And certain substances do make this
sacrifice. More specifically, no crystals are formed by molecules
that are so shaped that their arrangement in long-range order is
more space wasting than their close-packed arrangement without
long-range order (the present author has found with organic
substances that the empty space in the crystals may not exceed
40%). Instead of forming crystals, these substances slowly thicken
and rigidify into the amorphous state.
Comparing substances built up of atoms, ions, ion radicals,
and molecules, We notice that those composed of molecules are
characterized by a high degree of order in a certain sense. This
is obviously because the molecules themselves represent ordered
arrangements of atoms; not only is the geometric disposition of
atoms constant in molecules of the same type, but the bond
lengths are mostly invariant as well. The constancy of these
bond lengths or intramolecular distances thus also confers a
higher degree of order on molecular liquids and amorphous sub-
stances. If, furthermore, the molecule itself exhibits repeating
42 Order and Disorder in the World of Atoms

features, the short-range order is even more conspicuous. This


is the case with amorphous carbon, an appreciable proportion of
which forms dense and periodic atomic layers. The larger and
the more numerous these are in stacks, the higher is the degree
of order in amorphous carbon (see Fig. 28).

Fig. 28. Graphite networks linked together by


chains of carbons in housecoal. Sometimes two
or three such layers are tightly pressed together

The amorphous state is exhibited by many organic substances,


such as plastics and organic glasses, composed of giant molecules
each consisting of several thousand atoms. These deserve special
attention and will be discussed later.

3. Liquid crystals
Isn't this a contradiction in terms? It means "liquid solids," and
is this any less ridiculous than "silent noise" or "bitter sugar?"
At first sight, maybe not. However, liquid crystals do exist.
Without going into their rather complicated chemical designa-
tions, we shall merely mention that one encounters them quite
often, particularly among organic compounds and biochemical
substances (e.g., viruses and lipoids in living tissues).
The molecules in liquid crystals are always elongated in
shape. Moreover, liquid crystals exist only within a certain
temperature range: When they are heated beyond the upper
limit of this range they become ordinary liquids, whilst cooling
Elements of Order in Disorder 43

below the lower limit converts them into ordinary crystals. Their
name has arisen from the fact that they combine oddly the prop-
erties of liquids and crystals. Thus, they How and form droplets,
but instead of being spherical, these droplets are sometimes
elongated. Each droplet is something like a piece of jelly. De-
tailed investigations have shown that the molecular arrangement
in these possesses an order unknown in the realm of ordinary
liquids.
Two types of liquid crystals are known: in one case the short-
range order is complemented by parallel alignment of all the
molecules, whilst in the other case the molecules first form layers
characterized by parallel molecular alignment and short-range
order (Fig. 29).

aaaa~aa8G8a
aQ~aaaa8~a8Q
QUQaaa~a~aaaa
~a a~ a~ ~ ~ Q~ qaaa
Fig. 29. The two types of molecular arrangement found in liquid crystals

As regards the degree of order, these arrangements are clearly


intermediate between liquids and solids. Thus, one may well find
for them a more felicitous name than "liquid crystals," but cer-
tainly not a more descriptive one.
Soap dissolved in water gives rise to liquid crystals, and this
property is closely connected with its cleaning action. The soap
molecule is shaped like a rod measuring 30-40 by four ang-
stroms. One end of the rod has a build-up of negative charge and
this polar "head" gravitates toward water molecules. As regards
the liquid crystal structure, we find in soap solution that the
soap molecules are organized in a tail-to-tail manner into a
double layer, forming parallel strata interspersed by layers of
water (Fig. 30). The polar ends of the soap molecules always
44 Order and Disorder in the World of Atoms

point outward and are in contact with water molecules. More-


over, the rod-shaped molecules are aligned densely and with
short-range order. Certain features of this sandwich structure
depend on the concentration of the soap in water. When this is
low, the soap layers are separated by thick water layers; when
it is high, the soap layers are more numerous and the water
layers are thinner. In saturated soap solution the thickness of the
water layers is about 20 angstroms.

I rI rI
1l l l l
}
II III ,}
l l l II
------

Fig. 30. Arrangement of soap molecules in


water. The tadpole-shaped molecules align
themselves in parallel, vertically to the sur-
face of water. The resulting rows are then
paired tail-to-tail, with the polar end of the
molecules pointing outward. Layers of water
are sandwiched in between such double
layers of soap molecules

The double layers of soap molecules forming a liquid crystal


are highly mobile: they can slide easily to and fro. The cleaning
action of soap lies in this mobility, because the polar ends of the
molecules pick up dirt particles and hand them over to water.

4. Thermal motion of particles in liquids


We have already mentioned in passing that the molecules in
liquids execute a basically oscillatory movement. This also ap-
plies to crystals, but here the particles vibrate about a rigidly
Elements of Order in Disorder 45

fixed equilibrium position. In liquids there is no such rigid


framework. The movement is rather reminiscent of a sort of
stationary marching known as marking time, which the mole-
cules of liquids perform in that small extra space alloted to
them; they do not leave their environment of nearest neighbors.
Sometimes a molecule may admittedly change place with a
closest neighbor, but thereafter the stationary march continues.
The ease with which a molecule may change neighbors is di-
rectly connected with an important property of liquids known
as viscosity; the rarer this event, the thicker (more viscous) is
the liquid.
It is easy to see that heating will increase the amplitude of the
molecular vibrations, and therefore decrease the viscosity. Vis-
cosity is further affected by the shape of the molecules: mole-
cules having a complex shape go hand in hand with high-
viscosity. This is the reason behind the high viscosity of glue,
honey, and oils. Many substances solidify before they can be-
come very viscous.
This brings us to the concept of plastic deformation. Plastic
deformation is brought about in flowing substances by the posi-
tional exchange of molecules. It is quick and conspicuous when
external forces are used (e.g., in stirring a jar of honey), and slow
and less noticeable when it is brought about only by the thermal
motion of the molecules. This exchange virtually comes to a '
standstill when the liquid rigidifies. Thus, in the ordinary state,
glass breaks under pressure; it exhibits plastic deformation only
when it is heated. This indicates that the thermal motion of part-
icles in solid glass is insufficient to effect positional exchange.
(More accurately, such an exchange does occur, but very infre-
quently. In fact, it even occurs in crystals, mainly as a result of
imperfections. )
We can now understand why it is so difficult to decide
whether glass should be grouped with solids or liquids: it
exhibits some characteristics of both states. As in all amorphous
substances, the structure of glass is like that of liquids in that it
is characterized only by short-range order. As regards the ther-
mal motion of the molecules, glass (and all other amorphous
materials) is like crystals, since the thermal motion in both is
insufficient to cause positional exchange of molecules.
CHAPTER 4

ELEMENTS OF DISORDER IN ORDER

1. Thermal motion of atoms and molecules in crystals


We must now confess that we have overglamorized reality by
describing crystals as a mathematician's dream frozen into
space: a geometrically perfect arrangement of regular rows of
atoms extending ad infinitum in severe order at equal distances
and in well-defined directions. In reality, crystals are less perfect
and more interesting than three-dimensional wallpapers. First of
all, they are not stationary. The atoms or molecules execute
various complicated movements about their equilibrium posi-
tions called lattice points, and if instantaneous photographs
could be taken of the particles in a crystal, they would all turn
out to be different because of the thermal movement of the
particles. The ideal framework of lattice points is only obtained
by mental averaging of all such positions. The photographs of
our crystal would also reveal that this motion has a fairly large
amplitude, and that the amplitude increases with rising tempera-
ture. At room temperature it amounts to 0.1-0.2 angstrom, i.e.,
to a few percent of the repeating period.
The amplitude and the nature of vibrations depend on the
nature of forces holding together the particles in the crystal.
When the crystal is made up of molecules, the most noticeable
motion is that of each molecule as a whole. The molecules exe-
cute both translational motion (forward and backward along
straight lines) and torsional motion (twisting out of a given
plane). This could give rise to absolute chaos, each particle
jiving on its own, were it not for the bonds between the particles.
As a result, the vibrations present a rather orderly picture of
concerted movement, rather like formation dancers. It is cus-
tomary to imagine the bonds as springs between spheres, as
shown in Fig. 31. When one sphere is displaced from its equi-

46
Elements of Disorder in Order 47
librium position, it is followed by the second, third, fourth, and
the other spheres, so that a wave ripples through the entire
chain. This is exactly what happens in crystals; instead of the

GJVW(Yv'.
I
I
I
,
I

Fig. 31. Spheres connected by springs behave as atomic rows in


crystals. The atoms are bound together by electric forces, which
may be likened to springs. The figure shows the positions at two
moments: in one case the middle set of three spheres is fully ex-
tended, in the other it is fully compressed

particles vibrating independently of one another, the individual


vibrations merge into a concerted wave motion.

2. The gas-crystalline state of matter


Camphor is a solid, and yet applying the word "solid" to this
substance we are acutely aware of the poverty of our language.
True enough, it's not a liquid-but it can easily be crushed in
our hands. And camphor is not the only one; there are a great
many such substances, mainly among organic compounds. They
have a waxy texture, a cloudy appearance, and generally give the
impression of being amorphous. The truth is that they are not
amorphous, and have nothing to do with glasses and the like.
They form an interesting group of solids in search of a name.
The main peculiarity of these substances is that their mole-
cules form a perfectly regular lattice as regards their positions,
but then they spoil the symmetry by assuming all kinds of
orientations. It is as if some imaginary sergeant-major had first
called a platoon to attention, to form beautifully straight lines,
and then "at ease," so that they could tum to one another, but
not leave their positions. Thus, the actual spacing is ordered, but
the orientation is disordered.
Investigations have shown that the disordered orientation is
due to the continuous rotation of the particles, though it might
theoretically also be brought about by freezing the molecules in
48 Order and Disorder in the World of Atoms

random rotational positions. This state clearly calls for a name,


and one may offer "rotational crystalline" or "gas-crystalline"
state, stressing in the latter the combination of gaseous and
crystalline features.
For a long time these substances attracted little attention and
were dismissed as rare examples of the anomalous behavior of
molecules in the crystal lattice. This view is no longer tenable
today, since recent investigations on many such substances have
revealed a number of common structural regularities, chiefly that
gas-crystalline substances are formed when the molecules do not
disperse their neighbors strongly during their rotation, and close
packing is thus not destroyed to any great extent.
When the molecules of the rotational crystal are spherical,
it is their centers that form the long-range order. The result is
one of the close-packed arrangements characterized by cubic or
hexagonal symmetry (e.g., camphor itself). When the molecules
are cylindrical, it is their axes that form the long-range order.
The arrangement is then similar to a bundle of pencils held
tightly in the hand. This has been found in some paraffins and in
many other compounds.
These substances are oily to the touch, and waxy in appear-
ance, except that they are transparent. They are not firm and
cannot keep a well-contoured shape. They exist in a definite
temperature range, just below melting point. This, is under-
standable, since their formation from ordinary crystalline sub-
stances is like the first stage of melting, a state that Frenkel
called "oriented melting." The melting of the rotational crystals
themselves proceeds in two steps: first, the molecular orientation
is destroyed, and then the long-range order of the molecular
positions.
QUite generally, atoms attract one another over long distances
and repel as they come together. The result of these two oppos-
ing forces is a compromise, i.e., the formation of equilibrium
distances at which attraction and repulsion just balance out.
This is a basic tendency, which creates long-range order in the
crystals, but which is complemented by an opposing but equally
fundamental tendency calling for disorder. This latter phenom-
enon originates from the fact that at all temperatures above
absolute zero (-273C) every substance contains some excess
Elements of Disorder in Order 49

energy which it must dissipate to maintain stability (lowest-


energy state). It does so most simply by the incessant vibration
( thermal motion) of the constituent particles. This is absolutely
necessary, and it naturally leads to disorder. It follows that the
higher the temperature, the greater will be the resultant dis-
order. Once again a compromise is reached, now on a higher
plane, between the two fundamental tendencies, and again a
compromise is reached between order and disorder.
At low temperatures, where the thermal vibrations are not
intensive, the ordered arrangement into stable crystals holds the
hegemony. However, as the temperature is increased, such an
arrangement leaves little scope for the essential vibrations, and
the frozen ideal order is destroyed. Even this is insufficient as
the temperature is increased further, and the substance even-
tually melts or evaporates. However, in some cases, namely, in
liquid crystals and rotational crystals, the compromise between
the two opposing tendencies has been consolidated into a fairly
stable state, which may therefore be thought of as nature's
attempt at reconciling incompatible partners.
In the case of the gas-crystalline state the round shape of the
molecules gives rise to a structure in which the tendency toward
order is satisfied by the regular arrangement of the molecular
centers into a lattice, and the tendency toward disorder by
rotation of the molecules. As the temperature is decreased, the <

molecular orientations harmonize, and at a certain point, order


begins to reign supreme: the rotational crystal suddenly changes
into an ordinary crystal. Thus methane, the chief component of
town gas, freezes into a rotational crystal at -182C, which in
turn changes into an ordinary crystal when the temperature has
dropped to -253C.
Thus thermal motion can limit the realm of order to the
centers of molecules in crystals, and create a situation in which
disorder is built in into order.

3. Block structure
Another manifestation of built-in disorder amidst order is due
quite simply to the difficulty of maintaining strict order over
long distances in crystals. The difficulty involved is best appre-
50 Order and Disorder in the World of Atoms

dated when one tries to build a "tall" tower out of children's


building bricks. The structure appears perfectly straight, and
yet it suddenly collapses. Closer inspection would reveal that a
slight bias in one direction crept in somewhere and then pro-
ceeded to increase additively and surreptitiously. In the case of
crystals, cohesion is too strong to permit total collapse, so the
structure just breaks down into ordered blocks or domains. This
has been shown experimentally to lead to the "sprained" blocks
illustrated in Fig. 32.

Fig. 32. Schematic representation of the block structure of


crystals. The angles of deviation from the straight lines are
exaggerated for the sake of clarity

On imagining these particles as beads and mentally stringing


them on a thread, we find that, after picking up on the thread a
certain number of beads lying in a straight line, we must turn
the needle to find the next one and others beyond, after which
we can again proceed in a straight line until the next turn. This
means that very long-range alignment in the crystal is broken
down and blocks are formed at angles of at most a few dozen
degrees with respect to one another. Each block is characterized
by ideal order. Furthermore, the blocks are twisted, now to the
Elements of Disorder in Drder 51

right and now to the left, so that, statistically speaking, straight


directions are maintained throughout the crystal. Nevertheless,
the breakdown into blocks does mean a breakdown of ideal
order.

4. Dislocations
Not so long ago William Bragg, who played a very significant
part in the investigation of the structure of crystals, proposed
a method of representing the arrangement of atoms in crystals
with the aid of soap bubbles. A photograph of the surface of a
liquid covered with soap bubbles (Fig. 33) illustrates remark-
ably well several features of crystal structure that we are about
to discuss. The photograph shows two differently oriented
"crystal grains." The boundary between them has an irregular
structure, contains much empty space, and the atoms in the
transition region are not closely packed. The presence of voids
indicates that foreign particles will be preferentially distributed
in these transition regions, as has already been mentioned above.
However, the main point of interest about this model is not
the boundary between the two grains, but the boundary between
two blocks of the same grain, shown on the right-hand side of
Fig. 33. This boundary is marked by an arrow. Careful inspec-
tion of the bubbles reveals that the row of bubbles slightly
changes direction on crossing the boundary. Following the
boundary of the blocks, we find a site of gross imperfection.
Closer examination shows that the reason for the destruction of
order is that the number of horizontal rows to the right of the
boundary is one less than the number of horizontal rows left of
the boundary. Inother words, an extra row has crept in into the
left-hand part of the block.
The photograph is only a two-dimensional model of the
crystal, but little imagination is needed to visualize a three-
dimensional crystal distorted in this way. Each row of circles is
treated as if it were the projection of a layer of atoms running
perpendicularly to the drawing. The site of strong imperfection
seen in the photograph becomes, in the three-dimensional
crystal, a linear region perpendicular to the plane of the paper.
Such a site is called a dislocation. The first theory of dislocations
~

"

o
a.
~
...
.....
CI
i"
a.
~
s
s:
CD

~
;s:
a
Fig. 33. Surface of a liquid covered with soap bubbles. There is a farreaching analogy be- ~
tween the distribution of the soap bubbles and the distribution of atoms in crystals :3
en
Elements of Disorder in Order 53

was worked out by Taylor and Deliger. The presence of disloca-


tions is one reason for the fact that crystals break down into
blocks.
It will be seen later that dislocations explain not only the block
structure of crystals but also many other phenomena, and there-
fore this peculiar defect of crystals deserves closer study.
We distinguish between edge dislocations and screw disloca-
tions. The dislocation visualized with the aid of the soap-bubble
model is an edge dislocation. An edge dislocation (marked by
an inverted T) is shown schematically in Fig. 34a.
/ - - - - ---:1 To become aware of the pres-
/ /1 ence of a dislocation, let us ex-
/ / I
/ / I amine the boundary between the
// / ril
/ / / I I two blocks (broken line): the
/ / / I \ atomic layer above the boundary
/ . / / / i
/ / I
contains one row of atoms more
/ / I than the adjacent layer below
/ /
[ i / / /
; "" . / / /
-- ------ ~ --------- - /
/
/
~~__9__+1~~~>__/
~~_4~H4~~~/
o} b}
Fig. 34. a) Edge dislocation. The layer above the broken line indicating the
boundary contains one row of atoms more than the layer below the bound-
ary b) Layers of atoms above (full lines) and below (broken lines) the
block boundary. This figure is highly schematic, and a picture closer to
reality is given by Fig. 33. The reader is invited to compare carefully the
two illustrations

the boundary. The dislocation region is near the line running


along this boundary. It is here that the distortion is greatest, and
from here, it quickly disperses in both directions as we recede
from the line of dislocation. Figure 34 also shows a top view of
two atomic planes on either side of the block boundary. The
upper, compressed plane (full lines) contains one row of atoms
more than the lower plane (broken lines).
Figure 35 shows an analogous scheme for a screw dislocation.
Here the lattice is split into two blocks, one of which has as it
54 Order and Disorder in the World of Atoms

were slipped by one period with respect to the other. The


greatest imperfection is concentrated on the axis shown in the
figure, and the region around this axis is called a screw disloca-
tion. The nature of this dislocation is best grasped by examining
Fig. 35b, which shows two atomic planes, one on either side of

0)

b)
Fig. 35. a) Screw dislocation b) Cross section
of diagram (a) (two atomic plane's adjacent to
the plane of the section; the black dots are
above and the white ones below the plane of the
paper)

the dislocation plane. The axis of the screw dislocation is the


same as in the three-dimensional drawing. Full lines indicate the
plane of the right-hand block, and broken lines the plane of
the left-hand block. These drawings show that the screw dislo-
Elements of Disorder in Order 55

cation differs from the edge dislocation. In the case of the


former, there is no extra row of atoms, and the :Haw is due to
the fact that, in the vicinity of the dislocation axis, the atomic
rows change their nearest neighbors: they sag and become level
with particles of the next layer below.
Why is this type called a screw dislocation? Let us mentally
shrink again to subatomic dimensions and begin walking round
the dislocation axis starting from the lowest plane. It is easy to
see that each higher level is reached by following a screw
thread. Finally, we reach the top of the crystal in exactly the
same way as we walk up a spiral staircase. In our drawing, this
spiral or screw runs up counterclockwise. Had the blocks slipped
the other way, a clockwise spiral would have been formed.
The screw dislocations in a given crystal may all follow the
same direction. However, an interesting defect arises when one
plane contains a clockwise and a counterclockwise screw dislo-
cation (see Fig. 36) .

'"
Fig. 36. Ramp formed by two screw dislocations,
one running clockwise and the other counter-
clockwise

5. The dislocations move


To observe the movement of a dislocation under a microscope
would be rather difficult, and the question whether our hypoth-
esis of the existence of dislocations in crystals is in fact true
must therefore be decided on the basis of deductions arrived
at by examining the appropriate models and schemes.
Let us first see how crystals would deform in the presence
56 Order and Disorder in the World of Atoms

and in the absence of dislocations. These situations are illus-


trated in Figs. 37 and 38.
Displacement requiring a large force

Fig. 37. Mechanism of the displacement of one atomic plane with respect
to another in the absence of dislocations. The displacement takes place in
steps. Initial, intermediate, and final positions are shown

Displacement requiring little force

Fig. 38. Probable mechanism of displacement. The displacement takes


place only when a dislocation runs through the crystal, and occurs sud
denly. The displacement is transmitted through the crystal just as com
,pression pulses are transmitted from sphere to sphere

Suppose that we want to move the upper half of a crystal


with respect to the lower half by one interatomic distance. For
this we must push an entire plane of atoms situated above the
line of motion, over the plane situated below this line. However,
our aim can be achieved in a completely different manner if
the crystal contains a dislocation.
Figure 38 shows a set of close-packed spheres (end spheres of
atomic rows) incorporating an edge dislocation. It is assumed
Elements of Disorder in Order 57

for the sake of simplicity that the dislocation region involves only
a small number of rows. The presence of the dislocation is then
tantamount to a linear fissure between the two rows of the upper
extended plane adjacent to the boundary between the blocks.
On the other hand, the lower compressed plane adjacent to the
block boundary contains an extra row of atoms. The result of
the incorporation of this extra row of atoms is that the two
rows of atoms immediately below the fissure are extremely com-
pressed, so much so that the atoms lose their spherical shape.
Let us now move the upper block to the right with respect to
the lower one. For the sake of clarity, the spheres have been
numbered, and the numbers of the spheres in the compressed
layer carry dashes. Originally; the fissure was between rows
2 and 3, and rows 2' and 3' were compressed. When the force
is applied, row 2 moves into the crack, sphere 3' can now breathe
freely, and sphere l' is compressed. What does all this amount
to? The whole dislocation has moved to the left, and it con-
tinues to move until it emerges on the surface of the crystal.
The result is a displacement by one atomic row, i.e., the same as
we have seen in the case of the ideal crystal in Fig ..37. In other
words, the displacement consists of a shift of the dislocation
line along the plane of the shearing force. It is not necessary to
prove that a much smaller force is required in this case than in
the case mentioned above, where the force had to be large
enough to overcome the interaction between all atomic rows
making up two planes. In the second case the force has to be
sufficient only to move one atomic row.
Calculations show that the strength of crystals in which shear-
ing displacement takes place in the absence of dislocations is a
hundred times as great as the experimental value. The presence
of a small number of dislocations appreciably lowers the strength
of the crystal.
However, a difficulty arises at this point. As can be seen from
Fig. 38, the applied force expels the dislocation out of the
crystal. This means that as the degree of deformation of the
crystal is increased, the structure should become stronger and
stronger, and should finally reach the theoretical strength when
all the dislocations have been expelled. The strength of the
crystal is in fact increased by the elimination of dislocations, but
58 Order and Disorder in the World of Atoms

by nothing like a factor of 100. The trouble is that crystals retain


their screw dislocations, as the latter are very difficult to expel
(this is difficult to illustrate by drawings, so the reader will have
to take it on trust). Furthermore, the displacement can be pro-
duced in the crystal through both types of dislocation. The
theory of dislocations is explained satisfactorily by the charac-
teristics of the displacement of crystal planes. According to
modem views, plastic deformation of crystals is based on the
migration of imperfections along crystal planes.
However, this does not mean that we should regard the block
structure as "frozen" when no force is acting on the crystal.
Thermal motion may propagate dislocations along the block
boundaries, so that the imperfections migrate through the crys-
tal and the block configuration is in a state of constant change,
leading to the formations of new boundaries and to annihilation
of old ones. Much work must still be done to elucidate com-
pletely the continuous struggle between order and disorder, the
very breath of life of the crystal.

6. Ideal crystals
It is in fact possible to prepare an ideal crystal and satisfy one-
self that its strength is about a hundred times as great as that of
a crystal containing dislocations. Such ideal crystals have re-
cently been prepared quite accidentally in the form of
extremely thin (less than a micron) monocrystalline tin wires
(tin whiskers). These whiskers are grown very slowly under
ideal conditions. Moreover, their small size is not conducive to
the accommodation of dislocations. Their strength is indeed
about a hundred times as great as that of ordinary large tin
crystals.
This discovery was int~resting in that it provided an impetus
to the search for other ideal crystals. It was thus found that
whiskers can be made from very many (most probably all)
substances, but no one has yet succeeded in growing defect-free
crystals of larger dimensions. It may be that ideality is a prop-
erty only of the whisker form, though no conclusive decision
can yet be reached on this point. It is known, however, that
the thinner the whiskers the stronger they are.
Elements of Disorder in Order 59

Many interesting experiments can be carried out with


whiskers, but we shall describe only one, illustrating their tend-
ency to retain ideal order. A copper whisker measuring 2.5
microns in diameter bends like a sword (Fig. 39a), and springs
back when released, indicating great elasticity (Fig. 39b) .
Figure 39c illustrates plastic bending of the whisker: in this
case when the force was removed the whisker kept its assumed
form. Then the bent whisker was heated for ten minutes at an
elevated temperature to facilitate thermal motion of its atoms.
The free motion of the atoms proved sufficient to restore the

, .

IJ'

'0) bJ H

d) e) f)

Fig. 39. Experiments with copper whiskers

ideal lattice, and the whisker straightened out without any


mechanical assistance (Fig. 39d). Moreover, the elastic proper-
'ties were also fully restored, as shown by repeating the first
experiment with the recovered whisker (Figs. 3ge-39f).
Thus, under some special conditions, one can eliminate almost
completely the features of disorder from crystals.
60 Order and Disorder in the World of Atoms

7. Defects within the blocks


We have already seen that crystal lattices are made up of
blocks, with dislocations sliding along the block boundaries. But
what-if any-deviations from ideal order are to be found
within these blocks?
Features of disorder are found inside the blocks as well; the
lattice, too, can incorporate defects in the form of holes and
impurity atoms, and very few of these suffice to cause appre-
ciable distortion. Figure 40 shows the effect of an empty node
amidst filled lattice points, of an atom introduced into the basic
lattice, and of an impurity atom occupying a normal site.

I.., 1 """'r-'-
h H
.....

~ ~'"
J

1
0)

,.. 1..,. , "I


'-1
~J
i Ii
tt- .----<
~ ~N
~
~ ~H ,.. "--"
.....
1 r J. J.
J
H
A I~
b) c)

Fig. 40. Lattice deformations caused by various factors: a) Large impurity


atom occupying a normal lattice site b) Inclusion of a foreign atom
c) Deformation of the lattice due to an unoccupied site
Elements of Disorder in Order 61

Such effects may operate within a radius of 5-10 lattice


periods. In the latter case the domain thus covers 10 x 10 x 10 =
1000 cells, and it follows that even 0.1% of an impurity can
fundamentally alter the properties of the crystalline substance,
though not by appreciable distortion of the lattice.
To make this point clear, consider a germanium (or silicon)
crystal, in which each atom has four closest neighbors, and each
pair of atoms is linked by a pair of electrons. Germanium has
four valence electrons, so all of them are utilized in this capacity.
Suppose now that one germanium atom has been replaced by an
atom of arsenic. This newcomer must conform to the bonding
rules of germanium and offer four of its valence electrons for
bonding with neighboring germanium atoms, but this will still
leave one electron because arsenic has five valence electrons.
This fifth electron has nothing to do, and can wander freely in
the germanium lattice. In this way, the impurity arsenic atom
confers on germanium n-type electric conductivity (the n stands
for "negative"). When a voltage is applied to this crystal, the
excess electrons begin to move in an ordered fashion, which
means that an electric current is produced.
If the germanium atom is replaced by a trivalent aluminum
atom, the latter can give up only three valence electrons. It is
thus bound only to three germanium atoms, and a hole is formed
where the fourth bond should be, i.e., at the site where the
electron is missing. It is well known from electrostatics that a
site from which an electron is missing behaves as a positive
electric charge, and this is the situation in the present case. Thus
when a voltage is applied to a germanium crystal containing
aluminum impurity, the holes begin to move in an ordered
manner, so that a current flows through the crystal. This is called
p-type conductivity (p for positive) .
Imagine a group of soldiers standing in formation. When one
of them in the front row leaves, the one behind him in the
second row steps into the vacant place to reestablish ordered
arrangement. This of course leaves a gap in the second row,
which is filled from the third row, and so on. Now, this can also
be described by saying that the gap is moving backward, and
it is exactly in this way that holes travel in a crystal. When a
potential field is applied to such a crystal, an adjacent electron
62 Order and Disorder in the World of Atoms

jumps into the hole, leaving behind a vacancy which in turn is


occupied by another electron-and thus the hole is set in motion
in the crystal.
If experiments prove that a certain germanium crystal exhibits
e.g., n-type conductivity, then the introduction of aluminum
atoms can reduce this conductivity. As the addition of aluminum
is continued, more and more excess electrons are "neutralized,"
until finally the crystal becomes nonconductive. A further addi-
tion of aluminum will change the germanium crystal into a
p-type conductor. Experiments show that the addition of a
single impurity atom to a billion germanium atoms can upset
the balance between free electrons and holes, and therein lies
the great importance of impurities.
The very young industry producing germanium crystals of
various kinds is of immense significance in modem technology.
A sandwich of a p-type crystal between two n-type crystals (or
vice versa) constitutes a remarkable device called a transistor,
which fulfills the function of a triode whilst offering great addi-
tional advantages. However, the mechanism of the action of
transistors and their uses lie outside the scope of this book.

8. Crystals with errors


The simplest dense arrangements of atoms in crystals are repre-
sented by hexagonal close packing, in which the third layer
coincides with the first, and by cubic close packing, in which
the fourth layer coincides with the first.
As mentioned earlier, there are innumerable possible arrange-
ments, differing in their sequence of layers. Looking at the
closest packing of spheres we see that the layers can be classified
into three types, A, B, and C, such that all layers A coincide
with one another, all layers B coincide with one another, and
all layers C coincide with one another. If the first layer is called
A, then layers Band C will be layers fitting in the light holes
of layer A (layer B) and in the dark holes of layer A (layer C).
No other layers can arise in close packing (cf. Fig. 11).
Any sequence of close-packed layers can easily be described
by letters. Thus, the hexagonal packing and the cubic packing
Elements of Disorder in Order 63

are characterized, respectively, by sequences ... ABABAB ...


and ... ABCABCABC ...
The repetition of any sequence is a characteristic of ideal
long~range order. Thus, the sequence ... ABABCACBCABABC-
ACBCABABCACBC ... corresponds to an ideal crystal in which
the repetition occurs after every nine layers.
Some close-packed systems of spherical particles are also
found in nature in which the sequence of layers is not repeated.
These "crystals" represent interesting examples of mixed order
and disorder, in that the atoms have an ordered arrangement
within each layer, but the distribution of layers is devoid of
order. This happens in the case mentioned earlier, in which
order prevails in two directions but not in the third.
We shall now examine more closely the structures of sub-
stances in which order is combined with disorder in this interest-
ing manner. Arrangements of this type occur in nature with
various degrees and forms of order. We can take as an example
a growing crystal in which the layer arrangement follows the
hexagonal pattern:
... ABABABAB ...

At some point, say, in the ninth layer, the crystal makes a mis-
take and the ninth layer follows the cubic pattern:
... ABABABABC ...

Further layering is again in accordance with the hexagonal


arrangement:
... ABABABABCACACACACACACA ...

until a mistake is made once more:


... ABABABABCACACACACACACACAB ...

but the "proper" sequence is then gamely continued:


... ABABABABCACACACACACACACABCBCBC ...

This process gives rise to a "crystal" whose individual sections


show a regular structure, while the errors introduce in it an
element of disorder. If these errors are fairly numerous, we
eventually obtain a substance characterized by order in two
64 Order and Disorder in the World of Atoms
dimensions (each layer is ordered) and disorder in the third
(the stacking of the layers is disordered). But are these sub-
stances ultimately ordered or disordered? The truth is that order
and disorder are combined in these cases, and we cannot
group these substances either with crystals or with amorphous
materials.
An example of such a substance is cobalt. Prolonged annealing
at a high temperature can convert finely powdered cobalt into
fairly large grains consisting of large sections with packing
. ; . ABCABC ... (the cubic phase) and small 9-10 layer sections
of type ... ABAB ... separated by "cubic" mistakes. A series of
such packets forms what may be called the hexagonal phase
of cobalt.

9. Order and disorder in binary alloys


The letter sequences discussed above are not merely academic
curios, because the structure of substances is closely linked with
their properties. Pure metals find much less application in prac-
tice than alloys. We shall now deal with binary (two-compo-
nent) alloys, which are used widely in metallurgy.
It has been found that the concept of the close packing of
spheres is applicable to alloys as well. Binary alloys often repre-
sent an arrangement of close packing of spheres. This arrange-
ment is easily realized when the atoms of the two rrietals are
roughly the same size, or when the atoms of one metal are
considerably larger than the atoms of the other metal. In the
second case the large atoms are arranged in close packing and
the small atoms occupy the cavities. These interstitial alloys
include steel, in which the larger atoms are iron and the smaller
ones carbon.
Binary alloys, with the two kinds of atoms having roughly the
same size, are also fairly numerous. An example of this group are
the alloys of copper and gold. When the atoms of the additive
replace those of the matrix forming a close-packing arrange-
ment of spheres the alloys are said to be substitutional, and it
is to this type of alloy that we shall now tum our attention.
Suppose that such an alloy is built up of layers of one type
of atoms. The model of the structure of the alloy can be repre-
Elements of Disorder in Order 65
sented by a close-packed assembly of spheres, the layers being
stacked in accordance with the hexagonal or the cubic rule.
Layers of black spheres are interleaved with layers of white
sph~res.
Anyone having a large number of black and white spheres
can build many different close-packed assemblies. Let us first
assume that we have the same number of black and white
spheres to fonn black-and-white assemblies. The simplest ar-
rangement is an alternating row:

o.oeoeo
.another simple arrangement being

ooooee.eoooo....oooo
i.e., one based on a regular alternation of the same numbers of
black and white spheres.
Such arrangements have indeed been found in several alloys.
The first type of arrangement is called the simplest ordered
structure, and the second type is called a superstructure.
However, alloys also exhibit disordered arrangements such as

eoeeoo~....oeooooeOOMOOM.oeeooeooOM

or

The complete disorder in the last but one case lies in the
fact that 10 out of the 20 black layers have white layers as their
right-hand neighbors, while the other 10 have black layers.
Conversely, 10 out of the 20 white layers are flanked on their
right by black layers, and the other 10 by white ones. The same
argument applies of course to the left-hand neighbors.
Partial ordering of the above structure begins to appear when
the number of black layers in contact with other black layers
and the number of white layers in contact with other white
layers are decreased. Such a situation is depicted by the last
arrangement above, with the result that the sequence does not
66 Order and Disorder in the World of Atoms

give the impression of utter chaos. In fact, only 5 of the 20 black


layers have black ones for their neighbors (on their right or
left). In other words, the black layers have "correct" neighbors
in 75% of the cases, and order predominates over disorder. When
there are equal numbers of black and white layers, the same
also applies to the white layers. We thus find ourselves halfway
between order and disorder: 20 "correct" neighbors for 20 layers
mean order, 10 "correct" neighbors disorder, and in the case
under discussion the number of the "correct" neighbors is 15.
The problems of order and disorder assume a more compli-
cated character in alloys having a more complex composition.
Consider, for example, the possibilities arising from having
four times as many white layers as black ones. The simplest
ordered structure and the superstructure are, respectively:

()()()()e()OO0e0()()()eOOOoe
and

Moreover, we can also have a structure of the type

in which no two black layers are in contact, and the number of


white layers having black neighbors is the same as the number
of black layers having white neighbors. These structures .have
no long-range order as regards the sequence of layers, but they
do exhibit perfect short-range order as regards the environment
of the black layers.
To find a disordered arrangement when the white-to-black
ratio is 4: 1, we consider that since there are four times as many
white layers as black ones, the proportion of black layers having
white neighbors should be four times as great as the proportion
of white layers having black neighbors. This is satisfied by the
following arrangement:
oooooeooooooeeooooeoooo.oooeoooeoooooooe.
00000.000
Elements of Disorder in Order 67

Here we have 40 white and 10 black layers. For the sake of


simplicity, we consider only the right-hand neighbors, and note
that 8 out of the 10 black layers (i.e., 80%) have white right-hand
neighbors and 8 out of 40 white layers (i.e., 20%) have black
right-hand neighbors. The ratio between 80% and 20% is 4:1,
and so the above condition is indeed fulfilled.
In contrast to the case in which the number of white layers
was the same as the number of black ones, the ordering of such
alloys can be effected in two ways, namely, by decreasing the
number of black neighbors so that short-range order in the
environment of the black layers may become ideal, or by equaliz-
ing the number of white layers between black ones so that
long-range order may arise.
The fact that we keep talking of disorder in the stacking of
layers may have given the reader the impression that order
necessarily prevails within these layers themselves. Although
elements of disorder in order are more readily traced in sub-
stances consisting of ordered layers, it is sometimes the actual
structure of the layers that is disordered. This is the case with
an iron-cobalt alloy which has the body-centered cubic lattice
shown in Fig. 41. Each atom-whether iron or cobalt-has
eight nearest neighbors. The arrangement of the atomic centers
in the crystal lattice is always completely ordered, in that the
atoms form the same body-centered ~ubic lattice under all con-
ditions. However, the situation is different as regards the dis-
tribution of iron and cobalt atoms over the two kinds of lattice
points, i.e., the apical and the central ones. In the case of
complete order, all the apical sites are occupied, say, by iron
atoms, and all the central ones by cobalt atoms. The ideal long-
range order of such a crystal gradually deteriorates when atoms
begin to appear on "usurped" sites. However, as long as the
number of atoms on "correct" sites differs from the number of
atoms on "usurped" sites, we may speak of long-range order in
the crystal, though it may not be complete or flawless. The
long-range order finally disappears when only half the atoms
occupy their "correct" sites.
When a crystal exhibiting complete order is heated, more and
more atoms occupy usurped positions, and the order is gradually
obliterated, disappearing completely at a certain characteristic
68 Order and Disorder in the World of Atoms

temperature. This temperature, above which long-range order


cannot exist, is called the lambda point of the material. For the
iron-cobalt alloy the lambda point is 770C. The transition from

0)

b) c)
Fig. 41. Structure of an iron-cobalt alloy, showing the atoms on the sites
of a body-centered lattice. The white and the black circles denote, re-
spectively, iron and cobalt atoms. The lattice points may be occupied in
different ways: a) complete order-all cube corners are filled by iron
atoms and all centers by cobalt atoms b) long-range order becomes
frayed c) total absence of long-range order

order to disorder means that thermal motion has overcome the


tendency of the atoms toward long-range order.
The processes of melting and of obliteration of the difference
between the occupancy of correct and usurped sites have much
in common, because both lead to the disappearance of long-
range order. However, melting is accompanied by loss of
Elements of Disorder in Order 69

long-range order in the arrangement of the atomic centers,


whilst transition through the lambda point eliminates only the
order in the arrangement of the atoms of different elements.
A basic feature of iron-cobalt type alloys is the possible
existence of partial long-range order, i.e., long-range order as
regards the distribution of the cobalt and iron atoms but not as
regards the arrangement of the atomic centers. As in the case
of melting, the destruction of long-range order does not give a
fully disordered structure, since short-range order persists.
This short-range order in the distribution of iron and cobalt
atoms lies in the "desire" of iron atoms to surround themselves

Fig. 42. Structure characterized by the presence of shortrange


order and the absence of long-range order

by cobalt atoms, and vice versa. Examining the eight closest


neighbors of any A-type atom, we find more than four B-type
atoms, e.g., five, six, or seven, depending on the degree of the
short-range order (see Fig. 42).
The short-range order in a copper-gold alloy is very high
and extensive: it is manifested not only in the number of nearest
70 Order and Disorder in the World of Atoms

neighbors, but also in the number of those next but one, next
but two, and so on. Drawing a series of spherical shells around
any gold atom we find that the first shell will consist almost
exclusively of copper atoms, and the second shell almost exclu-
sively of gold atoms. As we proceed outward the order becomes
progressively blurred, but it is still noticeable even in the tenth
spherical shell.
Accurate investigations with the aid of x rays have revealed
how long-range order is produced in alloys. Thus, in the case of
cobalt-platinum alloys, domains of long-range order grow in
the disordered crystal as nuclei of crystals grow in a liquid, these
nuclei being oriented in a well-defined manner with respect to
the crystal axes.
Elements of order and disorder often coexist in laminar min-
erals such as graphite, mica, and chlorites. The structure of
these substances obeys the rules discussed above.
Spatial disorder has been investigated most extensively in al-
loys, bringing to light a number of important factors concerning
the forces binding the particles in metals, inorganic substances,
and organic compounds. This binding results from the inter-
actions between adjacent ions in inorganic crystals, and between
adjacent molecules in organic crystals. In contrast, in metals the
cohesive binding is mainly between the positive metal ions and
the negative electrons, the latter wandering more or less freely
and chaotically through the ion lattice. Thus, metallic crystals
differ basically from the others by the absence of definite bonds
between atoms. It is therefore not surprising that it is in sub-
stances based on the metallic bond that the deviations from
ideal structure are particularly frequent. Clearly, since the
atoms in metals are "depersonalized," they can easily be dis-
placed and substituted.

10. Magnetic order


Have we not yet exhausted the types of order? Are there still
other forms besides short-range order, long-range order, order
as regards the array-formation of atomic centers, and order in
the distribution of atoms of different elements? The answer is,
yes, there are.
Elements of Disorder in Order 71

Before discussing the main topic of the present section, we


shall mention briefly the concept of isotopic order. Isotopes are
species of the same element that are chemically identical but
differ in the number of certain particles in their nuclei, and
hence have different atomic weights. The word "isotopic" means
"belonging to the same place," specifically in the periodic table
of elements. From the chemical point of view, an alloy com-
posed of two isotopes of the same element is not an alloy at all
but pure metal, since no one using purely chemical means could
distinguish between the two components. However, from an-
other point of. view, it is indeed an alloy, as indicated by the
fact that various degrees of order or disorder may exist in it
with respect to the distribution of the two atomic species. This
is what is meant by isotopic order.
Another very interesting possibility is magnetic order. In this
connection, it should be noted that a few substances possess
remarkable magnetic properties. These substances include iron,
cobalt, nickel, gadolinium, compounds of these elements, as well
as certain compounds of manganese and chromium. Such mate-
rials are called ferromagnetic, after their chief representative,
iron (ferrum). The reason for the exceptional properties of these
substances lies in their remarkable atomic structure and in the
general properties of electrons. Not only do electrons rotate in
definite orbits around the nucleus, but they also spin clockwise
( t) or counterclockwise (t) about their own axes. Singly spin-
ning electrons generate minute magnetic fields, and can there-
fore we regarded as micromagnets. However, in most atoms the
electrons tend to pair off, their spins couple in a so-called "anti-
parallel" manner (N )-a clockwise spin coupling with an
counterclockwise one-and their magnetic fields neutralize one
another. This phenomenon may be likened to the cancellation
of magnetism observed when we place two magnets side by
side in such a way that the North pole of one is level with the
South pole of the other. Thus, electron pairs with antiparallel
spins are magnetically "dead." It follows that to show appre-
ciable magnetic properties a substance must possess unpaired
electrons. For example, iron atoms contain four magnetically
"live" (unpaired) electrons. Such atoms are said to possess a
permanent moment. However, though the presence of unpaired
72 Order and Disorder in the World of Atoms

electrons is necessary, it is not sufficient for the appearance of


ferromagnetism. A further condition is an ordered arrangement
of the atomic magnets, because if the atoms were randomly
oriented their magnetic fields would cancel out. To possess
ferromagnetic properties, a substance must therefore be in a
state in which the atoms have fixed positions and their magnetic
moments can be oriented in parallel. Only the crystalline state
can allow such tidy accommodation. The regions of strictly
ordered magnetic orientation, containing millions of atoms, are
called domains, and the theory dealing with these is called the
domain theory of ferromagnetism. Nevertheless, the domain
theory does not provide a satisfactory answer to the question of
why certain atoms with unpaired spins (e.g., iron) form magnetic
domains in crystals, whilst others (e.g., chromium) do not. The
problem is difficult, and considerable work must still be done
before we can fully explain the reasons for the formation of
domains.

Fig. 43. Distribution of the magnetic arrows of


elementary magnets, as illustrated in older text
books. This picture does not correspond to reality

Until not so long ago, the domains in an unmagnetized piece


of iron were believed to be arranged as shown in Fig. 43, in
which the arrows indicate the overall magnetic orientations of
the domains. It was further thought that during magnetization
the domains rotated through a certain angle so as to align
Elements of Disorder in Order 73

themselves mainly in the directioIt of magnetization, and that


the stronger the magnetization, the greater was the order in the
arrangement of the domains. However, reality turned out to be
rather more interesting.
The first question was how to detect the magnetic domains,
or more specifically, how to detect their boundaries. The do-
mains are large enough to be seen under the microscope, but
they are not discontinuities in the same obvious manner as
crystal grains or even crystal blocks, so that crystallographi-
cally quite homogeneous regions can contain several magnetic
domains.
The problem was solved by sprinkling extremely fine powder
of a magnetic substance over a piece of iron to be examined.
Since the domains behave as miniature magnets, with their
boundaries corresponding to magnetic poles, the tiny powder
grains become concentrated at these poles and map out the
domains.
The first important result was that the order in the arrange-
ment of atoms in the crystal is closely connected with magnetic
order. The domains are magnetized not at random but in direc-
tions connected in a certain way with the arrangement of atoms
in rows and layers.
The situation is particularly simple in the case of cobalt, form-
ing a close-packed assembly ... ABAB ... (d. Fig. 12). Experi-
ments show that the magnetic moments (i.e., the domain arrows
in the model representation) are always perpendicular to the
close-packed layers, but can point in either direction. In an
unmagnetized crystal the number of arrows pointing in one
direction is the same as the number of arrows pointing in the
opposite direction. When the specimen is magnetized these
arrows become aligned in the same direction, this phenomenon
being quantitatively proportional to the magnetizing force. As
the magnetization is increased the magnetic moments become
directionally aligned in more and more domains, until eventually
all domains are oriented in the same directions, and no further
magnetization can take place. It follows that magnetization
will be unsuccessful in the direction parallel to the close-packed
layers, and only partly successful in the oblique directions. These
phenomena do not of course take place in microcrystalline
74 Order and Disorder in the World of Atoms

cobalt, in which the microcrystals (and thus also the domains)


are distributed at random.
The chief representative of magnetic materials, iron, crystal-
lizes with a cubic lattice. Experiments show that the magnetic

Fig. 44. Two-dimensional representation of the


orieJ:]tations of magnetic arrows of domains in an
iron crystal. The dots and strokes denote, respec-
tively, directions toward and away from the reader

moments are the easiest to align along the six cube axes, and
the latter are correspondingly called the easy di:t;ections of
magnetization. The domain structure of iron is shown sche-
m~tically in Fig. 44.
When in a piece of unmagnetized iron we mentally pair off
each arrow with one pointing in the opposite direction, and then
ignore them as neutral sets, we end up with zero resultant mag-
netization. The process of magnetization consists of aligning the
magnetic arrows, but this need not take place at once over the
whole domains, and is in fact more likely to proceed gradually.
Figure 45 shows two photographs, taken at a high magnifica-
tion, of the surface of an iron crystal sprinkled with magnetic
powder to outline the domain boundaries. Above and below the
zigzag the arrows (domain orientations) point, respectively,
right and left. Figure 45b shows the same section after the
application of a magnetic field directed to the right. As a result,
the domains with the arrows pointing to the right (here the
Elements of Disorder in Order 75

upper domains) encroach upon the lower ones, and the bounda-
ries are lowered.
Magnetic order and order with respect to the distribution of
atonis, i.e., crystal blocks and magnetic domains, have much in
common. As we have already seen, crystal blocks are bounded

oj bJ
Fig. 45. Domain boundaries a) before and b) after magnetization .
Application of a magnetic field directed to the right causes the domains
with fieldparallel arrows to encroach upon the domains with fieldanti-
parallel arrows. The boundaries are consequently lowered

not by sharp lines but by defect regions. Exactly the same ap-
plies to magnetic domains: they are bounded by regions in
which the arrows gradually change from one easy direction of
magnetization to the other. Such a transition region is shown
in Fig. 46.
At a certain well-defined temperature (the melting point),
the long-range order in the arrangement of atoms is destroyed.
Temperature has exactly the same effect on the orientation of
the magnetic arrows; order in this respect is destroyed at another
well-defined temperature, called the Curie point, and the mate-
rial loses all its magnetic properties. The Curie point of iron
lies at 700C.
It has recently been shown that the long-range order of the
magnetic arrows in domains can assume a more complex form
than in the cases of iron and cobalt. Thus, in magnetic sub-
stances composed of different types of atoms one set of atoms
76 Order and Disorder in the World of Atoms

can point its arrows in one direction and a second set of atoms
in another, antiparallel with respect to the first.
It might seem at first that the phenomenon would be unde-
tectable if half the arrows point in one direction and half in

iii i \ ,""-......--~
i r r i \ ,~--~
i ri i \
iii i
1 iii
\ ,"'-----
,""-......--~

\ "'-""-......_--
iii i \ ,""-......----
f iii \ ,~----
Fig. 46. Formation of a transition zone between
adjacent domains. In actual fact the domains are
thre'edimensional

another. However, it can be detected in the presence of mag-


netic order, provided that the latter does not coincide with the
long-range order in the distribution of atoms, which we shall
now call chemical order. This condition is satisfied, for example,
by manganese oxide, whose structure (Fig. 47) shows that the
crystallographic repeat unit is half the size of the magnetic
repeat unit. The difference can be ascertained by accurate
measurements.
Mixtures of oxides, such as that formed by iron oxide and
nickel oxide, offer very interesting examples of complex mag-
netic order. One molecule of iron oxide contains two iron atoms,
and one molecule of nickel oxide contains only one nickel atom.
Crystals of the mixture consist of a close-packed oxygen lattice
in which the hollows accommodate one nickel atom or two iron
atoms. We have already seen in Chapter 2 that there are two
types of voids in close-packed assemblies, tetrahedral and octa-
hedral (cf. Fig. 29). The atoms occupying tetrahedral sites
are surrounded by four neighbors, and the atoms occupying
Elements of Disorder in Order. 77

octahedral sites by six. It has been found that the iron atoms are
distributed over both the tetrahedral and the octahedral sites.
In both cases the magnetic arrows of the iron atoms form a
fully ordered arrangement, but the arrows of iron atoms in
tetrahedral hollows are antiparallel to the arrows of iron atoms

/ J / I /
/ / / / I

il / / V /
/ / ,; 1/ /

!/ / / / V
/ V V V V
--I Chemical period I--
Magnetic period
Fig. 47. The lattice of manganese oxide. The mag
netic period is seen to be twice as long as the
chemical period

in octahedral hollows. Thus these cancel out and the magnetism


of such a mixture of oxides is due to only the nickel atoms, the
magnetic arrows of which all point in the same direction. These
remarkable substances are called ferrites. The great practical
importance of ferrites stems from the fact that in addition to
their magnetic properties they are electrical insulators.
CHAPTER 5

ORDER AND DISORDER IN THE WORLD


OF LARGE MOLECULES

1. long and branched molecules


First of all, what do we mean by a large molecule? The largest
molecules known today are of the order of 0.1 mm, a colossal size
on the atomic scale. For comparison, we may recall that the sizes
of atoms and interatomic dista~ces are close to one angstrom, i.e.,
10-' mm. If we further consider that the transverse dimensions
of the molecules referred to above are only about three or four
angstroms, we can see clearly just how improbably long they are.
They might be compared to rail tracks 10 cm wide and 10 km
long.
Such long molecules, sometimes consisting of hundreds of
thousands of atoms, are often encountered in nature and made
in lahoratories. Examples that may he quoted are rubber, cellu-
lose, and some proteins. More modest-though still immense-
lengths are found in plastics known commercially as polythene,
nylon, and so on. These great lengths of molecules are character-
istics of the dass of suhstances known as polymers.
Let us 1100V turn to some further aspects of the long molecules.
The ahove comparison to a rail track is not very apt because
the molecules are flexihle. A hetter analogy would he a rope,
except that ropes are not often made in such lengths.
The flexihility is due to the very interesting property that parts
of such molecules can rotate about interatomic bonds. Not all
honds permit such rotation, only those formed when the atom is
associated with the maximum possible number of other atoms.
Such bonds occur in an overwhelming number of molecules, and
the degree of flexibility will depend on their actual proportion
in tbe given suhstance. In all, the long molecules are not flexible
78
Order and Disorder in the World of large Molecules 79

like a length of cane but rather like a very extensive succession


of ball and socket couplings.

Fig. 48. Molecules of polyvinyl alcohol, (a) extended


and (b) coiled. The flexibility is due to rotation about
bonds joining carbons (black)

Figure 48 shows a molecule of polyvinyl alcohol, used in the


production of artificial silk. The molecule consists of a group of
seven atoms repeated along the chain, and thus exhibits long-
range order. This long-range order disappears when the molecule
tangles up into a coil. Depending on circumstances, flexible
molecules may be wavy, fully coiled, or fully extended (Fig. 49).
However, long linear molecules are not the only representa-
tives of the class of large molecules. In some naturally occurring

Fig. 49. Model of a flexible molecule in three positions


80 Order and Disorder in the World of Atoms

and synthetic products the bonds between atoms propagate in


all possible directions, giving rise to branched structures. In the
linear type, each group of atoms is connected with two other
groups, at its head and its tail. In the branched molecules each
atomic group connects up with its neighbors in at least three di-
rections. Obviously, the resulting maze of bonds will endow the
system with rigidity. A very common example of such branched
molecules are the hard formaldehyde resins.
Some recently synthesized molecules exhibit a tree-like struc-
ture intermediate between the linear and the randomly branched
molecules. In these, branches of more or less equal length issue
from a stem constituted by a linear molecule. In turn, the
branches may be to some extent joined by interatomic bonds.
Theoretically, there is thus an infinite number of possible grada-
tions between the extreme linear and randomly branched types.
Most giant molecules exhibit repeating units. These may be
a single group or several groups of atoms (in particular, the
second situation is found in proteins). In the treelike molecules,
different groups may make up the branches and the stems. These
groups are disordered in randomly branched polymers made up
of several groups of atoms. An interesting possibility appears in
long molecules containing more than one kind of atomic groups:
the different groups may follow one another in a strictly defined
sequence along the chain, or they may be arranged haphazardly.
This difference can exert a considerable influence on the
polymer's structure.

2. Bundles of long molecules


In spite of their great size, many millions of long molecules are
still contained in a tiny piece of nylon fiber or in a crumb of
rubber. The question is thus, how are these molecules packed
together?
We have already seen in many examples that one of the main
tendencies in the formation of solids and liquids is the tendency
toward close packing. In this case the problem is how to pack
tightly awkward, long, and flexible objects. One solution would
be to roll them up into balls and pack these in the same way as
we packed the spheres representing atoms in Chapter 2. Another
solution is to extend the objects to their full length and to ar-
Order and Disorder in the World of Large Molecules 81

range them in bundles like pencils. No intermediate arrangement


will satisfy so well the condition of close packing, as it will in-
evitably lead to complex tangling of the molecular chains and to
loss of packing density. These two solutions are in fact found in
practice. The long molecules in polymers are either rolled up
into coils or extended in bundles. The first situation occurs in
various protein substances, in particular, in viruses, and the
second in artificial fibers.
We may now ask how the molecules are packed in bundles,
and how the individual bundles
lie with respect to each other.
Moreover, what happens to the
concepts of long- and short-
0)
range order, and of the crystal-
line and the amorphous states?
Do they still have any meaning
in the case of polymers?
If we cut a molecular bundle
across, we can find out how the
cross sections of the long mole-
cules are arranged within this
particular bundle. The cross
sections of individual molecules
can of course be quite com-
plicated, and will vary from
case to case, but we can postu-
c)
late here for the sake of sim-
plicity that they are elliptical.
We are now faced with three
possibilities ( Fig. 50 ). In the
the first place, the centers of Fig. 50. Three possible kinds of
the cross sections can form a arrangement of long chain mole-
cules: (a) full order (crystalline),
regular lattice, and the ellipses (b) order in the arrangement of
can all be oriented in the same the centers of chain cross sec-
way. This is completely anal- tions, combined with disorder in
azimuth, (c) "liquid" arrange-
ogous to crystals, i.e., to long-
ment of the molecular cross
range order. In the second sections
arrangement, the centers still
form a regular lattice but the orientation of the ellipses is ran-
dom. In this case we can speak of long-range order in the ar-
82 Order and Disorder in the World of Atoms

rangement of the axes of the chain molecules, and short-range


order in the distribution of the chains in azimuth. Finally, we
can have the situation shown in Fig. 50c, in which there is only
short-range order in the mutual disposition of the chains.
A point of view that has been proposed by the present author
and later demonstrated experimentally is that the bundles can
contain the molecules arranged in crystalline, gas-crystalline,
and liquid-crystalline ("liquid") packings. These three types
can even be found in one and the same polymer. Moreover,
there is strong evidence that the three types can appear in
the same molecular bundle. Thus, after lying rigidly in a row
over a certain distance, the molecules may for a time fall out of
step, return to an ordered arrangement for a further distance,
and so on.
As regards the dimensions of the molecular bundles, we can
say that their thickness varies from a few hundred to a thousand
angstroms. The changes in the order along the bundles can also
stretch over hundreds of angstroms.
3. Behavior of bundle polymers
Our knowledge of the structure and the behavior of polymeric
substances consisting of long molecules is much less complete
than in the case of crystals, and considerable work will still have
to be done before the information can be checked and the prob-
lem worked out in reasonable detail.

Fig. 51. Model of the bundle structure in polymers


Order and Disorder in the World of Large Molecules 83

The arrangement of molecular bundles in a solid polymer is


shown schematically in Fig. 51. It should be pointed out once
more that the bundles are flexible, so that they can move about
and even coil up. To what extent have the properties of the
individual molecules been passed on to the bundles?
Although this is not shown in the above diagram, the bundles
are in close contact with one another. Little can as yet be said
about the nature of the regions of transition between the
bundles.
The behavior of such bundles has many aspects, but we shall
consider here only the thermal and the mechanical properties.
Most polymers undergo a number of structural alterations
before melting. Heating increases the degree of disorder, but
there are so many gradations between order and disorder in a
polymer that heating at various temperatures will "trigger off"
various disorders in succession.
If at low temperatures most of the bundle has a crystalline
character, progressive heating will initiate rotation of individual
sections along the molecular chains, then transitions of crystal-
line regions into gas-crystalline ones, and finally transitions into
liquid regions. Thus the melting point alone tells us little about
the thermal behavior of a polymer.
If the starting polymer has no crystalline regions, it will of
course have no sharp melting point. It will merely become softer
and softer with rising temperature, and its behavior can be com-
pared to that of glass.
Polymers built up of molecular bundles, such as rubber, poly-
thene, or wood, exhibit a great variety of mechanical properties.
How does rubber acquire its remarkable elastic properties,
permitting it to be stretched to five times its length and to return
exactly to its original state? More specifically, what happens to
the molecules during such reversible deformation? It is very
probable that the bundles of rubber molecules are coiled in the
form of ship's cables, and that they simply unwind when
stretched. The coils must be quite large since experiments show
that the bundles cannot be strongly curved over 100-200 ang-
stroms.
Many polymers of this type are susceptible to plastic deforma-
tion. Thus, polythene films can be deformed irreversibly by
84 Order and Disorder in the World of Atoms

stretching. In this case the most likely explanation is that parts


of the bundles are torn and they slide over themselves; it will be
recalled that a somewhat similar phenomenon occurs in crystals
( Chapter 4).
Rotation and orientation of the bundles, with the appearance
of texture, can also occur more or less in the same way as in
the microcrystalline substances.

4. Alignment of polymer molecules to form single crystals


Electron microscopic studies unexpectedly revealed in 1957 the
presence of regular crystals in polyethylene and then in other
linear polymers. The formation of such crystals is incompatible
with the structural model showing tangled chains. This is because
regular crystal form should be a reflection of strict internal order
in the arrangement of atoms or molecules. The discovery men-
tioned above attracted much interest, mainly because the in-
ability to form regular single crystals had been regarded as a
typical feature of high polymers.
Before discussing the single crystals of linear polymer mole-
cules, we must say a few words about the structure of the
single crystals formed by normal (straight-chain) paraffins,
which are low-molecular chain compounds. The reason for this
apparent change of the subject is that normal paraffins differ
from linear polyethylene mainly by their shorter length. Thus, a
crystal of a normal paraffin, which is exactly like the crystal
shown in Fig. 63c, is constructed from molecules whose zigzag
chains contain 36 atoms of carbon. The crystal consists of thin
layers winding in a spiral around an invisible axis and giving
the surface a stepped appearance. It has been established that

flIIIIIIIIIIIIII
oj bJ
Fig. 52. The thickness 8 of a single layer: (a) in paraffin; and (b) in
polyethylene
Order and Disorder in the World of Large Molecules 85

the molecular chains do not lie in the plane of these layers but
are vertical to them, the layer thickness being roughly equal to
the length of one paraffin molecule (Fig. 52a).
Dissolution of polyethylene in a hot solvent followed by slow
cooling results in the precipitation of tiny crystals of the polymer
which are easily observable under the electron microscope.
These crystals have proved to be remarkably similar to the single
crystals of normal paraffins, being also diamond-shaped and
composed of individual layers. The thickness of these layers is
constant and amounts to 100-120 angstroms (Fig. 53).

Fig. 53. Electron micrograph of a polymer crystal (x


35,000)

It was also firmly established that, as in the case of paraf-


fins, the molecular chains lie at right-angles to the plane of each
component layer. The difference is that in the paraffins the layer
thickness is determined by the length of the molecules, while
this is clearly not the case with polyethylene, in which the layers
are only 100-200 angstroms thick and the molecules about 6000
angstroms long.
The only explanation is that the polyethylene molecules in
the crystals have a folded configuration (Fig. 52b). However,
the reason for the folding of the molecules at constant intervals
of 100-200 angstroms is still open to discussion.
86 Order and Disorder in the World of Atoms

The folded configuration of molecules in single crystals and in


other forms (e.g., in fibers, see Fig. 54) has now been found
in several other linear polymers.

Fig. 54. Electron micrograph showing a band of individual


polymer fibers resting on a supporting film (x 300,000)

5. The structure of polymers


Three main types of structure can be distinguished among the
immense variety of artificial fibers and plastics-the above-dis-
cussed bundle polymers, globular polymers, and branched. poly-
mers. The main representatives of the large class of bundle
structures are rubber, cellulose, nylon, and perlon. This class has
already been dealt with in some detail in the preceding section.
The globular structures are often encountered in protein sub-
stances. Consider the electron microscope photograph of tobacco
mosaic virus shown in Fig. 55. It is perfectly clear that this is
a highly ordered structure, though this does not necessarily
mean that we are dealing with a true crystal. The point is that
the ordering in such substances may not extend to the atoms
within molecules or within fairly large atomic groups. The
whole assembly may be compared to a strictly ordered arrange-
ment of sacks of potatoes: the sacks form a regular lattice but
Order and Disorder in the World of Large Molecules 87

the disposition of the contents of each sack differs from case to


case.
The third class of structures, the strongly branched systems, is
best regarded as amorphous. We have already mentioned such
molecules, but perhaps in a somewhat oversimplified manner.

Fig. 55. Electron micrograph of the tobacco mosaic


virus

In the solid state-exemplified by a piece of plastic-the con-


cept of molecules becomes meaningless. The atomic groups are
joined continuously and we cannot really say where one mole-
cule ends and another begins. The situation is like the one we
have already encountered in the discussion of the crystals of
sodium chloride.
Before coming to discuss polymers, we encountered a number
of points which may have shaken our faith in the classification
88 Order and Disorder in the World of Atoms

of solids as crystalline and amorphous, but nevertheless this


division seemed to possess a basic validity. We did find elements
of order in disorder and vice versa, but these were rather ob-
stacles and imperfections, and the classification remained basic-
ally sound.
Now the situation becomes radically different. Polymer struc-
tures just cannot be described in terms of the simple concepts
of crystalline and amorphous states, and we shall not solve the
problem by trying to estimate the degree of crystallinity of
every polymer.
Instead, we can say that polymers contain combinations of
molecular arrangements exhibiting various degrees of order. The
terms "order" and "disorder" should now be defined more pre-
cisely if the polymer structures are to be correctly described,
but this is far from being easy. The order may concern the ar-
rangement of the axes of chain molecules-it is found in all
bundle-type polymers and in the disposition of the chain cross
sections. Several types of order are possible even in the same
substance. The disorder may be found in the arrangement of
atomic groups along a molecular chain, in the orientation of
molecular bundles, and so on. Clearly, we cannot talk our way
out in a few words by referring to polymers as crystals, amor-
phous substances, or mixtures of amorphous and crystalline re-
gions. Reality proves too complicated for this simple classifica-
tion, and we can only admit that order and disorder are both
present in substances constructed of large molecules.

6. The living cell


All animal and plant tissues are built up of cells. The cells of
various species-and even the cells of various tissues in the same
organism-differ sharply from one another, although all cells
possess certain common features.
The cell is bounded by a membrane about 100 angstroms
thick, enclosing a liquid substance known as the cytoplasm. The
cytoplasm contains solid structures, such as the nucleus, chloro-
plasts (in the cells of green plants), and mitochondria. All these
organelles can be observed under a microscope at only moderate
Order and Disorder in the World of Large Molecules 89

magnification. Great progress has recently been made in eluci-


dating the structures and functions of cell components.
A living cell is rather like a highly complicated and fully
automated factory. It produces substances needed by the or-
ganism and can receive and fulfill commands. To do all this,
the cell requires a supply of energy. The energy is extracted
directly from solar radiation by the chloroplasts, and from food
by the mitochondria. The commands to produce molecules re-
quired by the organism as a whole are issued by the cell nucleus,
which also contains the inheritance material handed down by
the cell to its descendants after mitosis.

Fig. 56. Electron micrograph of a chloroplast (x 90,000)

It would be beyond the scope of this book to consider in any


detail the structure and operation of the living cell, as we are
only interested in the structure of matter from the point of view
90 Order and Disorder in the World of Atoms

of order and disorder, and we shall confine the following dis-


cussion to this aspect.
The drawings of cells in old biology textbooks hardly stirred
the curiosity of a physicist enquiring into the structure of matter.
Before the advent of the electron microscope, capable of im-
mense magnifying power, the cell seemed to be a disordered
aggregation of nearly featureless organelles. Only recently has
a high degree of order been found in the individual cell com-
ponents.
Figure 56 shows the power plant of the cell, converting solar
energy into mechanical, chemical, and electrical work as dic-
tated by the needs of the parent organism. The molecules
transforming light into work belong to the substance called
chlorophyll. The cell contains many such molecules, but their
work would be very difficult-perhaps even impossible-if they
were distributed randomly throughout the cell. The answer
provided by nature is to sandwich the chlorophyll molecules be-
tween alternating layers of lipids and proteins. These are the
layers visible in the photograph. The whole system is composed
of packets of such layers. This is obviously a pretty foolproof
arrangement-if any packet goes out of order, there are many
others to carryon.
Figure 57, resembling at first sight a metallic crystal, is even
more striking. It is in fact a cross section of a lying muscle of an
insect. The muscle appears to be made up of two kinds of fila-
ments (protofibrils); thicker ones, shown as dark spots in the
photograph, .and thinner ones, shown as smaller and lighter
spots. The protofibrils are arranged with a high degree of order,
distinguished by a hexagonal motif, indicating a gas-crystalline
state which was described earlier in this chapter. The axes of
these filaments form a regular lattice, while the azimuths of the
constituent molecules are arranged at random (this has been
demonstrated by physical investigations). There is again a good
reason for this ordered arrangement. According to a widely held
hypothesis, muscle contraction occurs as a result of the sliding
of the thin protofibrils with respect to the thicker ones-one
system of filaments as it were moves into the other. Such a mech-
anism would be clearly impossible in the absence of an
ordered arrangement.
Order and Disorder in the World of Large Molecules 91

Fig. 57. Electron micrograph of a cross section of muscle


fibril (x 250,000)

These two illustrations do no more than touch on the im-


mense subject of the structure of living tissue. Our aim was
only to indicate that the problems of order and disorder in the
arrangement of atoms and molecules have a direct bearing on
biology.
CHAPTER 6

TRANSITIONS BETWEEN ORDER


AND DISORDER

1. Iron vapor and solid air


Although this may seem a strange combination of words, iron
vapor and solid air actually exist in nature. Not, of course, under
what we would regard as normal conditions.
Observations show that the state of matter is fully determined
by only two parameters, namely, temperature and pressure.
Life on Earth exists under reasonably constant conditions.
Atmospheric pressure varies only to the extent of a few percent
about its mean value of about 14 pounds per square inch, and
the temperature, say, in New York, between about 24 and 82F.
On the absolute scale of temperature, this latter variation is also
not more than a few percent. Very naturally, we are so accus-
tomed to these conditions that we often forget to specify them,
and regard certain combinations of terms, such as "solid iron" as
right and proper.
When iron is heated, it will first melt and eventually evaporate.
Conversely, air will first liquefy and finally freeze on intense
cooling. Even if thc reader has never actually seen iron vapor
and solid air, he can probably take it on trust that any substance
at all can be obtained in solid, liquid, or gaseous state (phase)
by changing its temperature. An example of a substance which
everyone must have seen in all three states is of course water.
The question now is, under what conditions does a substance
pass from one state into another? Such phase transitions are in
fact transitions between order and disorder, and the rules which
govern such changes constitute the primary subject of this book.
Consider first of all evaporation, the transition from the liquid
to the vapor state. In brief, evaporation consists of the escape of
a molecule from the liquid surface. Such an event can occur if the
92
Transitions Between Order and Disorder 93

molecule collects sufficient momentum in'the right direction by


collisions. The impetus required for the molecule to break free
of the forces holding it in the liquid must of course be very
powetful. Since increased temperatures correspond to faster
molecular motion, on heating the molecules in the liquid surface
undergo stronger and more frequent collisions, and the evapo-
ration process is accelerated.
Another conclusion, which is fully supported by experimental
observations, is that since the evaporating liquid loses preferen-
tially its fastest molecules, the average velocity of the remaining
liquid molecules is decreased, and this in turn means a drop in
temperature. Thus a liquid can evaporate at a constant tempera-
ture only if heat is supplied to it continuously from the outside.
This is why we feel cold on coming out of a swimming pool-the
evaporating water draws heat from the body, and since the
molecules escape the process continues until we are completely
dry.
Evaporation may be practically instantaneous, as in the case
of ether, or it may take days or even years. It all depends on the
intensity of the thermal motion of the molecules and on the
forces holding these molecules together.
What happens if the evaporation takes place in a closed
vessel, or even in a room with its doors and windows tightly
shut? This too is a situation familiar to everybody-the evapora-
tion slows down and eventually comes to a halt. The point is that
in an enclosed space the evaporating molecules cannot escape
far enough from the liquid surface, and, still undergoing random
collisions, are eventually recaptured (condense). The rate of
condensation increases with increasing amount of the vapor,
and eventually the two opposing processes-evaporation and
condensation-balance out exactly. The system is then said to be
at equilibrium, and no further evaporation can take place-the
liquid and the vapor exist side by side with continuous exchange
of molecules. Another name for this situation is state of saturation.
The degree of evaporation required for the establishment of
equilibrium varies from liquid to liquid. In other words, the
density of the saturated vapor can assume an immense range of
values.
At room temperature, the density of saturated water vapor is
94 Order and Disorder in the World of Atoms

about 0.00002 g/cm 3 ; the corresponding figure for mercury is


0.00000001 g/cm 3 Thus a tightly shut room having a floor area
of 25 m 2 and a height of 4 m can hold 2 kg of water and 1 g of
mercury in the form of saturated vapor. This means that a vessel
holding two liters of water would evaporate to dryness in such
a room, while a water cask would only lose 2 liters by evapora-
tion, independently of its size.
The same argument can of course be used for mercury, except
that the answer strikes a little closer to home. Mercury vapor is
highly toxic, and anyone working with this substance must take
great care not to spill even a drop. Note that it does not matter
whether the spilled amount is large or small-in either case only
1 g of the mercury will evaporate. All this applies of course only
if we postulate a completely sealed room; if we actually spill a
little mercury we need only open the window.
During evaporation in the open air the molecules are not quite
free to escape in any direction from the surface of the evapora-
ting liquid, since the whole process takes place in the terrestrial
gravitational field, which affects equally all matter including the
molecules of the vapor. If we consider large masses of water on
the Earth's surface gravity can no longer be neglected, and in
this sense the behavior of water vapor in the atmosphere resem-
bles the conditions prevailing in a sealed room. Air always con-
tains some water vapor, often at the point of saturation. Clearly,
no water would evaporate from the beaker into such an atmos-
phere.
Let us return for the moment to the case of a closed vessel
in which the water-vapor system has reached equilibrium, except
that now we shall close the vessel not with a stopper but a close-
fitting plunger. Moreover, we shall slowly push the plunger in.
Since liquids are practically incompressible, the compression
proceeds only at the expense of the vapor, so that the latter's
density is increased. However, before the compression the
density of this vapor was already as high as it could be at
equilibrium, and it needs little imagination to see that now the
excess of the vapor must somehow be removed. In other words,
the vapor begins to condense more and more as the space avail-
able to it is decreased.
Continuing to decrease the volume above the surface, we
Transitions Between Order and Disorder 95

could eventually convert all the vapor back into liquid water.
The converse is of course also true. Moving the plunger up we
increase the free space available to the vapor, and we could
evaporate the liquid completely provided the vessel is large
enough. Note that the vapor will be saturated (and therefore its
density will be constant) as long as even a drop of water remains
in the liquid phase, and that only further expansion resulting in
total evaporation will make the vapor unsaturated so that its
density will begin to decrease.
It will be clear that gases can be liquefied by compression at
constant temperature (provided this temperature is below a
certain level).
It has already been said that liquids evaporate faster at higher
temperatures. However, if the evaporation takes place in an
enclosed space, the temperature will give rise to yet another
important effect-the density (and thus the pressure) of the
saturated vapor will become higher. The reason is easy to see.
When a closed liquid-vapor system at equilibrium is heated the
molecules of both phases are accelerated, but the only important
effect is to facilitate the escape of liquid molecules into the
vapor. The vapor molecules condense on contact with the liquid
surface independently of their velocity, and are little affected by
the rising temperature. The saturation vapor density is thus
increased.
The difference between the arrangement of molecules in
liquids and gases is that while liquids exhibit short-range order,
in gases the molecular distribution is fully disordered. On the
other hand, just how fundamental is this difference? It is only
a consequence of a difference in density. When the molecules of
a gas are pushed closer together, order emerges more and more
clearly out of chaos, and we can expect that if the vapor density
approached that of the liquid we would no longer be able to
distinguish between the two phases.
The above situation is easily realized in practice. When a
tightly closed vessel made of thick glass is filled almost com-
pletely with a liquid and then heated, the small free space soon
becomes saturated with the vapor and further heating leads to
an increase of the vapor density above the liquid surface. On
further heating the vapor becomes as dense as the liquid from
96 Order and Disorder in the World of Atoms

which it escaped, and the boundary between the two phases dis-
appears. The temperature at which this phenomenon takes place
varies from liquid to liquid and is called the critical temperature.
Suppose the temperature is raised still further. Is the material
in the vessel a liquid or a gas? If the vessel is closed by a
plunger, and this plunger is now raised, the density decreases
and eventually becomes so low that the substance can no longer
be regarded as a liquid-its molecules are fully disordered. On
the other hand, what happens if the plunger is lowered? We
can compress the gas to a very high density and set up short-
range order in the arrangement of its molecules, but have we
converted it back into liquid? If we try to pour this fluid into a
beaker, we must first raise the piston and release the pressure,
whereupon the "liquid-gas" substance will promptly lose its
short-range order and becomes a gas.
Thus, gases cannot be liquefied in the ordinary sense of the
word by the mere application of pressure above the critical
temperature. This was the reason for the early failures to liquefy
oxygen, nitrogen, hydrogen, and other "permanent" gases, whose
critical temperatures are very low (-146C for nitrogen,
-llSoC for oxygen, and -240C for hydrogen).
The reader has probably noticed that the above discussion
made free use of the terms "gas" and "vapor" for the state in
which the molecules are in disorder. The two terms are more or
less synonymous (we can say that water gas is the vapor of
water), but there is a certain traditional difference. The word
"gas" is usually applied to substances whose critical temperature
is far below the range of temperatures to which we are accus-
tomed, while the word "vapor" is used when we speak of sub-
stances which can also exist as liquids at ordinary temperatures.
The phenomenon of evaporation is not restricted to liquids;
solids too can evaporate, though it is more usual to call this
process sublimation. One of the oldest examples of the evapora-
tion of solids is provided by sublimation of naphthalene (this
property of naphthalene is responsible for its use in mothballs,
since the vapor in very small concentrations is poisonous to the
insects) .
All solids evaporate to some extent, sometimes appreciably
enough to produce an odor, more often imperceptibly, and oc-
Transitions Between Order and Disorder 97

casionally so slightly that even careful measurements fail to


detect the vapor. Such failure, however, reflects only on the
sensitivity of the method.
The very slight degree of evaporation of solids is quite natural.
After all, the molecules are held much more firmly in a solid
than in a liquid. Moreover, the molecular motion in solids is very
ordered and thus considerably less susceptible to random events
capable of ejecting a molecule out of the surface.
The density of a saturated vapor in equilibrium with a solid
decreases with falling temperature, just as it does in the case of
a liquid. Thus, many substances completely lose their odor at
low temperatures. The fact that a solid can evaporate implies
that the reverse phenomenon is also possible. Vapors can be
made to condense into solids. This method is used in the pro-
duction of very pure crystals, for example, by depositing the
vapor on lightly cooled glass.

2. Water-an exception to the rule


The transition from the liquid to the solid state (crystallization)
and the reverse process (melting) involve a fundamental struc-
tural transformation. Melting destroys the long-range order. At a
given pressure, melting occurs at a strictly defined temperature
corresponding to the point at which the motion of atoms or
molecules becomes too extensive for long-range order to be
maintained any longer.
If the supply of heat to the substance during melting is such
that it just balances cooling, the liquid can coexist at equilibrium
with the crystal, and the crystal will neither grow further nor
melt. An example of such equilibrium is a piece of ice floating
in water held very near ODe.
The melting temperature is affected by external pressure, as
a rule becoming higher as the pressure is increased. However,
all rules have their exceptions. One of these is ice, the melting
of which is promoted by a rise in external pressure. This behavior
of ice is connected with another anomaly-ice is lighter than
water, whilst the majority of substances are heavier in the solid
than in the liquid phase. It is clear that increased pressure,
favoring the denser packing of molecules, will make the ice melt.
98 Order and Disorder in the World of Atoms

The anomalous properties of water are of immense im-


portance to life on Earth. Consider what would happen if water
behaved like any other liquid. First of all, rivers would freeze
to the bottom because the ice would sink. Second, the very
existence of rivers would be imperiled, because many rivers
originate in glaciers lying high in the mountains and melting
at the bottom under the pressure exerted by the weight of snow
in the upper layers.
The properties of ice are due to its peculiar structure, which
does not obey the rule of close packing of particles. Therefore,
disturbances of long-range order lead in ice not to the usual
decrease but to an increase in the density. The situation will
become clear on closer inspection of Fig. 19. During the melting
of an ice crystal, water fills the broad tunnels passing through
the structure. Not surprisingly, the density is increased.
Thus (for a given pressure) each substance possesses char-
acteristic temperature ranges at which liquid-vapor, liquid-solid,
and vapor-solid equilibria can be established.
This leads to the interesting question whether there is a point
at which all three states of matter can exist in equilibrium with
one another. The answer is yes; it is called a triple point.
If water and ice are contained in a closed vessel at OC, the
free space above the mixture will become progressively richer
in water vapor (and "ice vapor") until saturation is reached at
4.6 mm Hg. Further evaporation will then cease, in accordance
with the earlier discussion. Ice, water, and water vapor are now
in equilibrium. This is the triple point of water.
All substances exhibit a triple point. For substances which
sublime only very slightly the triple points are at a pressure
practically equal to zero.

3. The growth of crystals


Crystals can be grown from melts, vapors, or solutions. The last
of these methods is the easiest and the most common.
Let us perform another experiment, this time with common
salt. At room temperature (20C), a tumbler of water will dis-
solve only about 70 g of salt, after which any salt added to the
Transitions Between Order and Disorder 99

system will fall to the bottom without dissolution. At this point


the solution is said to be saturated. The solubility varies with
temperature-as everybody knows, most substances dissolve
more readily in hot water than in cold.
Imagine a sugar solution saturated at a temperature of 30C
and slowly cooled to 20C. For every 100 g of water, 223 g of
sugar dissolves at 30C and only 205 g at 20C. The above solu-
tion must get rid of 18 g of sugar on being cooled to the latter
temperature, and it does so by precipitating the excess solid.
One method of obtaining crystals is therefore to cool their
solutions.
Our resources are not yet exhausted. Crystals can be made
to grow simply by preparing a saturated solution at room tem-
perature and leaving it to stand in an open vessel. The reason is
that the liquid evaporates and the remaining solution becomes
more and more concentrated. Since, however, it was already
saturated in the first place, the excess solid crystallizes out.
The initial stages in the growth of a crystal cannot of course
be observed, but it may be imagined that some of the disorderly
moving atoms or molecules of the dissolved substance begin the
process by collecting more or less in the order which will later
characterize the crystal lattice. Such a group, usually called a
nucleus, forms very often if minute foreign particles are present
in the solution. The most rapid and easiest crystallization occurs
if the process is triggered off by dropping a tiny crystal (seed)
of the dissolved substance into the saturated solution. Under
these conditions no new crystallites are found, and only the
seeding crystal exhibits growth.
The growth of the seed crystal is of course the same as the
growth of crystal nuclei. The only advantage is that the seed
"attracts" to itself the material coming out of the solution and
thus prevents the formation of large numbers of nuclei. In other
words, seeding allows us to obtain one reasonably large crystal
in place of many small ones.
Let us now turn to the question of how new atoms or mole-
cules arrange themselves on the surface of a crystal growing from
a solution. Experiment shows that the process consists as it were
of a displacement of faces in a direction perpendicular to them-
selves, so that the angles between the growing faces remain
100 Order and Disorder in the World of Atoms

unchanged. This constancy of the angles is an important char-


acteristic of crystals, and is due to their lattice structure.
Figure 58, taken from Shubnikov's "Formation of Crys-
tals," shows three possibilities in the growth of a crystal of one
and the same substance. In the case shown on the left the num-
ber of faces is preserved. In the middle diagram, a new face

Fig. 58. Left: growth of a crystal with preservation


of the number of faces. Center: disappearance and
reappearance of faces during growth. Right:assump-
tion of the correct form by an initially shapeless
crystal fragment

appears for a while (in the northeast corner) and later dis-
appears. All these features can be seen under a microscope.
It is important to note that the different faces do not all grow
outward at the same rate, so that the fastest ones tend to dis-
appear (like the one in the southwest corner of the center draw-
ing in Fig. 58) and the slow ones grow largest, i.e., are the best
developed. This last point is illustrated in the right-hand draw-
ing in Fig. 58, where, owing to anisotropic growth, an initially
shapeless crystal fragment assumes the same form as the other
crystals. Thus, certain faces develop more extensively than
others and give the crystal its characteristic form.
All the above features of growth apply irrespective of whether
the crystal is grown from a solution, a melt, or a vapor. The
parallel displacement of the faces suggests that the crystal grows
by acquiring new substance in layers, in such a way that one
layer is finished before the next begins. This situation is shown
in Fig. 59 at an intermediate stage at which the new layer is
only half completed. An incoming atom is most likely to attach
itself at site A, because there it is attracted from three sides,
Transitions Between Order and Disorder 101

Fig. 59. Growth of a monatomic crystal. The in


coming atom is most likely to attach itself at A,
and least likely to favor site C. This scheme shows
that crystals grow layer by layer

while it is attracted from only two sides at B and from one side
at C. The usual growth therefore proceeds layer by layer, start-
ing from a "column" one atom or molecule high, which then
spreads in an orderly manner until the surface is covered.
However, suppose that, owing to some random factors, the
new layer begins to grow simultaneously from several sites. We
recall that in the formation of a close-packed assembly of spheres
the spheres of a new layer may have a choice of two kinds of
hollows to occupy. If the preceding layer was of type A, the
next one may be of type B or C. Thus, if the new crystal layer
starts to grow from several positions, it can end up with regular
regions of types Band C and imperfections at the boundaries
between these regions (Fig. 60).
If further layering also proceeds with mistakes, the crystal
will contain whole three-dimensional islands of regular packing,
i.e., it will have a kind of block structure with defects along the
edges. These imperfections are more "serious" than imperfec-
tions in the formation of layers. If the crystallization is slow,
such faulty growth is rare, and is in any case confined to the
less serious type. Each incoming atom manages to find its proper
place. Imperfections arise when the crystallization is very fast.
The above-described formation of islands characterized by
regular spherical packing has been observed in the case of
102 Order and Disorder in the World of Atoms

cadmium iodide. The dense packing is formed by atoms of


iodine, with the smaller cadmium atoms filling the gaps. The
observed imperfections can occur both in the formation of the
dense packing and in filling-in by the cadmiums.

t
Fig. 60. Initiation of a new layer by two atoms occupying at random differ-
ent hollows. Each atom grows its own layer giving rise to two regions with
a "defect" boundary

4. Spiral growth
On closer examination, the picture of crystal growth given in
the preceding section proves to be a little too oversimplified.
Calculations show that atoms are very unwilling to occupy posi-
tionsof the type denoted by C in Fig. 59, and if an atomic plane
is fully built up, the formation of a new layer is very improbable.
Such growth can occur only if the substance surrounding the
crystal nucleus is highly supersaturated. However, it was shown
as early as 1931 that crystals grow fairly fast even at very low
degrees of supersaturation, exceeding the calculated growth
rate by a factor of 101000. This is probably the greatest dis-
crepancy known to physics between theory and experiment.
The unexpected ease with which new layers are formed on
Transitions Between Order and Disorder 103

growing crystals remained a mystery for many years until, in


1945, it was found that rapid growth is possible only if the
crystal contains spiral (screw) dislocations, allowing continuous
formation of new layers by the addition of atoms to sites of

0)

,.
d) ..

Fig. 61. Successive stages in the growth of a crystal with a screw disloca
tion, showing the formation of a spiral ramp

type A. The situation is shown schematically in Fig. 61, in which


the shaded wall of the ramp offers sites very suitable for the
attachment of incoming atoms. During growth, this surface con-
tinuously accepts atoms of liquid or vapor, so that the ramp
moves over the crystal face. This displacement is not parallel to
the original ramp, because the end of the ramp, lying on the
axis of the dislocation, remains stationary, and no atoms can be
deposited at that point. As a result of this the ramp begins to
curve and eventually becomes a spiral, the central part growing
highest above the face of the crystal. The process can continue
indefinitely, with the top of the spiral as it were pulling the
turns upwards and the turns growing out sideways until-one
by one-they reach at the base the boundaries of the crystal
face.
104 Order and Disorder in the World of Atoms

A similar spiral pyramid, though


of different construction, arises if
the crystal possesses two screw dis-
locations side by side, spiraling in
opposite directions. Note that in
this case there are no spiral steps,
and the pyramid is built up of a
oj i
number of closed surfaces. In the

:~\.
~V
c}
Fig. 62. Successive stages in the growth of a crystal with two oppositely
directed screw dislocations

course of further growth each surface extends sideways, and


new surfaces are started continually at the top.
At this point the reader might well begin to doubt whether
these idealized illustrations, such as Fig. 63, have any connec-
tions with reality. However, his suspicions are unfounded. In
fact, Fig. 63 shows photographs of real crystals, in beautiful
agreement with the dislocation theory of crystal growth.
Crystals grow in round spirals or surfaces when all growth
directions in the plane of the growing surface are equivalent.
This situation is encountered in many cases, but is not the only
possibility. On the contrary, it is more natural to imagine the
crystal as growing in rows.
Spiral growth in rows is shown in Fig. 64 on the example of
a growing plane of a cadmium iodide crystal. It is not hard to
guess that the growing plane is close-packed. The four succes-
sive photographs also illustrate clearly the growth of the top of
the spiral. The central stroke in the photographs is the region
of dislocation and the numbers denote growing rows of atoms.
The' height of the step in the spiral can be measured with
'g
::I
II>
a:
::I
..
II>
Q:I

i......
::I
0
a.
!?l
::I
......
52
:g
a.
"',''i- --.~~ ;r I !?l
.1',.", \.
-,., I ;"
~ " "
~.~- ..... /:,
L' ,.
/ /1
a

Fig. 63. Photographs of crystal surfaces, showing


patterns arising by mechanisms illustrated in Figs.
61 and 62: a) Surface of a crystal of silicon carbide
b) spiral growth on silicon carbide (oblique illumina-
tion) c) a crystal of stearic acid (growth around two
oppositely directed screw dislocations)
....
Q
U1
106 Order and Disorder in the World of Atoms

great accuracy, and, in full agreement with expectations, it


proves to be equal to the lattice period in the direction perpen-
dicular to the growing plane (or to a multiple of this period).

Z
1 J

0) c)

2 z
J J

b) d)

Fig. 64. Successive photographs showing the spiral growth


of a crystal of cadmium iodide

A height equal to one period corresponds to a single dislocation.


The fairly high steps occasionally found simply indicate the
presence of several dislocations having the same direction.
We shall now return for a moment to the structures of crystals
with faults in the superposition of layers mentioned briefly in
Section 3 of this chapter. It is not difficult to imagine a regular
crystal constructed on the principle of two or three repeating
layers, e.g., . . . ABAB ... or ... ABCABC ... , and it is equally
easy to accept that disordered layering can readily occur. The
problem is that in some crystals the layers begin to repeat only
at rare intervals, say, after only 10-20 layers. In some specimens
Transitions Between Order and Disorder 107

of silicon carbide the first layer is repeated only after 243 layers.
How, then, does the 244th layer "know" the arrangement of the
first layer? The forces of interaction between atoms cannot of
course operate over such distances.
The answer lies in spiral growth. A series of screw disloca-
tions can result in a spiral step equal to the height of several
layers, and this step constitutes the repeating period of the
structure.
Thus, dislocations control crystal growth and appear in the
course of the growth process. As to their origin, one possibility
is the inclusion of a foreign particle, such as a speck of dust, in a
growing crystal lattice (Fig. 65).

Fig. 65. Crystal growth around a dust particle


(large irregular body)

5. Conversions between crystal structures


It is well known that one and the same structure can give rise
to crystals having completely different structures. Some exam-
ples of this phenomenon were already given in Chapter 2 (gray
and white tin, red and yellow sulfur, two modifications of iron,
graphite and diamond, and so on).
108 Order and Disorder in the World of Atoms

How are these different structures formed, and how can we


obtain one modification rather than the other? We can prepare
crystals by cooling a liquid or providing a vapor with a cold
surface on which to condense, but how do we know which
modification will be formed? It turns out that the modification
depends on the pressure under which the crystallization is
carried out.
Thus, yellow rhombic sulfur is formed when liquid sulfur is
cooled under a pressure higher than 1288 atmospheres, while
the same process at ordinary pressure results in red monoclinic
crystals. An interesting phenomenon occurs at a pressure of
1288 atmospheres and a temperature of 151 ee. Under these
conditions liquid sulfur gives rise to the two kinds of crystals
side by side.
When crystallized from the vapor phase, sulfur gives red crys-
tals at higher pressures and yellow crystals below 1 mm Hg.
Here too there is a point at which the two types are obtained
together, namely, at a temperature of 95.5e.
Similar examples could be given for other substances.
The interesting point about all this is that one crystal struc-
ture can be transformed into the other. However, conversions
of white tin into gray, yellow surfur into red, etc., occur only
under certain conditions. In this respect, the situation is closely
analogous to the process of melting. For red sulfur to change
into yellow at ordinary pressure, the temperature must be below
lOoee. Above this temperature and up to the melting point
the atomic arrangement of the red form is the stable one. As the
temperature falls, the atomic oscillations are reduced, and, be-
ginning at Hoec, nature finds a more convenient type of pack-
ing for the sulfur atoms. This results in the conversion.
Each crystal phase has its range of stability, defined by tem-
perature and pressure. The laws governing the transitions be-
tween different crystal modifications are the same as those of
melting and evaporation.
For any given pressure there is a temperature at which the
two crystal modifications exist in equilibrium with one another.
If the temperature is raised, one type of crystals will change
into the other. The reverse transition will occur if the tem-
perature is reduced.
Transitions Between Order and Disorder 109

This last point needs to be qualified: in some cases solid state


transitions occur easily only in one direction. Thus, diamond is
easily converted into graphite but the reverse process remained
impossible until very recent times. At present the conditions
for the graphite -+- diamond transition appear to have been
found.
Solid state transitions involve a conversion from one kind of
long-range order into another, and their mechanism is of con-
siderable interest. The conversion should take place in such a
way that the atoms move as little as possible from their old posi-
tions, i.e., it should take place with the minimum effort.
The simplest conversions between solid phases of simple sub-
stances occur if the structures of the two phases are of the type
of close-packed assemblies of spheres. Consider a few examples.
Cobalt and thallium can have the structure of the type
... ABCABC ... (cubic face-centered) or ... ABABAB ... (hex-
agonal close-packed). How do the transitions take place? Figure
66 shows once more the familiar characteristics of spherical pack-
ing. The assembly is built up of layers. All possible stackings of
layers (see Chapter 2) reduce to the three illustrated in the
figure with the aid of crosses, circles, and squares (A, B, and C) .

,
I( )( )( It 1/
0 0 0
0 0 0 0 B
x x x
0
t
0
c
0
'0"" 0
tJ C

x/'x )( )(

Fig. 66. Scheme of the conversion of a close-


packed assembly ___ ABCABC ___ into __ . ABABAB
... The three layers are denoted by different sym-
bols

In the ... ABCABC ... packing the arrangement is repeated


after every three layers, and in ... ABABAB ... after every two
layers. How does one system change into the other?
By slightly displacing the layers in the direction of the arrows
in Fig. 66, we can convert layer A into B or C, layer B into A or
C, and layer C into A or B. The transition from ... ABCABC ...
110 Order and Disorder in the World of Atoms

into ... ABABAB . . . stacking, amounts to changing in a series


of steps the first row into the second:
... AD CA BC AB CA BC .. .
... ABABABABABAB .. .

Clearly, out of six layers two can be left undisturbed and the
remaining four require changes: C into A, A into B, B into A,
and C into B.
This can be done in several ways by moving the layers in
various directions, but we must find the way that requires the
least work in the crystal. The conversion would be easiest if
the four layers ... CABC ... could be moved into position
... ABABAB ... in a single displacement, but the figure shows
that this is impossible. Nevertheless, we can save work by mov-
ing the layers in pairs. The arrows leading from a square (C) in
Fig. 66 show three directions along which displacement of two
adjacent layers '" CA. " will convert them into ... AB ...
(squares will become crosses, and crosses of the neighboring
layer circles). The arrows starting from a circle (B), run in
directions opposite to those of the arrows starting from the
square and show three directions along which we can move
... BC ... pairs to convert them into ... AB ... pairs.
Thus, every six layers require two displacements to transform
packing ... ABCABG. .. into packing ... ABABAB . ", and
each of these displacements can occur in anyone of three
directions.
Conversions of this type have even made it possible to grow
a single crystal of the phase ... ABAB ... from a cubic crystal.
Such an operation is generally unsuccessful because the growth
of the new phase begins simultaneously at several centers, and
a microcrystalline system is formed instead of a single crystal.
Most often the crystal simply falls apart after such a phase
transition, but sometimes the old external form is preserved,
even though the material is now microcrystalline.
The reason for this difficulty is that the crystals of the new
phase can begin growing from different sites. In cubic face-
centered packing we can trace out four systems of close-packed
layers. In the crystal shown in Fig. 12 the close-packed planes
Transitions Betwee.n Order and Disorder 111

are perpendicular to the cube diagonals. There are four such


diagonals (because a cube has eight comers), so that a crystal
.. ABCABC ... can give rise to crystals belonging to the type
.. ABABAB ... with four different orientations.
On the other hand, the transitions of cobalt and thallium from
the packing ... ABABAB ... should always lead to the forma-
tion of a new crystal having one and the same orientation. Hex-
agonal close-packed cobalt cannot give rise to an unoriented
microcrystalline material. Most frequently, the transition will
result in a crystal with "faulty" layering, i.e., in a packing of
the type
... ABCABCABCACABCABCABCABABCABCAB '" ..

The study of atomic reshuffles in phase transitions began


with Kurdyumov's work on the conversions of iron and steel.
At high temperatures, iron has the cubic face-centered lattice
.. ABCABC ... , while at low temperatures the structure is
body-centered cubic. The transition from one into the other is
immensely important in practice, and deserves a more detailed
discussion.
The effect of rising temperature is
illustrated diagrammatically in Fig.
67. Close packing according to the
scheme ... ABCABC ... is shown on
the left. On the right, there is a some-

fig. 67. The conversion of iron. The close-packed structure on the left is
the state stable at high temperatures. The diagram on the right is a projec-
tion of a body-centered cubic lattice (compare Fig. 15) along a face diagonal

what unusual view of body-centered packing, shown as a pro-


jection along a face diagonal of the cube.
The two structures seem to have little in common. The main
112 Order and Disorder in the World of Atoms

difference is that the structure on the left is three-layered, whilst


that on the right is two-layered (the small triangles lie on the
second layer). A less important dissimilarity is the difference in
the rhomb angles, though this is not shown in Fig. 67.
As the crystal is heated the atomic vibrations increase, and
when the temperature reaches 906C the less dense body-
centered packing becomes disadvantageous. The two-layered
structure changes into the three-layered one by the alternate
displacement of the layers represented by small triangles in
Fig. 67. For example, layers 1, 3, 5, etc., move to the left, and
layers 2, 4, 6, etc., to the right. This displacement, which pro-
ceeds along the rhomb diagonals, is accompanied by a change
in the rhomb angles.
Unlike the case of cobalt, a single crystal of one type of iron
seldom gives on phase transition a single crystal of the other
type. It is easy to see that in the phase transition of iron the
crystals of the new phase may assume different orientations.
When a single crystal of iron is cooled from the transition point,
the crystallites of the new phase begin to grow in no less than
24 directions. This figure is obtained as follows. The face-
centered cubiC lattice contains four densely packed planes (cf.
Fig. 12) perpendicular to the four
cube-diagonals. Since the trans for- "~/)(I
mation starts at various sites, the... I\ ~x// )(.
reshuffle may start with any of the
close-packed layers. A crystal of / .
/)(~\
~ /
\1
/)(~
~

II \ I
the new phase may thus begin to x~/)(~/x
grow in six directions on each of )( I I \ )(
the four close-packed layers, and )( . /x.~ . )(
so we get 6 X 4 = 24 directions in / ~
all. The six possible orientations are x )(
shown in Fig. 68. Fig. 68. Possible orientation
So far, we have discussed two of growing crystallites of body
centered iron (stable at low
examples of oriented phase transi- temperatures)
tion consisting of the ordered re-
shuffle of atoms, in which the atoms follow suit, i.e., the
rearrangement of the first atom is copied by the second atom,
the third atom, and so on. The question now arises, is this the
rule or the exception? Until quite recent times it was thought
Transitions Between Order and Disorder 113

to be the rule, and even in the last edition of this book the
author believed the above examples to be typical. However,
experiments carried out in the hist few years have shaken this
once Widespread belief to its foundations.
As mentioned above, single crystals generally disintegrate
on phase transformation, so that it is difficult to establish whether
the new phase has the same orientation as the old one. If it does,
the phase transition is clearly oriented and ordered. But what
if it does not?'
Organic single crystals of one phase were converted in the
author's laboratory into single crystals of another phase, under
mild conditions ensured by inserting the single crystal into a
drop of glycerine or a similar substance which formed a protec-
tive envelope around it. This method enabled us to grow a crys-
tal within a crystal, and even to "goad" the transformation in
either direction, shifting the phase boundary in one direction
or the other by increasing or decreasing the temperature. Such
a phase boundary is clearly shown in Fig. 69. However, the most
interesting thing was that once we had prepared several such
phase-in-phase samples, we could subject them to many physical
tests to answer unequivocally the question of orientation corre-
spondence. The unanimous answer now found in several sub-
stances was that there is no connection between the orientation in
the parent phase and that in the daughter phase. Thus, the orien-
tation of the daughter phase is absolutely fortuitous: there is no
predictable connection between it and the parent orientation-
there is no oriented growth. In short, the growth of one crystal
from another follows the mechanism of the growth of a crystal
from a liquid.
The packing in solids is tight, and the mobility of atoms and
molecules is restricted. How then can the nuclei of the new
phase be formed and carryon growing? In the first place, this
mobility is not so small as one might think. Second, the crystals
contain a great many sites-"-voids, fissures, dislocations-where
the particles can easily disperse, and it is exactly on these sites
that the new phase begins to grow. Not much room is required
for an atom or molecule to leave one phase, migrate over a short
distance, and become attached to the other phase.
This is how things stand today. Further investigations on a
114 Order and Disorder in the World of Atoms

Fig. 69. Growth of a crystal within a crystal, as seen un


der an optical microscope. The substance is pdichloro
benzene.
Transitions Between Order and Disorder 115

wider range of substances will show whether oriented phase


transformation is the rule or the exception. At present, it cer-
tainly seems to be rare.

6. Delayed transitions
However quickly or slowly we make a solid approach its melt-
ing point, it can never jump over it and just become a hotter
solid instead of melting. For example, ice cannot be kept over
ODC without melting. However, the situation is different in the
reverse case, i.e., in liquid ~ solid transitions, and we know a
great number of instances in which a liquid is supercooled,
sometimes without any special measures. In fact, the phe-
nomenon can be a nuisance to the preparative chemist who has
cooled a melt far below its melting point and who is vainly
waiting for crystals to appear. Glycerine, for example, greatly
increases its viscosity on supercooling, and can be kept in this
amorphous state for months or years.
However, how do we know that a certain substance should
solidify at a certain temperature? Where do we get this "correct"
freezing point? Perhaps some substances simply crystallize out
when they want to, and not when we think they should? No,
this is not the case. At a given pressure, all liquids possess a
well-defined crystallization temperature. To prove this point, it
is sufficient to bring together a liquid and a crystal of the same
substance and observe what happens. There are three possibil-
ities: the crystal is absorbed by the liquid, the liquid crystallizes
out on the crystal, or the two phases coexist peacefully without
any changes. If the second phenomenon takes place, and the
crystal is seen to grow fairly fast, we know that the liquid is
supercooled. When the liquid is greatly supercooled the crystal
grows at a spectacular rate. Thus, if we drop a snowflake into
supercooled water, ice needles shoot out around it immediately
and the whole volume of water turns into ice within seconds.
Particular interest is attached to delays in crystal-crystal trans-
formations. Unlike the case of solid ;::= liquid transformations,
crystal-crystal transitions can be delayed in either direction.
Thus, yellow sulfur should change into red sulfur at 95.5 C, but
D

on rapid heating it jumps over this transition point and it is yel-


116 Order and Disorder in the World of Atoms

low sulfur that eventually melts at 113C. If we now cool the


melt, red sulfur crystallizes out at 113C; this form can be cooled
to room temperature without turning into yellow sulfur at
95.5C, but the conversion will eventually occur, even though
it may take a day or two.
Again, how do we know that the change should occur at
95.5C? A very simple experiment provides the answer. Tie to-
gether a piece of yellow sulfur and a piece of red sulfur and
start heating. At 95C, the yellow form "devours" the red, and at
96C the red "devours" the yellow. This shows that the red
modification should not exist at 95C; it has wandered over
foreign terrain, is treated as an intruder, and is converted into
the yellow modification. The opposite argument applies to the
temperature of 96C.
In some cases we encounter certain phases at temperatures at
which they are not supposed to exist, and it is important in this
connection to know whether we are dealing with stable crystals.
Thus white tin turns into gray tin at 13C, and yet the white
tin, which is the common modification, exhibits no change in
the winter. In fact, white tin will easily withstand supercooling
by 20--30C. During a severe winter, however, it does begin to
change. This was unfortunately unknown to Captain Scott, and
the belated phase transition of white tin ruined his expedition
to the South Pole in 1912. In this case fuel oil was carried in
containers soldered with white tin which under the influence of
the intense cold turned into gray tin: the containers came apart
and the expedition lost its fuel.
However, as in the case of sulfur, we can determine the
transition point of tin by placing a grain of the gray form on a
piece of the white; the latter changes into the former at a
temperature slightly below 13C.
To understand the causes of delay transitions consider the
difference between liquid - crystal and crystal - crystal con-
versions, on the one hand, and crystal-liquid conversion, on the
other. Transformations from a crystalline to a liquid phase are
accompanied by destruction of long-range order, but liquid -
crystal and crystal - crystal transformations imply the creation
of long-range order from scratch or from another type of order.
Crystals melt without any delay because the destruction or
Transitions Between Order and Disorder 117

order is relatively easy: the atoms simply leave the lattice


one after the other, starting at the surface. The reverse process
is more difficult. Crystallization requires the formation of long-
range order from short-range order. The process starts at the
surface and advances into the substance. The atoms must es-
tablish strict order in a crowded place, which requires coordi-
nated movements. Still more difficult is the task of reorganizing
one long-range order into another, a task that requires the
ordered migration of atoms from one set of sites to another. The
difficulty of this reorganization explains the delays in crystal -+
crystal phase transitions as well as in liquid -+ crystal trans-
formations.
Solid state transitions always begin on grain boundaries, be-
tween blocks, on dislocations and empty sites-wherever it is
easiest. The first step is the most difficult: after even only a few
dozen atoms have occupied sites of the new system the rest
readily follow suit, and the nucleus grows in an oriented manner,
i.e., more and more atoms go over to it from the previous less
suitable order (or, in the case of crystallization, from disorder).
This is how a seed directs the growth of crystals, and this is
why a glass of supercooled water turns into ice crystals when
seeded with a snowflake.

7. Particles do change place in crystals


Recent investigations have revealed that the mobility of atoms
and molecules in solids had been seriously underestimated, and
that these particles may leave their lattice points, force their
way through their less adventurous neighbors, and migrate over
the entire lattice. In fact, if this were impossible, diffusion
through crystals would never be observed. Diffusion implies
penetration, and the diffusion of foreign atoms through crystals
has long been known. Thus, the surface layer of steel can be
saturated with various substances such as carbon, nitrogen, or
boron. When carbon is used, the steel is said to be case-
hardened. In addition to this, there is also the phenomenon of
self-diffusion: iron atoms can diffuse through the iron lattice,
copper atoms through the copper lattice, and so on. It is this
118 Order and Disorder in the World of Atoms

self-diffusion that had until recently remained an unproven


hypothesis.
To show, e.g., that carbon has diffused into steel, all we have
to do is to inspect the sample under a microscope, or carry out a
chemical analysis, or examine those physical properites that are
changed by the presence of foreign atoms. Any of these tech-
niques will readily demonstrate that carbon atoms have indeed
penetrated deep into the steel. It is equally easy to verify
diffusion in other cases. Thus, a great number of silver atoms
penetrate at 200-300C into lead, reaching a depth of a few
centimeters in one hour.
But how can we prove, e.g., that copper atoms diffuse through
a piece of copper, or iron atoms through a piece of iron? The
answer is the use of "labeled atoms." Any substance can be made
radioactive by exposing it to neutron bombardment in a nuclear
reactor, and will then emit characteristic penetrating radiation
detectable by sensitive instruments. If a specimen of, say, radio-
active copper is placed for a time in close contact with ordinary
copper and then removed, it will be found that some of the
radioactivity has been passed on to the once inert metal.
This proves beyond doubt that some radioactive copper atoms
have diffused out into the sample. This technique enables us
to study the self-diffusion of any substance with the same ease
as the diffusion of "foreign" substances.
It is not so easy to explain the mechanism by which the atoms
migrate in a close-packed crystal lattice. The first requirement is
the thermal vibration of the atoms; in the absence of this,
diffusion is absolutely unimaginable. However, we know that the
atoms do vibrate about the lattice points. Thus, when one atom
is "off center" another may slip into its place. All the first atom
can do now is to occupy the previous place of the intruder. The
result is that the two atoms have exchanged sites. Such an ex-
change is of course difficult if only two atoms are involved.
Suppose we are trying to push our way through a crowded sub-
way car (i.e., to diffuse through the crowd). If only our im-
mediate neighbors are considerate enough to give way we shall
not get very far. However, if most passengers are willing to do
the same, they only have to move a little, and we can pass
Transitions Between Order and Disorder 119

through the car from end to end. It seems probable that this is
what happens during diffusion through solids: an exchange of
sites between two atoms involves a whole group of neighbors.
The diffusing atoms can proceed only when the other atoms of
the medium move out of their way by virtue of thermal vibra-
tions.
There is no doubt that all types of imperfections (dislocations,
holes, fissures) in the crystal play an important part in diffusion.
Thus, the presence of holes enables atoms to advance through
the lattice step by step. If the diffusing atoms are small, the
presence of holes is in itself sufficient to permit diffusion, without
the lattice atoms having to exchange positions: the small diffu-
sing atoms just skip along the holes and the close-packed ar-
rangement remains intact.
Diffusion is a two-way process. When a zinc plate and a cop-
per plate are clamped tightly together, zinc atoms penetrate the
copper and copper atoms penetrate the zinc. However, the
diffusion may be much faster in one direction than in the other.
Diffusion through crystals depends on many factors, but we
shall only mention that the process is fastest when the diffusing
atoms differ in all respects from the lattice atoms. It is as if the
lattice atoms were most anxious to send on their way those in-
truders which are least like them. Consequently, self-diffusion is
the slowest process, followed by diffusion between elements in
the same column of the periodic system.
Since imperfections facilitate the journey of atoms through
the lattice, diffusion is faster in metals subjected to deformation.
While diffusion of atoms has long been known, the movement
of large molecules through crystals is a recent discovery. It has
been shown by a very interesting new techniques, called nuclear
magnetic resonance spectroscopy, that as the temperature is in-
creased the molecules often change their oscillatory motion into
progressive motion, abandon their positions, assume new ones,
and advance along fissures changing sites and orientations.
Crystal defects probably play a fundamental role in molecular
diffusion, but oddly enough the positional exchange of large
molecules can take place to some extent even in ideal crystals.
We have thus seen that, in one way or another, particles in
120 Order and Disorder in the WlIrld of Atoms

crystals are fairly free to move and undergo rearrangements.


Thus we should not be surprised by the large number of transi-
tions from one solid phase into another, and a similarly large
number of chemical reactions in the solid phase. The movement
of atoms and molecules in crystals is of course not some kind
of whim, but a physical necessity governed by physical laws.
What these laws are, and the drive behind this motion, will be
described in the followfng chapter.
CHAPTER 7

ORDER OR DISORDER?

1. Probability and disorder


If molecules are not subject to external forces and there are
practically no cohesion forces, their distribution will fall into
ideal disorder. While bonding forces are easily annihilated by
heating, melting, or evaporating the substance, external forces,
and particularly gravitational ones, are difficult to eliminate.
Difficult, but not impossible. Thus, the force of gravity has very
little effect in a thin vertical layer of gas, and so the particles in
such a layer are characterized by ideal disorder. The reason for
this is that in the absence of all forces the arrangement of par-
ticles will be decided by chance, as a result of which, the most
probable type of spatial distribution will prevail-total disorder.
This will be demonstrated by the following argument.
Suppose we are to distribute six animals (say, a dog, a cat, a
hare, a rabbit, a fox, and a hedgehog) among three cages, plac-
ing them in two at a time. The first cage can be filled in many
ways-we can put in the dog with the cat, the dog with the hare,
the cat with the hare, and so on. This gives 15 different arrange-
ments. When the first cage is filled, there are only four animals
left to choose from as we come to the second cage, and the choice
is simplified. If the first cage contains, for example, the dog and the
cat, into the second one we can put the hare with the rabbit, the
hare with the fox, the hare with the hedgehog, the rabbit with
the fox, the rabbit with the hedgehog, or the fox with the hedge-
hog. There are thus only six arrangements. The total number of
variations is therefore 15 x 6 x 1 = 90 (the number 1 means that
after we have filled two cages there are only two animals left,
and these go into the third cage without any choice) .
There are thus as many as 90 ways of distributing uniformly
only 6 subjects among 3 sites. If we are not frightened of large
121
122 Order and Disorder in the World of Atoms

numbers, it is easy to scale these calculations up. Thus the


number of equivalent ways in which 1000 grains can be distrib-
uted over 100 squares is given by 1 followed by thousands of
zeros.
Considering that a cubic centimeter of gas contains a billion
billion molecules, we can see that the number of equivalent ways
of distributing them over the available space is beyond imagina-
tion. We call these equivalent arrangements, because one is as
good as any other, and just as likely. However, when there is
some bias in the distribution, a form of order will emerge. The
number of these preferential arrangements is considerably
smaller than the number of completely random ones, and the
more preferential the arrangement (i.e., the greater the degree
of order), the fewer the number of ways in which it can be
brought about.
Returning to the above example, there are only 16 arrange-
ments (instead of 90) if two of the cages are to hold only one
animal each and the third cage is to hold four. If we concentrate
900 out of 1000 grains on one square and distribute the re-
mainder singly, then the number of ways in which this can be
done is still very large, but much smaller than it was previously
(1 followed by hundreds of zeros, instead of thousands).
In the case of a billion billion molecules the difference be-
tween a uniform and a nonuniform distribution will be even
sharper. From the above-numerical examples it follows that if
the molecular distribution is governed by chance then the most
"common," most easily realized, and most probable distribution
is one fully isotropic and characterized by uniform density. In
other words, the most probable distribution is an ideally dis-
ordered one.

2. The tendency toward disorder


We have thus seen that if the molecules are left to their own
resources and are not subject to any influence interfering with
their thermal motion, then their most probable arrangement is a
disordered one.
Does this mean that spontaneous deviation from disorder is
unlikely to occur? Does it follow that there is an underlying
Order or Disorder? 123

tendency toward disorder? Yes, indeed it does. To appreciate


this issue, consider two questions.
The first one is, can we freeze water by heating it? Naturally
not. But why not? This question seems nonsensical at first sight,
but only at first sight. We see in every phenomenon a manifesta-
tion of the general laws of nature, which govern all events in the
physical world that surrounds us. Which law is it then that
forbids the spontaneous freezing of water on heating? Perhaps
it is the law of the conservation of energy? No, this could
actually be obeyed in our preposterous case. We may thus
imagine a vessel filled with water and placed on a large metallic
plate heated to 300C. Eventually the block reaches 400C and
the water freezes. The law of conservation of energy is not
necessarily violated, since it could be said that the water has
given off heat and the block has absorbed it. We must there-
fore find another law forbidding the above event.
Let us examine the mechanism of heat transfer between mole-
cules. The molecules are known to move faster in a hot body
than in a cold one. When two substances at different tempera-
tures are placed in contact, the slower "cold" molecules will
collide with the faster "hot" ones, as a result of which the molec-
ular velocities of the two groups will be equalized after some
time. This is supported both by theory and by experiment.
Now consider the molecular state of these substances before
and after the equalization of the temperature. If black and
white balls form a disordered pattern in a box, then the probabil-
ity of finding a white ball on a given site is the same as the
probability of finding a black one. However, we have seen that
we must always ask "order or disorder with respect to what?"
In the present case we are thinking of order and disorder with
respect to the distribution of the molecular velocities. From this
point of view, the presence of faster molecules in one region of
space and of slower molecules in another does not constitute
disorder. In contrast, a disordered arrangement is one in which
we find fast and slow molecules with equal probability in any
region of space.
Thus, two substances in contact at different temperatures do
not represent a disordered arrangement of the particles from
the viewpoint of the molecular velocities, and we must conclude
124 Order and Disorder in the World of Atoms

that the transfer of heat from a colder to a hotter body con-


stitutes a transition from disorder to order. We have seen that
a disordered arrangement is the more probable one, and there-
fore a transition from disorder to order would mean transition
from the more probable to the less probable. For this reason,
such processes generally do not take place.
The second question is, can a flywheel gather speed on its
own? No, of course not. It must be driven by energy. Well then,
suppose that the flywheel is in a room and the two constitute an
engine: the temperature of the room drops and the wheel
rotates. This is not contrary to the law of conservation of en-
ergy since the mechanical energy of the rotating wheel does
not come from nowhere: it comes from the thermal energy of
the room.
The impossibility of obtaining mechanical work by cooling the
surroundings is not at all obvious, and many hopeful inventors
have wasted much time trying to devise an engine driven by
energy derived from cooling the oceans. This is a great pity, for
we could thus obtain enormous amounts of energy. If we could
only cool by one-thousandth of a degree centigrade the roughly
one billion cubic kilometers of water on the Earth's surface
and utilize the heat thus liberated, we would obtain a million
billion kilowatt-hours of energy, sufficient at the present rate of
consumption for about 1000 years!
However, this cannot be done. Heat does not spontaneously
change into mechanical energy, because this is an improbable
process. It is improbable, because it means a spontaneous change
from the disordered thermal motion of the particles in the
medium into the fully ordered mechanical movement of a
machine. And this just won't happen.
If we empty a sackful of white spheres and a sackful of black
ones into a box, and shake well, the spheres will be intermingled
in the way in which thermal motion mixes atoms and molecules.
After a while, a handful of spheres withdrawn from the box will
show roughly equal numbers of black and white ones. Thus, the
order of the two separate sacks has been converted into disorder.
However long the box is shaken, the two colors will not unmix.
The more or less uniform distribution of the two colors is the
Order or Disorder? 125

more stable state. In the case of atoms and molecules, such a


state is called thermal equilibrium.
The tendency toward disorder explains many of the phe-
nomena discussed before, notably the diffusion processes. It is
the tendency toward disorder that causes the molecules from
the lump of sugar in a cup of coffee to move upward, although
they are heavier than water, and become evenly mixed with the
water molecules. It is the same tendency that forces the atoms
of zinc and copper plates fastened together to interpenetrate.
If we ignore this law of nature, we cannot explain the phe-
nomena of phase transitions and the stability of phases. When
no force acts on the molecules, disorder will prevail. If the mole-
cules can assume various arrangements, then, other conditions
being equal, there is a preference for that arrangement which
enables the thermal motion to exert itself and which thus helps
the realization of disorder.

3. The tendency toward order


How would atoms and molecules be arranged in substances in
the absence of thermal motion? This is not a purely speculative
question, since thermal motion does stop at absolute zero, and
another way of phrasing it is, what is the atomic arrangement at
a temperature of absolute zero?
The characteristics of the world of atoms are dictated to a
great extent by the presence of thermal motion. If thermal mo-
tion were absent, the laws of the microcosm would resemble
those of the macrocosm. Therefore let us for the moment turn
to the latter and examine the conditions under which large
bodies are in a stable state of equilibrium.
Equilibrium prevails when the sum of the forces acting on
the body in question is zero. A body is at equilibrium when the
force pulling it to the right is equal to the force pulling to the
left, and when the pressure acting on it downward is the same
as the reaction of the support acting on it upward.
However, not all equilibria are stable-try to place a ball on
a curved lampshade or stand an egg upright on the table.
Theoretically, this is possible. In practice, such unstable equili-
bria exist only momentarily.
126 Order and Disorder in the World of Atoms

When is an equilibrium stable? As regards the example of


the ball, the answer is simple: it must be put in a hollow. The
reason for the stability of this particular position is that to dis-
place the ball it must then first be raised, and this would neces-
sitate performance of work to overcome the gravitational force.
This means that the sphere will not spontaneously leave the
hollow, since the work needed for this is not available and it
cannot be conjured up. Thus, the stability of the ball in a hollow
is guaranteed by the law of conversion of energy. If we analyze
any stable equilibrium, we always find the same reason for its
stability: an appreciable amount of work is required to disturb
the equilibrium.
A suspended pendulum is at equilibrium, since to deflect it
work must be expended against the force of gravity. A suitcase
lying on a seat in a train represents another stable situation. The
seat is pressed down under the weight of the suitcase, and to
change the latter's position we must perform work against
gravity or against the springs in the seat, depending on whether
the suitcase is to be lifted or pressed down.
The degree of stability of an equilibrium differs widely from
case to case. Thus, the ball is at equilibrium both in a shallow
and in a deep hollow, but its displacement from a shallow hollow
requires very much less work. The degree of stability is here
measured by the depth of the "well" accommodating, the body.
Figure 70 shows two spheres joined r--\A ~
by a spring and a rubber cord in such ~
a way that the spring is slightly
Fig. 70. Equilibrium be
compressed and the rubber slightly tween two spheres
stretched. If this system is in equilib-
rium, the two opposing forces just balance out. The equilibrium
is stable, and any departure from it (additional stretching of the
rubber or additional compression of the spring) requires work
applied from the outside. The degree of stability of this equilib-
rium depends on the strengths of the rubber and the spring.
The work of rolling the sphere out of the hollow in the first
example now corresponds to the work of rupture of the rubber
cord.
Another common property of bodies or groups of bodies in
stable equilibrium is that these bodies can oscillate about their
Order or Disorder? 127
equilibrium position. Such oscillations can be brought about by
slightly pulling the pendulum out of the vertical position, by
giving a fillip to the ball in the hollow, or by compressing the
system shown in Fig. 70. In the absence of friction, the result-
ing vibration would go on as long as we liked.

Fig. 71. Close packing in a layer-the spheres


are at equilibrium positions

Consider a system consisting of a large number of spheres,


springs, and rubber cords; each sphere has six hooks, so that it
can be connected with springs and rubber cords to other
spheres (Fig. 71). The arrangement shown in this figure repre-
sents the stable equilibrium position of these spheres. The cen-
ters of the spheres are disposed in space in the same way as in
a close-packed assembly.
It is so evident that stable equilibrium leads in this case to
long-range order that no rigorous proof is needed. If all the
spheres, springs, and cords are identical, then any deviation
from the regular arrangement shown in Fig. 71 requires work
to be expended on compressing the springs or stretching the
cords.
Now we are ready to return to the world of atoms. Atoms too
are subject to attractive and repulsive forces (cord and springs),
128 Order and Disorder in the World of Atoms

except that these forces are not mechanical, but electrical. We


cannot elaborate on this here since the matter cannot be ex-
plained in a few words, and would be beyond the scope of this
book, so suffice it to mention that attractive forces get the upper
hand when the interatomic. distance is relative large (when as
it were the rubber cord is stretched and the spring is not com-
pressed), and repulsion forces begin to predominate as the atoms
approach one another. Somewhere between the two extremes
there is a position in which attraction just balances out repulsion,
and there the system is at equilibrium.
Matter consists of an immense number of atoms; which par-
ticular atomic arrangement is stable? The most stable arrange-
ment, as we have already seen with the aid of our sphere model,
is the spatial distribution of atoms in accordance with long-
range order. The tendency toward order is a manifestation of the
same law by which the ball rolls down a hill and comes to rest
in the valley. The tendency toward order is the tendency toward
achieving a stable equilibrium position.

4. The struggle between order and disorder


We have already seen in the previous sections that disorder in
the spatial arrangement and in the velocity distribution of atoms
and molecules represents the most probable state of affairs. This
indeed is the case until forces acting on the particles enter the
field. The action of forces is directed at establishing order. If
the atoms or molecules are in thermal motion, and if forces are
acting upon them, the disordered arrangement will no longer
representthe most probable state of affairs.
Innumerable examples could be quoted to illustrate this strug-
gle between order and disorder. We must conclude that we are
confronted with a law of nature in the form of a compromise
between two opposing tendencies, i.e., between a tendency to-
ward order (achievement of a stable equilibrium), on the one
hand, and a tendency toward disorder (achievement of the most
probable distribution characteristic of particles in thermal mo-
tion ), on the other.
A simple example is the distribution of molecules in a vertical
Order or Disorder? 129

column of air. In the absence of thermal motion the tendency


toward equilibrium would force the molecules to the ground.
In the absence of gravity the density would be uniform (fully
disordered arrangement). In fact, the atmospheric pressure
(and therefore also the density of the air) "decreases with in-
creasing altitude, being only half the sea-level value at an alti-
tude of 5600 m. This is of course the compromise referred to in
the preceding paragraph.
It is easy to understand from this point of view why the most
probable state of affairs is sometimes disorder (gases), some-
times short-range order (liquids), and sometimes long-range
order (crystals )-all depending on the conditions.
At high temperatures the particles move fast, and the forces
of interactions between them do not noticeably affect the spatial
arrangement. As the temperature is lowered, the thermal motion
becomes less violent and, at a certain point, cohesive forces be-
gin to collect the particles into drops. The most likely arrange-
ment now is one with short-range order. Further cooling leads
to a point at which the vibrations of the particles are so slow
that the formation of a regular lattice is possible. It is now the
arrangement with long-range order that represents the most
probable state.
Transition from the liquid to the solid state can be visualized
with the aid of the following model. Imagine a box with small
hemispherical cavities containing one sphere each. The number
of cavities is equal to the number of spheres, and the cavities
are set out in regular rows to form a plane lattice.
When the box is subjected to continuous shaking, the spheres
perform small vibrations, leave their sites, and generally give
the impression that is typical of liquids. As the shaking is grad-
ually slowed down, a time comes when the spheres can no
longer get out of their recesses. Without changing the intensity
of the shaking (constant temperature), we eventually arrive at
a point when all spheres are back in their recesses (a crystal is
formed).
If the shaking is reduced suddenly, then only a few spheres
end up in their cavities. No crystallization takes place, and we
end up with a liquid-type picture characteristic of amorphous
materials.
130 Order and Disorder in the World of Atoms

Thus, when the box is shaken only gently the most probable
state is an ordered arrangement of spheres, i.e., a crystal lattice.
If the recesses are shallow, then the picture in which some of
the spheres roll around without settling in the recesses may per-
sist for any length of time. In this situation we see order con-
taining elements of disorder. If, therefore, the probabilities of
ordered and partly disordered states differ little from each other,
order and disorder will coexist.
How can two phases of the same substance coexist at equilib-
rium? Consider, for example, the equilibrium between a crystal
and its saturated vapor. The crystalline state with its long-range
order is stable, and work is required to detach atoms from the
lattice and bring them out into the apparently less stable gaseous
state. And yet, atoms do escape from the lattice and form a
gaseous cloud around the crystal. The factor compensating for
the lower stability of the gaseous state is that the tendency to-
ward disorder is ideally satisfied only in this state. The tendency
toward order finds its best realization in crystals, which at the
same time frustrate the tendency toward disorder, for the atoms
are tightly arranged and their movement is curtailed. By con-
trast, the gaseous state offers the best opporunities to the tend-
ency toward disorder: each particle is given ample scope, and
thermal motion can fully assert itself. When the "sums" of order
and disorder in both phases are the same, the two phases are in
equilibrium.
We have already seen that the pressure of a saturated vapor
varies with temperature. The lower the temperature the smaller
is the pressure, i.e., the smaller is the vapor's density. As the
density decreases the volume of space allotted to each atom or
molecule is increased, and so is the degree of disorder in the
vapor. Since the crystal does not contract appreciably as the
temperature is decreased, the volume of space allotted to each
atom or molecule-and the degree of disorder-remains much
the same. On the other hand, there is an increase in the degree
of stability of the crystal, i.e., in the tendency toward order. The
lower the temperature, the greater is the work required to de-
tach atoms or molecules from the crystal.
As the conditions of equilibrium between the crystal and the
Order or Disorder? 131

saturated vapor are changed, a series of compromises are set up


between order and disorder: nature balances out a higher dis-
order in one phase with higher order in the other.
What is the position as regards phase transitions in the solid
state? Using again the model of the box with cavities and
spheres, we note that the difference between two phases lies
primarily in the difference between the depths of the cavities.
One of the phases is more stable than the other.
But why are there two phases at all? Here, too, the stability
(i.e., the tendency toward order) is balanced out by the tendency
toward disorder (i.e., by possibilities for thermal motion). At
high temperatures disorder predominates, and the most probable
state of affairs is represented not by the deepest but by the
widest cavities, for the atoms or molecules need more space to
move. This means that the solid phase more probable at the
higher temperature is the one in which the atoms are bound
more weakly and can vibrate more intensely. When the cavities
are at the same time wide and deep, both the desire for stability
and the necessity for thermal motion are satisfied in a "package
deal." The need for two modifications is thus abolished, and the
same phase will be stable under all circumstances. In the case
of phase transitions, the cavities on one side of the transition
point are narrow but deep, those on the other side being wide
and shallow. When the greater stability of one phase (tendency
toward order) is balanced out by the greater possibilities for
thermal motion (tendency toward disorder) in the other phase,
the two phases are in equilibrium. If we now increase the tem-
perature, the atoms or molecules opt for the state that offers the
better opportunities to realize the tendency toward disorder, one
phase is fully abandoned and he other fully embraced, and one
solid phase is transformed into the other. Conversely, as the
temperature is reduced below the equilibrium value the atoms
or molecules choose the state that offers the greater opportunities
to realize their tendency toward order, which now gets the up-
per hand, and the particles assume the positions in the low-
temperature modification.
We have now finished our journey through the phenomena of
order and disorder in the structure of matter. We can conclude
132 Order and Disorder in the World of Atoms

that, just as the spectrum contains all the colors from red to
violet, so does nature exhibit all shades of order and disorder,
ranging from the ideal order of perfect crystals to the ideal
chaos in gases.
INDEX

Anisotropic bodies 2 -,calcite 14


-,distributions 2 -,elementary cell of 14
Annealing 35 -,growth of 98
-,lattice structure of 10
Belov, N.V. 22 -,liquid 42
Binary alloys 64 -,magnesium 19
Block structures 49, 50 Curie point 75
Bovine tendons, molecular Cytoplasm 88
structure of 33
Bragg, W.H. 15 Diamond, structure of 26
Bragg, W.L. 15,22,51 p-Dichlorobenzene 114
Bragg's method .51 Dislocations, edge 5.3
-,screw 5.3
Cadmium iodide 102 -,theory of 51
Calcium carhonate 14 Disorder 1
Camphor 47 -,characteristics of 2
Carhon, amorphous 42 -,ideal, criteria of 2
Carbon dioxide, crystals of 29 Distributions, disordered 2
-,molecules of 29 -,ordered 2
Case-hardened steel 117 -,uniform density 2
Chlorophyll 90 Domains, defined 72
Closest packing, rule of 24 -,formation of 72
Collisions, random 5 - ,magnetic 73
Conductivity, n-type 62 -,detection of 73
-,p-type 62
Electrical insulators 77
Copper-gold alloy 69
Electron gas 23
Couplings, ball and socket 79
Electrostatic repulsion 24
Critical temperature 96
Envelopes, nonspherical 25
Crystalline iodine,
Equivalent arrangements 122
structure of 28
Crystallography, structural 15 Face-centered cuhic structures
Crystals 14 20
133
134 Index

Fedorov, F.S. 15 -,space 16


Federov's theory 15 Laue, M. 15
Ferrites 77 Liquid-gas substances 96
Ferromagnetic properties 72 Liquids, structure of 36
Ferromagnetism, appearance
of 72 Magnetic order 71
-,domain theory of 72 Manganese oxide, lattice of 77
Free electrons 23 Matter, gaseous state of 2
- ,behavior of 23 Maxwell, J. 8
Frenkel's oriented melting 48 Maxwell's theoretical
calculation 8
Gas-crystalline state 48 Mercury, structure of 25
Gases, thermal motion in 5 Metals, divalent 23
Germanium lattice 61 -,monovalent 23
Graphite, structure of 26 Microcrystalline bodies,
Gray selenium 28 order in 33
Molecular vibrations,
Haiiy, Pere 14
amplitude of 45
Ice, crystal structure of 27 Molecules, density of in gas 4
Iodine crystals, molecular -,kinetic energy of 6
nature of 30
Ions, cadmium 22 Naphthalene, sublimation of
-,chlorine 15, 22 96
-,sodium 15, 22 Nitrogen molecules, velocity
Iron-cobalt alloy 67 distribution for8
-,structure of 68 Nuclear reactor, neutron
Iron, conversion of III bombardment in 118
-,Curie point of 75 Paraffins, normal 84
-,domain structure of 74 Particles, geometrical
-,vapor 92 arrangements of 2
Isotopic order, concept of 71 Pauling,L.22
Isotropy, defined 2 Plastic deformation,
Konobeevskii, S.T. 34 concept of 45
Polyethylene, dissolution of 85
Laminar minerals 70 -,regular crystals in 85
Lattice deformations 60 Polymers 78
Lattices 15 -,bundle, behavior of 82
-,crystal, model of 16 -,linear 84, 86
-,invisible 1.5 -,structure of 86
Index 135

Polythene 78 Stearic acid, crystals of 105


-,films 83 Structural analysis, x-ray 15
Polyvinyl alcohol, molecules Sublimation, defined 96
of 79 Symmetry, mirror plane of 11
Probability, theory of 7
Taylor-Deliger theory of
Protofibrils, arrangement of 90
dislocations 53
Quartz, amorphous 40 Texture, appearance of 34
-,crystalline 40 Thermal equilibrium 125
-,structure of 40 Tobacco mosaic virus 87
Transitions, delayed 115
Resins, formaldehyde 80
-,phase 92
Scott 116 Translations, primitive 16, 18
Self-diffusion, phenomenon of
Uranium, structure of 26
117
U spenskii, N.E. 34
Shearing force 57
Shubnikov, A.V. 10, 100 Viruses 42
Soap molecules, arrangement Viscosity 45
of 44
Wallpaper patterns, symmetry
Solid state, amorphous 39
of 9, 10
Spectroscopy, nuclear
Water-vapor system 94
magnetic resonance 119
Whiskers, copper 59
Spheres, close packing of 17
-,crystals of 17 X-ray diffraction by crystals
State of saturation 93 14

Вам также может понравиться