1 s2.0 0502820558900054 Main

3
THERMODYNAMICS AND KINETICS OF

MARTENSITIC TRANSFORMATIONS*
Larry Kau]mant and Morris Cohen$
1. INTRODUCTION
THe. term "martensite" was first used in honour of Professor A. Martens
to identify the plate-like microconstituent in hardened steel. Subse-
quent x-ray measurements b y ~ and CA~PB~.LL~1~ showed this
structure to be body-centred tetragonal. I n 1946 T ~ o ~ o and
GRE~r~G~R ~2~pointed out that the hardening reaction in steel had much
in common with the shear-like, diffusionless transformations in other
alloy systems. Accordingly, the meaning of the word "martensitic"
was revised to describe this t y p e of transformation rather than the
character of the transformation product. In a practical sense, the steel
hardening reaction and the properties of the body-centred tetragonal
product have continued to enjoy great technological importance, b u t
we shall be concerned here with the broader usage of the term "marten-
sitic" as designating a class of solid-state transformations.
Within the past decade, noteworthy progress has been achieved in
the understanding of martensitic transformations. B y comparing these
reactions in various systems (including pure metals), it has become
possible to strip away the details that lack generality and thereby
arrive at the essential features. For example, just as high hardness
and body-centred tetragonality had to be discarded as necessary
properties of the transformation product, it is now evident that
athermal kinetics and insuppressibility must likewise be removed from
the definition of a martensitic transformation. While most martensitic
reactions do display athermal characteristics, starting at a well-defined
temperature (even during drastic quenching) and continuing only while
the temperature is falling, isothermal martensitic transformations
have now been discovered in several cases, both with and without
prior athermal transformation. The isothermal component is suppres-
sible b y rapid cooling, and dropping temperature is obviously not a
requirement for the progress of this reaction.
* This review was conducted as a part of a research programme being sponsored at
the Massachusetts Institute of Technology by the Office of Naval Research under
Contract No. N5ori-07896.
t Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, Mas-
sachusetts.
Department of Metallurgy, Massachusetts Institute of Technology, Cambridge,
Massachusetts.
165
PROGRESS IN METAL PHYSICS
The use of single crystals of the parent phase and the resulting
formation of martensite in large units have permitted quantitative
determinations of the kinetic and crystallographic aspects of the
transformation to a degree not yet attained with most other solid-state
reactions. Of particular significance are the observations on reaction
rates, lattice relationships, habit orientations, and macroscopic dis-
placements. The fact that martensitic reactions m a y occur near or
below room temperature often facilitates the measurements and
enhances the experimental accuracy. In addition, the absence of
diffusion processes allows a simpler theoretical treatment of the
nucleation kinetics than is possible in the case of precipitation and
eutectoid decomposition which are complicated by changes in com-
position. In many ways, the thermodynamic analysis of martensitie
transformations is likewise more straightforward than that of other
solid-state reactions, and furthermore it can be extended down to
subatmospheric temperatures where diffusion processes become
inoperative.
Several reviews are already available ¢S-6~ that deal with the details
of specific martensitic transformations; hence no attempt will be made
here to cover the ground in that way. Recently, the crystallographic
aspects were treated comprehensively b y BrLBr and Cm~ISTIA~, ~ and
the reader is referred to their excellent paper for that part of the
subject. The object of the present paper is to offer a comparable
review of the thermodynamics and kinetics of martensitic reactions.
Table 1.1 contains a list of the symbols employed in subsequent
equations. Whenever a symbol is not defined in the text, Table 1.1
should be consulted for its meaning.
T A B L E 1.1
List of Symbols
c¢' martensite
ferrite, or the stable low-temperature phase
austenite, or the parent phase
F~ chemical free energy of austenite, cal/mol
F ~' chemical free energy of martensite, cal/mol
A F ~'-~ difference between chemical free energy of austenite and
martensite ( F y -- F~'), cal/mol
To temperature at which A P ~'-r ~ 0, i.e. the metastable
equilibrium temperature between austenite and marten-
site, °K or °C
M8 martensite-start temperature on cooling, °K or °C
A8 austenite-start temperature on heating, °K or °C
166
MARTENSITIC TRANSFORMATIONS
TABLE l.l--(O~.)
highest temperature at which the austenite-to-martensite
reaction is induced by plastic deformation, °K or °C
A~ lowest temperature at which the martensite-to-austenite
reaction is induced by plastic deformation, °K or °C
X atomic fraction of alloying element
difference in free energy between f.c.c, and b.c.c, iron,
cal/mol
difference in enthalpy between f.c.c, and b.c.c, iron, cal/mol
FR free energy of mixing of austenite, cal/mol
AF~ difference in free energy between the f.c.c, and b.c.c.
modifications of component A, cal/mol
AF~ difference in free energy of mixing between austenite and
martensite, cal/mol
X~ equilibrium composition of ferrite coexisting with austenite
at any temperature, atomic fraction
equilibrium composition of austenite coexisting with ferrite
at any temperature, atomic fraction
difference between the heat of solution of component A in
austenite and in martensite, cal/mol
activity coefficient of component A in austenite
activity coefficient of component A in martensite
activity coefficient of iron in austenite
activity coefficient of iron in martensite
AF* difference in free energy between ordered and disordered
arrangement of carbon atoms in iron, cal/mo]
AS~'~ difference in entropy between austenite and martensite,
cal/mol °K
AS 7 difference in entropy between f.c.c, and b.c.c, iron, cal/mol
°K
AF~e-~.)~.) difference in free energy between f.c.c, and b.c.c, iron due
solely to thermal (non-magnetic) effects and to the
enthalpy difference at O°K, cal/mol
AFye(Mag) contribution to A F t e r arising from magnetic changes in the
ferromagnetic b.c.c, iron on heating, cal/mol
AFter[T] value of A F ~ v at °K, cal/mol
y weight per cent of alloying element
167
I~ROGRESS Ilg M E T A L FHYSICS
T A B L E IAw(Contd.)
AT(~s) change in the "magnetization uncoupling temperature"
produced b y the solution of one weight per cent of
alloying element in iron, °K
AT(~urt) change in M 8 temperature (AMs) produced by the solution
of one weight per cent of alloying element in iron, °K
0~ Curie temperature of ferrite, °K or °C
A/~7~ difference in free energy between f.c.c, and b.c.c, nickel,
cal/mol
AH r-~' heat liberated b y austenite-to-martensite reaction, cal/mol
L force, dynes
s stress or pressure, dyne/cm 2
a cross-sectionai area, cm ~
Al change in length, cm
V volume, cm 3
AH heat liberated b y austenite-to-martensite reaction, cal/cm 3
transformation strain in axial direction of specimen
p density, g/cm 3
AG~'-~ nonchemical free energy difference between austenite and
martensite, cal/mol or cal/particle*
A W¢'~ net free energy difference between austenite and martensite,
cal/mol or cal/particle*
AG8 surface energy contribution to AG~'-*~, cal/mol or cal/
particle*
AGs strain energy contribution to AG~'-~, cal/mol or cal/
particle*
r radius of lenticular or ellipsoidal embryo, cm
c semi-thickness of lenticular or ellipsoidal embryo, cm
A r AG,
elastic constant defined b y c" V ' dyne/cm2 or erg/cm 3 or
cal/cm 3
0 elastic constant defined as A/4, dyne/cm 2 or cal/cm 3 or
erg/cm a
a specific interfacial energy of the austenite-martensite inter-
face, erg/cm 2
* The same symbol is employed here for the different qu~ntitles cal/mol and eal/
particle. However, the text a n d equations clearly specify which quanti~y is intended
in each instance.
168
MARTENSITIC TRANSFORMATI01WS
TABLE 1.1--(Con~.)
enthalpy of austenite, cal/mol
enthalpy of martensite, cal/mol
AHI AH~'-~ at 3/,, cal/mol
molar volume, cmS/mol
~o macroscopic shear strain of the austenite-martensite
reaction
eo macroscopic dilatation of the austenite-martensite reaction
angle between stress axis and normal to habit plane, degrees
AW. activation energy for the formation of a martensite embryo
of critical size, erg/event or cal/(mol of events)
,xw , at 0°K
v,, volume of embryo of critical size, cmS/embryo
rgc radius of embryo of critical size, cm
0
r~ at 0°K
C~ semi-thickness of embryo of critical size, cm
initial nucleation rate of isothermal martensitic reaction,
particles/cm3 sec
characteristic frequency of lattice vibration, sec-1
Avogadro's number, tool -1
k Boltzmann's constant, erg/°K
R gas constant -----N0k cal/mol °K
AW~ increase in free energy attending the removal of all carbon
atoms from a region of average concentration in steel, ergs
B activation energy used in Orussard's theory of martensite
formation (analogous to AW~), erg/event
h Planck's constant, erg sec
N~IF number of embryos per unit volume having free energies
between F and F -5 gF, cm-a
~,W.(T) experimental activation energy of nucleation for isothermal
martensiZic reaction at °K, cal/(mol of events)
shear angle in the reaction-path theory, radians
¢o observed macroscopic shear (tan -1 Y0)
Q activation energy for diffusion, cal/(mol of events)
d spacing between screw dislocations in Frank's model of the
austenite-martensite interface, cm or A
~2 169
TABLE 1.1--(Contd.)
a~, lattice parameter of austenite, cm or
angle between [110] and [55~]
Ag value of AG~"~/V~, cal/cm a
Agmi. minimum value of 5g, eal/em 3
G shear modulus, dyne/era ~"
# Poisson's ratio
Fs strain energy per unit length of dislocation, erg/cm
b Burger's vector of dislocation, cm
ro radius of dislocation core, cm
r1 radial extent of "active" dislocation strain field, cm
tan -1 (b/d), radians
r0 embryo radius at which Aw and A W are equal to zero, cm
r~ minimum embryo radius at which cataclysmic growth can
occur, cm
F~ line tension of the equivalent circumferential dislocation
loop surrounding the embryo, dynes or erg/cm
Burger's vector of equivalent circumferential dislocation
loop, cm
W~ energy of equivalent circumferential dislocation loop,
erg/loop
shear stress acting on equivalent circumferential dislocation
loop, dyne/cm 2
maximum energy of equivalent circumferential dislocation
loop, erg/loop
AW ~ activation energy per unit step during thermally activated
spontaneous growth of embryo, erg/unit growth of loop
maximum activation energy per unit step during thermally
activated spontaneous growth of embryo, erg/unit
growth of loop
r.~ embryo radius where activation energy per unit growth of
loop is maximum, cm
re radius of largest embryo present, cm
AW activation energy of largest embryo present, eal/(mol of
events)
170
M A R T : E N S I T I C TI~Alq S F O R M A T I O ~ S
2. NATURE OF MARTENSITIC TRANSFORMATIONS
2.1. Essential Features

Martensitic reactions are displacive or shear-like. They entail a co-
operative movement of atoms such that the region in question undergoes
a transformation strain or change in shape. Fiducial lines on a pre-
polished surface are sheared over, and the free surface of the region
is tilted out of the plane of polish. No diffusion or interchange of atoms
is involved in this process. The transformation product inherits the
composition of the parent phase, ~) and m a y be atomically ordered if
the latter contains a superlattice. (s~
The observed displacements indicate that the transformation strain
is homogeneous within any one plate, b u t GRENINGER and TROIA~O (9~
have shown that this strain (frequently called the "first" strain) does
not describe the atom movements because ff the austenlte lattice were
to be deformed in this way, the atoms would not be carried into their
final positions on the martensite lattice. Therefore, it is necessary to
postulate the occurrence of some heterogeneous deformation (fre-
quently called the "second" strain) due to twinning or slip on a fine
scale in the martensite. Then, the lattice deformation arising from the
primitive atom movements is homogeneous only over very small
portions of a single plate, and the apparent discrepancy between the
lattice deformation and the overall change in shape is accounted for
b y the heterogeneous deformation. We shall have occasion to recognize
these factors in treating the kinetics of nucleation.
Another important crystallographic aspect of these transformations
is that the martensitic units usually develop as lenticular or ellipsoidal
plates because of the opposing stress fields in the surroundings. The
main interface is often considered to lie parallel to the habit plane.
In some cases, the interface m a y be quite fiat, as when a martensitic
unit spreads across the whole section of a single crystal or when nearby
units cause displacements in the same direction, thereby minimizing
the back stresses. The nature of the interface is of particular signifi-
cance in our considerations. Inasmuch as it connects the displaced
region (martensite) with the surroundings, the interface must remain
invariant, i.e. unrotated and undistorted when averaged over macro-
scopic distances. This means that the shear component of the
macroscopic strain lies parallel to the interface (or habit) plane,
whereas the dilatational component is normal to it. The interface
must also provide a suitable linkage between the two phases such that,
as it moves forward, the atoms are transferred systematically from
the austenite to the martensite lattice in the right orientation. In
general, the interface possessing these properties, if it can exist at all,
171
PROGRESS I N ]~IETAL P H Y S I C S
has a "complex orientation; the habit planes are then said to be

irrational, meaning that the MJllerian indices are not simple.
Much progress has been achieved in rationalizing these crystallo-
graphic features by BOWLES and I~/LkCKENZIE, (10) W E C H S L E R et al. ~n)
and BILBY and CH~IST~. (~) According to recent views both the
overall homogeneous displacement and the heterogeneous strain are
accomplished at the interface as it spreads (rapidly or slowly) through
the parent phase, leaving mal~ensite behind. The interface m a y be
considered to comprise dislocation arrays which travel along with the
interface and produce the heterogeneous strain, as first proposed by
F ~ K ; a~) thus perfect coherency between the two lattices does not
prevail. Moreover, the orientation of the dislocation interface that
couples the two phases will depend on the mode of slip or twinning in
the heterogeneous strain and, in turn, on the temperature at which the
martensite is generated. This is reflected by the change in habit
orientation with transformation temperature in l~ig. 3.12.
I t is a str~iug fact that martensitic reactions can propagate at
temperatures approaching absolute zero, as demonstrated by KuLrl~
and COHENua~ for iron-nickel and iron-'nickel-carbon alloys; the
mobility of the interface does not seem impaired by the obvious lack
of thermal activation. Evidently, the interface motion is a coordinated
or wave-like phenomenon, not involving atom-by-atom transfer across
the interface, and neither short-range nor long-range diffusion is
required for the process. The actual movement of each atom relative
to its neighbours is very much less than one lattice spacing.
B ~ N s H ~ and MEHL(14~ have found that the propagation rate of
martensitic plates in an iron-29-5 per cent nickel alloy is about 105 cm/
sec, with a time of formation of 0.05-0.5 ,usec. This corresponds to
one-third the velocity of sound. Of equal significance is their observa-
tion that the propagation velocity of isothermally-formed mar~ensite
is the same as for the athermal product, and is virtually independent
of the temperature over the range studied (-- 20 to -- 195°C). The
sound velocity is likewise insensitive to temperature in this alloy.
It m a y be concluded that the activation energy for the growth mech-
anism of martensite (except for the bainite and uranium-chromium
cases discussed in Section 2.3) is effectively zero.
2.2. Athermal Characteristicz

Although athermal-kinetie behaviour is not an indispensable mani-
festation of martensitic reactions, nevertheless this feature is often
observed. Two examples are shown in Fig. 2.1 for iron-nickel uS) and
gold-cadmium ue) alloys, which also display athermal kinetics in the
reverse transformation on heating. In both cases, the athermal
reactions are not suppressed by rapid cooling or heating, and proceed
172
while the temperature is changing. I n the past, it was common to

state t h a t martensitic transformations cease when the cooling is
stopped, and start again when the cooling is resumed. However, in
the light of later findings, both of these statements need some
qualification.
Clearly, the martensitic reaction m a y go on after the cooling is
interrupted in those instances where isothermal transformation occurs.
On the other hand, if the athermal mode is deemed to require no
thermal fluctuations for its progress (in contrast to the isothermal
mode), then the athermal transformation p e r 8e can be considered to
stbp if the cooling is halted.
When the cooling is resumed, the athermal transformation m a y not
start immediately if the phenomenon of stabilization (1~) is operative.
I n fact, stabilization m a y impede the heating ~15) as well as the cooling
reaction, and the isothermaF TM as well as the athermal reaction. The
m a n y examples of stabilization reported in the literature are not
necessarily attributable to the same cause. One possibility is that
solute or impurity atoms undergo short-range diffusion to the strain
centres in the parent phase, and thereby lower the driving force for the
transformation in the regions which might otherwise serve as preferred
nucleation sites. (19, m) Alternatively, the interface of an embryo or
plate of martensite m a y become immobilized (m) by a slight concentra-
tion build-up in a manner analogous to dislocation pinning. I t has also
been suggested t h a t the close coupling between austenite and marten-
site, which confers high mobility upon the interface, m a y be partially
destroyed b y time-dependent plastic yielding or relaxation.OZ') More
recently, it has been proposed that stabilization results from an
extraneous strengthening of the parent phase, caused by the formation
of Cottrell atmospheres (~) or precipitation, ~ which makes the matrix
more resistant to the displacements, attending martensitic tr~.nsforma-
tions.
The quantitative details of stabilization will not be reviewed here
because, for present purposes, it m a y be regarded as a parasitic
phenomenon which occasionally interferes with the nucleation or
propagation mechanisms at play, and then only introduces deviations
from the primary thermodynamic and kinetic characteristics of the
transformation.
There are important differences in the reaction behaviour between
the two cases illustrated by Fig. 2.1. The iron-nickel alloys transform
to martensite on cooling by the successive formation of new plates,
with each one shooting out suddenly to its final size. Consequently,
the course of the transformation as a function of dropping temperature
is dependent upon the rate of nucleation and the size of the fully
grown plates, but not upon their rate of growth. Such reactions are
173
PROGRESS I1~ M E T A L PHYSICS
typified by a high degree of supercooling and hence a relatively large

driving force below the martensite-start temperature (Ms). Once
nucleated, each martensitic unit propagates quickly until it collides
with a structural barrier or its mechanism becomes jammed (Section
6.4). On continued cooling, there is no additional growth of the plate
despite the increased driving force; further nucelation occurs elsewhere
in the parent phase.
In contrast, the gold-cadmium alloys generate martensitic plates
that do not reach their final size at once. Instead, although each unit
i
I. 00!
I
o o.v5
gs=5S
o.5o
/
I /As:'~ c ~
1\1
I 7~.30 I
re"
0.25
-I ~ I l t t l l l
-100 0 100 200 300 400 500
Temperature °C
Fig. 2.1. Electrical resistance changes during the cooling and heating of
an iron-nickel ~16)and a gold-cadmium (16jalloy, illustrating the hysteresis
between the martensitic reaction on cooling and the reverse transforma-
tion on heating
seems to nucleate suddenly and "pop out" to a detectable size within

a small fraction of a second, further propagation in length and thickness
continues (frequently in jerky fashion) with the decreasing tempera-
ture until collision or jamming takes place. Martensitic reactions in
this category start to transform with comparatively little super-
cooling (Section 4.5), and the driving force is insufficient to supply
the strain-energy and other nonchemical requirements of fully grown
plates. In other words, the growth of a martensitic unit m a y be halted
at a given temperature because of the limited driving force available,
and a state of thermoelastic balance is approached. As the temperature
falls, the enhanced driving force permits the growth to proceed while
additional nucleation also ensues. The progressive propagation is
intrinsically a rapid process in that it is able to pace the dropping
temperature and maintain the athermal character of the reaction.
B u t the rate of propagation is then controlled b y the rate of cooling,
unlike the case for reactions of the iron-nickel type.
Corresponding differences carry over to the reverse transformation
174
on heating. In systems of the gold-cadmium class, the reverse trans-
formation occurs with relatively little superheating; the martensitic
plates shrink more or less progressively and disappear approximately
in the reverse order of their formation. There is no obvious nucleation
process involved in this reversion. However, in the iron-nickel case,
appreciable superheating is necessary to start the reversal, and the plates
do not snap back out of existence, b u t often transform piecewise in
smaller platelike units, c~) This process evidently does entail nucleation.
There is a phenomenological similarity between the two types of
martensitic reaction and mechanical twinning, with applied stress
taking the place of the chemical driving force. (26, 2v~ When polycry-
stalline zinc is slowly compressed, thin mechanical twins first form
discontinuously and then thicken progressively as the stress is increased.
Removal of the stress causes the twins to shrink back and even disap-
pear. On the other hand, if the zinc is shock loaded or over-stressed,
the mechanical twins do not grow reversibly and they remain on
unloading. !~Iore generally, various combinations of both behaviours
m a y be encountered, and the same is true for martensitic reactions.
The iron-nickel and gold-cadmium prototypes cited here constitute
particular cases of athermal kinetics that will be found to varying
degrees in other martensitic systems.
2.3. Isothermal Characteristics

Although numerous isothermal martensitic transformations have not
been found as yet, enough information is already at hand to rule out
athermal kinetics as a necessary manifestation of a martensitic reaction.
In most martensitic systems, the isothermal component is either not
operative or is obscured b y the predominant athermal behaviour. Where
isothermal reaction has been detected it usually occurs below Ms, b u t
some cases are known ~28-3°, 94, 96~ in which martensite can form iso-
thermally above M~. The first instance was reported b y KURDJUMO¥
and MAxr~0vA ~94) for an iron-0"6w/o carbon-6.0w/o manganese
alloy.* Subsequently these investigators discovered isothermal trans-
formation in an iron-nickel-manganese alloy. (96~ Hence, athermal
martensite is not an essential precursor for isothermal martensite.
However, because martensitic reactions are strain sensitive and auto-
catalytic, athermal martensite m a y stimulate the nucleation of iso-
thermal martensite when the cooling is interrupted below M,, provided
not too much of the parent phase has been consumed b y the athermal
transformation.
The basic nature of the isothermal reaction is best studied in the
absence of athermal martensite, and in later sections more weight
* Private communication from G. V. K u r d j u m o v : this alloy also contained 2 w]o
copper, which was inadvertently omitted in the published analysis.
175
PROGRESS Ii~ M E T A L PHYSICS
will be placed on those isothermal reactions which occur above M~.

Important information can also be obtained in the other case, but it is
essential to ensure that the effect of temperature on the isothermal
kinetics is not clouded b y the presence of different amounts of athermal
martensite. This problem is handled b y pre-quenching below the lowest
isothermal level to be studied, thus introducing a fixed quantity of
athermal mar~ensite, and then quenching upwards to the various
temperatures for the isothermal kinetic measurements. (zs~
Whether the isothermal reaction takes place in the absence of
-tO0
1 5 10 50 100 500 1000

Isothermal holding time (t + 1) rain
Fig. 2.2. C curves for i s o t h e r m a l t r a n s f o r m a t i o n of m a r t e n s i t e in a n

i r o n - n i c k e l - m a n g a n e s e alloy. (After CECH a n d HOLLO~ON (~z))
athermal martensite or in the presence of a fixed amount, the isothermal

kinetics display a C-curve behaviour, with the maximum rate of trans-
formation coming at some intermediate temperature. Quite often, the
active temperature range extends well below room temperature, and
the maximum isothermal transformation rate is found in the vicinity
of 100-150°K. This was demonstrated b y CEcil and HOLLOMO~(sl~ for
an iron-nickel-manganese alloy, as shown in Fig. 2.2. Up to 40 per
cent isothermal transformation has been observed under these
conditions.
In almost all instances the isothermal reaction proceeds b y the
nucleation of new plates, rather than by the growth of existing ones,
and the plates reach full size within a small fraction of a second. The
appearance of the martensite and its mode of formation are quite
similar to those of the athermal transformation, as described in Section
2.2 for the iron-nickel alloys. Such isothermal reactions are obviously
controlled b y the rate of nucleation.
Only two cases have been reported in which isothermal growth
occurs, in analogy with the gold-cadmium type of progressive athermal
176
growth. When the latter propagation comes to a halt at a given

temperature because the available driving force is balanced by the
required non-chemical factors (Section 4.5), further growth of the
martensite at the temperature in question can only take place if an
increase in the driving force or decrease in the restraints comes into
play as a function of time. The isothermal bainite reaction has been
suggested (3~ as an example of this possibility, in which carbide pre-
cipitation within the transformation product or carbon diffusion in the
surrounding austenite increases the thermodynamic tendency for the
bainite to continue growing. According to this interpretation, the
isothermal growth of bainite should be controlled by the rate of carbon
diffusion either in the austenite or bainite, whichever is slower. However,
it has not been established t h a t bainite reactions always have the
displacive character t h a t would properly classify them as martensitic
transformations.
Another instance of isothermal growth in a martensitic transforma-
tion has been reported by HOLDEI~(29) and MorT and HA~ES c33~ for
uranium-chromium alloys. On a prepolished surface, the martensitic
displacements not only produce surface upheavals, but the tilted
regions expand as a function of time. This growth is observed at
room temperature where the diffusion rate is too slow to account for
the isothermal kinetics. I t has been proposed (~9~t h a t the strain-energy
factor opposing the transformation is progressively relieved by time-
dependent yielding due to twinning in the martensitic product or slip
in the surroundings. Still another possibility is t h a t the interface
between the two phases is not highly mobile and that thermally
activated dislocation adjustments have to take place therein in order
for the atoms to move co-operatively.
2.4. A utocatalytic and Partitioning Effects

Once a martensitic transformation sets in, either athermally or iso-
thermally, the subsequent course of the reaction m a y be greatly
influenced by autocatalysis. Preferred nucleation sites are developed
in the parent phase by the disturbance around an existing plate, and a
chain reaction m a y ensue with the plates forming from end to end in
zig-zag fashion. MAC~n~ and COHEN(a4) have shown that, in special
cases, a burst of transformation m a y occur as a result of co-operative
displacements among several plates. If potential plates in the vicinity
of an actual plate tend to cause the same displacement in the inter-
vening austenite, then the back stress on the transformation is reduced
and m a n y plates can propagate almost at once. Such bursts not only
create discontinuities in the reaction kinetics, but occasionally comprise
a large part of the total transformation.
A quantitative theory governing these autocatalytic phenomena is
177
lacking, and consequently, it is only the initial rates of nucleation that

we shall deal with here. B y the same token, when testing various
models of nucleation, we shall devote more attention to the ~]/~ tem-
perature than to the subsequent transformation kinetics on cooling.
Another complication that arises after the reaction gets under way
is the diminishing size of fully grown plates, due to partitioning of the
matrix b y the increasing number of plates. This makes it difficult to
relate the transformation rate to the nucleation rate. FISHER et al. (aS~
have treated this interesting problem in spatial geometry, but the
regularity of the shape and formation-sequence of the plates is not
sufficiently precise to yield explicit results. It becomes necessary to
introduce an adjustable parameter which is evaluated empirically b y
fitting the transformation curve, and which therefore depends upon the
autocatalytic effects as well as the partitioning factor. In view of
these uncertainties, it seems advisable to rely on initial rate measure-
ments when comparing kinetic theories with experiment.
3. THE CHE1WICALFREE ENERGY CHANGE

3.1. The Concept of an Equilibrium between Austenite
and Martensite
In view of the diffusionless nature of martensitic reactions, martensite
and its parent phase (austenite) have the same composition. Each
phase has a chemical free energy that varies with temperature and
composition, and for a given alloy, there m a y be a temperature (To)
at which the two free energies are equal (FY-- - E~'). * At any other
temperature, the difference in free energy m a y be expressed as
F ~ -- F ~' = A F ~'-~ cal/mol (3.1)
This quantity is positive when martensite is more stable than austenite
and negative for the opposite. It is a quantitative measure of the
driving force for the martensitic transformation, being larger in a
positive sense the greater the driving force. (Note that A F ~'~r is the
negative of the free energy change attending the martensitic trans-
formation.)
Fig. 3.1 is a schematic representation of F ~' and $ " as a function of
temperature. To is defined as the temperature at which A F ~'-*r ---- 0.
In the cooling of any given alloy from the austenitic range, M~ lies
below T O because the austenite-to-martensite reaction does not
necessarily start when the martensite becomes more stable than the
austenite. The reasons for such supercooling are discussed in Section
• The prime is used to designate the martensitic phase. Usually, but not necessarily,
this phase is a modification of the low temperature equilibrium phase ~. The high
temperature phase is denoted by y.
178
MARTENSITIC TRAI~SFORMATIO]qS
4.1. The reverse transformation on heating, starting at A s, m a y also

require superheating as indicated in l~ig. 3.1.
Unfortunately, sufficient thermodynamic data are not available for
many alloy systems exhibiting martensitic transformations. One of
the primary difficulties is that diffusionless reactions often occur at low
temperatures where equilibrium cannot be achieved. However, con-
siderable progress has been made along these lines in the iron-
carbon(aS, 19, 3~) and iron-nickel (aS, 15) systems, and consequently they
~' s~,ol~e thereto- ~stoble t h e r m o -

dynomlcoily clynornlcolly
relative t o ~ relot~ve to ~ '
ro A5
Temperot ure * °K
Fig. 3.1. Schematic represenf~ion of F =' and F~ vs. temperatm~e

for an iron-base alloy
will receive most attention here. Iron-manganese alloys (3s) display

certain similarities to the iron-nickel case.
The experimental determination of T 0 depends on bracketing it
between M s and A s. In iron-base alloys, the hysteresis between M s
and A s m a y be as large as 400°C; examples are shown in ~ig. 3.2. The
reverse transformation is not commonly observed in iron-carbon
alloys, presumably due to the rapid decomposition of the martensite
via tempering reactions on heating. However, GRID~EV and TRE-
FILOV(a°) have recently reported A s determinations for iron-carbon
martensites (rapidly heated at ~-~ 600°C/sec) lying 300-400°C above
M s•
In general, the hysteresis between M s and As can be narrowed b y
plastic deformation. M s is thereby raised, becoming Md, and As is
lowered, becoming A~. Using cobalt-nickel alloys, HEss and
BARRETT(4°) were successful in closing the gap almost completely with
this procedure, and the resulting T o determinations are plotted in
179
Fig. 3.3. I f M~ and ~1~ cannot be brought into coincidence by plastic
deformation, T0 is taken as ½(J~d ~- A~). This case is shown in Fig. 3.4
for iron-nickel alloys ;(15) actually ½(Md ~ Ad) turns out to be about
i-- 19'0 ,,JiN; 14"5

. . . . % N;"C g ' 5 % NI•
/
I-3=.7-I, Ni M,=2~O'C M=.==u M~=S2~'C
/l/, : - - I S O ' C 23.7~i "/! NI [ /
- / 2g.7%NI • M, =1 ~~ u^ ' ~i . " ~" t .,,,,,.,~
2 x l O " .(=) 30"7%Ni Specimenlength 1.4in

Ms = - 7 2 ° C O;ameter 0 . 0 6 2 i n
I I I I I I I I
-250 -tSO -5O 50 ~50 250 350 4,50 55O
Temperature °C
= 3i5°C
3 3 . 0 and 32.7"/,Ni
As =, 3 0 0 °C
g.~%Ni
• - iI
t71 ~ i i i i Ill i i lg.O,I, Ni
II ,~, . , ,_. , . , ._, , / ~ 29.3"~N," ,t,:SlO"C " "

I
I,
- - ~ ........
I- "-"-" I
I I I 1, i I I I ~ I I I i I I I I I
0 1(30 2 0 0 300 400 500 600 700 800
Tempel~ture °C
Fig. 3.2. Composite resistance-temperature curves for iron-nickel

alloys(16)
equal to ½(•, -~ A,) for this system. When the difference between ~l~',
and A , is only several degrees, as in the indium-thallium (~) and gold-
cadmium (le) alloys, plastic deformation is not necessary to bracket ~0-
However, there are alloys, such as beta brass, (t2) in which the trans-
formation temperature range on cooling overlaps that on heating, and
hence As lies below 5I,. These are undoubtedly cases in which the
strain energy built up by the dlsplacive nature of the martensitie
180
MA-RTENSITIC TRANSFORMATIONS
transformation is sufficient to start the reverse transformation "pre-
m a t u r e l y " on heating. Then T Ois not bracketed by M , and A,, and its
experimental determination becomes uncertain.
The role of plastic deformation in arriving at T O is a kinetic one;
it stimulates the nucleation process so t h a t the more stable phase
can form in the temperature range where the martensitic transfor-
mation or its reverse does not take place merely on cooling or heat-
ing. However, the question does arise as to the possible effect of
the concomitant elastic stresses (strain energy) on the equilibrium
°u 4oc ~ " ~ - -
=u 550 i '
30C 500 . -.~
45C
400
oM# 35O
"~ ~ 300 A~......
~ I00
E ~, 2oo ~,,~+A,)C:;"-'--
150.
0
100 I
°Q--
I
T
; 50 ,.. "~
-IOC
-50
-20C --I00
~0 3o 40 27 28 29 ,30 31
%Ni Nickel at.%
Fig. 3.3. Bracketing of To in cobalt- Fig. 3.4. B r a c k e t ~ g To b y plastic

nickel sTsbem by plastic deformation. deformation in iron-mcke] alloys (I~
(After HEss and B ~ (4°))
temperature. BURKART and I~EAD(41) have demonstrated that, in an

indium-thallium alloy, M , and A 8 (and therefore To) are shifted by
elastic stress, while PATEL and C O H ~ (4s) have shown an analogous
influence on M8 in iron-nickel alloys. This shift in the equilibrium
temperature due to mechanical factors is a thermodynamic effect
governed by the LeChatelier principle, in contrast to the kinetic effect
of plastic deformation.
However, when the temperature hysteresis between M , and A, is
large, as in the iron-base alloys, the thermodynamic shift in T o due to
the m a x i m u m elastic energy increment t h a t can be supported by the
system is small compared to M~ -- M , and A, -- A~ as produced by
plastic deformation. Furthermore, it is not obvious t h a t the complex
181
pattern of elastic stresses set up during plastic deformation would

produce any net shift in the observed T o.
On the other hand, for systems where A s -- lkr 8 is small, the shift
of M s, A s and T 0 due to stresses m a y be sufficient to raise the ~hr above
the "unstressed" T 0. In such cases, the T o bracketed b y M~ and A~
would not be the equilibrium temperature being sought. However,
as was pointed out previously, it is not necessary to use plastic deforma-
tion to determine T 0 when A s ~ M 8 is small.
An important product of the thermodynamic treatment of marten-
sitic transformation is the calculation of T 0. This has been done for
iron-carbon and iron-nickel alloys, as described in the following
sections.
3.2. Correlation between T o and the Equilibrium Diagram

Using the iron-nickel system for purposes of illustration, Fig. 3.5
shows the free energy relationships between T o and the equilibrium
phases upon which the thermodynamic calculation of To is based.
At a high temperature T4, the stable state for an alloy of com-
position x is the ~ phase (f.c.c.) as indicated b y point A. I f the alloy
is cooled rapidly to T a, the free energy per tool is given b y point B
although the equilibrium state now corresponds to a two-phase mixture
of low-nickel alpha (b.c.c.) and high-nickel gamma (f.c.c.) having the
lower free energy per tool designated b y point B' on the common
tangent line.
With further rapid cooling to avoid compositional changes or
separation processes via diffusion, T~ is reached where the F~ and F r
curves intersect at the given composition x. This is the temperature
T o at which alpha and gamma of the same composition have identical
free energies. At any lower temperature T1, the gamma phase {point
D) has the possibility of transforming martensitically into alpha
(point D') of the same composition (in which case it is called ~') or to
move toward the more stable mixture {point D ~) of low-nickel alpha
and high-nickel gamma. However, the latter reaction requires diffusion
and is often so slow at the temperatures involved that the martensitic
transformation, being diffnsionless, intervenes. I f the latter starts at
T 1 in Fig. 3.5, then M , = T 1 and the driving force at Ms is DD'
according to Eqn. 3.1.
Fig. 3.6 is a composite three-dimensional diagram ~15~of free energy
versus composition and temperature. The horizontal base is a portion
of the conventional phase diagram. Free energy curves of the type
shown in Fig. 3.5 are obtained b y passing vertical sections through the
alpha and gamma free energy surfaces in Fig. 3.6. The points of
common tangency define the compositions of the coexisting alpha and
gamma phases in the equilibrium diagram. The line PQ is the
182
MARTEI~SITIC TRANSFORMATIONS
intersection between the two free energy surfaces, and defines the locus of
F ~ = F r. This locus, projected to P'Q' in the temperature-composition
plane, gives T Oas a function of composition. The points A BCD trace
the free energy path for the gamma phase in a 31 a/o nickel alloy on
rapid cooling from 800 to 200°K. At temperatures below point C, the
F y surface lies above the F ~ surface and there is a finite driving force
for forming martensite. At 200°K, which happens to be M~ for the
F° '
~ F ~
500 800
I=D X ct/o Ni Ni 7 ~ °K
Fig. 3.5. Schematic repre- Fig. 3.6. Schematic representation of Fr

sentation of chemical free and surfaces as a function of x
energy versus composition and T (15~
for the alpha and g~.mma
phases in the iron-nickel
system at four tempera-
ture levels
31 a/o nickel alloy, the free energy can drop from D on the F r surface
to D' on the F : surface by means of a diffusiordess transformation.
3.3. Thermodynamic Properties of Pure Iron

In calculations of the free energy changes accompanying the martensitic
reaction in iron-base alloys, the thermodynamics of the alpha-gamma
transformation in pure iron plays an important part. JOm~NNSO~,c44)
ZENERc3s) and D~LRKEN and SMITH(45) have reported on A F ~ "r.
FISHER (37) utilized the following equation below 500°K:
A F ~ ~ = 1238 -- 0.395T -- 12.5 × 10-4T 2 cal/mol (3.2)

183
P R O G R E S S IN M E T A L P R Y S I C S
whereas KAUFMAN and COHE~ (is) have proposed:

A F ~ *v = 1202 -- 2.63 × 10-ST 2 + 1.54 × 10-ST a cal/mol (3.3)
between 200 and 900°K, which is the range of martensitic interest.
The latter equation will be adopted here because it conforms reasonably
well with Johannson's data and also satisfies the third law of thermo-
dynamics.
The enthalpy change as a function of temperature can be derived
from Eqn. 3.3 b y using the relationship:
a S ~ ; ~ = ~ F ~ ? ~ - T 3 A~----f--
F~Y (3.4)
, I i I
1sooi'~\ I
~eoo \ \ i i
1400 ~ %'
",,/. Darken and Smith~'S)
~2oc~ \
*, Johannsorr ~ \
,,~ 800,
I I I , ~.,~ \
Kaufman and Coheri.~...~. ~ ''
/1F'r,~'~' =1202-2. e3x10"3T2
60(:]- +1.54 xlO-e T3
40C
0 100 200 300 400 500 600 700 800
Temper'c~ure =K
Fig. 3.7. The free energy" change accompanying the alpha-gammA trans-
formation in pure iron as a function of temperature according to various
investigators
and the results are in fair agreement with Johannson's values for
~/~'~.
As shown in Figs. 3.7 and 3.8, the Darken and Smith curves for
A ~ V ~ and AH~eY are at considerable variance with the data adopted
here. Because these differences would introduce serious discrepancies
in the computations to be discussed, it is necessary to arrive at some
basis for making a choice. The main reason for the difference lies in
the temperature dependence that is used for the specific heat of pure
gamma iron, which is subject to considerable uncertainty because it
cannot be measured directly over the temperature range of 0-1183°K,
and extrapolations cannot be safely made from the higher temperatures
where gamma is the stable phase. Fortunately, SCHEIL and NOR-
~_.ANN(4s) have recently determined A H ='~r in a series of iron-nickel
alloys (Fig. 3.9) b y measuring the specific heat versus temperature for
184
b o t h t h e a l p h a a n d g a m m a phases. A s t r a i g h t line is o b t a i n e d w h e n
A H = ' ~ a t a n y g i v e n t e m p e r a t u r e is p l o t t e d as a f u n c t i o n of (w/o Ni) ~,
~2oo0 f 1
,ooo / I
140C ,.~ . ' "
60C- .,4H :1202+2.63x10"~T c ~ - - "

- 3.o8 x ~ - o r~-----------~: \ I
4 0 0 - (Kaufmon ,and Cohen'S 4
adjustment of donannson's \ \
clara to comply with third ~\ '1
2O0 400 60O 800 1000 ~ o o ~4oo

Temperature °K
Fig. 3.8. Enthalpy change for the alpha-gamma transfm~aation in

pure iron versus temperature according to various investigators
~400
1200
|
"" 15" ~ I 0 ',,5 tNIW/O
1000
_-"
~
.....
800
u
~, oOc
400
t '., "',, ,,.,'.
200 t , "t, ",, ,',"
0 ~0 200 300 400 500 ;00 700 800 900 E~o
Temperotuve °C
Fig. 3.9. Enthalpy change accompanying the martensitic trans-
formation in iron-nickel alloys measured as a function of tem-
perature by N O R M ~ and S C H E I L {46)
a n d e x t r a p o l a t i o n to 0 w/o Ni yields t h e values s h o w n in Fig. 3.8.

I t is e v i d e n t t h a t t h e y d i s c r i m i n a t e in f a v o u r of t h e J o h a n n s o n c u r v e
(and t h e p r o p o s e d m o d i f i c a t i o n to c o m p l y w i t h t h e t h i r d law) o v e r t h e
~ 185
PROGRESS IiW M E T A L PHYSICS
lOW temperature range where the martensitic transformation comes

into play.
3.4. Free Energy Changes in Iron-base Alloys
The chemical free energy of the austenitic phase in a binary F e - A
alloy can be expressed as
F v -----(1 -- x)F~e -~- xFr~ -~- FrM cal/mol (3.5)
A similar equation can be written for the martensitic phase, and on

subtracting from Eqn. 3.5,
A F ~'-*~ = (1 -- x)AF~-~ r -p x A F ~ ~r -p A F ~ ~r cal/mol (3.6)
For multicomponent systems in general,
A F ~'-'r -~ x F o A F ~ ' -~- Z x ~ A F ~ ~v -+- Z A F ~ - y cal/mol . (3.6a)
The quantity A F ~ r has already been treated in Section 3.3; AF~-*r

is the difference in free energy between the gamma and alpha modifi-
cations of pure component A; it is negative for f.c.c, elements and
positive for b.c.c, elements. In the case of elements which do not have
stable f.c.c, or b.c.c, allotropes, the sign of AF~-" cannot be inferred,
but real values do exist in principle.
The difference in the free energies of mixing A F ~ - r is difficult to
estimate, even as to sign. It can, of course, be obtained from activity
measurements of the iron and component A in the austenitic and
martensitic phases as a function of temperature and composition.
Accurate calorimetric measurements (4~, ~8) can also yield information
concerning A F ~ -~. However, ordinarily, such data are not available,
and assumptions must be introduced to bridge the gap.
ZEIffER{36} has derived the thermodynamic properties of medium
alloy steels on the basis that the carbon and alloy concentrations are
sufficiently dilute to allow the alpha and gamma phases to be regarded
as ideal solutions. According to this assumption, the mixing term
A F ~ ~ in Eqns. 3.6 or 3.6a is zero and A F ~~ -~ I~T In xJx~. Since
AF~- r = A H ]~v -- T A S ] -*r, the quantity R T In xJx~ is not generally
constant, b u t Zoner takes it to be so, implying that AS~-r = 0. (49) In
this case,
AH~~ -~ R T In x~,/x~, = A H ~ cal/mol (3.7)
with the latter equality arising from the assumed ideality of the
solutions. A//~ -*r is the difference between the heats of solution of
component A in the austenitic and in the martensitic phases. Eqn. 3.6
now simplifies to:
AF~-*~ = (1 --x)AF~-~ r A- x A H ~ --'r cal/mol (3.8)
186
MARTENSITIC TRANSFORMATIOI~S
Once AH~'-~ is evaluated, t h e chemical free e n e r g y change for the
m a r t e n s i t i c reaction can be calculated f r o m E q n . 3.8, a n d t h e n T O is
o b t a i n e d b y setting A F ~'~v equal to zero. Since A F ~ ~ becomes larger
(in t h e positive direction) as t h e t e m p e r a t u r e decreases, the elements
for which AH~ ~ is negative ( e x p a n d e d gamma-field phase diagrams)
lowers To, while elements w i t h AH~ -*~ positive (gamma-loop phase
diagrams) raise T o.
More recently, JONES a n d PU~PHREY cSs) h a v e utilized Zener's
a p p r o a c h to derive the free e n e r g y changes for t h e martensitic reaction
in t h e i r o n - n i c k e l a n d i r o n - m a n g a n e s e systems. H o w e v e r , in order to
m a k e AH~*~ a n d A H ~ ~ come o u t r e a s o n a b l y constant, it was necessary
t o m u l t i p l y A F ~ e r b y two different empirical constants. This p o i n t e d
o u t a basic deficiency in the t h e r m o d y n a m i c t r e a t m e n t because
A F ~ ~ should be identical for all iron-base systems. T h e discrepancy
arises f r o m t h e f a c t t h a t t h e austenitic a n d martensitic phases are n o t
ideal o v e r t h e range of composition involved, n o r is ,~A rq~'--.~Ni or AH~-*~
c o n s t a n t o v e r t h e r a n g e of t e m p e r a t u r e s considered. Nevertheless,
E q n . 3.8 is useful for obtaining r a p i d a p p r o x i m a t i o n s over the limited
t e m p e r a t u r e range in which AH~ -*~ has been c o m p u t e d . T h e AH~ -*v
values in the l i t e r a t u r e are listed in T a b l e 3.1.
TABLE 3.1
Values of AH~'-~ for Approximate Calculation of Chemical Free Energy Changes
Accompanying the Martensitic l~eaction in Iron-base Binary Alloys
Binary component AH~"~ (cal/mol) Reference
C -- 8100 (36)
N --5360 (36)
Mn - - 2440 (36)
Mn -- 2700 (38)
Ni --1600 (36)
Ni - 2500 (38)
Cu -- 1280 (36)
Zn - - 590 (36)
Si + 475 (36)
Be + 810 (36)
A1 + 1300 (36)
W + 1360 (36)
Mo + 1360 (36)
V ÷ 2830 (36)
P + 4180 (36)
Sn + 5500 (36)
Ti 9000 (36)
Cr 1200 (35)
I t should be n o t e d in T a b l e 3.1 t h a t the value for c h r o m i u m is nega-

tive despite t h e fact t h a t c h r o m i u m is a g a m m a - l o o p element. AH~'~ ~
187
was determined by FmHER et al. (3~) at low temperatures from con-

siderations of the martensite kinetics, but would have been a positive
number (as for the other gamma-loop elements in Table 3.1) ff the
equilibrium diagram and Eqn. 3.7 were adopted for the evaluation.
This emphasizes the risk in using the AH~'~' quantities for calculating
the free energy changes over too wide a range of temperature, particu-
larly for the gamma-loop elements wherein the alpha-gamma equilibria
lie far above the martensitic range.
FISH~,~ ~aT) has made a detailed analysis of the thermodynamics of
the martensitic transformation in steels, employing the following
assumptions: (a) the x A F ] - r term in Eqn. 3.6a is taken to be negligible;
(b) R T Inf~/f~" is assumed constant* and f~e----f~e----- 1; when the
latter is true, it can be shown that x R T l n f ~ / f ~ ' = A F ~ ~ r in Eqn.
3.6a; and (c)f~/f]" is regarded as independent of concentration when
alpha and gamma have the same composition as in a 4iffusionless
transformation. In the case of carbon, a temperature dependence is
introduced for A F ~ -~ along with an order parameter A F * which takes
into account the preferential arrangement of the carbon atoms in the
body-centred tetragonal lattice of the martensite in steel. (There are
three sets of octahedral interstices for the carbon atoms, which are
equivalent in b.c.c, ferrite, but one of these sets becomes preferred in
b.c.t, martensite.) Fisher's free energy equation is (in the present
notation) :
AFt';', = xF,/~Fr~ ~ + x,(-- 10,500 + 3.425T)

-- AF* + ZxaRT ln f~t/f~" cal/mol (3.9)
COHEN et al. (19) have also reported on the free energy changes in
iron-carbon alloys. Their results agree quite well with Fisher's, and
are shown in Fig. 3.10. From the slope of the linear portion of the free
energy curves, ~ A F ~ ' - * ' / 2 T = - - 1.4 cal/mol °K, and therefore AF ='-~
at M s is 1.4(T 0 --M~) independent of the carbon content. The T O
value for each composition is given by the intersection of the corres-
ponding free energy curve with the A F ~'-~ = 0 axis.
In comparing Zener's equation (3.8) with the more general equation
(3.6), it can be seen that
AH~ - r = AF~-'~-+- AF~-~-----Z
cal/mol (3.10)
x
Thus, AH~-Y is not likely to be constant because AF~t- r depends on

temperature and A F ~ -~ depends on both temperature and composition.
Moreover, AH~'~' m a y or m a y not have the same sign as A F t - '
* f ~ a n d f ~ ' are the activity coefficients of component A in the austenitic and marten-
sitic phases of the same composition.
188
depending on the sign and magnitude of AF~-*~. This explains why

AH~-'~ is not always negative for f.c.c, elements or positive for b.c.c.
elements. :For example, alumlnlum is f.c.c. (A_F~r is negative) and
yet aluminium is a gamma-loop element ( A H ~ TM is ~- 1300 cal]mol
according to Table 3.1). To circumvent this difficulty, KAU~_AN and
~00
\
"~ C 1.1w/o Carbon ~ ....._
-4~ and "l'refilov)Bg) _
200
I 400- 600
} 800
t\.--,:
1000 1200
Temper atut'e OK
Fig. 3.10. C h a n g e i n c h e m i c a l free e n e r g y a t t e n d i n g t h e a u s t e n i t e -

m a r t e n s i t e r e a c t i o n i n i r o n - c a r b o n alloys as a f u n c t i o n of t e m p e r a t u r e
a n d c a r b o n c o n t e n t (19)
Co~v.N[15) have adopted a more rigorous treatment for the iron-nickel

system, and have arrived at explicit values for the individual terms on
the right side of Eqn. 3.6. The steps are outlined below.
Equations are set up for the free energies of the alpha and gamma
phases, treating them as regular solutions of iron and nickel. The
rule of common tangents is then applied to the expressions for F = and
F ~, and by utilizing the known phase boundaries, it is possible to
evaluate F~, although not F ~ or AFt7 v at this stage. Nevertheless,
because of the way the latter two terms happen to appear in the exact
equation for AF ='-*r, an approximate solution for A F ~'~;' can be
derived for compositions up to 25 a/o nickel. This solution shows that
To is closely equal to ½(Ms -t- As), and that the driving force to start
189
PROGRESS I1~ METAL PHYSICS
the m a r t e n s i t i c t r a n s f o r m a t i o n a t M~ is n u m e r i c a l l y equal to the

driving force to reverse t h e t r a n s f o r m a t i o n a t A s. W i t h t h e r e a s o n a b l e
a s s u m p t i o n t h a t these correlations also p r e v a i l in the c o m p o s i t i o n
r a n g e a b o v e 25 a/o nickel, it becomes possible to e v a l u a t e the p r e v i o u s l y
u n k n o w n t e r m s F ~ a n d A F m , a n d t h u s o b t a i n a n explicit expression
c~
~ ' , , , . i f x =0.05 ! ! -T200 S

I I i
-1000
=0.20
-800
+ 600
IX \ kSc-~x=o.3q
+ 400 , -0-35
t~
u,,.
u,.
-20,
-41 +400
I ! i I +600
O I I
0 200 400 600 800 1 0 0 0 1200
Temperetur'e °K
F i g . 3.11. C h e m i c a l f r e e e n e r g y c h a n g e a c c o m p a n y i n g t h e
m a r t e n s i t i c t r a n s f o r m a t i o n i n t h e i r o n - n i c k e l s y s t e m (16}
for A F ='-*~ as a f u n c t i o n o f t e m p e r a t u r e (below 1000°K) a n d nickel

content:
F~t = - - x(1 - - x)[3400 -f- 0.75T(1 - - In T)]
-F R T [ x l n x --f- (1 - - x) In (1 - - x)] cal/mol (3.11)
F ~ = x(1 - - x)[200 - - 0.17T(1 - - In T)]
+ RT[x In x + (1 - - x) In (1 - - x)] cal/mol (3.12)
A F ~ 7 " = - - 3700 ~- 7.09 X 1 0 - a T 2 -4- 3.91 x 10-~T a eal/mol (3.13)
A F ~'~y = (1 - - x)(1202 - - 2.63 x 1 0 - a T 2 + 1-54 × 1 0 - e T a)
x ( - - 3700 + 7.09 x 10"4T 2 + 3.91 × 10-TT a)
+ x(1 - - x)[3600 + 0-58T(1 - - In T)] eal/mol (3.14)
190
The chemical free energy changes indicated by Eqn. 3.14 are plotted
in Fig. 3.11. F r o m the slopes of these curves, AS ='-'~ = 1-45 cal/mol °K
Nickel ato/o
25 30
1200 i t
100£
~fT, for Fe- NI 1100
go( °1 ~" 1000' ~,k*\ I ,, ~,(M+. A,)
800 ~.--o
g00 -\o\i .-..] . v
700
/ ~1 \fX "/////~,
~O00
E ~e-C and "3,, I \~,,
50C ' ii-. I X vb'~"
i t,~ i
400 300 ok:.)
30C
1 [ %.t ~"$ "j~ I
1oo I I
}i
o , \
20C
0.2 0-6 1.0 1.4 1.8 C 10 20 30 40 5O
Carbon wt.% Nickel at. %
Fig. 3.12. M, and T o fop iron-carbon Fig. 3.13. Experimental and theoretical
and iron-nickel alloys determination of To in the iron-nickel
system (zS)
50(3
Weight % CorZx~n
1.T2 1:0 0:8 0;6 0'.4 o F'e-C
400 • Fe-Ni
300
t~ 2OO -
u. 100
30 25 20 15 1 •
I I I I :
Atomlc % Nickel 5
O--
I I ,. I I I i 1
2OO 300 400 500 C:~30 700 800 9(30 1000
Temperature oK
Fig. 3.14. The effect of temperature (or composition) on the

driving force at M, for iron-carbon and iron-nickel alloys
for iron-nickel alloys containing 15-30a/o nickel. Consequently

AF ='~v at M , is 1.45(T 0 - - M , ) . However, this proportionality con-
stant becomes smaller at lower nickel contents. The M+ and A+
191
temperatures are also shown in Fig. 3.11 for the iron-nickel alloys,
and it can be seen that A F ='-~ at M s = AFr-'=' at A 8.
The T o values at AS'='~ ~ 0 are replotted in Fig. 3.12 along with
those from Fig. 3.10 for iron-carbon alloys. M s falls 200°C below T Oin
the iron-carbon system, and 70-220°C below T o in the iron-nickel
system, depending on the nickel content. For the iron-nickel case,
experimental values of T Oare available from ½(M 8 ~ A s) and ½(Md t a d )
measurements, and excellent agreement with the calculated T O is
found, as shown in Fig. 3.13.
The driving force at l]/8 is given in Fig. 3.14 as a function of carbon
and nickel contents. It does not vary appreciably with carbon content,
but clearly increases with increasing nickel content. The physical
significance of these findings is discussed in Section 6.4.
3.5. Magnetic Effect in the Martensitic Thermodynamics

of Iron-base Alloys
The use of a single parameter AH~ -~r = R T in (xJxy) (Eqn. 3.7) to
calculate the martensitic thermodynamics of an F e - A system leads to
a serious dilemma: several gamma-loop elements, which obviously
raise To, also lower Ms. The question then arises as to whether these
gamma-loop elements impede the martensitic transformation unduly
and require extraordinarily large driving forces to start the reaction,
thereby leading to lower M8 temperatures despite the raised T O
temperatures.
To account for this apparent anomaly, ZE~ER (5°~ has proposed a
two-parameter treatment that recognizes the role played b y magnetic
changes in iron-base alloys. The free energy change in pure iron
A F ~ ~ is first decomposed into a magnetic and a non-magnetic term:
AF~ = A~=-*~
'--'~ Fe(~.~.) ~ A~=-*y
'-'~' Z ~ e ( ~ ) cal/mol (3.15)
The significance of this separation is shown in Fig. 3.15. Curve A is
the actual free energy change versus T, indicating alpha-gamma
equilibria at the two temperatures where A F ~ ~ ~ 0. The linear
portion of this curve below 500°C is taken to represent the free energy
changes over a temperature range in which there is no appreciable
uncoupling of the magnetic moments of the iron atoms in the alpha
phase. Hence, the extended curve B is regarded as the "non-magnetic
component" of the free energy change; this curve would denote the
free energy changes if the magnetic moments in the alpha phase
remained coupled over the entire temperature range. In this event, the
free energy curve would not bend back and only one alpha-gamma
equilibrium would be observed. The equation of linear curve B is
-- AF~-~.~.) = 1.41(T -- 740) cal/mol . (3.16)

192
Curve C is the difference between curves A and B, and constitutes

the portion of the free energy change that is attributable to magnetic
effects; it arises from the energy absorbed in uncoupling the magnetic
moments in the alpha phase.
The addition of component A to iron is considered to influence
magnetic and non-magnetic parts of the free energy change indepen-
v t \
140C ~
1200
I
• 1000
p 8oc f
~r Fe (NM~
600
/
400 J
2OO ~
./
+ 1000 +(500 +200 0 -200 -~0
Fig. 3.15. S e p a r a t i o n of A / ~ - * r i n t o m a g n e t i c a n d n o n - m a g n e t i c
terms~So)
dently by vertical shifts along the temperature axis, each of which is

proportional to the weight per cent of A. In other words, the non-
magnetic term is changed from
AF~s(~.~.)[T] to ~-~ --
and the magnetic term is changed from
AFFe(~g)[T] to AF~e(M~g)[T -- yAT(M~)]
where y is the weight per cent of A in the alpha and gamma phases.*
AT(N.~.) and AT(~g) are the shifts in the respective terms per 1 w/o A,
and are the two parameters that Zener introduces to replace the
inadequate single parameter AH~'~r.
AT(~.~.) is equated to the change in Ms per 1 w/o A on the assump-
tion that the martensitic reaction reflects the compositional dependence
of the free energy change in the low temperature range where there is
no magnetic contribution; thus AT(~.~.)= AM s. The AT(Ma~) term
* The t e m p e r a t u r e q u a n t i t y in brackets is the a r g u m e n t of the free energy function;
it specifies the t e m p e r a t u r e at which the free energy change is taken.
193
is a measure of the effect of 1 w/o A in shifting the t e m p e r a t u r e range

in which the magnetic uncoupling occurs. T h e free e n e r g y change for
the diffusionless reaction at t e m p e r a t u r e T can now be w r i t t e n :
A.F~"~[T] ---- A F ~ . ~ . ) [ T 1 ] ÷ AF~e-~g)[T~] cal/mol (3.17)
where T 1 = T -- yAT(~..~.) = T -- yAM,
and T 2 = T -- yAT(~g)
Adding and subtracting AF~?(~.~.)[T2] on the right side of Eqn. 3.17,
and n o t i n g t h a t
A F ~ . ~ . ) [ T 1 ] - - AF~(~.~I.)[T2] ---- - - 1.41(T 1 - - T2) cal/mol °K
from E q n . 3.16:
AF='-r[T] = -- 1.41(T I -- T2) ~- AFFe(~Iag)[T2] -~- AFFo(N.M.)[T2]
------ l'41y(AT(~ag ) -- A M , ) ~- AF~?'[T2] (3.18)
At the equilibrium temperature To, A F a'-r ----0, and
l'41y(AT¢~ag ) -- A M , ) = AE~?r[Ts] (3.19)
For the gamma-loop case, w e can take y and T 2 corresponding to
the limiting composition of the loop,* from which AF~e'[T2]-----
- - 18"5 cal/mol (from curve A in Fig. 3.15): Thus,
AT(~g) = AM, - - 18"5/(l'41ymax). (3.20)
Using the i r o n - c h r o m i u m s y s t e m as an example, Y,,ax = 1 3 w / o
c h r o m i u m , * A M , ---- -- 19°C per 1 w/o chromium, and t h e n AT(~ag ) ----
- - 20°C per 1 w/o chromium.
TABLE 3.2
A T ( ~ ) Parameters and Curie Temperatures for Gamma-loop Elements
Element AM,(°C) (s*' AT(m,)(°C) (5°~ Effect an 8~ (6"'
Cr -- 19 -- 20 Cr raises then lowers

0~. 10 w/o Cr causes
almost no change in e~
-- 10 -- 15 No effect
W -- 5 -- 8 No effect
V -- 35 -- 48 15 w/o V raises e~ from
770 to 839°C
Si -- 6 10 w/o Si lowers a~ from
770 to 600°C
A1 + 30 -b 17 10 w/o A1 lowers 0~ from
770 to 700°C
* The maximum composition is taken halfway between the alpha and gamma phase
boundaries.
194
Table 3.2 lists the calculated AT(~ag ) values for several gamma-loop
systems, using the AM 8 data tabulated by HOLLOMON and JAFFE. ~51)
Unfortunately, AT(Mag) is a parameter t h a t cannot be verified inde-
pendently. As Zener points out, the effect of alloying elements on the
magnetic uncoupling actually spreads out over a temperature range,
and cannot be regarded as a mere temperature shift; therefore,
additional parameters would be required for a more precise description
of the magnetic changes. (81)
The comparison of AT(~ag ) and the alpha-phase Curie temperature
in Table 3.2 shows no obvious correlation. Aluminium and vanadium
are quite anomalous in t h a t ATo~ag) and 0F indicate magnetic shifts
in opposite directions. Although Z ~ . R ~2~ has suggested t h a t the
effect of these elements on magnetic ordering m a y be a factor, the
physical meaning of the calculated AT(~g) values remains obscure.
Nevertheless, this t r e a t m e n t is an interesting example of how M ,
determinations can be employed to provide thermodynamic information
at temperatures where the classical methods of measurement are not
feasible.
I t is curious t h a t 11~8 decreases with increasing concentration of
some gamma-loop elements, despite the fact t h a t T o increases. (This
is in contrast to the gamma-stabilizing elements such as carbon and
nickel for which T e and M , both decrease and hence (T o - - M s ) is
relatively insensitive to composition.) According to Eqn. 3.18, how-
ever, the slope of the curve of A F ~ ' ~ versus temperature decreases with
increasing alloy concentration of those gamma-loop elements that
lower M,. Therefore, the driving force does not increase rapidly with
decreasing temperature in the range between T 0 and 11I, in such cases,
and so substantial supercooling is required before the driving force is
sufficient to start the transformation. The driving force at M , for
gamma-loop elements can be computed from Eqn. 3.18 and the data
in Table 3.2. The results show that the driving force at M , does not
vary abnormally with composition, and is of the same order as for
gamma-stabilizing elemehts.
3.6. Additional Thermodynamic Information from Martensitic

Transformations
In view of the foregoing thermodynamic considerations relative to
diffusiouless reactions, it becomes possible to utilize certain transforma-
tion studies to provide thermodynamic data at low temperatures
where the attainment of equilibrium is out of the question. Also,
calculations can be made concerning metastable states that cannot be
treated otherwise.
One interesting by-product is Eqn. 3.13, which gives the free energy
change ( A F t ? ~) from b.c.c, to f.c.c, nickel. C15) This quantity is negative
195
over the entire solid-state temperature range, signifying that f.c.c.

nickel is more stable than b.c.c, nickel. Furthermore, inasmuch as
A F ~ *r becomes increasingly negative with decreasing temperature,
the f.c.c, phase of pure nickel becomes relatively more stable, and there
is little chance of producing b.c.c, nickel by plastic deformation even
at temperatures approaching absolute zero.
In contrast, the free energy change ( A F ~ r) from b.c.c, to f.c.c.
iron (Eqn. 3.3) becomes more positive as absolute zero is approached,
and hence b.c.c, iron increases in stability with decreasing temperature
in this range. Actually, these findings on both nickel and iron are to be
expected on theoretical grounds. ~s~
The free energy changes versus temperature for pure nickel and pure
iron are plotted in Fig. 3.16, along with the free energy changes for all
,4F a'-*.-,,~
,~r~o?.---~"
-I
T=.,.~°K J Temperature OK r=1200°K
Fig. 3.16. S c h e m a t i c r e p r e s e n t a t i o n of t h e A_~'-*y, x, T

co-ordin~.te s y s t e m (zs~
the intermediate compositions, t h e latter being the surface generated

by Eqn. 3.14. This is the chemical free energy change AE ~'-'r versus
temperature and composition for diffusionless reactions of b.c.c, alpha
to f.c.c, gamma solid solutions. The intersection of the surface with the
plane A F ~ ' ~ = 0 depicts the equilibrium temperature T Oas a function
of composition.
The enthalpy change accompanying the martensitic transformation
in iron-nickel alloys can also be computed from Eqn. 3.14 since
AH ~''~ = A F ~ ' ~ -- T 3AFc"~r . (3.21)

~T
196
and the result is

~r/~'-*v = (1 --x)(1202 + 2.63 x 10-ST 2 -- 3.08 × 10-eT a)
- - x(3700 + 7-09 X 10-4T ~ + 7.82 × 10-TT 3)
+ x(1 -- x)(3600 + 0.58T) (3.22)

This expression for the enthalpy change is of particular significance
because it permits direct comparison with the heat effects measured
recently b y SCHEIL and NOR~L~'~. (46) Table 3.3 lists the calculated (aS)
and observed (46) values of L//~'~v at A, and AH ~-*:' at M , for several
iron-nickel alloys; the agreement is obviously quite satisfactory.
T A B L E 3.3
Calculated aS~ a n d O b s e r v e d ~46~ H e a t Effects A c c o m p a n y i n g t h e ~'--~
a n d F "~ ~' R e a c t i o n s in I r o n - N i c k e l Alloys a t t h e A, a n d M ,
Temperatures
A H ~ ' ' ~ (callmol) ~t'v"~' (cal/mol)

Atomic As Ms
fraction Ni (°K) Cak, ulated Observed
(°K)
Calculated Obs~vcd
0.05 1020 +574 + 540 890 -- 1018 -- 840

0.10 950 + 720 + 630 750 -- 1160 -- 980
0.15 890 +748 + 650 610 -- 1144 - - 1080
0.20 835 +714 + 520 480 -- 1000 - - 1000
0.25 765 + 669 + 380 365 --810 --820
0.30 660 + 587 235 -- 593
.Note: AH~''*r positive slgnlfle~ heat absorbed b y t h e m a r t e n s i t e - t o - a n s t e n i t e t r a n s f o r m a t i o n , a n d

~H ~ ' ~ " negative signifies h e a t e v o l v e d b y t h e a u s t e a t t e - t o - m a r t e n s i t e t r a n a f o r m a t i o n ,
Another noteworthy outcome of this thermodynamic treatment is

that the iron-nickel phase diagram can now be extended down to
subzero temperatures where equilibrium cannot be established. The
free energies of the alpha and gamma phases as a function of com-
position at any given temperature can be calculated from equations
of the form of Eqn. 3.5 b y substitution of Eqns. 3.3, 3.11, 3.12 and
3.13. Applying the rule of common tangents yields the coexisting
compositions of the alpha and gamma phases over the entire tempera-
ture range, as shown in Fig. 3.17. At the higher temperatures, the
calculations are necessarily consistent with the known phase diagram
because the experimental boundaries are used in evaluating some of
the numerical constants. Below room temperature, a regression of
the ~/(~ + 7) boundary is indicated b y the calculations, while the
(~ + r)/r boundary moves well over to the nickel side of the diagram.
BURKART and I~EADm) have investigated the effect of applied stress
on the martensitic transformation in indium-thallium alloys to
determine the latent heat of the transformation. The M , and A,
197
temperatures are found to increase in parallel fashion with uniaxial

tension and (at a somewhat different slope) with uniaxial compression.
The hysteresis between M~ and A 8 is only 4-5°C, so that T 0 m a y be
considered to lie midway. Thus, the equilibrium temperature is
raised by either tensile or compressive stress. The latent heat of the
reaction is computed from a modified Clausius-Clapeyron equation in
o~ ;200
HO0 - -
00wen and Liu ~
1000
900
800
700
600
& 5oo
E
~. 400
30O
2OO
100
~0 20 30 40 50 60 70 80 90 100
Fe Atomic °/o N~ckel Ni
Fig. 3.17. I r o n - n i c k e l e q u i l i b r i u m p h a s e d i a g r a m cxs~
which length changes are substituted for volume changes and force is
substituted for pressure:
dTo &l. T~ 8T 0
d L . . . . -~. _ °K/dyne . (3.23)
VAH aAH
or
alto eTo OK/dyne/cm2 (3.24)
ds AH
where L = sa = force, a = cross-sectional area, Al/V = the change
in length per unit volume of transformation, e = transformation
strain, and AH is in cal/cm 3. AH/p comes out to be 2.66 × 10-3 cal/g
for i n d i u m - 21 a/o thallium ~a) and 0.45 cal/g for g o l d - 47.5 a/o
cadmium.( xe}
I t is worth emphasizing t h a t the Clapeyron effects of hydrostatic
pressure and tension would be of opposing signs, whereas un]axial
compression and tension both raise T O. This is due to the multiplicity
of transformation modes which permits the transformation strain to
have the same sign as the applied stress; thus the specimen elongates
under uniaxial tension, and shortens under compression, when the
reaction takes place. On the other hand, the volume change attending
198
the reaction can have only one sign, so that if hydrostatic pressure
were to raise To, then hydrostatic tension would lower it.
4. THE NONCHEMm~ FREE ENERGY CHANGE
4.1. Supercooling of Martensitic Transformations
I t goes without saying that an austenite-to-martensite reaction cannot
take place spontaneously at a given temperature and pressure unless
the temperature is below To where the driving force A F ~'-~ ---- 0. On
the other hand, despite the fact that there is no need to wait for
diffusion, such transformations do not occur immediately below T o.
In some instances, the supercooling (T o - - M , ) m a y be as large as
200°C (and the corresponding driving force about 300 cal/mol) before
the reaction sets in. Although there are also cases in which the super-
cooling is only a few degrees, it is never zero. The iron-carbon system
is of particular interest in that, notwithstanding the marked depen-
dence of T O on carbon content, the M 8 changes in a parallel fashion,
making the driving force to start the reaction virtually independent
of the carbon content.
There are two general reasons for expecting martensitic trans-
formations to supercool even though they are not suppressed by rapid
quenching. In the first place, the nonchemical factors such as inter-
facial energy and strain energy must be p u t into the overall free energy
balance to obtain the necessary condition for the martensitic reaction
to occur:
AW ~'-'~ ~- A F ~'-'~ + AG~'~r > 0 cal/mol (4.1)
or
A F ~'-'~> AG~-'=" eal/mol (4.1a)
In this notation, the nonchemical free energy change AG m a y be
regarded as a "restraining force" that opposes the chemical driving
force when the latter becomes positive below To. Thus, for martensite
to form, the temperature must be decreased sufficiently to make
A F ='-~ not only greater than zero, b u t greater than some finite
quantity. This thermodynamic aspect of the problem is treated in the
present section.
The second factor in the supercooling is related to the nucleation
process. The nucleation barrier is considered to control the reaction
kinetics since the rate of propagation is almost instantaneous, and
hence the degree of supercooling (T O - - M s ) m a y be determined b y
kinetic as well as thermodynamic variables. These implications are
discussed in Sections 5 and 6.
4.2. Nonchemical Factors in Martensitic Transformations
As in the case of other reactions, new interfaces are produced when
martensitic transformations take place, and interracial energy is
199
associated with these regions. For a lenticular plate of radius r and

semi-thickness c, in which r ~ c, the interfacial free energy is taken as
AG~r-'~' ~ 2wr~a cal/particle (4.2)
where 2rrr 2 is the approximate surface area of the plate and a is the
specific interracial free energy.
KURDJUYIOV{3} has regarded the austenite-martensite interface as
coherent, and on similar grounds FISHER et al. ~35~ have arrived at a
value of 24 erg/em 2 for ~(5.7 × 10 -7 cal/cm2). This is of the same
order as the interracial energy of a twin boundary. On the other hand,
FRANKc12~and READ~m have proposed models of the interface in which
the two lattices are not coherent on an atomic scale, although there is
macroscopic matching to minimize the average strain. Under such
conditions,~19, 55~ ~ m a y be as high as 500-1000 erg/cm ~ (1.2 -- 2,4 ×
10-Scal/cm2), commensurate with ordinary interphase boundaries.
K~LPp and DE~rrX~GER~Se~have recently calculated a value of 200 erg/
cm2, * based on Frank's model for the austenite-martensite interface
in an iron-carbon alloy with a (225)r habit plane at 25°C.
Mechanical free energy is also an important factor in martensitic
reactions because of the displacements involved. Elastic strains are
developed, both in the transformation product and the surroundings,
by the dilatational and shear components of the displacements. Usually
the dilatational contribution is neglected in comparison with the shear
contribution, and the strain energy then takes the form
AG~r'*~" ~ ~rrc2A ~ ~rr2c(Ac/r) eal/particle (4.3)
where ~rr~c is the approximate volume of the lenticular plate and
(Ac/r) is the strain energy referred to a unit volume of martensite.
(FISHER et al. (as~ use a strain energy constant 0 = A / 4 . ) For the case
of martensite in steel, FISHERc5~)reports A -~ 40 ----- 2 -- 6 × 10~° dyne/
cm 2 or 480-1440 cal/cm a, and with somewhat different assumptions,
K~LPP and DE~rr.r~GERcSs) arrive at A ~ 500 cal/cm a at 25°C.
Other nonchemical effects such as acoustical vibrations ("clicks")
and plastic deformation m a y occur during martensitic transformations.
The associated work must come from the chemical free energy A F ~ ' ~ ,
and is either lost as heat or stored in the form of lattice imperfections
or localized strains. Generally, such stored energy is negligible and is
not available to aid the reverse transformation, unlike the non-
chemical factors previously mentioned.
COHEN et al. c19~ estimated that, if one considers the final size of
martensitic plates in steel, the interfacial and strain energy terms
amount to 9-1 cal/cm 3 or 65 cal]mol. This should lower T O by about
• T h e actual value of a published b y K~.~eP a n d D~T~r~G~R ~*~ is 100 e r g / c m ~, but
a factor o f 2 w a s overlooked in their paper.
2OO
MART]~NSITIC TRANSFORMATIONS
45°C (Fig. 3.10) and T o -- Ms would then be 155°C instead of 200°C.

It is quite evident, then, that such lowering of T o due to the non-
chemical factors in Eqn. 4.1 does not account for the observed super-
cooling. However, there is the possibility that the transformation
could proceed slightly below the modified T o if it were not for the tem-
perature rise caused b y the sudden generation of heat from the rapidly
propagating plates; this alternative is treated below.
~_ "~P~'tr,
Temper'ot~uPe °K
Fig. 4.1. Graphical rep~esen~on of the adiabatic theory of the

martensitic transformation in iron-carbon alloysc6s~
4.3. The Adiabatic Hypothesis

Inasmuch as martensitic plates form suddenly at M s, K.RISEMENT et
al. ¢Ss) have proposed that Ms must be low enough to prevent the heat
liberated b y the transformation from adiabatically raising the local
temperature above the modified T o. Let T 1 be this temperature, l~ow
consider that a mol of martensite formed at the Ms temperature in
Fig. 4.1 generates a quantity of heat equal to AHI~ ° or A A ' . Under
adiabatic conditions, this would raise the temperature of the martensite
to point E, b u t because of the interracial and strain energies re-
quired b y the transformation, the temperature can only attain some
lower level which Krisement et al. assume to be T 1. They estimate
that the interfacial and strain energies associated with a fully grown
plate amount to 50 cal/mol, and b y setting this equal to A F ~'~r in
middle panel of Fig. 4.1, T 1 can be determined on the supposition that
z4 201
PROGRESS IN iYI:BTAL PHYSICS
at least 50 cal/mol must be available for doing work at T 1. The next
step is to relate t h e M , temperature to TI.*
Based on the foregoing assumptions, Krisement et al. derive the
equations below:
AF='-YJ T, = HrIM, - - H='I ~ cal/mol (4.4a)

and
Hr[ T, - - H ' I M ° -~ T zXS='-~ cal/mol (4.4b)
Since AS ~'-r is known from the slope of the A F ~'-~ curve in Fig. 4.1,
T 1 can now be located graphically from the austenite enthalpy-
temperature curve in Fig. 4.1 by using Eqn. 4.4b and the known M s
temperature. A F ~'~' at T 1 determined in this w a y is found to be
80 cal/mol for iron-carbon alloys, independent of the carbon content.
This can be considered as a better value for the interracial and strain
energies than the 50 cal/mol previously estimated.
Conversely, if one now starts with the assumption that the driving
force at T 1 (the modified To) is 80 cal/mol, M 8 can be calculated as a
function of carbon content by utilizing Eqn. 4.4a and the bottom
panel of Fig. 4.1. At T~, point C is fixed b y setting C D equal to
80 cal/mol, and then the horizontal line through C intersects the
H-curve at M s.
However, this explanation of the supercooling below T~ m a y not be
realistic because the value of 65-80 cal/mol for the nonchemical
energies is based on fully formed plates, while the conditions that
govern the start of the transformation must be related to plates of
much smaller size.
4.4. M e c h a n i c a l Interactions with M a r t e n s i t i c T r a n s f o r m a t i o n s

I t is well known (59-el) that the Ms temperature can be varied b y applied
stress. PATEL and COHEn(4s~ have utilized this effect to demonstrate
the interaction of mechanical energy with the thermodynamics of
martensitie reactions. The acting stress system is resolved into com-
ponents parallel to the shear and dilatational displacements of the
transformation, being respectively parallel and normal to the habit
plane. The mechanical work done on or b y the transforming region,
as the resolved components of the acting stress are carried through
the corresponding transformation strains, is added algebraically to the
chemical free energy change of the reaction in order to compute the
alteration in temperature at which the critical value of the driving
force to start the transformation is attained.
The M s is either raised or lowered depending on whether the
* This assumption has been criticized by M. Hrr,r.~.RT.~)

202
mechanical work aids or opposes the chemical driving force. For

uniaxial tension or compression,
AG'~t-*¢'' ~- ½V,,,s [Y0 sin 23 ± ~0(1 ~- cos 23) ] cal/mol (4.5)
where V~ is the molar volume, 8 is the acting stress in suitable units,
fl is the angle between the stress axis and the normal to the habit
plane, 70 is the transformation s h e a r strain (0-20 for iron-nickel
martensite), and so is the dilatational transformation strain (0.04 for
iron-nickel martensite). The plus sign in Eqn. 4.5 is used for uniaxial
tension, and the minus sign for uniaxial compression.
The orientation fl of particular interest is the one t h a t makes AG~r-'~'
a maximum, since the M~ temperature is affected by the plates t h a t
form first under the influence of the applied stress. The condition for
maximizing AG~"-*~' is ~AG~'*~'/~fl -~ 0, whence
sin 2____f_~
l • 7o (4°6)
cos 2fl s0
or
fl ---- 39.5 ° (tension) and 50.5 ° (compression) (4.6a)
l~or hydrostatic pressure, there are no shear-stress components to
influence the reaction; thus the pressure interacts only with the
dilatational strain. Eqn. 4.5 then becomes
AG~r'*~" ~ - - Vmse o cal/mol (4.7)
and the orientation dependence disappears.
The magnitudes of AG~r-'~' for the three cases at hand are:
A G ~ ' ( 1 0 0 0 lb/in 2 uniaxial tension) ---- -~ 122 in lb/in a
-- -~ 1.42 cal/mol
AG~r-*~'(1000 lb/in 2 uniaxial compression) ~- -F 83 in lb/in 3
---- ~ 0.96 cal/mol
AG~r-*~'(IO00 lb/in 2 hydrostatic pressure) ----- -- 40 in lb/in a
~- -- 0.47 cal/mol
Fig. 4.2 demonstrates how the effect of these mechanical energies on
the M , temperature is determined. Curves 1, 2 and 3 represent the
variation of A F ~ ' ~ ' with temperature for a series of alloys in a given
system, while curve 4 denotes the driving force at M~. Taking alloy 2
as an example, the dashed curve is drawn parallel to curve 2, but
displaced by AG~-~' cal/mol. The displacement is upward if AG} '~¢'' is
positive, and downward if negative. The M~ temperature is shifted
to M's by the applied stress and is given by the intersection of the
dashed curve with curve 4; in other words, M~ is the temperature at
203
PROGRESS IN I~ETAL PHYSICS
which the algebraic sum of the chemical driving force and the mechanical
work is equal to the driving force required to start the transformation.
With hydrostatic pressure, the dashed curve is displaced downwards
and the results in a corresponding lowering of M s.
A comparison is presented in Table 4.1 of the calculated and observed
effects of applied stress systems on the M8 temperatures of iron-nickel
and iron-nickel-carbon alloys. (da) The agreement is good, and
undoubtedly better than is justified by the uncertainties involved.
:~ Ms w i t h o u t applied s t r e s s
~-~ ~ \ ~ o/~s with appl;ed s t e e s s
o
o
>,
u.
\\\
Fig. 4.2. Schematic diagram showing how the mechanical energy
A,G~"~' due to the applied stress system changes the Ma temperature of
alloy 2 by contributing to the thermodynamic driving force. The
driving force to staxt the transformation varies with the ~l/, temperature
according to curve 4. (After PATF_J.and COHEN(43~)
4.5. Thermoelastic Martensite

Beta brass (a2) is an interesting example of an alloy in which there is
overlapping of the cooling and heating transformation ranges; the
A s on heating lies below the M8 on cooling. This is in contrast to most
martensitic systems in which A 8 definitely lies above M , and also
above T o. Presumably, at least part of the elastic energy expended by
the martensitic reaction in beta brass is available to start the reverse
transformation on heating at a temperature below the "unstressed"
T 0. (Interracial energy considerations can be neglected here because
fuU-grown martensitic plates are involved.) This is another case in
which mechanical energy plays a role in the thermodynamic balance.
In fact, the build-up of elastic energy during a martensitic transforma-
tion m a y be one cause for its self-stopping nature at any given tem-
perature; t h a t is, if AG,r-*~' becomes equal to or greater than AS'~'~,
the reaction will cease. This condition is more likely to be realized if
M s is not far below T o where the chemical driving force is small.
204
T A B L E 4.1
Effect o f A p p l i e d S t ~ o n t h e M , T e m ,erature (after PATEL a n d CO~EN |i$1
Stress 8yztem, Uniaxial tension Uniaxial compression Hydrostatic pressure

Material 0 . 5 % C, 2 0 % Ni, hal. Fo 0.5% C, 20% Ni, bal. Fo 7 0 % Fe, 3 0 % N i
¢91 dM_j ICalculate d . + 1.07°C]I(P l b / i n t + 0"72°C]10 s lb/in 2 -- 0.38°C]10 s lb/b~ t
ds [Experimental + I-0°C/10 s lb/in t + 0"65°C/10 s Ib/in s -- 0-57°C/10 * lb/in z
Change in ~Calculated @ + 16°C + 10"6°0 -- 5.7°C
M, per I[Experimental + 15°C + lO°C _ 8.5°C
15,000 lb/in 2
Dropping temperature, which increases A F ~'~r, permits the transforma-

tion to proceed, provided that the interfaces are free to migrate.
Accordingly, a state of thermoelastic balance can be visualized between
the chemical driving force and the nonchemical restraining force.
Ku~DJu~ov and KHANDaOS(~2~ have found graphic evidence of this
thermoelastic behaviour in a copper-aluminium-nickel alloy. The
martensitic plates were observed to grow progressively on cooling
through the transformation range. When the cooling was stopped, the
growth ceased; when the cooling was resumed, the growth continued
until impingement occurred. On heating, the plates decreased in size,
with the last ones formed on cooling being the first to disappear on
heating. It should be emphasized, however, that this thermoelastic
balance is not observed in iron-base alloys and m a n y other martensitic
systems. In such cases, the martensitic units propagate quickly to
their final size, and further transformation on cooling is accomplished
only by the formation of new plates. Here, the plates do not grow or
shrink in balance with the changing driving force.
One might expect that, when the thermoelastic balance prevails,
applied stress should influence the balance in a manner similar to that
of "elastic twinning. ''(6s~ This has been shown in beta brass by
REr~OLDS and BEVER(64) who noted metaUographically that strain-
induced martensite at room temperature partially disappears when the
operating stresses are removed.
-Along similar lines, BURKART and READ (4a~ found that complete
transformation above the "unstressed" M~ can be induced in indium-
thallium alloys by applied stress, and then complete reversal is obtained
by releasing the stress at the same temperature.
5, CLASSICAL I%[UCLEATION CONCEPTS A P P L I E D TO

I"~A_RTENSITIC TRANSFORMATIONS
5.1. Homogeneous Nucleation Theory

An excellent review of this subject has been published by HOLLOMON
and TI3RNBULL,(~5~and therefore only the main features need be covered
here.* Using Eqns. 4.2 and 4.3 for the increase in interracial free energy
(AG~r-'~') and in strain energy (AG~r-'~'), the total change in free energy
attending the formation of a lenticular particle (see Fig. 5.1) of
martensite is
&W ~ " = ~rr2e&fr-'" + 27rr2a + ~rre2A (5.1)
: 2~rr2o• + ~77.2C( AfV--~a" + rCA) cal/particle (5.1a)
• The application of classical nucleation theory to inaxtensitic transformations has

been conducted along similar lines b y Kurdjumov, and has been reviewed b y Lv.~ENT. (°a~
206
MARTEI~SITIC TRANSFORMATIO]qS
where Afr-'=' = AFr-~='I V~ = chemical free energy change per cm 3 of

martensite formed. This quantity is opposite in sign to that of the
chemical driving force discussed in Sections 3 and 4. Below To, Afr-*='
is negative, whereas a and A are always positive.
The critical values of c and r at the saddle point of A Wr-*=" are
obtained b y setting
3(AW--'=')
= 0 (5.2)
3c
and
3(A W,'---=')
= 0 (5.3)
~r
from which (remembering that A = 40)
v o l u m e -" ='~" c
S u ~ a c e =r,ea~2='r 2
Fig. 5.1, Schematic drawing of mu,etensite embryo
e. = - 8./a±f cm (5.4)
r,~ = - 2Ac~/Af = - 8oc,dAf cm (5.5)

V~, = ~'r~c~, = - - 4 W ~ , / A f = - - 32,768~rO~a3/27Af 5 cm s (5.6)
A W ~ = 8192rrO2aa/27 A f 4 cal/particle (5.7)
Note that the superscripts Y -~ ~' have been dropped for convenience,
and Af is negative within the temperature range of interest for the
--~ ~' transformation.
AW~ is the free energy of nucleation; it constitutes the nucleation
barrier that must be mounted in order for embryos to reach the critical
size leading to spontaneous growth. In the classical theory, the energy
to activate the nucleation event comes solely from thermal fluctuations.
I f the nucleation is assumed to be entirely random, each atom is
regarded as a potential nucleation site, and
= (No/V,~)~ exp (-- A W ~ , / k T ) nuclei/cm ~ see (5.8)
207
where No/V,, is the number of atoms per cm a. The lattice vibration

frequency ~ is taken to be the number of nucleation attempts per
second about each atom, and exp ( - - A W ~ / k T ) is the "successful"
fraction of these attempts. The activation energy for growth that
ordinarily appears in the nucleation-rate equation for diffusion-
controlled reactions is neglected in Eqn. 5.8 because it is presupposed
that the martensitic interfaces are sufficiently mobile without thermal
activation to maintain an equilibrium distribution of embryos down to
very low temperatures.
The implication of this theory is that martensite nucleation occurs
only isothermally. Eqn. 5.8 indicates that ~ varies with temperature
in a C-curve fashion, increasing to a maximum with decreasing tem-
perature and then diminishing with further decrease in temperature.
To account for the athermal transformation characteristics in iron-base
substitutional alloys, FISHER~6s) has identified M , with the temperature
at which _gr _= 1 nucleus/cm s sec. At somewhat higher temperatures,
is so small that isothermal nucleation is not observed within reason-
able times; and at slightly lower temperatures, ~ is so large that iso-
thermal nucleation is not suppressed even on rapid cooling. On this
basis, Eqn. 5.8 reduces to
AW~ = 82.9 at M , (5.9)

kT
and AW~ can then be substituted into Eqn. 5.7 to yield a value for
0~aa, once Af is known. Using the data of JO~ES and P u t , PHOnY(aa~
for Af in iron-nickel alloys, FIS~,R (Be) finds that
02aa = 9.92 × 10~a ergS/cm n = 7.75 × 10-1~ calS/cm12 (5.10)
The O2aa parameter can now be used in Eqn. 5.7 to compute AW~ at
any temperature for which Af is known; and when A W~ is substituted
in Eqn. 5.8 the nucleation rates plotted in Fig. 5.2 are obtained as a
function of temperature for a series of iron-nickel alloys.
The C-curve behaviour is evident. Furthermore, the rate of nuclea-
tion is so dependent on temperature when ~ = 1 that M , defined on
this basis is quite insensitive to the cooling rate. In fact, if isothermal
nucleation is measurable only when 10-1 < N < 10, as suggested b y
FISHER, (as) there is only a narrow range of temperatures and nickel
contents in which isothermal nucleation can be detected, even though
it is always an isothermal process in this view.
As the nickel content is increased, the C-curves in Fig. 5.2 are
displaced to lower nucleation rates, and the temperature at which
= 1 decreases accordingly. However, this " 2 / , temperature"
should vanish at 110-120°K where the vertical line at N = 1 is tangent
to one of the C-curves (,-- 30.1 a/o nickel). In alloys above this nickel
208
c o n t e n t , m a r t e n s i t e should n o t f o r m on cooling even to 0°K, since the

n u c l e a t i o n r a t e falls off v e r y rapidly below t h e nose t e m p e r a t u r e . T o
t e s t this point, K A ~ z ~ a n d COHEN c6~) h a v e d e t e r m i n e d t h e M ,
t e m p e r a t u r e s of a s y s t e m a t i c series of iron-nickel alloys; as shown in
Fig. 5.3, t h e M , t e m p e r a t u r e s are f o u n d to lie a t levels well below the
x 320 I i I
o
Appm"ent Atomk
- 29"0
240
J \
io0
~ f I 31.0
f/
E
0
16 08 - 8 -16 -24 -32
log nucleation rote
Fig. 5.2. Calculated isothermal nucleation rate as a function of tempera-
ture for l~e--Ni alloys in the composition range 29.0 ~_ atom per cent
Ni <: 31"0. Rates less than 10 -1 and greater than 10= nucleiper cm8 per
sec are difficult to measure experimentally. (After FZSHERcu*)
3 0 0 . ~
(g
] =
~ 150
DO J
No fc:x'vrtot~r
of morCensite~
set'red or~
uenchtngt.¢
50
0
27 28 29 30 31 32 33 34 35
Nickel at.=/°
Fig. 5.3. Ms temperature versus nickel content for iron-nickel

aUoys~")
nose of the C-curves in alloys containing considerably more t h a n the
limiting 30.1 a/o nickel m e n t i o n e d above. I n t h e light of these results,
it is e v i d e n t t h a t t h e a s s u m p t i o n of h o m o g e n e o u s classical nucleation
for t h e m a r t e n s i t i e reaction in the iron-nickel alloys is unrealistic, and
t h a t t h e r m a l a c t i v a t i o n does n o t basically a c c o u n t for t h e a t h e r m a l
features of the t r a n s f o r m a t i o n . E m p l o y i n g the more r e c e n t t h e r m o -
d y n a m i c d a t a on the iron-nickel s y s t e m ~15~p r e s e n t e d in Section 3 does
n o t alter these conclusions.
209
PROGRESS IN :~¢IETAL P H Y S I C S
FISHER (e~ has also applied Eqns. 5.7 and 5.8 to the isothermal
transformation data of CECH and HOLLO~ION(al~ on iron-nickel-
manganese alloys. In this case, it is necessary to adopt a 02aa value six
times t h a t given by Eqn. 5.10 to obtain good agreement with the
isothermal kinetics. For the athermal transformation in 3 w/o
chromium steels, (aS) 02~a comes out to be twenty-nine times larger.
Returning to the iron-nickel system and applying Eqn. 5.6, the
critical nucleus volume V~ in a 30 a/o nickel alloy is calculated to be
0.8 × 10-~° cm a at the ~]/8 temperature. The critical nucleus would
then be 1-2 atoms in semi-thickness, 10 atoms in radius, and contain
nearly 700 atoms. BIL]3r and CHRISTEN (7) have questioned the use of
a constant interracial energy for particles t h a t are so thin, the identity
of the interface under such conditions being in doubt.
The physical significance of the 02~a parameter is also worth con-
sidering. From elasticity theory, FISHER £t a~. (3s) drive a value of 0
having the order of 101° dyne/cm~; if 02a3 is 9.9 × 10~ ergS/cm1~,
comes out to be 4.6 erg/cm 2. This low value is of the same magnitude
as the specific energy of a twin boundary, and seems to suggest t h a t the
austenite-martensite interface is completely coherent at the nucleation
stage. However, according to crystallographic analysis, (12) the interface
contains dislocations and is by no means completely coherent. KNXPP
and DEHLn~GER('~e) have recently calculated a value of 200 erg/cm 2
for a*, based on Frank's dislocation model (12~ of the interface. At the
same time, they made a detailed analysis of the strain energy factor,
and found 0 ~ A / 4 to be 5.2 × 10~ erg/cm 3. Substitution of these
numbers, along with Af _-- -- 44 cal/cm a from I ~ v ~ ' ~ and COHEN,aS~
into Eqn. 5.7 gives
AW~ ---- 1-8 × 10-8 erg/nucleus---- 2.7 × 108 cal/mol
for a 30 a[o nickel alloy at its M~ temperature. This magnitude (104 eV)
for the nucleation energy is prohibitively high.
The primary difference between the martensite nucleation described
above (ca) and t h a t developed by KU-~DJUMOV(~ lies in the formulation
of the equation relating N to T. In contrast to Eqn. 5.8, Kurdjumov
states t h a t
~" = K exp [-- Q/IcT] exp [-- A W ~ / k T ] nuclei/cm a sec (5.11)
where K is a constant and Q is the activation energy for the growth of

martensite plates. The factor K exp [ - - Q / R T ] replaces No~,/V m in
Eqn. 5.8. KURDJUMOV and M-~xI~ovA(95) have evaluated K and Q
from a study of the isothermal formation of martensite in a number of
steels. They find t h a t K ~ 108 cm -a sec -1 and Q ~- 600 cal/mol. In
* See f o o t n o t e , p. 200.
210
addition, they find values of AW~ equal to 1400, 700, 100 and 0 cal/mo|
at temperatures of 0°C, -- 30°C, -- 40°C and -- 50°C respectively for
a steel containing nickel and manganese.
This value of Q predicts a tenfold increase in the rate of martensite
propagation at -- 20°C over that at -- 196°C. However, B u N s ~
and MEHLC14) found that the rate of martensite propagation is inde-
pendent of temperature over this temperature range, implying Q ~_~ 0.
On the other hand, FISHER et al. (77~ assume that the activation energy
for coherent growth of martensite is nil, i.e. exp [-- Q / k T ] -~ 1.
5.2. Influence of Stress on Classical ~Vucleation

In Section 4.4 it was shown that the effect of applied stress on the M ,
temperature can be interpreted in terms of the mechanical work done
on or b y the transformiug region as the acting stresses are carried
through the transformation strains. M , is then quantitatively raised
or lowered, according to whether this energy term adds to or subtracts
from the. chemical free energy change of the reaction. On the other
hand, F I S H E R a n d T U R N B U L L (68) have accounted for the stress effect
b y considering the nucleation process itself.
The free energy of nucleation according to Eqn. 5.7 depends upon
Af. In the presence of external stresses, this quantity is modified b y
the mechanical interaction of the transformation strains with the
acting stresses, in a manner quite similar to that outlined before.
FISHER and TURNBULL~eS) have carried out these calculations for a 30
per cent nickel alloy, and the results are compared with the experi-
mental findings of Patel and Cohen in Table 5.1. Although the agree-
ment is not as good as that shown b y the athermal calculations in
T A B L E 5.1
Comparison o f Calculated ~6s~ a n d E x p e r k n e n t a l c'a~ Variations o f M , w i t h
Stress (after F ~ s m ~ and TU~rSVLL)
J
Stre~s mJstem SSress (lb/in 2) [ Ca~cula~ AM, Measured AM,
[
Uniaxial tension. . 15,000 ~- 25 °C W 15°C

Uniaxial compression . : 15,000 -t- 17"5°C W 10°C
H y d r o s t a t i c pressure 15,000 -- 7-5°C -- 8.5°C
Table 4.1, the correlations seem satisfactory. However, this circum-

stance does not validate the classical nucleation Eqn. 5.7 because the
calculations were based on the assumption that M , occurs at fixed
value of the net driving force, independent of temperature. On the
contrary, for the iron-nickel alloys, classical nucleation requires that,
because of Eqns. 5.7 and 5.9, A f 4 T ~ constant at M,.
911
5.3. HeterogeneousNucleation
The failure of homogeneous classical nucleation to account for M,
temperatures approaching absolute zero (eg) causes one to question
either the "homogeneous" part of the theory, or the "classical" part,
or both, as applied to martensitic transformations. There is abundant
evidence that these reactions are, in fact, heterogeneous.
The most straightforward clues that martensite nucleates at preferred
sites, rather than in a random fashion, come from metallographic
studies. With martensitic transformations that can be conveniently
reversed, as in beta brass,(~°~ the position and formation-sequence of the
plates on cooling are repeated almost precisely in the next cooling
cycle, after complete reversal of the transformation by heating between
the cooling cycles. This behaviour certainly suggests that preferred
nucleation sites exist in the parent phase. Along similar lines, the
martensite produced on cooling a single crystal of g o l d - 47.5 a/o
cadmium(11~consists typically of many plates; but ff the single crystal
is first well-annealed at a relatively high temperature in the austenitic
phase, the nucleation pattern on cooling is so changed that a single
unit of martensite can be generated, with only one interface sweeping
through the specimen. It is evident here that the nucleation is not a
random process.
Another stril~ing series of experiments by CEcil and TIYRNBULL(~2~
bears on this subject. Tiny particles of an iron-nickel alloy were
austenitized and quenched into the martensitic temperature range. It
was found that the reaction started at widely different temperatures
among the particles, and some did not transform at all down to the
lowest temperature reached even though they were of the same size
and composition. The conclusion to be drawn is that the nucleation
was quite heterogeneous in this alloy; some of the particles evidently
contained fewer effective sites than others, and hence tended to super-
cool further. It may be noted that although the particles were suffi-
ciently small to disclose the heterogeneity of the process, they were
more than large enough to allow homogeneous nucleation to come into
play if it were operative. Particles larger than 100 microns had M,
temperatures comparable to that of bulk specimens.
Accordingly, we shall now drop the "homogeneous" assumption,
and pursue the classical approach on the basis of preferred nucleation
sites. In a purely formal way, localized variations in free energy are
considered to be "frozen-in" during the cooling of the parent phase, and
are thus available to promote nucleation in a non-random manner.
In principle, these embryos may become critical in size when a suitably
low temperature is reached, without requiring thermal fluctuations,
although there may be instances in which the fluctuations can help
212
MARTENSITIC TRANSFORMATI01~S
the nucleation process. However, since thermal activation is not

necessary, the nucleation kinetics become essentially athermal, and the
extent of transformation is then governed primarily by temperature
rather than by time. During the cooling, the preferred sites are acti-
vated as the temperature drops, with the larger embryos being consumed
first and the smaller ones requiring progressively more supercooling.
5.4. Compositional and Structural Heterogeneity

In dilute solutions generally, and in iron-carbon anstenite particularly,
FtSHV.~ et al. ~3s~ have suggested that compositional fluctuations may
provide the preferred nucleation sites for the martensitic transformation.
These fluctuations are visuA.li~ed to be the statistical distribution of
compositional variations that exist in equilibrium at the austeniti~ing
temperature. If AW~ is the increase in free energy attending the
removal of all i carbon atoms from a region of average concentration,
then the number of these carbon-free regions per unit volume decreases
with increasing size of the region as follows:
n~ = No/V,~ exp (-- A W~/kT) embryos]cm3 (5.12)
where -~o/V,,, is the number of lattice sites per cm 8, and AWt depends
on the number of carbon atoms removed and hence on the size of the
depleted region. This leads to a steady-state size distribution of
carbon-free embryos which, of course, are all subcritical in size relative
to martensite formation at the austenitizing temperature.
Such compositional variations stand a chance of being "quenched
in" on rapid cooling because of the time required for diffusion, and
may become available to help nucleate the martensitic reaction
inasmuch as the driving force for the transformation will be greater
for carbon-poor embryos of a given size than for carbon-rich ones.
Moreover, because of the range in sizes of the carbon-free embryos,
the larger ones will achieve the critical size first on cooling below T O,
while the others will become critical at correspondingly lower tempera-
tures. Thus, in the martensitic temperature range, thermal agitation
is not necessary to produce volume fluctuations to reach a given
critical size; instead, the embryo sizes can remain stationary while
the required critical size for nucleation decreases with dropping
temperature (for example, see Eqn. 5.6). In this way, the nucleation
process takes on an athermal character.
On the assumption that the critical nucleus size at Me is independent
of composition (which is equivalent to a constant driving force at M~
according to Eqn. 5.6), I~ISHER et al. c3~ have used the foregoing treat-
ment to calculate the compositional dependence of the M e in iron-
carbon-chromium alloys. In this interpretation, no sharply defined
Me should exist because Eqn. 5.12 shows there is always a finite
213
(though very small) probability of having extremely large carbon-free

embryos in the austenite and these should become supercritical as soon
as the temperature falls below T 0. Consequently, it is necessary to
assume a value for the minimum number of martensitic plates per unit
volume that can be detected experimentally, and the largest embryo
size having this "abundance" is taken to be the critical size at ~ .
Still larger embryos can presumably nucleate into martensite above
Ms, but the quantities would be too small to find.
There are two limitations in regarding compositional fluctuations as
the basis for athermal nucleation. First, many martensitic reactions
take place in concentrated solid solutions where the dilute-solution
concepts are invalid. In fact, FISHER cs61 points out that for the iron-
30 per cent nickel alloys, the compositional fluctuations are unimportant
in the martensite problem. Furthermore, no evidence is available to
indicate that nucleation kinetics is fundamentally different in dilute
and concentrated alloys. Secondly, the hypothesis of compositional
fluctuations should lead to r a n d o m nucleation in view of the dynamic
nature of the embryo distribution prevailing at the austenitizing tem-
perature; on the contrary, the nucleation is decidedly heterogeneous.
COHEN et al. ~19~ attempted to circumvent the latter difficulty b y
postulating the existence of structural heterogeneities, consisting of
non-equilibrium lattice imperfections, internal surfaces, and local
strains due to accidents of crystal growth or plastic deformation.
Such "strain embryos," like arrays of dislocations, m a y well survive
the austenitizing treatment and thus provide centres of high free energy
for the heterogeneous nucleation of martensite. On the other hand,
heat treatment at extremely high temperatures might remove or re-
arrange many of these imperfections, thus leading to a reduction in
the number of preferred sites. In this way, READ and co-workers ~4~,71~
were able to convert single crystals of gold-cadmium and indium-
thallium alloys into single (or twinned) crystals of martensite.
Real crystals or polycrystals m a y be presumed to contain a spectrum
of these high-energy sites which, in turn, become active nuclei without
need of thermal fluctuations when the free energy required for nuclea-
tion (see Eqn. 5.7) is progressively lowered b y dropping temperature.
M~ is then determined b y the local free energy of the most potent
strain embryo or b y that number of high energy embryos that will
yield a measurable amount of martensite.
The condition of heterogeneous nucleation is adapted to the theory
of classical nucleation in Fig. 5.4, which shows the free energy change
per particle of martensite formed in an iron-30 a/o nickel alloy as a
function of size and temperature. In line with Section 5.1 the values
for a and 0 are taken to be 200 erg/cm z and 5.2 × 109erg/cm 3
respectively, and A W r--'~" is calculated from Eqns. 5.1 and 5.2. At
214
MARTENSITIC TRAI~SFORMATIONS
0°K, the critical nucleus size corresponds to a plate of r~ = 2.7 ×

10- e c m and AW~ ~ 4.6 x l0 T cal/mol, while at 250°K, r~ ~ 3.0r~
and AW~ ~- 9-0AW~. A frozen-in embryo of size 1.Sr~ is subcritical
at 250 and 200°K (points A and B), but becomes critical at 150°K
(point C) and is supercritical at 100°K (point D). Smaller embryos
achieve the critical size at progressively lower temperatures, and thus
Fig. 5.4. T h e free e n e r g y of f o r m a t i o n of m a v t ~ n s i t e p a r t i c l e s i n a n

i r o n - 3 0 p e r c e n t n i c k e l alloy
athermal nucleation can extend down to absolute zero if some of the

available embryos are sufficiently small.
However, a difficulty is encountered in connexion With isothermal
transformations in view of the fact t h a t AW increases very rapidly
with temperature. SH.~ et al. mm found that, in an iron-nickel-
manganese alloy, isothermal transformation occurs at 100°K which is
above M s , and the observed activation energy for the initial rate of
nucleation is about 7700 cal/mol. The calculated free energy of
nucleation at this temperature (Fig. 5.4) is 1-591J~ ~ 7.1 x l0 T cal/
tool. In order to account for the observed activation energy, which is
smaller t h a n the calculated value by a factor of almost 10a, it is
necessary to assume t h a t the most potent of the existing embryos are
virtually critical in size at 100°K, the latter size being 1.2r~ ~ 3.3 x
10 -6 cm. I f this is the m a x i m u m size of the embryos present in the
specimen, it is then possible to compute the activation energy t h a t
215
PROGRESS I~ METAL PHYSICS
should be observed at a higher temperature such as 150°K. This

quantity comes out to be 1.2 × l0 T cal/mol; it is the difference between
the free energy of nucleation at 150°K and the free energy of the
existing embryos of radius 3.3 × 10-~ era.* Such a high value would
preclude isothermal transformation at 150°K; nevertheless, isothermal
transformation is actually found at this temperature, and the activation
energy for the initial rate of nucleation is only 11,000 cal/mol. (3°~
To account for this discrepancy, the question arises as to whether
structural heterogeneities, like internal surfaces and dislocation arrays,
can effectively lower the interracial energy requirements in the forma-
tion of a martensite nucleus. It is evident from Eqn. 5.7 that, if a
were sufficiently small, A W~ could take on values which are more
consistent with the activation energies observed in isothermal trans-
formations. Such lowering of a would be phenomenologically similar
to the catalysis of droplets on dust particles in the condensation of
liquids from vapours. (eS~ However, even if a were decreased b y a factor
of 5, the foregoing difficulties would still prevail. Furthermore, it is
doubtful whether a can be reduced so much in the case of martensite
because the optimum shape of the plate is governed by strain as well as
interracial considerations. A droplet condensing from the vapour
phase can take advantage of an available free surface in the nucleation
process b y adjusting its shape to minimize the total interfacial energy,
b u t martensite nucleation would not appear to have this flexibility.
6. NOlO'CLASSICAL N U C L E A T I O l ~ T H E O R I E S OF
]VIAR T E N S I T I C TRAl~SFORM_ATIOl~S
6.1. HomogeneousNucleation
The use of the Boltzmann probability factor [exp (-- AW~,/kT)] in
nucleation-rate equations of the t y p e of Eqn. 5.8 has been criticized by
CRUSS~LRD.(~3) The Boltzmann factor stems from the customary
assumption that the atoms can be regarded as separate oscillators
having a characteristic frequency. Crussard states that the energy
for the thermally activated process under consideration is really
supplied b y statistical reinforcements of elastic waves, and proposes an
alternative probability factor based on quantum theory:
where B is equivMent to the free energy of nucleation (AW~ in Eqn.

* T h e a b o v e c a l c u l a t i o n s refer to a n i r o n - 3 0 a/o nickel alloy. F o r a 25 a/o nickel
alloy (as in Sec. 6.6), t h e e m b r y o r a d i u s t u r n s o u t to be 1.8 × 10 -s cm, a n d t h e corres-
p o n d i n g a c t i v a t i o n e n e r g y a t 150°K is 1 × 10' cal/mol.
216
MAI~TENSITIC TRAI~SI~ORMATION S
5.8), and ~ is the active frequency which depends on the size of the
nucleus.
Eqn. 6.1 is of particular interest because it predicts a finite rate of
isothermal nucleation at 0°K. I t has been suggested b y CRuSS~D (~4, ~5)
that the athermal and isothermal martensitic transformations at sub-
zero temperatures can be explained in this way, making it unnecessary
to resort to the concept of heterogeneous nucleation (Section 5.1). On
the other hand, the presence of lattice singularities m a y concentrate
the elastic waves and thereby enhance the probability of nucleation.
LruBOV and OsIP'¥A~ t~e) have recently shown b y quantum-
mechanical calculations that the transition probability for isothermal
nucleation via "tunnelling" at 5°K m a y be much larger than for the
case of classical nucleation, assuming a nucleation barrier of AW~
1000 cal/mol independent of temperature. However, the divergence
between the quantum and classical probabilities becomes small above
25-50°K. Furthermore, the classical nucleation barrier is 104 larger
(Section 5.1) than the A W~ value taken b y L y u b o v and Osip'yan, and if
the revised value were used in their calculations, the difference in the
transition probabilities would become very small indeed, even at
temperatures well below 25°K. Therefore, the "tunnel effect" cannot be
employed to support the concept of homogeneous nucleation in the
vicinity of, say, 120°K where the maximum rate of isothermal trans-
formation is observed. ~a°, 31~
At temperatures approaching absolute zero, Eqn. 6.1 might con-
ceivably have some advantage over the Boltzmann factor in permitting
a thermally activated process to occur. B u t the observed transforma-
tions at this very low temperature level have been athermal rather
than isothermal, and here again the homogeneous nucleation concept
turns out to be inadequate.
6.2. The Reaction-19ath Model

In classical nucleation, an embryo is promoted to the status of a
nucleus when it achieves the critical size at any given temperature. As
an alternative to this concept, it was proposed b y COHEN et al. (Ig) that
within the tiny volume where the nucleus forms, the lattice passes
through a succession of states as the atoms go through their primitive
motions to convert the parent phase into martensite. These move-
ments are visualized to take place in a co-ordinated fashion, with
the lattice strain providing a rapid sequence of intermediate structures
in any one region and then propagating out like a strain wave. The
succession of states is regarded as a reaction path which contains a free
energy barrier between the initial and final states. P r o m this stand-
point, activation is achieved b y fluctuations in the atomic configuration
of embryos, rather than in their size. Of course, interracial and strain
~5 217
energies must also be considered in the magnitude of the free energy

barrier which determines the free energy of nucleation.
COH~.N et al. (19) suggested that the embryos for this process are
strain centres comprising arrays of dislocations. Such strain embryos
are assumed to be "sheared" part way along the reaction path, thus
constituting local regions of high free energy. A distribution of these
embryos is depicted in the right-hand curve of Fig. 6.1. N F is so
defined that N~, d F is equal to the number of embryos per unit volume
having local free energies between F and F ~ d F . Thus, the area
/
Free energy of / Free energy
,%
4 distribution
~ of embryos
w~
I
I
r -----~ Ms
Fig. 6.1. Relationship between the embryo distribution and the

dependence of nucleation free energy on temperature (18)
ABC represents the number of embryos per unit volume with free
energies above the level F.
The left-hand curve in Fig. 6.1 indicates schematically the variation
of the nucleation free energy with temperature. Above Ms, the free
energy required for nucleation is in excess of that of the most potent
embryos, b u t on cooling below Ms, the free energy for nucleation drops
sufficiently so that some of the embryos "find themselves over the
barrier" and spontaneously move the rest of the w a y into martensite.
The athermal nature of the transformation is accounted for by embryos
of decreasing potency becoming operative as the nucleation barrier
diminishes with lowering temperature. Formally, this is exactly the
same approach as that used for heterogeneous classical nucleation
(Section 5.4) although it was suggested first in connexion with the
reaction-path model. (19)
The shape of the curve relating free energy to temperature in Fig.
6.1 can be ascertained from isothermal transformation kinetics.
NIxC~rr,r~ and COHEN(18) found that martensite forms isothermally in
an iron-30 per cent nickel alloy below M s. In order to avoid having
different amounts of athermal martensite at the various isothermal
holding temperatures, specimens were prequenched in liquid nitrogen,
thus converting all embryos with free energies above W a (77°K) to
athermal martensite, and then the specimens were up-quenched to
218
predetermined temperatures for the isothermal reaction measurements.

Under such conditions, and assuming that N F is independent of F in
Fig. 6.1, it can be shown that the initial rate of nucleation is
and
~(77°K) = N~,vk 77 nuclei/era ~ sec (6.3)
Dividing Eqn. 6.2 by Eqn. 6.3, and noting that at each holding
temperature the initial rate of nucleation is proportional to the initial
rate of isothermal transformation:
[ initial transformation r_aat_eat _T_ ]
In [initial transformation rate at 77°KJ
77 W~(T) -- Wa(77°K)
× -T---- - - RT (6.4)
The free energy difference [Wa(T ) -- W,(77°K)], referred to here as

the relative activation energy of nucleation, can be evaluated from
Eqn. 6.4. In reaction-path terminology, [W~(T) -- Wa(77°K)] is the
increment of free energy t h a t must arise from thermal fluctuations at
T in order to activate embryos of free energy Wa(77°K) which are the
most potent ones remaining after the prequench in liquid nitrogen.
The thermal vibrations at T are visualized to superimpose fluctuations
in strain, or displacement along the reaction-path, on the existing
strain embryos.
The isothermal kinetics of the iron-30 per cent nickel alloy display
a C-curve behaviour, there being a m a x i m u m rate of nucleation at
about 130°K. This leads to a temperature dependence of the form
given in Fig. 6.1 for both AWa[T] and Wa[T], since Wa(77°K) is
constant.
I n systems where isothermal martensite can be formed above M,,
the situation is somewhat more straightforward because athermal
martensite does not interfere with the observed kinetics. The C-curve
behaviour is then quite evident, as reported by S - , - et al. ~3°~ for an
iron-nickel-manganese alloy in Fig. 6.2. Particular pains must be
taken to avoid athermal martensite due to quenching stresses or
manganese loss from the surface of the specimens. The initial nuclea-
tion rate can be computed from the initial transformation rate, ~3°1and
then set equal to
= ntv exp [-- A W , ( T ) / R T ] (6.5)
where AWa(T)----[Wa(T ) - - F ] is the relative activation energy of
nucleation, referred to the free energy (F) of the most potent embryos
219
PROGRESS IN .WIETAL PHYSICS
t h a t p a r t i c i p a t e in the initial nucleation, n~ is the n u m b e r of these
p r e f e r r e d e m b r y o s p e r unit v o l u m e , a n d ~ is the lattice v i b r a t i o n
frequency. T h e initial n u c l e a t i o n r a t e s a n d relative a c t i v a t i o n energies
190 I ' o
-=" , I
,ac i
E
~- 11o
able amount of ~ "~r-,~
?o (°ba,,t 0.2"/o.~t ~ "~ ....
10 100 1000 10 0 0 0
Time sec
Fig. 6.2. Transformation-temperature-time curves showing progress
of isothermal martensite formation in samples of an iron-23.2 per cent
Ni-3"62 per cent Mn ahoy containing no initial mar*~ensite (after S~ti,
A v ~ A C E and COHEN (s°))
14oooI 1 ~ I ,320 ~,
i/)
i Booo f~ Wo 280 E
2,m}
co
-_~
-~ ~2ooc I
~ 11000 /y L:'O0~
2
160 ~"
E
•~= 9000 / \ 120~

g
£
8000 ," ~.~
\ 40
6ooo I I , 0
60 80 ICE) 120 140 160 180 2 0 0
Temperature oK
Fig. 6.3. The effect of temperature on the initial rate of isothermal

nucleation and the activation energy of nucleation in an iron-base alloy
containing 23.2 w/o I~i and 3.6 w/o lYln(81)
are p l o t t e d in Fig. 6.3, on t h e a s s u m p t i o n t h a t t h e r e is one m o s t - p o t e n t

e m b r y o p e r a u s t e n i t e grain t h a t nucleates first in the i s o t h e r m a l
reaction, a n d hence, n~ can be c o m p u t e d f r o m t h e grain size. I t comes
out to be 105 p e r c m 3 for a grain size of A S T M 1-2, in c o n t r a s t to 102s
p e r c m 3 for h o m o g e n e o u s n u c l e a t i o n (Eqn. 5.8). F o r a g i v e n initial
220
MARTENSITIC T:RANSFORMATIOl~'S
rate of nucleation, an increase in the assumed value of n t would have
little effect on the calculated magnitude of the relative activation
energies in Fig. 6.3. An increase in n~ from 105 to 10~8 per cm a would
only double AW~.
HOT.LOMON and TURNBULL(eS) have correctly pointed out that the
reaction-path model is incomplete ff the interfacial energy is neglected.
In the reaction-path concept, the coherency and misfit strains that
couple the embryo to the surrounding matrix m a y be regarded as being
localized in an interface, as suggested b y Frank (Section 6.3). ~ S H E ~
and T V - ~ B ~ L (~) treated the problem b y assuming the interfacial
energy to vary with the macroscopic shear angle of the martensitic
embryo, i.e.
o = +.o(¢/¢o) ,+ (6.6)
where a 0 and ~0 are the final interfacial energy and shear angle
respectively.
The strain energy is also a function of the shear angle :
rrr~c
AG = ]lrrc*G~ * ~- - ~ - [G(~bo -- ¢)~] cal/embryo . (6.7)
where the first term on the right is the strain energy in the surroundings
and the second term is the strain energy in the embryo. Eqns. 6.6 and
6.7 can now be substituted into Eqn. 5.1 to give the net change in free
energy (AW ~-~=') accompanying the formation of lenticular embryos
of dimensions r and c and shear angle ¢. I f ¢ = ¢0, AG, ~ lrrc2A since
A -~ ]Gel in Eqn. 5.1.
The critical values, r+, c+ and ¢ , can be derived b y setting
~AW~+"~'/~r = ~AW~+~']~c = ~AW~-+~']~¢ ~ 0. At the M~ tempera-
ture, ¢ is 96-91 per cent ore0 for values of n in Eqn. 6.6 lying between
0 and 2, and taking ¢0 to be about 116. Thus, unless the exponent n
happens to be unusually large, the critical shear angle is not appreciably
different from the ultimate shear angle. Although the corresponding
nucleation energies range from 93 to 63 per cent of the classical nuclea-
tion value as given b y Eqn. 5.7, FIs++~+~ and Ttm~+BmUt+(++) conclude
that these differences are not significant and that the two theories
yield essentially the same result.
MAC~ (5") has argued that FISHER and TUR~BVLL <vv) have not
really dealt with the reaction-path concept in the foregoing calculations,
but with a shear-strain corrected nucleation-and-growth model. I f the
configuration fluctuations are pictured to occur in regions of various
sizes, it is evident that nucleation can only occur along the reaction
path in volumes larger than that given b y classical theory (Eqn. 5.6);
otherwise the martensitic particles would tend to shrink because of the
interfacia] energy requirement. Hence, the free energy barrier along
221
the reaction path must be at least as large as that indicated by classical

theory. In order to calculate the actual reaction path, it would be
necessary to know the free energies of all the intermediate states, but
this information is neither available nor in sight because of the com-
plexity of the problem.
Nevertheless, in Machlin's view, the nucleation rate may be faster
via the reaction path because of a larger frequency factor. Here, the
number of nucleation-attempts per embryo per second can be identified
with the lattice vibration frequency because configuration (rather than
volume) fluctuations are involved. According to MACHLL~,~55J the
classical treatment requires that atoms be transferred from one side of
an interface to the other, and hence the frequency of nucleation
attempts depends upon a probability factor exp (--Q/RT), where Q
is the activation energy for the atom-transfer process. Even though
may be much smaller than the activation energy for normal diffusion,
the probability factor is considerably less than unity and the frequency
factor for nucleation is correspondingly less than ~.
This argument has been countered by FISHER and TURNBULL~s~ who
state that an atom-by-atom transfer process is not involved in the
volume fluctuations of the martensitic embryo (only the motion of a
coherent interface); hence Q ~ 0, and the above probability factor
approaches unity. Unfortunately, the stalemate persists because up to
the present time, experiments have not critically distinguished between
the classical and reaction-path models.
6.3. Frank's Model of the A ustenite-Martensite Interface

K ~ P P and Dmrr~TNG~.~(Se~ have recently treated the kinetics of
athermal martensite formation in steel by adopting FRANK'S{12}model
of the austenite-martensite interface. The geometry of this interface
will be reviewed here in order to provide a suitable basis for discussing
the Knapp-Dehlinger kinetic analysis.
Considering that the interface lies parallel.to the (225)r habit plane,
FRA~-K~12~ has suggested that close-packed planes of the two phases
meet along close-packed rows. In converting one lattice (f.c.c.) into
the other (b.c.t.), a shear of 1/~/32 and a dilatation of about 5 per cent
give both the observed macroscopic shear and the proper spacing
between the close-packed rows when the contact plane is (225)~. How-
ever, this does not yield the correct final structure because the atomic
arrangements in the close-packed rows do not match properly. The
latter adjustment can be accomplished by a heterogeneous shear due
to an array of screw dislocations, lying parallel to the close-packed
direction, one between every sixth close-packed plane and comprising
the (225)~ interface. Because of its heterogeneous nature, this shear
produces no macroscopic change in shape, and the observed 1/~/32
222
shear is not affected thereby. The growth of the martensitic plate

consists of a "thickening" process in which the interface of screw
dislocations moves ahead into the austenite. The transformation
strains take place in the interface as the austenite is converted into
Virgin I (m)y (a)

ausf.enite I
i
Extens~c~10]~[ (m~ 1 (~)
Shear (¢)
plExt
usesheor
nsion (cO
--.-[11~]~
If l l I I / T
in~-,rfoce
Fig. 6.4. l ? l ~ ' s model of ,,he austenif,e-m~wtensit~ interface(ll)
martensite, with the interface furnishing a highly mobile linkage

between the two phases.*
The nature of the dislocation interface in Frank's model is illustrated
schematically in Fig. 6.4. Fig. 6.4A is a close-packed plane with the
close-packed rows running vertically. The principal contribution to
the overall dilatation results from an extension of the close-packed
rows, which is indicated in l~ig. 6.4B. The 1/~/32 macroscopic shear is
* T h e screw d i s l o c a t i o n s n i c e l y a c c o u n t for t h e h e t e r o g e n e o u s s t r a i n , b u t do n o t
p r o v i d e a m e c h a n i s m for t h e m a c r o s c o p i c d i s p l a c e m e n t s , cgs~ S o m e s u g g e s t i o n s r e g a r d i n g
t h e l a t t e r h a v e b e e n p r e s e n t e d r e e e n t l y , e m p l o y i n g a s e c o n d s e t o f dislocations. ~7. sa, .l~
223
PROGRESS IX ~ETAL PHYSICS
not shown, because this occurs in another dimension. Figs. 6.4c and D
indicate the heterogeneous shear achieved by the screw dislocations.
giving no macroscopic change in shape. The orientation of the disloca-
tions in the (225)v interface is illustrated in Fig. 6.4E; except for these
dislocations on every sixth plane, there is coherency between the two
lattices, and the interface is a plane of zero average strain.
The atomic arrangements in the (li0)v and (llT)~, planes which are
o .... o (~ustenite)o- --a (M~rten£ite) /~¢
Fig. 6.5. 1~artensite--austenite inl;erface as seen in the [~10]v a n d

[II1]~, directions (after KN.~PP(~6'
normal to the dislocations in question, are shown in Fig. 6.5. It is

seen that the spacing between the dislocations is
d = 6a~ 1 . . . . (6.8)
%/2 " cos
where cos y = 0-87. F o r a~ = 3"6 •, d = 17"8 .~. The macroscopic
shear of 1/%/32 arises from the conversion of triangles of type Y in the
austenite to triangles of type Z in the martensite. This shift also
involves the aforementioned dilatation.
The next step is to compute the interracial energy from the disloca-
tion geometry in the interface, and also the strain energy arising from
the dilatation and macroscopic shear. Although the following calcula-
tions b y Knapp and Dehlinger are worked out only for the (225)~
interface, the (111)~ and (259)v cases could also be treated similarly,
in principle, b y the three-dimensional prism-matching technique* of
B I n B Y et al. ~7, 92~ With this procedure, one attempts to derive an array
* The concept of'prism matching was first proposed by Bilby and Frank at the Third
International Congress of Crystallographers in Paris, July 1954.
224
MARTENSITIC TR~.I~SFO]~MATIONS
of interface dislocations consistent with the observed habit orientations,

lattice relationships and interplanar spacings.
It should be noted that the F r a n k model presupposes the existence
of a suitable interface, and deals only with its crystallographic aspects.
The problem of nucleation, or how this interface comes into being, is
not considered.
6.4. Knapp-Dehlinger Treatment of the Athermal Transformation

Using the dislocation model for the austenite-martensite interface,
K ~ P and Dv.~v~eER c~) extended ~m~cx's (1~) analysis by regarding
d= 2-~ C0$~ and [11@
t~=2~2 2C l ~0]'
' Sir
I
Positive . ~ L J~ ~.Dislocation
screw loops
Fig. 6.6. KNAPPand D~Hr.r~GER'smodel of the martensite
embryoCSe~
the embryo of martensite as a thin oblate spheroid, surrounded by

loops of dislocations (Fig. 6.6). These loops are assumed to consist
primarily of screw dislocations, with short edge components joining the
positive and negative screws. Growth of the embryo in the [li0]v and
[225]v directions is achieved by expansion of the dislocation loops.
However, growth in the [55~]r direction requires the generation of new
loops. The co-operative movement of these loops leads to the thickening
and radial propagation of the martensitic embryo as the dislocation
interface passes through the austenite.
The energy to form and expand the dislocation loops (inteffacial
energy) must be supplied by the chemical driving force Af ='-~v. This
is also the case for the strain energy set-up by the macroscopic dis-
placements. KNAPP and D E H L I N G E R (56) a s s u m e that an existing
225
embryo will "trigger" spontaneously into full-size plates on cooling as

soon as the chemical driving force exceeds the required interracial and
strain energies. In other words, the initiation of the athermal trans-
formation is determined by a free energy balance at A W ---- 0 (Eqn. 5.1)
rather than by the maximum or saddle point in the free energy barrier
at d ( A W ) = 0 (Eqns. 5.2 and 5.3).
Eqn. 4.1 can be rewritten:
AW ~-~' = A F r-*~" + AGr-*~' cal/particle . (6.9)
or referring to a unit volume of martensite, and dropping the ~ -~ ~'
superscripts for convenience:
Aw = Af + Ag cal/cm a (6.10)
Af is now the change in chemical free energy attending the formation
of 1 cm 3 of martensite, and is negative at temperatures below T 0. Ag
is the corresponding nonchemical free energy change, and is positive.
Aw is the net change in free energy per unit volume of martensite
formed.
The volume of the embryo is ~rr2c and the inteffaeial area is approxi-
mately 27rr2, where r is the radius of revolution and c is the semi-
thickness.* According to Eqns. 4.2 and 4.3,
A G = A G s + AG,
Ac
----- a × (interracial area) -+- - - X (volume) cal/particle (6.11)
r
Dividing through b y the volume of the martensitic embryo, we have
3a Ac
Ag = ~ - --r cal/cma (6.12)
For a given radius, Ag can be minimized relative to the semi-thickness

(~Ag/~c) = 0), and the minimum value of Ag is found to be
Agmi n ~ ~ --
C
when
[3ar] 112
c = [~-~) cm (6.14)
However, Agmin is more properly evaluated for a given volume of the
martensitic embryo ( in which case dr = -- 2c

r dc) " E q ns" 6 " 1 3 a n d
6.14 then become
* I t s h o u l d be n o t e d t h a t t h e m a r t e n s i t i c e m b r y o is o f t e n r e g a r d e d a s a l e n t i e u l a r
p l a t e o f v o l u m e ~rr~c i n s t e a d of ~rr2c for t h e o b l a t e s p h e r o i d a s s u m e d h e r e b y K n a p p
and D e h l i n g e r . T h e a p p r o x i m a t e i n t e r r a c i a l a r e a 2=r I is t a k e n t o be t h e s a m e for t h e
t w o cases w h e n c ~ r.
226
MARTENSITIC TI%AN S F O R M A T I O I ~ S
(6.15)
and
c= cm (6.16)
I t is evident that there is relatively httle difference between the two

sets of equations; the latter set is adopted here although Knapp and
Dehlinger have employed the former.
Based on values of o = 200 erg/cm 2. and A = 500 cal/cm~ =
2.1 × 1010erg/cm 3, as derived b y Knapp and Dehlinger, Ag~n can
t ] i5°°
'~,.,=-} ~ - 4oo e
_ -IIIIIII1~
~ c~
0 = 2 0 0 ~--~--- 3 0 0
A-- 5 0 0 c---
¢~~
-- 2 0 0
(v
¢v
g
~ o ~
r at Ms"
- 14
=4t" .d ~ -~0
-200
-300
iO-~ 10~ iO-S i0-~ 10-3 I0-~
Embryo radius, r cm
Fig. 6.7. E f f e c t of e m b r y o size o n t h e r e s t r a J n l n g force Agmin a n d t h e

n e t d r i v i n g force ~ w a c c o r d i n g t o KI~APP a n d DEKLINGER (56)
now be calculated as a function of embryo size. A is obtained from

elasticity theory, following the treatment of KRO~V.R.c~9~ Because of
the agreement between FmKER et al. c35) and K n a p p and Dehliuger
on the magnitude of A despite quite different methods of approach,
the details will not be considered here. However, the disparity in
a(200 versus 24 erg/cm~) , is too large to be ignored; this point is
taken up in the Appendix. Fig. 6.7 shows the variation of Ag~n with
r, assuming the c-dimension given b y Eqn. 6.16 is maintained. There is
a monatouic decrease of Agmin with increasing size of the embryo.
This means that (a) the larger the embryo, the smaller is the restraining
force or back stress which opposes the transformation, and (b) once an
embryo starts to grow, the restraining force falls off progressively.
For a given embryo, one visualizes that A f t = ' in Eqn. 6.10 becomes
* See footnot~e, p. 200.
227
PROGRESS I1~ I ~ I E T A L P H Y S I C S
more negative as the temperature drops until - - A f r-~' or ~f~'-*r

exceeds ~g~n for the size in question; then the chemical driving
force overbalances the nonchemieal restraining force, and a net driving
force arises (~w r-*~' becomes negative). This can be likened to a physical
stress on the dislocation interface, which is thereby moved outward to
generate martensite from the existing embryo. At the temperature
where this event takes place, the propagation can be quite rapid
because, not only is the interface mobile, but the net force on it
increases as the particle expands. (The chemical driving force remains
constant at a given temperature, while the nonchemical restraining
force decreases with particle size according to Fig. 6.7.)
The embryos that participate at the M~ temperature are presumably
the largest ones available. Since Af~'~*r at M~ is about 42 cal/cm a
(300 eal/mol) in steel, A g ~ for the largest embryos is also equal to
this value, and the corresponding radius from Fig. 6.7 is 8.5 × 10-e cm
(380 atoms). Eqn. 6.16 then gives the semi-thickness as 3.5 × 10-7 cm
(15 atoms). Such embryos therefore contain about 8 × l0 s atoms.
Smaller ones are restrained from transforming (because of larger
Agn~n values) until the temperature is lowered to the point where Af ~'-*~
becomes sufficiently large. Thus, the athermal nature of the reaction
is explained in terms of a distribution of embryo sizes.
In line with the proposal of COH~.~ et al., (19~ Knapp and Dehlinger
regard the embryos as preferred nucleation sites originating from
dislocations in the austenite. However, the latter authors have put
this concept on a more quantitative basis. They suggest that the
interaction between certain pairs of edge-screw or edge-edge disloca-
tions generates a region in the austenite that is geometrically similar
to martensite. Such embryos have a relatively high value of 5g
because they are more-or-less cylindrical in shape, but this non-
chemical free energy can be reduced to Ag~n if the embryo then
changes its shape to a thin oblate spheroid, with a dislocation interface
of the type described in Section 6.3 and dimensions satisfying Eqn. 6.16.
At temperatures above To, these martensitie embryos in the austenite
cannot grow because both the chemical driving force and the net
driving force favour the formation of austenlte. However, the embryos
do not disappear because the dislocations cannot be "relaxed out";
since dislocations persist despite extensive annealing, the proposed
embryo configuration m a y be pictured as the most tolerable way of
accommodating these imperfections. Below T O but above Ms, the
chemical driving force becomes favourable for martensite formation,
but the net driving force is still unfavourable; that is, even for the
largest embryos, the nonchemical restraining force Agm~ exceeds the
chemical driving force. Below M~, the chemical driving force exceeds
Agmia for the larger embryos, and the net driving force then acts as a
228
stress (erg/cm3 = dyne/cm s) to move the dislocation interface out

into the anstenite.
The temperature dependence of A/~'~y has been treated in detail in
Section 3. Agm~ also varies with temperature because it is proportional
to the shear modulus which, in the range of 250-I000°K for iron,(8°)
can be expressed approximate]y as
G = ~[1.5 + (1000°K -- T°K)10-3]10 ~ dyne/cm 2 (6.17)
This variation is usually neglected in calculating the nonchemical
factors, b u t it probably accounts (at least in part) for the fact that the
driving force at M , increases with decreasing M~ temperature in iron-
nickel alloys (Section 3.4). As the nickel content is raised and M a
displaced to lower temperatures, Agmin for any given embryo becomes
larger in view of the larger shear modulus at the lower temperature.
Thermal activation plays no role in "triggering off" the embryos in
this model, and hence does not control the kinetics. Accordingly, in
principle, the transformation can continue down to 0°K, with smaller
and smaller embryos becoming operative. At the temperature where a
given embryo starts to propagate, about 60 per cent of the chemical
free energy change goes into enlarging the dislocation interface and
40 per cent into strain energy. B u t as the martensitic particle expands,
with a corresponding decrease in nonchemical requirements per unit
volume, there is a rapid acceleration of the moving interface because
of the progressive increase in the net stress on the dislocations. B y
utillzlng BURI~1~A~DT'S{sl) relationship between the velocity of disloca-
tions and the applied shear stress, K n a p p and Dehlinger were able to
calculate the velocity of the martensitic interface as a function of
particle size. From this, the time of formation of a plate was computed
to be 10 -~ sec, in good agreement with the values of 0.5 -- 5-0 ~ 10-?
sec measured b y BUNS~AH and MEHL. a4) f-~ .....
W h a t stops a martensitic plate, once started ? An obvious answer is
t h a t physical barriers, such as other plates, grain boundaries, sub-
boundaries, and inclusions can block the propagation. However, this
cannot be the full answer because when r is prevented from increasing,
the plate should be able to thicken (thus departing from Eqn. 6.16
which minimizes Ag) and the corresponding increase in Ag should
ultimately equal Af ~'-~, thereby reducing the net driving force to zero.
This is the condition of thermoelastic martensite; a decrease in tem-
perature should then cause the plate to thicken further as the chemical
driving force is increased. Although such thermoelastic behaviour is
observed in some martensitic systems (see Section 4.5), it has not been
found in iron-base alloys.
I t is suggested b y the present authors that propagation-stoppage can
also occur irreversibly b y jamming of the dislocations in the interface.
229
PROGR:ESS I~ METAL PHYSICS
Imperfections in the path of the interface or accidents in the co-

operative movements m a y well tangle the dislocations, thus helping to
reduce the close-coupling of the two lattices and the mobility of the
interface. Such processes can lead, not only to large deviations from
the optimum shape and hence to an excessive back stress, b u t also to
the destruction of the propagation mechanism itself. When stoppage
occurs in this case, no thermoelastic behaviour is to be expected.
This mode of self-stoppage probably accounts for the fact that
martensitic plates formed in single crystals of austenite often do not
extend across the entire crystal, and yet there m a y be no microsco-
pically visible barriers. Stabilization of martensitic transformation
b y plastic deformation of the parent phase ~s2~ is likewise attributable
to the jamming of the propagation linkage between the two phases.
6.5. Comparison of the Knapp-Dehlinger Model with

Clazsical Nucleation
I t was shown in Section 5.4 that the existence of a spectrum of pre-
ferred nucleation sites due to structural imperfections permits the
classical nucleation theory to account for the athermal mode of the
martensitic transformation, b u t there are quantitative difficulties when
applied t o the isothermal kinetics. Knapp and Dehlinger did not put
their model to this test. The present authors carry out this step in the
following section, using arguments that have not been published
elsewhere. However, it is first necessary to highlight the essential
differences between the Knapp-Dehlinger model and the classical
nucleation concept.
In classical nucleation theory, the net free energy change A W ~'*~"
accompanying the formation of a martensite embryo or particle is
expressed as a function of r and c, taking the chemical, interfacial
and strain energies into account as in Eqn. 5.1. This equation was
wL-itten for lenticular embryos; for oblate spheroids, it has the form:
AW ~-*~' ---- 2rrr2~ + -~ r2c A f r-*~" + cal/particle . (6.18)
The AW r-*~' surface, plotted in Fig. 6.8, has a saddle point whose
height is AWa~ ---- 32rra3A~]3Af 4. In order for a very small embryo to
attain detectable values of r and c, the system must pass over a free
energy barrier, and the saddle point is the minimum (therefore, the
most probable) barrier. For a given embryo, nucleation is achieved
when the dimensions become commensurate with the r and c co-
ordinates of the saddle point (Eqns. 5.2 and 5.3).
The most probable path for growth, i.e. the relationship between
the dimensions r and c as the volume increases, is defined b y the
230
minimum value of A Wr-*~' for a n y given embryo volume. This relation-
ship is obtained by expressing A W~ ' as a function of the embryo
volume and, say, c and then setting ~AWr-*~'/~c at constant volume
equal to zero. Regarding the embryos as lenticular, then
[ 4crr~112
c = om (6.19)
Fig. 6.8. Schematic representation of the free energy of a ~ n s i t e

embryo as a function of embryo radius and semi-thickness showing
saddle point and minimum energy path
I f the embryos are taken as oblate spheroids, the above equation

becomes
c = cm (6.20)
Fig. 6.9 shows the AW y-~' values as a function of volume along the
most probable growth path, calculated from Eqn. 6.18 under the con-
dition of Eqn. 6.20. The saddle point of Fig. 6.8 lies at the top of this
path. Larger embryos of this shape are supercritical, and can grow
spontaneously with decrease in free energy. The act of jumping this
barrier by existing embryos, whether via thermal activation or decrease
in the barrier height because of dropping temperature, is a true
nucleation process. However, in order to make this a likely event, it is
231
PROGRESS IN 5fETAL PHYSICS
necessary to presuppose such low values of a that they are inconsistent

with the dislocation model of partially coherent interfaces.
In contrast to the classical treatment, Knapp and Dehlinger do not
really deal with the problem of nucleation at all, but with the con-
ditions for propagating an existing embryo. An embryo is regarded as
supercritical when the net driving force favours the formation of
martensite, that is when A W~ ' < 0. Since the n u c l e a t i o n of a particle
is not involved, the thermodynamic condition for propagation can be
put on a unit volume basis: Aw ~ Af-~ Ag < 0, dropping the
X --~ ~' superscripts for convenience. Any given embryo is pictured to
adopt a shape that minimizes Ag, and it turns out that the dimensions
are then defined b y c/r -~ ( a / A r ) 112 (see Eqn. 6.16), exactly as in the case
of classical nucleation (Eqn. 6.20). However, even when Ag is thus
minimized (Eqn. 6.15) for a given embryo volume, it opposes the
transformation, and exerts a back stress that tends to prevent the
dislocation interface from moving out. The latter can only propagate
with appreciable velocity when the chemical driving stress exceeds the
back stress. In Fig. 6.9, this condition is achieved, not at the top of the
barrier, but at the larger embryo volume where AW (or Aw) passes from
positive to negative. Then the interface, if it is mobile, starts to move,
and because the back stress decreases with increasing particle size,
the growth accelerates rapidly.
This model requires the pre-existence of embryos, since the back
stress is infinite for embryos that have to start from "scratch." This
is indicated b y Eqn. 6.15 which shows that A g ~ approaches infinity
as r and e approach zero. Accordingly, nucleation in a perfect
lattice becomes impossibly difficult, and one does not deal here with a
barrier-type of free energy curve, as in Fig. 6.9, but with a mono-
tonically decreasing curve, as in Fig. 6.7. In this view, if the
martensitic transformation can take place at all on cooling, it must
result from meeting the conditions for "triggering off" existing
embryos, not from getting the embryos over a nucleation barrier.
Some insight into the nature of the pre-existing embryos m a y be
gained from the following considerations. Above To, an embryo of
martensite would appear to be unstable, first because bulk martensite
has a higher free energy than bulk austenite at such temperatures, and
secondly because both interracial and strain energy would be associated
with martensitic embryos. However, the austenite undoubtedly con-
tains dislocations which increase the free energy of this phase, and it is
conceivable that the collection of these imperfections into an interface
(which would greatly reduce the strain energy surrounding the indi-
vidual dislocations) might result in a sufficiently large decrease in free
energy to permit the formation of martensitic embryos. Pursuing this
line of attack, we m a y assume that the core energy per unit length of
232
dislocation is the same whether the dislocations are arranged at random

(state 1) or in an austenite-martensite interface of the ~ N K type (1~)
(state 2). Taking the strain energy surrounding the interface disloca-
tions to be essentially nil compared to t h a t of the random case, we have
---- --: -- niP, per embryo (6.21)
where n = 2r/d ---- number of dislocation loops per embryo, [ ---- ~r =

mean length of the dislocation loops, and F~ =- strain energy per unit
~w-vl
--
5 ~3~2
[ 2 ~ ~ 3
4_~hJ/~A,
~ -- *J'J
I
1 Vt r~8~'a3a2 4 4w.
= 3.4 f5 = .4f
~'~ At Ms
~ = x:)-s ergs
~ V , = ----.2.3x10-17cm3
~0.' ,~W. \
.£
v~
Fig. 6.9. Free energy of isothe~'n~l nucleation of m ~ . ~ i t e as a func-
tion of embryo volume according to the classical nucleation theory.
(Calculated for eUipsoidal embryos)
length of random dislocation. Substituting for nl in Eqn. 6.21, we

have
A i i n ~ = ~rr% ( A f r'*~" ~-c--~Ar) -- 21rr~PJd p e r e m b r y o (6.22)
which is analogous to Eqn. 5.1a except t h a t Af ~ " is now positive

because T > T 0, and the last energy term is negative because of
the available dislocations.
For the case of embryos of optimum shape, Eqn. 6.16 holds, and
then the optimum size of the embryos at any given temperature is
found by setting
~V 1~2 ~ ~/V1 ~ 2
ar ac
6FJd -- 4a
c, = 5Aft--'=" cm (6.23)
r , = c~, A / ~ cm (6.24)
z6 233
/~ W 1-*s is a minimum for these conditions, and thus c, and r . m a y be

regarded as the dimensions of the most probable embryo. I t m a y be
noted that c. and r . become larger with decreasing temperature (still
above To) since Af ~-*=' diminishes. The coalescence of embryos on
cooling indicated by Eqns. 6.23 and 6.24 depends, not only on the
existence of dislocations in the austenite, b u t also upon their ability
to "diffuse" into suitable arrays for the interfaces. Consequently, the
optimum dimensions can only be attained at temperatures where the
dislocations are sufficiently mobile. In order for embryos of a given
size to coalesce on cooling, it is necessary for the dislocations to
rearrange themselves into a new set of interfaces corresponding to the
larger embryos. This is in contrast to the spontaneous growth of
embryos into martensitic plates below M , where the chemical driving
force becomes large enough to expand existing dislocation loops and
create new ones. Since the coalescence of embryos m a y actually
cease at a certain temperature above To, they m a y become "frozen in"
and be available for propagating into martensite when l)/, is reached.
According to COTTRELL, (83)
F , - Gb2 [ln r~ _ 1] erg/cm (6.25)

4tr L r°
Taking r° ---- 10-7 cm for the dislocation core and r 1 ---- 0.5 × 10 -4 cm for
one-haft the mean distance between dislocations in an annealed metal,
F, is computed to be about 5 × 10-5 erg/cm. In Section 6.4, c at M ,
for an iron-30 per cent nickel alloy is found to be 3.5 × 10-7 cm, which
can now be substituted into Eqn. 6.23 to ascertain Af r*~'. The latter
corresponds to a temperature of about 515°K (75°C above To) which
m a y be construed as the temperature level at which the embryos
effectively stop coalescing on cooling and become "frozen in." Although
these calculations are extremely rough, they indicate the feasibility of
forming "chemically unstable" embryos of martensite in the parent
phase at temperatures well above T O.
The interface of the embryo m a y be described as a special type of
polygonized boundary; it must contain arrays of dislocations to
achieve both the macroscopic and heterogeneous strains. These
strains are taken into account in the energetic calculations, with the
former contributing to A and the latter to a. The atoms surrounded
by the network of dislocations are thereby displaced into a configura-
tion similar to that of martensite. Although this embryonic structure
m a y not be identical to martensite, it is so assumed in Eqn. 6.22 in
order to have some basis of calculation.
The dislocations in the austenite are strained regions of high local
free energy, (56, s6) even higher than that of martensite at the austenitiz-
ing temperature. If only a few of the proper types of dislocations
034
interact, this could reduce the local free energy in two ways: (1) b y
converting the strained region to martensite, and (2) b y eliminating
the strain fields around the individual dislocations. Thus, there is a
tendency for the participating dislocations to "stick," and for others
to be added to the array until an optimum size is reached at the given
temperature (Eqns. 6.23 and 6.24). Further lowering of the free energy
due to collapsing or self-annihilation of the dislocations m a y be
prevented b y the interlocking of the various types involved.
6.6. Athermal and Isothermal Formation of Martensite at

Preferred Nucleation Sites
According to classical nucleation theory, the critical embryo radius for
spontaneous growth of martensite (ellipsoidal shape) is
4aA
r~ = A~---~cm (6.26)
~at
When r = r~,,
32~aaA ~
AW~ AW~ - - - - cal/particle (6.27)
3~f 4
The basis for these relationships is given in Section 6.5. On the other
hand, the theory proposed b y K n a p p and Dehlinger (Section 6.4)
implies that although spontaneous growth can occur when r > r~,
cataclysmic growth of martensite cannot take place until Aw ----- A W
0, where
25aA
r o = 4~f~ cm (6.28)
From a mechanistic point of view, cataclysmic growth involves the

creation of new dislocation loops as the plate expands in the [55~]~
direction (Fig. 6.6) as well as the expansion of existing loops.* The
creation of these loops cannot occur, according to K n a p p and Dehlinger~
unless the "chemical" stress Af exceeds the backstress Ag~n, or the
net stress, Aw, is negative.
Although it is recognizably difficult to create a dislocation loop
spontaneously in a virgin lattice, the question naturally arises as to
w h y is it possible to form one when r ~ r 0 and not when r ~ r~.
When regarded in this perspective, the Knapp-Dehlinger condition of
r ~ r e as the prerequisite for cataclysmic growth appears to be an
intuitive choice. However, the Knapl>-Dehlinger approach does
* T h e r e is n o p r o b l e m h e r e a b o u t e x p a n d i n g t h e e x i s t i n g loops b e c a u s e t h e i r e n e r g y
is i n c l u d e d in t h e interfacial energy, and c a n be s u p p l i e d f r o m t h e overall decrease in
free energy attending spontaneous g r o w t h o f particles, w i t h r > r . .
235
"open the door" to a more general treatment of heterogeneous

nucleation encompassing both the athermal as well as the isothermal
transformation.
Suppose there is an embryo radius r~, such that when r ~ re cata-
clysmic formation of martensite can occur, i.e. new dislocation loops
can be generated athermally (in the Knapp--Dehlinger hypothesis
r~ ~ ro). I f r~ ~ r~, then there exists a range of sizes r~ ~ r ~ r~ in
which the embryo can grow spontaneously (i.e. lower the overall
free energy A W, by expanding) if thermal fluctuations facilitate the
formation of new dislocation loops. Embryos having sizes less than r~
cannot expand due to the extremely high value of A W~ (Section 5.4).
Consequently, ff a specimen containing a distribution of embryos is
quenched to a temperature where some of the embryos have sizes
greater than r~, athermal formation of martensite will occur. Subse-
quent isothermal holding may result in the isothermal growth of
embryos with r , ~ r ~ r~ to size r~ at which point the embryos could
transform cataclysmically into martensitie plates. Clearly the problem
before us is to consider the conditions under which an embryo can
grow cataclysmically.
The most direct method would be to consider the formation of a
dislocation loop in the parent phase at the tip of an embryo having a
radius r and semi-thickness c at a temperature where the driving force
is A / . FRANK and STROH(s4~ have considered an analogous problem for
the case of the propagation of a kink-band having the shape of a thin
elliptical cylinder. These authors show that the maximum shear stress
occurs at a distance equal to c2/r from the tip of the cylinder and is
equal to G~,o/V/-3, where Y0 is the kinking angle. Since c2/r is constant
for the martensitic problem at hand (approximately 1 A) and ro-----
1/%/3-2, it follows that the maximum shear stress is G/IO, and is con-
centrated near the tip of the ellipsoidal embryo. This would constitute
the effective stress acting to form a new dislocation loop.
This approach leads to many complications since all of the individual
loops are coupled together. Considered individually, some of the larger
loops m a y be supercritical, i.e. if separated from the rest of the embryo
they would expand spontaneously, while others near the ends of the
embryo m a y be subcritical, i.e. they would tend to shrink ff isolated
from the rest of the embryo.
In order to circumvent these difficulties, it is convenient to take
advantage of the fact that in dealing with embryos having r ~ r~ we
are concerned with thin ellipsoids having c/r ratios less t h a n l ] 2 ~
Under these circumstances, the array of screw dislocations m a y be
replaced (s4~ by a circular dislocation loop lying in the (225)~ plane which
follows the edge of the embryo. This equivalent loop of radius r
contains, in effect, the interracial energy of the embryo, and has a
236
MARTI~NSITIC TRANSFORMATI01~S
Burger's vector which is a p p r o x i m a t e l y equal to cb/d. The energy of

such a " g i a n t " loop W t is
W ~ = ~(2rF~ ~ ~z~r~) erg/loop (6.29)
where ~t is the active stress on the loop and F~ is the line tension of
the loop. Since (sT)
r,= G22 erg/em of loop (6.30)
and
c2 = ~- cm ~ (6.31)
then
Wt = ~b [ Gb ~ ) (6.32)
"-d [x-~ ar -- "rtr2c erg/loop
The active stress rz m a y be e q u a t e d to the chemical driving force

(-- Af) minus the strain energy expended per u n i t volume:
T, = -- Af -- _cA = -- Af -- _a dyne/era2 (6.33)

T C
or
Wt = ~b [ ( Gb ) / a \~;2 "1 (6.34)
~ + 1 ar ~ + Af ( ~ ) r6/~J erglloop
Substituting G = 8 × 10n dyne/era ~, A = 2-09 × 10a° dyne/era ~,

b ~- 1.27 × 10 -a em a n d d = 1-78 × 10 -~ cm yields
~rb ( l a ~ 1/~
W'---- -~- 3.75ar' + Af ~ ) r 5/2) erg/loop (6.35)
W z has a m a x i m u m value, W~, when
r 1/2 .~ 3(aA)1/2 cm 1/2 (6.36)
Hence,
9aA
r c ~- - - cm (6.37)
Note t h a t r c = ~3% o, where r o is the cataclysmic size proposed by

•K n a p p a n d Dehlinger (Eqn. 6.28). W h e n r = r c, A w = Af/6 a n d
W~ = 4"5~raSAg/Af 4. rc represents the e m b r y o size at which the
" g i a n t " loop can lower its energy b y expanding; no t h e r m a l activation
is required at this point. Once the e m b r y o has a t t a i n e d this size,
cataclysmic growth can occur i n a s m u c h as the overall free energy (A W)
decreases with growth of all embryos larger t h a n r~.
237
Fig. 6.10 shows A W and W z as a function of r for ellipsoidal embryos.

For sizes in the range of r , < r < r~, the embryo can expand sponta-
neously and lower its total free energy, if thermal fluctuations provide
sufficient energy to make each unit step between r and re. However,
when r _~ ro, the spontaneous growth becomes cataclysmic.
We must now consider the problem o£ the unit step and the activation
Therrnally I Cataclysmlc
No growth ~rowthsP°ntane°uathePrnal
activated
s
gPowth
Embryo radius r.
I Overallfree
Energy/embryo
1 o\
/
C~
Energy of equivalent
circumferential
dislocation loop
~ w~
f.
C Energy incrementfor
thermally activated
growth step I .-
I""/IW~ ;1.2 .=V

_.....I " I 1 Jf
1 2 3 4
Normali~.~l embryo radius paramete~-~A)V?'(~f)
Fig. 6.10. OveraU free energy" (~W), ener&:~Z of eqnivMent circum-
ferential d i s l o c a t i o n loop (WZ), a n d e n e r g y i n c r e m e n t for e a c h t h e r m a l l y
a c t i v a t e d g r o w t h s t e p (AW'), s h o w n as a f u n c t i o n of t h e e m b r y o
ra~lins p a r a m e t e r (r/~A)l/2(--Af), for eUipsoidal e m b r y o s . ( --A f ) is t a k e n
as 3 × 10 9 e r g / c m a, w h i c h is i n t h e t e m p e r a t u r e r a n g e of a n i r o n - n i c k e l
m a n g a n e s e a l l o y w h e r e i s o t h e r m a l t r a n s f o r m a t i o n occurs
energy per unit step during the isothermal growth from r to r c, where
r > r~. From Eqn. 6.35,
_~( /~,1/2 \
AW' = 7.5ar + 2.SAf ~ ] ) r a/~) Ar erg/unit growth of loop
(6.3s)
The unit expansion o£ the circular loop Ar is assumed equal to ~,
just as in the ease of slip, the unit displacement is the Burger's vector.
238
Consequently,
A W ~= d--T - 3 a t 3/2 + Af
(6.39)
The activation energy per unit step (AW ~) has a maximum value of
°aA 2"25(aA)X/2 cm 1/2 (6.40)

= o.11. when :,2 = = _ af
Table 6.1 summarizes the various key values of r and their

significance.
TABLE 6.1
Key Values of Embryo Radius r
f. 1/|
Significance
32~r aSA ~ r ----- r . ---- m i n i m u m size for

2-00 A W = A W , - - 3 A] 4 spontaneous growth
r = r+ = size w h e r e a c t i v a -
2"25 A W'. = o. x ] IV--------
_ ~ t i o n e n e r g y o f u n i t s t e p is
a In~irn~tn
2"50 AW ----- O; A w = 0 r----r0
3"00 A W z ---- O; W z ---- W ~

r = r© = mlnlmulla size for
~#A t
= 4 " 5 ~ cataclysmic growth
A f,
l~ig. 6.11 presents a plot of the normalized activation energy

parameter --AW(Afa/eaA) versus the normalized radius parameter
(r]aA) 1/~ ( - - Af). F r o m l~ig. 6.11 AW z can be deduced as a function of
radius as well as temperature. I t should be noted t h a t for r , < r < r+,
AW z increases before decreasing. Hence the actual activation energies
for embryos within this range would be equal to AWZ+,as in Fig. 6.11.
Athermal formation of martensite can occur when a specimen is
quenched to a temperature where some of the embryos have sizes
greater t h a n r c. Isothermal holding m a y result in the spontaneous
growth of embryos having r , ~ r < rc to size r~, followed by the sub-
sequent cataclysmic transformation of these particles into martensite.
The activation energy for the isothermal transformation is given by
Eqn. 6.39. This presupposes t h a t autocatalytic effects are absent so
t h a t subcritical embryos are not stimulated. Hence, the equations are
applicable to the initial nucleation process.
239
P R O G R E S S IN M E T A L P H Y S I C S
I t should be noted that A W ~is a function of size as well as tempera-

ture. I f AW~ is the observed activation energy for the initial rate of
isothermal formation, then A W~ must reflect the spontaneous thermal
growth of the most potent or largest embryos present having a radius
r, where r~ _< r, < r e.
In order to test Eqn. 6.39, the isothermal transformation data
reported b y S,~I~ et al.(a°) can be used since no prior athermal martensite
was detected in this case. Eqn. 6.39 can be written for the largest
Embryo radius
~0.300
"~0.100
"o
o
:'.0 2.2 2.4. 2.6 :"8 3"0
Normalized radius parameter (=-cz)v~(-/If)
Fig. 6.11. Activation energy parameter as a function of embryo radius

parameter for thermally activated spontaneous growth of marteusitic
embryos
embryos (r,) available in the range between r~ and r e. Inserting some

of the numerical quantities:
AW~=4 × 10-~ ( ~ ) ( 3 a ~ I~ ~- Af ~ ~ erg/unitgrowth o f l o o p

(6.41)
Before A W~ can be compared with the observed values of A W,, it is

necessary to assume a size for r,. An excellent fit between theory and
experiment is obtained ff ro is set equal to 2.3 × 10- e c m in the iron-
nickel-manganese alloy of S~r~r et al., (a°) assuming that the composition
is equivalent to 25 a]o Ni for ascertaining the thermodynamic data.
The value of r, was selected to make the calculated temperature for the
maximum rate of isothermal nucleation coincide with the observed
temperature of 130°K. The results are given in Fig. 6.12. The tem-
perature dependence of A W~ is almost identical with that of A W,.
With the adopted value of a -----200 erg/cm ~, the calculated activation
240
e n e r g y is a b o u t twice t h e e x p e r i m e n t a l value at each t e m p e r a t u r e .
H o w e v e r , if a is decreased b y 20 per cent, which is within t h e limits of
u n c e r t a i n t y (Appendix), a n d if r+ is likewise t a k e n to be 20 per c e n t
smaller, v e r y close a g r e e m e n t is found, as shown in :Pig. 6.12.
F o r t h e situation u n d e r discussion, t h e m a x i m n m size e m b r y o
contains a b o u t 2 × 105 atoms, a n d a t 130°K m u s t grow isothermally
t o a size of 5.5 × 106 a t o m s (r c = 3 × 10 -6 cm) before cataclysmic
/
o = I~(measur~cl) /
21 0 0 0 ,'IW++ (calculated)
_ o" =200 er~s/crr# _
19 OOC
re = 2"3 x lO-°cm. ~ ,
E
17000
15 OOO
u 13 0 0 0
P 11000
/
/ /
g
9 000
/
J
7OO0
f
,4~(¢oi~o+,e<:i)
(7=160 e+"~.~cm~'
5000 P, = I " 8 x I 0"-~ cm - -
3OO0
J
0 50 100 150 20(
"remperotLPe 'OK
Fig. 6.12. Calculated and measured values of the activation energy for
isothermal nucleation of martensite in an iron-nickel-manganese
alloy
g r o w t h takes over. I t is e v i d e n t t h a t this is n o t a process of " t r u e

n u c l e a t i o n " in t h e classical sense because the critical condition is n o t
a t r+ where g r o w t h begins to lower the overall free energy, b u t a t an
a p p r e c i a b l y larger size of pre-existing e m b r y o s where e i t h e r t h e r m a l l y
a c t i v a t e d or cataclysmic g r o w t h can ensue.
This t r e a t m e n t of m a r t e n s i t e nucleation predicts a n u p p e r limit to
t h e t e m p e r a t u r e range o v e r which isothermal t r a n s f o r m a t i o n can t a k e
place (in the absence o f prior a t h e r m a l martensite). T h e highest t e m -
p e r a t u r e for isothermal nucleation occurs where the largest e m b r y o
size r e is equal t o r+, or (from E q n . 6.26) - - A f ( T ) = 2(aA/re) ½. This
c i r c u m s t a n c e m a y explain w h y isothermal C-curve kinetics a b o v e M+
is such a rare occurrence.
241
APPENDIX
A value of 200 erg/cm 2 for the anstenite-martensite interfacial energy
in iron-base alloys is obtained as follows. B R o o K s ~85~ and K N A P P and
D~,HLI~GER ~56~have utilized the Read-Shockley equation
=~0n l - l n ~ m -49(1-~) ~m
where ~ ~__ b/d, on the assumption that the screw-dislocation interface
of FRA~K a2~ can be treated as a low-angle boundary.* Substituting
b~ a~ ~
2v'2 1.27 ~ and d ~ 17.8 ~ (Fig. 6.6), G = 8 × l0 n dyne/cm ~,
and/~ ~ 1/3 and ~ ~ 0.47 radians cs6~ into Eqn. A.1 fields 250 erg/
cm 2. However, the (1 ~ #) factor in Eqn. A.1 applies to edge disloca-
tions, and should become unity for screw dislocations. In this case, the
interfaeial energy is computed as 150 erg/cm 2. A value of 200 erg/cm ~
is adopted here, using only one significant figure because of the approxi-
mations involved.
A second w a y of arriving at the interracial energy is to assume that
it is comprised mainly of the core energy of the screw dislocations, the
elastic strain energy being taken as negligible because of the interphase
nature of the boundary. Then ~sT~
Gb ~
a ----- ~ erg/cm 2 (A.2)
or 330 erg/em ~.
Thus, both of these methods give a values an order of magnitude
larger than that deduced from nucleation kinetics (Section 5.1).
BROOKS{s6} has suggested that the latter determination of 24 erg/cm 2
m a y refer only to the coherent part of the interracial energy on the
supposition that the interface is completely coherent during the
nucleation event. However, from the arguments developed in Sections
5.1 and 5.3, martensite nucleation cannot be regarded to take place
"from scratch" in an ideal lattice, but seems to propagate from available
embryos of substantial size. If a dislocation interface already exists
around the embryo before M , is reached (Section 6.5) it is doubtful
that an interracial energy derived from "nucleation kinetics" under
such conditions could have the significance of a coherency energy.
Furthermore, if it is postulated that the interface is fully coherent
and contains no dislocations, then a coherency strain energy term must
be added to the general free energy Eqn. 6.18. This new term may be
* A n y a d d i t i o n a l set o f dislocations t h a t m a y be i n v o k e d to a c c o u n t for t h e m a c r o -
scopic d i s p l a c e m e n t s will c o n t r i b u t e p r i m a r i l y to t h e s t r a i n e n e r g y r a t h e r t h a n t h e
interracial energy-, see f o o t n o t e o n p a g e 223.
242
M A R T E N S I T I C TRANSFORMATIONS
e s t i m a t e d b y considering t h e (225)r interface p l a n e as in Fig. 6.4c,
r a t h e r t h a n one w i t h t h e screw dislocations as in Fig. 6.4D. U n d e r t h e s e
c i r c u m s t a n c e s , t h e c o h e r e n c y s t r a i n e n e r g y p e r u n i t v o l u m e is G ~ / 2
w h e r e ~ ~ bid ~ 1/14. H e n c e
4~ 4~
A W = - ~ r 2 c ( A f -~ G ~ / 2 ) -~ 2~rr~a ~- --~ rc~A . (A.3)
4~r
T h e u s u a l s t r a i n e n e r g y - ~ rc2A is v i r t u a l l y u n a f f e c t e d here because it
is a s s o c i a t e d w i t h a m a c r o s c o p i c d i s p l a c e m e n t which is essentially
i n d e p e n d e n t o f t h e c o h e r e n c y strain.
F o r t h e conditions a t h a n d , G ~ / 2 is a b o u t 350 cal/mol, a n d t h e
d r i v i n g force m u s t b e c o n s i d e r a b l y g r e a t e r t h a n this q u a n t i t y for t h e
t r a n s f o r m a t i o n t o p r o c e e d e v e n if a were zero. H o w e v e r , as s h o w n in
Fig. 3.14, M , t e m p e r a t u r e s in t h e i r o n - n i c k e l s y s t e m occur a t driving
forces m u c h less t h a n 350 cal/mol. C o n s e q u e n t l y , t h e a s s u m p t i o n of a
c o h e r e n t i n t e r f a c e w i t h t h e a t t e n d a n t r e d u c t i o n of a does n o t a c c o u n t
for t h e o b s e r v e d kinetics b e c a u s e of t h e high c o h e r e n c y s t r a i n e n e r g y
involved.
ACKNOWLEDGEMENTS
T h e a u t h o r s wish t o e x p r e s s t h e i r a p p r e c i a t i o n t o D r . H . K n a p p w i t h
w h o m t h e y h a d several s t i m u l a t i n g discussions. T h e y also w a n t t o
t h a n k P r o f e s s o r E. Scheil a n d D r . W. N o r m a n n for m a k i n g a v a i l a b l e
t h e h i t h e r t o u n p u b l i s h e d d a t a s h o w n in Fig. 3.9.
R~r~NCES
tl) FIWK, W. and CA~tPB~.T-T.~E.; Trans. Amer. Soc. 8reel Treat. 9 (1926) "/17.
(s) T R O ~ o , A. R. and GRENINGER, A. B. ; Metal Progr. 50 (1946) 303.
c8) KURDJ~OV, G. V.; J. Tech. Phys., Moscow 18 (1948) 999.
t4) Co~z~, M.; Trans. Amer. ,.%c. Metals 41 (1949) 35.
(5) ; Phase Transformo;tions ~n Sol,is, John Wiley and Sons, New York;
Chapman and Hall, London, 1951.
(6) BOWLES, J. S. and BARRETT, C. S.; Progress ~n Metal Physics, Pergamon
Press, London, Vol. 3, p. 1, 1953.
~) BILBY, B. A. and CHRISTIAN, J. W.; The Mechanism of Phase Transfor-
mat/ons ~n Metals, Institute of Metals Monograph and Report Series
No. 18, p. 121, 1955.
cs~ KUP.DJU~OV, G. V., MIRE~]r~, V. and S.rJr~rSKAYA, T.; J. Tech. Phys.,
Moscow 8 (1938) 1959.
; Trans. Amer. Inst. Min. (Metall.) Engrs 133 (1939) 222.
(~) Gm~NINGF_~,A. B. and Tm)Z•NO, A. R.; Trans. Amer. Inst. Min. (Mdall.)
Engrs 185 (1949) 590.
(10) B o w ~ s , J. S. and l ~ c ~ N z r ~ . , J. K. ; Acta Met. 2 (1954) 129.
cm W E C H S ~ , M. S., T,w.BE~N, D. S. and R~LD, T. A. ; Trans. Amer. Inst.
Min. (Meta//.) Engrs 197 (1953) 1503.
~Xa) FR~'~-K, F. C.; Aeta Met. 1 (1952) 15.
243
(18) K U I ~ , 8. A. and COKEN, l~I.; Trans. Amer. Inst. ~]lin. (Metall.) En~rs 188
(1950) 1139.
(14) BUNSHAH, R. F. and MEHL, R . F . ; Trans. Amer. Inst. ~lin. (Mdall.} Engrs
197 (1953) 1251.
~15) K ~ U F ~ , L. and C o m ~ , M.; Trans. Amer. Inst. Min. (MetaU.) Engrs 206
(1956) 1393; J. Metals, N.Y. 8 (1956) 1393.
~16) LLEBER~A~, D. S.; The Mechanism of Phase Transformations in Metals,
Institute of Metals Monograph and Report Series No. 18 (1955) Discus-
sion.
(*~) H ~ J ~ s , W. 5. and COHEN, M.; Trans. Amer. Inst. Min. (Metall.) Engrs
185 (1949} 447.
(18) M ~ C H ~ , E. S. and CoHEn, Yl.; Trans. Amer. Inst. Min. (MetaU.) Encrs
194 (1952) 489.
(19) COHEN, M., M A C O N , E. S. and PARAI~JPE, V. G.; Thermodynamics in
Physical Metallurgy, American Society of Metals, 1949, p. 242.
(20) D~s GUFrA, S. C. and L ~ N T , B. S.; Trans. Amer. Inst. Min. (Metall.)
Engrs 197 (1953) 530.
~ ) CHANG, L. C.; J. Appl. Phys. 23 (1952) 727.
~t2) HOLLO~O~r, 5. H., 5~VFE, L. D. and B w ~ ' u ~ , D. C.; J. AppL Phys. 18
(1947) 780.
~28) MOR~_~, E. R. and Ko, T.; Aeta Met. 1 (1953) 36.
ts~) CRUSSARD, C.; The Mechanism of Phase Transformations in Metals, Insti-
tute of Metals monograph and Report Series No. 18 (1955) Discussion.
(~a) ED~ONDSO~¢, B. and Ko, T.; Acla Met. 2 (1954) 235. K ~ , L.;
Thesis, M.I.T., 1955.
(~) B u R ~ , X. E.; Atom Movements, American Society of Metals, 1951, p. 209.
(~n C ~ , R. W. ; Suppl. Nuovo Cim. 10 (1953) 350.
~ls) KuLII~, S. A. and SPEICH, G. R.; Trans. Amer. Inst. Min. (MetalL) En~rs
194 (1952) 258.
~ HOLDS.N, A. N. ; Acta Met. 1 (1952) 617.
(-~0) S ~ C. H., fl~ERBACH, B. L. and COHEN, ~ . ; Trans. Amer. Inst. Min.
(Metall.) Engrs 203 (1955) 183.
(~) CECIl, R. E. and H o I ~ o ~ o ~ , 5. H . ; Trans. Amer. Inst. Min. (MetaU.)
Engrs 197 (1953) 685.
(~) Ko, T. and COTTP~EIJ~,S. A. ; J. Iron St. Inst. 172 (1952) 307.
(~) Moor, B. W. and H ~ E S , H. R . ; Rev. Mdtall. (1954) L I No. 9.
(s~) M~CHI~, E. S. and COHEN, 1~. ; Trans. Amer. Inst. Min. (MetalL) Engrs
191 (1951) 744.
(~) FISHER, J. C., HOL~O~O~, Y. H. and T u r n o u t , D.; Trans. Amer. Inst.
M~n. (Metall.) Engrs 185 (1949) 691.
(~) ZENER, C.; Trans. Amer. Inst. Min. (Metall.) Engrs [67 (1946) 513.
(l~) FISHER, J. C.; Trans. Amer. Inst. Min. (Metall.) Engrs 185 (1949) 688 and
P r i v at e Communication.
(~s~ $ o ~ s s , F. W. and PU~PHR~Y, W. I. ; J. Iron St. Ins$. 163 (1949} 121.
~9~) GRrD~EV, V. N. and TREF~OV, V. I. ; DokL Akad. N a u k S S S R 95 (1954)
741 (Brutcher Translation 3385).
(~0) HEss, J. B. and B A P ~ ' r r , C. S.; Trans. Amer. Inst. Min. (Metall.) Engrs
194 (1952) 645.
(a~) BURK~T, ~Yl. ~V. and RE~D, T. A. ; Trans. Amer. Inst. Min. (MetalL) Engrs
197 (1953) 1516.
(~) TrrcHE~vF_~, A° L. and BEVER, M. B.; Trans. Amer. Inst. Min. (Mdall.)
Engrs 200 (1954) 303.
(~* P~,TEL, J. R. and CoHEn, M.; Acta Met. 1 (1953) 531.
244
MARTENSITIC TRANSFORMATIOI~S
t,4) JOKa~TNSON, C. H . ; Arch. Eisenhiittenw. 11 (1937) 241.
~,5) D ~ _ R ~ , L. S. and S~.rH, R. P.; Industr. Engng Chem. (Industr.) 43 (1951)
1815.
(as) SCHEI~, E. and No~.~x~N, W.; P r i v a te Communication.
~4~) M F ~ - ~ O , J. L.; Aria Met. 4 (1956) 331.
~48) O ~ E N , W . ; Arch. Eisenhiittenw. 26 (1955) 253; /b/d. 26 (1955) 519.
(4~ D ~ N , L . S . ; Trans. Amer. Inst. Min. (Metall.) Engrs 167 (1946) 468.
cao) ZEN~R, C.; Trans. Amer. lnst. Min. (Metall.) Engrs 203 (1955) 619.
~1) HOLLO~ON, J. H. and :l~-~.~, L. D.; Ferrous Metallurgical Design, John
Wiley and Sons, New York, 1947, p. 47
(~) ttOESLF-~, U., S.(To, H. and Z E ~ R , C.; Symposium on Theory of Alloy
Phases, American Society of Metals, 1956, p. 225.
~e~) A S M Metals Handbook, 1948.
~s4) C O n r a i L , A.; Theoretical Structural Metallurgy, Edward Arnold CO.,
London, 1948, p. 125.
(56) I~CHLIN, E . S . ; Trans. Amer. 1net. Min. (Metall.) Engrs 200 (1954) 684.
~56) K ~ P , H. and D ~ c r x ~ G ~ , U. ; Acta Met. 4 (1956) 289.
(67) FISHF_~, J. C.; Thermodynamics in Physical Metallurgy, Americaza Society
of l~etals, 1949, p. 201.
(68) K~SF-~ENT, O., H O I I D ~ , E. and W~.vJa~, F.; Rev. MdtaU. 51 (1954)
401.
(~) SvJ~_~L, E . ; Z. Anorg. Chem. 207 (1932) 21.
(eo) MCREYNOLDS, A. W . ; J. Appl, Phys. 20 (1949) 896.
(~) KU~ N, S. A., C O ~ N , 1K. and AV~RB~C~, B. L.; Trans. Amer. Inst. Min.
(MetaU.) Engrs 194 (1952) 661.
(~) KUI~DXV-~OV,G. V. and K~x~DI~OS, L. C. ; Dokl. Akad. N a u k SSSI~ 66 (1949)
211.
~ea) G . ~ J ~ % R . ' I . ; Dokl. Akad. N a u k S ~ S R 21 (1938) 229; J. _Phys. Moscow
2 (1940) 313, 319; '/b/d. 11 (1947) 55; Dokl. Akad. N a u k S S S R 5 7 (1947)
555.
(e~) REYNOLDS, J. E. and B ~ v ~ , M. B.; Trans. Amer. Inst. Min. (Metall.)
Engrs 194 (1952) 1065.
(e~) HOLLO~O~, J. H. and TIYRNBIYLL,D. ; Progress in Metal Physics, Pergamaon
Press, London, Vol. 4, p. 333, 1953.
~a~) l~sa~a~, J. C.; Trans. Amer. Inst.Min. (Metall.) Engrs 197 (1953) 918.
( ~ ) ,,, , Aeta Met. 1 (1953) 32.
~s) F~s~a~, J. C. and ~ U L L , D.; Aria Met. 1 (1953) 310.
~) KAy, L. and COHEN, ~ . ; The Mechanism of Phase Transformations ~n
Metals, Institute of Metals Monograph and R e p o r t Series No. 18 (1955}
p. 187.
(~0) G R ~ O F ~ R , A. B. and M O O ~ , V . C . ; Trans. Amer. lnst. Min. {MetaU.)
Engrs 128 (1938) 337.
(~) CH_~(}, L. C. and R~.~D, T. A.; Trans. Amer. Inst. Min. (MetaU.) Engrs
189 (1951) 47.
~ ) CECH, R. E. and TUR~ULL, D.; Trans. Amer. Inst. Min. (Metall.) Engrs
206 (1956) 124; J. Metals, N . Y . 8, No. 2 (1956) 124.
(~) CRUSS~RD, C.; Physica, 's Gray. 15 (1949) 184.
(~) - ; Trans. Amer. InsL Min. {Metall.) Engrs 191 (1951) 552.
~ ) -. ; The Mechanism of Phase Transformations in Metals, Institute of
Metals Monograph and R e p o r t Series 1~o. 18 (1955).
(~e) LYUBOV, B. Y~. and OSIP'Y.klq,YU. A. ; Dokl. Akad. N a u k S S S R 1Ol (1955)
853.
245
¢~') FISB:EI~, J. C. and TUI~BULL, D.; Trans. A m e r . Xnst. M i n . (Metall.) E n g r s
197 {1953) 921.
c+s+ ; Trans. A m e r . I n s t . M i n . (Metall.) E n g r s 200 (1954) 685.
(tg+ KRONER, E.; A e t a Met. 2 (1954) 30.
<s0) KDsT~-~, W.; Z. Metallic. 39 (1948) 1.
(sl) B ~ T , W. ; Diplomaxbeit, Stuttgart, 1954.
+Sin) ~ L P - ~ , H. C., A V ~ A C H , B. L. and Comm+N,5I.; Trans. A m e r . Soc. Metals
47 (1955) 267.
(sa) CoTrlm~r+T+, A. H . ; Dislocations a n d Plastic F l o w i n Crystals, Clarendon
Press, Oxford, 1953, p. 38.
[sl+ FmtN~:, F. C. and ST~OH, A. N.; Proc. P h y s . Soc. Zond. B65 (1952) 811.
<as) B~ooKs, H.; M e t a l Interfaces (1952)20, AS51, Cleveland, Ohio.
(se~ READ, W. T. ; Dislocations i n Crystals, McGraw-Hill, New York, 1953, p. 192.
(s+> Fismmx+, J. C.; Trans. A m e r . Soc. Metals 47 {1955) 457.
(ss+ 5ASWON, M. J . ; The M e c h a n i s m of P h a s e T r a n s f o r m a t i o n s i n Metals,
Institute of 51e+als 51onograph and Report Series No. 18 (1955) p. 173.
(st) ~ , F. C.; The M e c h a n i s m of P h a s e T r a n s f o r m a t i o n s i n Metals, (1955)
Institute of Metals Report and Monograph Series No. 18 (1955) Discussion.
(9o) Owm~, E. A. and LIu, Y. H.; J . I r o n S t . I n s t . 163 (1949) 132.
(91) WEISS, R. Z. and TAU~R, K. J , ; P h y s . Rev. 102 (1956) 1495.
(92) BULLOUOH, R. and B ~ Y , B. A.; Proe. Roy. Soc. B69 (1956) 1275.
(ga) ~ N T , B. S.; N u o v o Cim. 1 (1955) 295.
(~4~ KURDJU~OV, G. V. and M ~ r ~ o v A , O. P. ; Dokl. A k a d . N a u k S S S I ~ 61
(1948) 83.
c95) . ; Dokl. A k a d . N a u k S S S R 73 (1950) 95.
(96~ ; Dokl. A k a d . N a u k S S S R 81 (1951) 565.
(aT, I~TT.T.~aT, M.; A c t a Met. 6 (1958) 122.
(o2~ MACKEI~Z~, 5. K. and BowI,ES, J. S.; A c t a Met. 5 (1957) 138.
246

1 s2.0 0502820558900054 Main

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 0502820558900054 Main

Uploaded by

Copyright:

Available Formats

3

THERMODYNAMICS AND KINETICS OF

2. NATURE OF MARTENSITIC TRANSFORMATIONS

2.1. Essential Features

has a "complex orientation; the habit planes are then said to be

2.2. Athermal Characteristicz

while the temperature is changing. I n the past, it was common to

typified by a high degree of supercooling and hence a relatively large

seems to nucleate suddenly and "pop out" to a detectable size within

2.3. Isothermal Characteristics

will be placed on those isothermal reactions which occur above M~.

1 5 10 50 100 500 1000

Fig. 2.2. C curves for i s o t h e r m a l t r a n s f o r m a t i o n of m a r t e n s i t e in a n

athermal martensite or in the presence of a fixed amount, the isothermal

growth. When the latter propagation comes to a halt at a given

2.4. A utocatalytic and Partitioning Effects

lacking, and consequently, it is only the initial rates of nucleation that

3. THE CHE1WICALFREE ENERGY CHANGE

4.1. The reverse transformation on heating, starting at A s, m a y also

~' s~,ol~e thereto- ~stoble t h e r m o -

Fig. 3.1. Schematic represenf~ion of F =' and F~ vs. temperatm~e

will receive most attention here. Iron-manganese alloys (3s) display

i-- 19'0 ,,JiN; 14"5

2 x l O " .(=) 30"7%Ni Specimenlength 1.4in

II ,~, . , ,_. , . , ._, , / ~ 29.3"~N," ,t,:SlO"C " "

Fig. 3.2. Composite resistance-temperature curves for iron-nickel

Fig. 3.3. Bracketing of To in cobalt- Fig. 3.4. B r a c k e t ~ g To b y plastic

temperature. BURKART and I~EAD(41) have demonstrated that, in an

pattern of elastic stresses set up during plastic deformation would

3.2. Correlation between T o and the Equilibrium Diagram

Fig. 3.5. Schematic repre- Fig. 3.6. Schematic representation of Fr

3.3. Thermodynamic Properties of Pure Iron

A F ~ ~ = 1238 -- 0.395T -- 12.5 × 10-4T 2 cal/mol (3.2)

whereas KAUFMAN and COHE~ (is) have proposed:

60C- .,4H :1202+2.63x10"~T c ~ - - "

2O0 400 60O 800 1000 ~ o o ~4oo

Fig. 3.8. Enthalpy change for the alpha-gamma transfm~aation in

a n d e x t r a p o l a t i o n to 0 w/o Ni yields t h e values s h o w n in Fig. 3.8.

lOW temperature range where the martensitic transformation comes

A similar equation can be written for the martensitic phase, and on

A F ~'-*~ = (1 -- x)AF~-~ r -p x A F ~ ~r -p A F ~ ~r cal/mol (3.6)

For multicomponent systems in general,

A F ~'-'r -~ x F o A F ~ ' -~- Z x ~ A F ~ ~v -+- Z A F ~ - y cal/mol . (3.6a)

The quantity A F ~ r has already been treated in Section 3.3; AF~-*r

Binary component AH~"~ (cal/mol) Reference

I t should be n o t e d in T a b l e 3.1 t h a t the value for c h r o m i u m is nega-

was determined by FmHER et al. (3~) at low temperatures from con-

AFt';', = xF,/~Fr~ ~ + x,(-- 10,500 + 3.425T)

Thus, AH~-Y is not likely to be constant because AF~t- r depends on

depending on the sign and magnitude of AF~-*~. This explains why

-4~ and "l'refilov)Bg) _

Fig. 3.10. C h a n g e i n c h e m i c a l free e n e r g y a t t e n d i n g t h e a u s t e n i t e -

Co~v.N[15) have adopted a more rigorous treatment for the iron-nickel

the m a r t e n s i t i c t r a n s f o r m a t i o n a t M~ is n u m e r i c a l l y equal to the

~ ' , , , . i f x =0.05 ! ! -T200 S

for A F ='-*~ as a f u n c t i o n o f t e m p e r a t u r e (below 1000°K) a n d nickel

Fig. 3.14. The effect of temperature (or composition) on the

for iron-nickel alloys containing 15-30a/o nickel. Consequently

3.5. Magnetic Effect in the Martensitic Thermodynamics

-- AF~-~.~.) = 1.41(T -- 740) cal/mol . (3.16)

Curve C is the difference between curves A and B, and constitutes

dently by vertical shifts along the temperature axis, each of which is