413

T H E T H I R T Y - N I N T H T H O M A S HAWKSLEY L E C T U R E

The Mechanism of Work-hardening of Metals

By Professor N. F. Mott, M.A., D.Sc., F.R.S.*
The most striking feature of the deformation of metals is the formation of slip lines. Recent investiga-
tions suggest that, when formed at low temperatures, each slip line is the result of a displacement of
the material along a single lattice plane through a distance of about a thousand atomic diameters.
Moreover, there is much evidence that the steps on the surface which appear as slip bands attain
their full height in a small fraction of a second, though their length may thereafter increase slowly.
At higher temperatures and at slow rates of strain the slip bands appear, under the electron micro-
scope, as clusters of lines about a hundred atomic diameters apart. The origin of slip lines, the reason
for this clustering and the cause of work-hardening are discussed. The two conceptions used in the
discussion are the dislocation line and the vacant lattice site.
Slip lines are believed to have their origin in a certain arrangement of dislocation lines of frequent
occurrence in the interior of the crystal. These are known as Frank-Read sources; their relation t o
recent work on the growth of crystals is shown. Where a slip line terminates dislocations must re-
main in the crystal; to the internal strainsround these is ascribed work-hardening, much as in Taylor’s
theory of 1934f. It is now, however, possible to explain what it is that stabilizes the dislocations and
prevents them from moving back when the stress is released.
Finally, vacant lattice sites are shown to be formed in a cold-worked material. If the temperature is
high enough for them to diffise, they soften the material round the slip band and allow the observed
clusters to form. They also play a part in producing the observed “fragmentation” of cold-worked
material.

INTRODUCTION recently, it has received experimental proof from several

My aim in this lecture is to present a model of what happens sources (Smigelskas and Kirkendall 1947; Correa da Silva
when a metal work-hardens. I shall try to base my model as and Mehl 1951 ;Barnes 1952; Nowick 1951), of which the most
much as possible on experiment, but none the less it will con- striking is perhaps some recent results obtained at Harwell by
tain a number of unproven hypotheses, and may be wrong in Adam, Green, and Dugdale (1952). The workers at Hanvell
certain points. My object in putting forward an unproven introduced vacant sites and, in addition, interstitial atoms into
theory is to stimulate experiment; that is the function of theories, metals by fast electron bombardment. On applying this treatment
and it is in the unproven state that they are most interesting; to a partially ordered copper-gold alloy of the type Cu3Au,
once they are shown to be right or wrong they can be relegated they found that the electrical resistance decreased. They inter-
to the text-book or to oblivion. preted this result as showing that the vacancies allow diffusion
The theory will be based on the assumption that the crystal-
line structure of a metal is marred, not only by obvious and
visible defects such as grain boundaries, but by defects of two
other types. These are (1) vacant lattice sites, and (2) disloca-
tions.
My first task will be to explain what I mean by these, and to
give experimental evidence that they exist.

VACANT LATTICE SITES

By a vacant lattice site I mean a point within the crystalline
structure of a metal from which the atom is missing. Such sites
can be envisaged to be formed at moderately high temperatures
at the surface of a crystal, at points such as A in Fig. 1, and to
migrate inwards. In a crystal in thermal equilibrium at a finite
temperature, a number of such sites must exist. If W is the
energy necessary to form a vacant site, the proportion of such
sites will be e-WIkT.
It is believed that self-diffision in metals, and indeed diffision Fig. 1. Vacant Site in Metal
of substitutional impurities generally, takes place through the
agency of vacant sites. The vacant sites wander about in the If W is the amount by which the energy is increased when a vacant
lattice, and an atom will change places only when it is next to site is formed, the proportion of vacant sites is r W , k T . At the
melting point this is 10-5-10-6, or less than 0.001 per cent of all sites.
one. This hypothesis was put forward first by Johnson (1939);
The M S . of this lecture was received at the Institution on 8th to take place at, or slightly above, room temperature, so that
August 1952. For the Minutes of the meeting in London, on 14th the degree of order can increase, and the resistance consequently
November 1952, at which this lecture was delivered, see Proc. decrease.
1.Mech.E. (A), vol. 166, p. 402.
* Director of the H. H. Wills Physical Laboratory, University of The importance of these considerations for the subject of this
Bristol. lecture is, as I shall show, that when a material is cold-worked,
t An alphabetical list of references is given in the Appendix. a large number of vacancies are formed so that self-diffision

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016

414 THE THIRTY-NINTH THOMAS HAWKSLEY LECTURE
is allowed at temperatures where this normally would not working produces vacancies ; the lower of these temperatures
be possible. So in examining self-diffusion, and the phenomena is therefore of importance in the theory of cold-work. Below it
which depend on it, such as recovery, there are two ranges the vacancies cannot move. We believe that a large part of the
of temperature to consider. enhanced electrical resistance of a metal cold-worked below this
First there are the temperatures which are high enough for temperature is due to vacancies, which increase the resistance
vacancies to be formed and to move about. At these tempera- in the same manner as would impurities. Fig. 2, due to Wilson
tures atoms can change places; self-diffusion is possible. At a and Thomassen (1934), shows how this increased resistance
lower temperature vacancies formed by some non-thermal diminishes, on heating, long before any recovery of the mechan-
process such as neutron bombardment can move, but new ones ical properties occurs.
will not be formed. Diffusion is, then, possible only if vacancies
are formed in some such way. At lower temperatures diffusion
is in no case possible. DISLOCATIONS A N D SLIP BANDS
The second type of imperfection is what we call the disloca-
tion. But before introducing this, I should like to remind you of
TABLE
1. APPROXIMATE
VALUES OF TEMPERATURES
Ti AND T2 what is known about the mechanism of slip in metals. In
mechanical deformation at ordinary speeds and moderate or
Metal ~ TI(deg. C.) ~ Tz (deg. C.) low temperatures, the deformation is all, or nearly all, con-
centrated in slip bands. I would like to say something about the
Copper 1 - 40 I 410 geometry of these slip bands. This is shown schematically in
I
‘
Lead 1 -130 150 Fig. 3, which represents a crystal under tension as shown by
Iron 80 I 800
300 f
Platinum 160 1,000
Nickel 120 800
--O I
Above TIvacancies can move if formed by neutron bombardment
or cold work, above T2 they are formed thermally, so that self-
diffusion and recovery are possible.
Table 1 shows the two temperatures, as closely as I have been
able to estimate them, for a number of metals. Later in this
lecture I shall propose-following Seitz (1952)-that cold- u Complete slip b Incomplete slip-screw dislocation

INDUCTION
I /
-(7
g &-_- - - --

- --A

c Incomplete slip-edge d Deformation during growth

HARDNESS dislocation
Fig. 3. Types of Slip in Crystals
R S ’ Screw dislocation. R‘Q’ Edge dislocation. AHCD Top surface.
EFCB Side surface.

the arrows ; PQ being the slip direction and PQRS the active
slip plane. Complete slip can be envisaged-as in Fig. 30 in
which the whole cross-section of the crystal has slipped-or
incomplete slip-as in Fig. 3b and c, in which the region that
has slipped is bounded by a line such as R’S’in Fig. 3b, or
R ’ Q in Fig. 3c. These lines are centres of strain in the crystal;
we call a line of the type S’R’ a screw dislocation and a line of the
type R’Q’ an edge dislocation.
Usually the term dislocation is used for the lines bounding a
plane that has slipped through only one atomic distance. The
slip lines observed in crystals are the result of slip through about
a thousand times that distance. The lines bounding such planes
I shall call dislocations of high order, or “piled up groups” of
dislocations.
Fig. 4 shows a representation of the position of the atoms in a
dislocation of an edge type. A point that should be emphasized
is that a plane of atoms terminates on a dislocation line, as in
Fig. 4.
Recent work with optical and electron microscopes has done
much to elucidate the nature of these steps on the surface of
deformed metals. According to Heidenreich and Shockley (1948),
and particularly to Brown (1949, 1951, 1952), at low enough
temperatures (-180 deg. C. for aluminium) the steps are
200 600 single and of height about 2,000 Angstrom units, which
ANNEALING TEI :RATURE-DEG. C.
corresponds to a displacement of about a thousand interatomic
Fig. 2. Effect of Annealing Temperature on Various distances. I n cubic metals, at any rate, the complete step is
Properties of Nickel formed in a very short interval (less than 0.01 second), and does
Annealed hourly. not thereafter increase (or decrease under reversed stress). The

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016

 THE MECHANISM OF WORK-HARDENING O F METALS 415
end of the step (point S‘ in Fig. 3) may, however, move slowly moderate temperatures, the region in the immediate neigh-
across the crystal as the stress is increased (Chen and Pond bourhood of a plane that has already slipped is particularly
1952). In terms of dislocations this would mean that screw dis- likely to slip again; thus, under the electron microscope, the
locations can move slowly but that edges normally move rapidly. “cluster” of slip bands illustrated in Fig. 5 is seen. Slip bands
do not, however, necessarily extend right across a surface ;they
sometimes terminate; and the line (within the crystal) on which
4 A k -
0 0 0 0 0 0 0 0 ~ 0 0 0 0 0 0 0 0 the slip terminates is called a dislocation.
The study of these steps on the surface of crystals, by means
O O O O O O O O O O Q O O O O O O of the optical and electron microscopes, forms one of the most
rewarding branches of solid-state physics. I t gives information
0 0 0 0 0 0 0 0 ~ 0 0 0 0 0 0 0 0 not only about slip, or plastic deformation, but also about the
0 0 0 0 0 0 0 0 ~ 0 0 0 0 0 0 0 0 growth of crystals. In growth from solutions or from vapour we
have found that crystals normally grow by rotation of the step
00000000 00000000
00000000 00000000 8 shown in Fig. 3 about the point where it is anchored to the screw
dislocation. Atoms coming in from the vapour or solution adhere
most easily to the step and, as they adhere, the step must move
0 0 0 0 0 0 0 0 ~ 0 0 0 0 0 0 0 0 forward, rotating about its fixed point. Each time it goes round,
0 00 0 o-6o-o~ob-o 00 00 a new layer is formed. As it rotates, it curls up into a spiral.
Spiral markings have been observed on a number of crystals
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 (Frank 1952), and in our laboratory it has been possible to make
0000000000000000 a film of the rotating spiral. We have distinguished cases where
the height of the step is just one lattice parameter, and others
0000000000000000 in which it is several hundred times as high, as are the steps
0000000000000000 observed after plastic deformation. The techniques by which
these steps are rendered visible are interesting; but I cannot
0000000000000000 go into that subject here. I would, however, emphasize that
0 0 0 0 0 0 G 0 0 0 0 0 0 0 0 0. crystal growth is associated with the movement of steps, while
plastic flow involves the movement of the dislocation-r, on
-+I- the surface, the movement of the point (S’ in Fig. 3) where the
Fig. 4. An Edge Dislocation (Seitz 1952) dislocation ends. If, therefore, on the surface is seen a pattern
X is the Burgers vector.
like that in Fig. 6, this must mean that the crystal has suffered
plastic deformation after growth has ceased. More complicated
The dislocation can be regarded as formed by adding the vertical patterns will occur if the crystal has been deformed during
dotted half plane or by subtracting the correspondinghalf plane in the growth. Now the study in our laboratory of growth steps on
lower half of the crystal. natural beryl crystals (Griffin 1952) has shown that that actually

-18ooc. 20oc. 150OC.

v
‘
, 2,000 A

€ N 0.05
I I I I I
E N 0.1
I l l 1I I I I l
E N 0.2
llllll IIIIll I/ 1 1 I/ Il
N o 4 1 / 1 II I I /II II Il /I Il 1 1 II Il l l Iliu Il1
u Schematic representation of fine structure. b Comparison of development at different temperatures of deformation.
Fig. 5 . Slip Bands (Brown 1952)

Slip bands formed at somewhat higher temperatures-for does occur; so these apparently non-ductile materials can be
example, room temperatures for aluminium-when examined ductile and, at any rate, have been so during their history.
under the electron microscope by means of oxide replicas, are I would like to show you now what happens to a step on the
shown to consist of a cluster of steps each of about the same surface of a crystal which terminates at both ends (Fig. 34.
height as that formed at low temperatures, separated by about
200 A, as illustrated in Fig. 5 (Brown 1952). The higher the tem-
perature and the lower the rate of deformation, the greater is the
number of such steps in each band. This has been ascribed
by Brown (1952) to an annealing which takes place after
slip in the immediate neighburhood of a slip band. I believe
that this is associated with the movement of the vacancies
formed by slip, in a way that will be described later in the
lecture.
The experimental facts about slip can thus be described as
follows : plastic deformation, at any rate at low temperatures, is
a succession of very rapid events, each lasting a small fraction of
a second. In each of these events, slip, over about a thousand
atomic distances, occurs on a given slip plane and extends over
an area of the order of one square millimetre. When this area is Fig. 6. Growth Steps on a Surface Where Slip has Occurred
bounded by a free surface a slip band is observed there. At After Growth is Complete

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016

416 T H E T H I R T Y - N I N T H T H O M A S H A W K S L E Y LECTURE
This can be understood from Fig. 7 ; by growing outward the
step can generate a complete ring, and, in doing so, reconstitute
itself. At Bristol University we have been able to photograph
the occurrence of this process on cadmium iodide, and take a
film of the process.
I: 1 dish Laboratory (Lomer and Nye 1952). Fig. 3 of their paper
a b C
e distances unless a stress were applied, as in Fig. 86, large enough
d
Fig 7. Crystal Growth on a Surface With Two Dislocations,
or the Action of a Frank-Read Source
PLASTIC DEFORMATION A N D WORK-HARDENING
After this digression on crystal growth, I must return to my
main theme, plastic deformation and work-hardening. The first
task of a theory must be to explain why slip occurs only on
selected planes that are at a distance of a micron or so from one
another, and why slip, especially in single crystals, occurs for
such low stresses. I think it is fair to say that this is now com-
pletely understood in terms of a model due to Frank and Read
(1950). This model has been discussed in a number of articles
and lectures, and I do not propose to spend much time on it
here. But it depends on the assumption that metals, whether
they are polycrystalline or single crystals, contain disloca-
tions; in an annealed material these are perhaps cm. apart;
and Frank and Read showed that with very reasonable assump-
tions regarding the arrangement of the dislocation lines, they
would act as sources-about 1012 of them per cubic centimetre
-from which slip could start and spread out. I n order to see
how this can occur, let us imagine that the line E F of Fig. 7 is a
dislocation line which at points E and F leaves the plane of the
paper. An arrangement of dislocation like this is what we call a
Frank-Read source. When a stress is applied to the material the
line will move in exactly the same manner as the step did; but
now the line is a dislocation line, that is to say, a line that
separates the region that has slipped from the region that has
not. The region that has slipped is shown shaded in Fig. 7.
The “source” can thus generate a ring and, indeed, an indefinite
number of rings, without being itself destroyed. Slip therefore
can start at these points, spread over the whole slip plane, and
continue until the displacement on the slip plane is indefinitely
large. Moreover, the stress necessary to make the process work
can be calculated and is about Gb/Z, where G is the shear modu-
lus, b the lattice parameter, and I the distance between the
two ends of the dislocation line. If I is taken to be 10-4 cm., this
gives values near to the yield points of single crystals of metals.
The model, then, appears satisfactory, provided we accept
the hypothesis that metal crystals, in the annealed condition,
contain dislocations. But this hypothesis is really not new, it is
very old. On the basis of evidence from the intensities with
which X-rays are reflected from crystals, it has for a long time
been accepted that most crystals are not perfect, but have some
Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016
sort of mosaic structure. But recent analyses (Read and Shock-
ley 1950, 1952; Shockley 1952) have shown that any mosaic
boundary, or boundary between two grains at which the change
of orientation is small, can, by geometry, be resolved into a row
of dislocations. This is particularly well shown by the “bubble
models” developed by Bragg and his colleagues in the Caven-
shows well how a mosaic boundary splits into a row of disloca-
tions.
The sources of slip lines are thus tolerably well explained;
the bulk of the lattice has the theoretical strength, but there are
sources-about 10’2 of them in a cubic centimetre-at which slip
will start when a much lower stress is applied; and once it starts
it will spread rapidly over the whole slip plane, unless something
stops it.
So far I have been dealing with ideas which, at any rate for
the last year or two, have been common knowledge among
metal physicists. Now I would like to turn to more speculative
matters-namely, how does slip on these planes increase the
hardness of the material? On this subject, I would remind you
first of Taylor’s famous model of work-hardening, which he
introduced in 1934 (Taylor 1934). Taylor introduced the hypo-
thesis that a cold-worked material contains dislocations, that
round each dislocation the material is elastically strained, and
that these strains hinder a dislocation from moving. This
model was generalized by Mott and Nabarro (1948), who
developed a theory according to which the high yield of an age-
hardened material was due to internal strains round the pre-
cipitates. The theory suggested that if a material contained
internal strains as in Fig. 8a, homogeneous over a few hundred
atomic distances perhaps, no dislocations could move over large
4
P
+
a Internal strains b Stress applied
Fig. 8. Internal Strains in a Metal Crystal
to ensure that the strain throughout was everywhere of the same
sign. This hypothesis concerning hardness can be applied to
Frank and Read’s model; unless the strains are as in Fig, 86,
none of the sources will generate slip bands.
STRAIN HARDENING
I will now outline the quantitative theory of strain hardening,
which is the main subject of this lecture. I believe that Taylor’s
model is essentially correct. Hardening is due to dislocations-
or rather to the strains round dislocations. But the dislocations
are not arranged at random throughout the solid in the way
that Taylor postulated; they are arranged, rather, at the ends of
d i p bands. It is frequently observed that slip lines terminate as
in Fig. 9; they do not always run to the edge of a crystal; and,
as I have already emphasized, at such points a dislocation of high
strength must enter the crystal.
Let me try to deduce a stress-strain curve from this hypothesis.
Suppose that the magnitude of slip on each active plane is n
atomic distances. Then the strain, at a distance r from the end
of one slip band is of the order,
u N nbGlr,
where G is the shear modulus and b the interatomic distance.
So if x is the distance between slip lines and L the mean length
of a slip line, the mean internal strain may be written as
u = nbG/v‘(sL) .. . . * (1)
 THE MECHANISM OF WORK-HARDENING O F METALS 417
This, according to my hypothesis, is the flow stress of the material regions do not, when the applied stress is released, go back to
for a given state of strain. The strain, on the other hand, is their original positions. The formation of sessile dislocations
E = nb/x . .. , .. (2) . gives, I think, the best explanation of this. After all, in these
localized regions where the internal stress is high, it is most
x can be eliminated between these two equations and
probable that slip on more than one set of planes will occur, so
g=J(F), . . . . . . (3) there is ample opportunity for sessile dislocation to form.
A fairly definite model of strain-hardening at low tempera-
obtained. tures can then be given. As the stress is increased, slip spreads
This begins to look like a parabolic stress-strain curve. It out from certain points in the crystal, which we call Frank-Read
would be so if I could prove that n/Lwere constant-indepen- sources. If it can spread right across a single crystal, little or no
dent of L and independent of the applied stress. I think this can hardening will occur. If, on the other hand, slip is stopped by a
be proved if certain hypotheses are made concerning the nature of grain boundary or a “sessile” dislocation, an intense centre of
internal strain is formed there. This is locked in position by the
-L- formation of further sessile dislocations. The random strains so
formed within the material are directly responsible for work-
J. hardening.

-----i-
Fig. 9. Idealized Slip Lines on the Side Surface of a Crystal,
and the Meaning of the Quantities x and L
the Frank-Read sources. I do not want here to go into the details*,
but the basic idea is something like this. Once a source has
initiated slip, it goes on generating “dislocation rings” according
to the mechanism ihstndted in Fig. 7. It will go on doing so
until the stress set up at the strained region at the end of the slip
band is big enough to stop it. Now clearly the bigger the value
of L, that is, the bigger the slip distance, the bigger must be the
distortion at the end to stop a source from acting in the middle.
This is why n is proportional to L. A detailed analysis gives
T H E R M A L S O F T E N I N G OR RECOVERY
I would like now to say a few words about the nature of
thermal softening or recovery-the softening of a work-hardened
metal that occurs before recrystallization begins. The clue to
understanding of this is provided by recent studies of the
phenomenon called “polygonization” (Calm 1951, Guinier and
Tennevin 1951, Crussard and others 1951). Briefly, the nature
of this phenomenon is as follows :-
When a metal crystal is bent, and then held for some time at
a sufficiently high temperature, it forms blocks-which are
relatively unstrained-separated by boundaries as in Fig. 10.
Now this cannot occur by a process at all similar to slip. T o see

where Y is the yield point of the unstrained material; in other

words. Y is the stress (of order Gbll)required to make the sources
act. Such a formula gives a stress-strain curve very much of the
observed type, since d / G is about 0.01, and values of o/G of
about 0.01-a hardening by a factor 100-are obtained in
single crystals for strains of the order of unity.
We now have to ask, what determines the slip distance L-or,
and it comes to the same thing-what determines the amount
of slip in each slip plane? We need in our theory some obstacle a As bent. b Annealed.
beyond which slip cannot extend. In polycrystalline materials
the grain boundaries may play this role; but in cubic metals Fig. 10. Polygonization of a Bent Crystal
single crystals harden in very much the same way as do poly-
crystalline macerials. Therefore, in single crystals, barriers to this, let us consider the arrangement of atoms in the bent
slip must exist. I think that the phenomenon of “easy glide” crystal. It will contain a number of the groups of dislocations
providcs the clue to this. Thanks to the work of Andrade and already discussed, which will be found at the extremities of
Henderson (1951), Masing and Raffelspeiser (1950), Lucke and slipped planes which terminate in the material. The form of one
Lange (1952), and others, it seems to be well established that of these is illustrated in Fig. 1la. After polygonization, however,
very pure single crystals of most metals can be deformed to a
considerable extent without appreciably hardening. Moreover,
when this occurs the X-ray asterisms associated with groups of
dislocations are not found, so that it may be assumed that the
dislocations escape out of the crystal. In other words, all slip
bands extend across the crystal, and do not terminate within it.
Moreover, Lucke and Lange have shown that this will happen
only if the crystal is so oriented that slip can occur only on one
set of planes. This, I think, shows clearly the nature of the
obstacles. When slip on two inclined sets of planes is occurring
at the same time, it is possible that the dislocations meet. Now
theoretical work (Lomer 1951 and Cottrell 1952) has shown that
when dislocations on inclined planes meet, they can join up and
form what we have come to call a “sessile” dislocation. This is a
dislocation which is unable to move. We believe, then, that in
addition to grain boundaries these sessile dislocations are the
obstacles that stop the glide on a given plane from proceeding
right out of the crystal. They doubtless play another important
part, too. It has never been immediately obvious, with any dis- a A piled-up group of dislocations b Suggested form of piled-up
in a work-hardened group after thermal softening.
location theory of plastic flow, why the whole process should
not be reversible i or why the dislocations in the highly strained ~ . Processes Responsible for Polygonization and
i 11.~ The
* A detailed discussion is given by Mott (1952). Recovery

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016

418 THE THIRTY-NINTH THOMAS HAWKSLEY LECTURE
move a vacancy. It was 1-1 eV. The activation energy was
the dislocations will be arranged as in Fig. 12, a configuration
measured also, for recovery of conductivity after cold-work.
of lower energy. But they can reach this configuration only by
This was found to be the same, suggesting that here also the
moving out of their slip planes. And t h i s they can do only by
resistance was due to vacancies.
absorbing or giving off vacant lattice sites. Polygonization, then,
A most interesting effect due to these vacancies is the way they
involves the formation of vacant lattice sites and their move-
give rise to local annealing and local polygonization. This can
ment through the lattice, and its activation energy will be the
same as for self-diffusion. occur if the temperature is high enough for them to move ;and,
thus, well below the temperatures for self-diffision and recovery.
Developing a suggestion due to Brown (1950), I have suggested
(Mott 1952) that the clustering of slip lines shown in Fig. 5 is
due to a purely local annealing round each slip band, which is
brought about by these vacancies. At low enough temperatures
this will not occur, and so there is no clustering. Particularly
wide clusters are observed in fatigue, where the specimen is
subjected to stress for a long time (Brown 1952).
More important perhaps is the block structure or crystallite
formation on a very small scale, which is now established to be
a consequence of cold-work (Gough and Wood 1936, 1938).
This is now believed to be the result of very fine scale polygon-
ization of bent regions, the polygonization being allowed owing
to the presence of the vacancies which are formed by the cold-
work.
APPENDIX
REFERENCES
ADAM,J., GREEN,A., and DUGDALE,
R. A. 1952 Phil. Mag.,
vol. 43, p. 1216.
ANDRADE, E. N. DA C., and HENDERSON, C. 1951 Trans. Roy.
SOC.,vol. 244, p. 177.
BARNES, R. S. 1952 Proc. Phys. SOC.By vol. 65, p. 512.
BROWN,A. F. 1949 Nature, vol. 163, p. 961.
1951 J1. Inst. Metals, vol. 80, p. 115.
1952 Advances in Physics, vol. 1, p. 427.
CAHN,R. W. 1951 Advances in Metal Physics, vol. 2, p. 151.
b Method of joining,
CHEN,N. K., and POND, T. A. 1952 Phys. Review (in press).
a Two grains with a CORREA DA SILVA,L. C., and MEEIL, R. F. 1951 Trans. h e r .
common crystal axis. and dislocation model.
Inst. Min. Metallurgical Eng., vol. 191, p. 155.
Fig. 12. Boundary Between Two Polygonized Regions COTTRELL, A. H. 1952 Phil. Mag., vol. 43, p. 645.
Showing how the boundary is formed of dislocations. CRUSSARD, C., AUBERTIN, F., JACOUL, B., and WYON,G. 1951
Advances in Metal Physics, vol. 2, p. 193.
DUGDALE, R. A. 1952 Phil. Mag., vol. 43, p. 912.
Thermal recovery, we believe, is a process of just this type. FRANK, F. C. 1952 Advances in Physics, vol. 1, p. 91.
The hardness of a work-hardened material is due to groups of FRANK, F. C., and READ,W. T. 1950 Phys. Review, vol. 79,
dislocations of the type illustrated in Fig. l l a . The material v . 922.
will soften if they separate from each other, as in Fig. llb. GOUG<H. J., and WOOD,W. A. 1936 Proc. Roy. SOC.,vol.
This they can do through the movement of vacancies, so soften- 159, p. 510.
ing can occur when self-diffusion is possible. 1938 Proc. Roy. SOC.,vol. 165, p. 358.
GRIFFIN,L. J. 1952 Phil. Maa., vol43, D. 827.
CONCLUSION GUINIER,A.-, and TENNEVIN; J. 195i- Advances in Metal
T o end my lecture, I would like to review the evidence that, Physics, vol. 2, p. 177.
when a metal is cold-worked, vacancies are formed in the lattice, HEIDENREICH, R. J., and SHOCKLEY, W. 1948 Bristol Confer-
as well as the piled-up groups of dislocations to which I have ence on the Strength of Solids, p. 57 (Phys. Soc.,
already referred. Some of the evidence for this has been pre- London).
sented by Seitz (1952) and some mechanisms suggested, which I JOHNSON, R. P. 1939 Phys. Review, vol. 56, p. 814.
shall not discuss here; it is as though a moving dislocation left LOMER, W. M. 1951 Phil. Mag., vol. 42, p. 1327.
vacancies behind it as a sort of debris. One of the most interest- LOMER, W. M., and NYE,J. F. 1952 Proc. Roy. SOC.A.
ing points is that the enhanced electric resistance which results LUCKE, K., and LANGE, H. 1952 Z. Metallkunde, vol. 43, p. 55.
from cold-working is almost certainly due to these vacancies; MASINC,G., and RAFFELSPEISER, J. 1950 Z. Metallkunde, vol.
they have just the same effect on the resistance of a pure metal 41, p. 65.
as have impurities. This is why a cold-worked metal recovers MOLENAAR, J., and h n ,W. H. 1950 Narure, vol. 166, p. 690.
its original conductivity, wholly or partly, after annealing at a Mom, N. F. 1952 Phil. Mag., vol. 43, p. 1151.
much lower temperature than is necessary for thermal softening MOTT,N. F., and NAEARR~, F. R. N. 1948 Bristol Conference
(Molenaar and Aarts 1950; Seitz 1952). For thermal softening on the Strength of Solids, p. 11 (Phys. SOC.,London).
the metal must be heated to a temperature which allows vacancies NOWICK,A. S. 1951 Phys. Review, vol. 82, p. 551.
o be formed thermally, while to get rid of the vacancies formed 1951 J1. Applied Phys., vol. 22, p. 1182.
by cold-work, the metal need only be hot enough to allow them READ,W. T., and SHOCKLEY, W. 1950 Phys. Review, vol. 78,
o mme. Table 1 shows these temperatures for various metals. p. 275.
These ideas have received very direct experimental support 1952 “Imperfections in Nearly Perfect Crystalsy’ (John
from some further work of Dugdale (1952). A platinum specimen Wiley, New York).
was subjected to bombardment by fast neutrons in the Harwell SEITZ,F. 1952 Advances in Physics, vol. 1, p. 43.
pile ;an increase of resistance was measured which could be due SHOCKLEY, W. 1952 Solvay Conference Report, p. 431.
only to the vacant lattice sites and interstitial atoms which are SMIGELSKAS, A. D., and KIRKENDALL, E. 0. 1947 Trans. Am.
formed when an atom is knocked out of position by a neutron. Inst. Min. Metallurgical Eng., vol. 171, p. 130.
On heating, the resistance recovered and, by measurements at TAYLOR, G. I. 1934 Proc. Roy. SOC.A, vol. 125, p. 362.
two different temperatures, the activation energy was deter- WILSON,J. E., and THOMASSEN, L. 1934 Trans. Am. SOC.
mined. This activation energy must represent that required to Metals, vol. 22, p. 769.

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 9, 2016