193
Materials Science and Engineering, 18 (1975) 193--198
© Elsevier Sequoia S.A., Lausanne -- Printed in The Netherlands
A Microstructural Study of Massive Transformations in the Fe--Ni System
T.B. M A S S A L S K I , J.H. P E R E P E Z K O and J. J A K L O V S K Y *
Metal Physics Laboratory, Mellon Institute of Science, Carnegie-Mellon University, Pittsburgh, Pa. 15213 (U.S.A.)
(Received July 19, 1974)
SUMMARY
The formation of the b.c.c, a phase from
the f.c.c. 7 phase by the process of a massive
transformation has been studied in the Fe - Ni
system over a range of composition and cool-
ing rates. The massive am product phase has
been observed metallographically in alloys
containing up to 8.7 at .% Ni. Microprobe
studies have confirmed that composition invari-
ance is maintained even in a two-phase region
b e y o n d the retrograde a/(a + 7) phase bound-
ary. In low Ni alloys the rate of transformation
was fast enough to form a completely massive
product at a cooling rate of a b o u t 3000 deg
C/sec. For alloys with about 8 at.% Ni a cooling
rate of a b o u t 7 deg C/sec is needed to produce
even a small a m o u n t of the massive product.
The occurrence of the massive transformation
in a two-phase field suggests that local equili-
brium is n o t established at the transformation
interfaces during the transformation.
INTRODUCTION
The high temperature f.c.c. 7 phase in Fe-
rich Fe - Ni alloys can transform on cooling
to the b.c.c, a phase b y a number of different
transformation mechanisms. Numerous phase
diagram determinations, as well as information
obtained f r o m the study of meteorites, have
shown that u p o n very slow cooling the 7 phase
undergoes an equilibrium solute segregation
in the (a + 7) two-phase field. In alloys whose
solute content exceeds a few at.% Ni this pro-
cess becomes very slow as the temperature is
* Present address: Alcoa Technical Center, Surface
Chemistry Division,N e w Kensington, Pa. 15068, U.S.A.
lowered, and can become arrested at tempera-
tures as high as 600°C. In practice, therefore,
the two-phase equilibrium separation into the
a and 7 phases is never completed. The form
of the phase diagram shown in Fig. 1 is that
proposed by Goldstein and Ogilvie [ 1 ]. The
extent of the a phase boundary, and its retro-
grade nature, have been well established b y
careful studies of metallic meteorites and
other conclusive evidence. In addition to the
equilibrium phase separation the 7 i~hase can
also transform b y composition-invariant pro-
cesses, t w o of which are martensitic and one
is massive. In the range of composition up to
about 4 at.% Ni the occurrence of the massive
transformation is well documented [2]. The
7 -* a massive transformation is accomplished
by a relatively rapid movement of incoherent,
high energy, boundaries of the nucleated a
phase, which is of the same composition as that
of the ~/phase matrix from which it forms. The
process is controlled by diffusion at the trans-
formation interfaces. The resulting microstruc-
9°° X
800
I Observed Limit of
I \ \ ossi e troo.,.... ,,oo
i.oo\
700 \ " \ aTA, O/oNi
~"50O 1
400 //I
!
300 I
lllJ jjiIi I
0 IO 210 50 4(3 50
Atomic % Nickel
Fig. 1. Phase diagram for the Fe - Ni system covering
the composition range of interest. Also shown is the
observed limit of massive transformation..
194
ture has characteristic features of a massive
transformation, revealing approximately equi-
axed am grains* surrounded by boundaries
that consist of both planar and curved portions.
In transmission electron micrographs [2] the
massive phase exhibits a low, and random, den-
sity of dislocations as may be expected from a
thermally activated and diffusion-controlled
process.
At compositions exceeding approximately
4 at.% Ni, up to about 25 at.% Ni, the decom-
position on cooling is known to occur by the
martensitic process, the nature of which changes
from a lath-type martensite to an acicular- or
lenticular-type martensite. The nature of these
martensites has been discussed in recent pub-
lications [2,4] including studies of relief, crys-
tallography, dislocation density and orienta-
tion relationships. However, there has been
relatively little study of the massive transfor-
mation in the Fe - Ni system. Swanson and
Parr [ 5] have reported that, above approxima-
tely 7 at.% Ni, the transformation in iron -
nickel alloys is always martensitic regardless
of the cooling rate. However, other studies
have shown that the critical cooling rate
capable of suppressing all diffusion-controlled
reactions, and hence aiding in the formation
of martensite, changes rapidly with composi-
tion. A cooling rate in excess of 3 × 104 deg C/
sec is needed to obtain martensite at 1 at.% Ni
[5]. In alloys containing between 18 and 30
at.% Ni, even following a slow cooling at
some 5 deg C/min [6], only the martensitic
transformation is observed. However, massive
transformation products have been reported
in alloys having less than 15 at.% Ni [7]. The
purpose of the present investigation was to
establish the range of composition in the Fe -
Ni system in which the massive transforma-
tion, as defined by its various detailed features
[3], can occur on cooling.
Recent studies of non-ferrous alloys have
shown that a massive transformation to a
single phase can take place also at composi-
tions corresponding to a two-phase field,
beyond the extent of a single phase field bound-
cry [8]. The implication of this result is
that, because the transformation is composi-
tion-invariant, a departure from equilibrium
must occur at the transformation interface.
* T h e s y m b o l a m is often used to denote the product o f
a m a s s i v e t r a n s f o r m a t i o n [ 3 ].
Because of the clearly established retrograde
boundary in the Fe - Ni system, it became of
interest to determine if identifiable massive
transformation products can be obtained also
in this system, beyond the retrograde a phase
solubility limit of approximately 7 at.% Ni.
EXPERIMENTAL DETAILS
The range of compositions studied in the
present work was approximately between 1.9
and 9.7 at.% Ni. Alloys containing 5.0, 5.5,
6.8, 7.0, 7.2 and 7.5 at.% Ni were prepared
from pure Fe - Ni components by levitation
melting under an atmosphere of argon. Addi-
tional alloys, containing 2.0, 3.9 and 9.1 wt.
% Ni, were received from the U.S. Steel Research
Laboratory*. In all cases the alloys were made
from materials of high purity exceeding
99.99 at.%. After casting and homogenizing
the alloys were cold-worked to a strip 0.015
inch in thickness. The specimens were austeni-
tized at l l 0 0 ° C in evacuated silica capsules
for periods ranging from 2 to 20 days, and
they were then quenched or cooled in a num-
ber of quenching media of decreasing severity:
NaOH, iced brine, water, oil and air. The spec-
imens were polished, either mechanically or
electrolytically, in a solution of 810 ml acetic
acid, 15 g Cr208 and 42 ml water and etched
with a solution of methyl alcohol plus 5% ni-
tric acid.
In order to determine chemical composition,
and to establish lack of composition gradients
between the massive a m phase and the ~f ma-
trix, a number of microprobe studies were con-
ducted. The usual technique was that reported
previously [ 8]. Measurements were performed
on many am-phase particles by a careful point-
by-point scanning at approximately 1 pm inter-
vals across the entire a m phase and the surround-
ing matrix. The data were corrected for back-
ground radiation, dead time of the counting
system, and the atomic concentration was es-
tablished using the Colby Magic IV computer
program including absorption, fluorescence
and atomic number effects. Following this
procedure an accuracy of 0.1 wt.% Ni was at-
tained.
* T h e authors are grateful to Dr. G . R . S p e i e h f o r p r o -
viding these samples.
 195

RESULTS AND DISCUSSION

Preliminary experiments have shown that

the time of homogenization in the ~ phase
field was very important because, on account
of the very sluggish rates of diffusion [9],
several days of homogenization at l l 0 0 ° C
were required in order to produce the initial,
well annealed, equiaxed, 7 phase. Following a
prolonged homogenization of some 20 days
under a protective helium atmosphere a char-
acteristic thermal grooving, representing the
grain boundaries of the 7 phase, usually de-
veloped on the previously polished specimen
surface. This grooving proved useful in distin-
guishing between the products of a martensi-
tic and a massive transformation. The former
was always restricted within the boundaries
of parent 7 grains, owing to the crystallo-
graphic relationship, but the growth of the
latter easily extended across such boundaries,
presumably because of the lack of orientation
relationship, and the resulting incoherent na-
Fig. 2. The typical martensitic structure obtained in a
ture of the transformation interfaces. 6.0 at.% Ni alloy following a relatively fast quench
In the currently studied composition range from the ~f phase field.
the transformation product was clearly related
to the cooling rate. The faster the cooling rate
the more likely was the occurrence of a mar-
tensitic phase because the thermally activated
massive transformation was suppressed. In
alloys with a low nickel content, a high rate
of cooling could suppress the massive trans-
formation altogether, and a typical lath-type
martensitic structure resulted. With decreas-
ing rates of cooling the a m o u n t of massive
phase increased; cooling at a rate of a few de-
grees per second produced a complete massive
transformation in the composition range be-
tween approximately 2 and 5 at.% Ni.
Figure 2 shows a typical martensitic struc-
ture obtained in a 6.0 at.% Ni alloy following
a relatively fast quench from the 7 phase field.
If the cooling rate is reduced, a characteristic,
100% massive, transformation is obtained. In
Fig. 3, following air cooling from the ~f phase
field at approximately 7 deg C/sec, massive
transformation a m grains are seen to cross
the prior ~ phase boundaries in a 5.0 at.% Ni
alloy. However, if the cooling rate is increased,
the massive transformation gives way partially to
a lath-type martensitic structure even in alloys
Fig. 3. The massive transformation 0~a grains are
poor in'Ni. For example, Fig. 4 shows a mix- seen to cross the prior 9' phase boundaries in a 5.0
ture of massive and martensitic structures in a at.% Ni alloy after air cooling from the 7 phase field
3.7 at.% Ni alloy following a quench in iced at approximately 7 deg C/sec.
196
Fig. 4. Typical a m and martensite (M) structure in a
3.7 at.% Ni alloy following a quench into iced water.
water. By contrast, when the Ni c o n t e n t is in-
creased, very slow cooling rates were needed
to produce even small amounts of the massive
a m phase. Figure 5 shows an example of some
massive a m phase produced mostly near prior
phase boundaries, in a sample that had been
cooled relatively slowly at 7 deg C/sec. This
alloy, with a composition of 8.7 at.% Ni, falls
b e y o n d the retrograde a phase boundary of
7 at.% Ni. Hence an equilibrium decomposition
of the ? phase that would result in solute par-
titioning is expected on slow cooling. However,
careful microprobe studies have shown that,
under the cooling conditions imposed in the
present experiment, this was n o t the case. In
Fig. 6, a microprobe scan taken across both
the martensitic product and the massive am
phase indicates that the composition of the
martensitically transformed matrix and that
of the a m particle are the same, and that no
solute segregation has occurred.
This result is not surprising ifone compares
the time available for the nucleation and
growth of the a m precipitates shown in Fig. 6
against the time needed to initiate solute seg-
regation. The time available is the time the
specimen spent between the To temperature
Fig. 5. Partial massive transformation structure in a
8.7 at.% Ni alloy cooled at 7 deg C/sec. Massive a m
phase occurs mostly near prior ~, phase boundaries.
(i.e. the temperature at which Gv = Ga) and
the M s temperature. Taking T o as the value
midway between the ~/I~, + a and a I~, + a phase
boundaries (T O -~ 900°K), and M s from the
data of Swanson and Parr [5] (M s -~ 650°K),
we obtain T o -- M s -~ 250 ° K for the alloy con-
taining 8.7 at.% Ni. At a cooling rate of 7 deg
C/sec the time interval available for the massive
transformation is about 35 sec. During this
time the am precipitates became approximately
20 #m wide. If it is assumed that the whole
time interval was spent for growth, the lower
limit of the growth rate, V, is approximately
6 X 10 - 7 m/sec*. The smallest size diffusion
layer, X, in which solute segregation can be
considered during growth ahead of the advanc-
ing interface in the parent phase may be taken
as one atomic spacing thick, or ~ 0.3 nm. The
rate of solute addition in such a layer is given
by D / X where D is the diffusion coefficient
* In fact, it has been s h o w n recently [10] that most
of the time available for a massive transformation
during continuous cooling is spent for nucleation.
This would tend to increase the estimated value for V.
However, for the purpose of estimating a lower limit
for V w e assume that all the available time was spent
for growth.
 197

O . o . • • *
. . r. •
.~ t
. • • • •,

E • • t" t.

g . # 8 • " ' • " "" " " '." • • "" t* • •

E'.'- 4~
" " i'" " °
" "
0 -- • J I I I ,I ,,, I I I ,I t .... i t l
8 16 24 32 40 48 56 64 72 80 88 ~ I~
Scan Distance (pLfrom stort of soon)

Fig. 6. Microprobe scan across a Fe - 8.7 at.% Ni specimen partially transformed to massive a m at 7 deg C/sec.
The composition of the martensitically transformed matrix and that of the 0lrn particle are the same.

of solute in the parent phase. If V is greater
than a b o u t 3 X 109 D m/sec., no significant
compositional buildup due to volume diffusion
can take place ahead of the interface. A simi-
lar condition has been discussed by Karlyn
et aL [11] for massive transformations that
occur in a single phase field in the Cu - Zn
system. In the present case, the diffusion co-
efficient of Ni in 7 iron may be taken as
D ~ i = 0.77 e x p ( - - 6 7 , 0 0 0 / R T ) c m 2 /sec [9].
Hence, at T O in the 8.7 at.% Ni alloy, the
minimum growth rate that still prevents com-
positional partitioning is 3 X 109D~i = 1.5 X
10 - i l m/sec, which is nearly five orders of
magnitude lower that the observed lower limit
value derived above. Hence, in agreement with
the microprobe results, long range redistribu-
tion of Ni cannot take place.
It is also of interest to examine the free
energy "gain", AGe, that would result if equi-
librium segregation occurred. The free energy
change accompanying this reaction is given by
[12]
&Ge = R T IX~°i In X~Ni
+ (1 - - X~NOi)In (1--X~i~
___
where X~°~ is the atom fraction of Ni in the
original 7 phase, and ~ i is the atom fraction
of Ni in the resulting 7 phase. Assuming that
both Henry's and Raoult's laws are obeyed,
over the temperature interval between T o and
M~ in the 8.7 at.% Ni alloy, AGe decreases
from --192 J/mole to --1920 J/mole. On the
other hand the free energy gain, AGr,, result-
ing from a composition invariant massive trans-
formation in the 8.7 at.% Ni alloy just above
M, is --1083 J/mole, on the basis of the ap-
proach developed by Kaufman and Cohen [6].
While in a single phase field AG m and AGe are
equivalent, in a two-phase field they differ.
In the present case, at the observed limit of
the massive transformation, the maximum
(negative) gain AGm is about one-half of the
value of AGe, resulting from the partitioning
reaction. Yet the massive transformation
takes place because of kinetic considerations
at the transformation interface.
The above conclusions are in agreement
with previous similar observations [ 8 ] in Cu -
Zn, Cu - A1 and Ag - Cd alloys in which a com-
position-invariant massive transformation pro-
duct is also observed to form in the equilibrium
two-phase region. In the present case similar
results were obtained also in alloys contain-
ing 7.2 at.% Ni and 7.5 at.% Ni. These re-
, sults clearly indicate that, even with relatively
slow cooling in the Fe - Ni system, of the two
alternative thermally activated processes that
are possible, namely equilibrium decomposi-
tion of the 7 phase with solute partitioning
and the composition-invariant massive pro-
cess, the latter occurs. Diffusion takes place
only at the transformation boundaries and
198
aids the transport of atoms across these bound-
aries, apparently without equilibrium at the
interface. This process continues until a
martensitic change, without diffusion, trans-
forms most of the remaining 7 matrix.
CONCLUSIONS
1. In the Fe - Ni system, under certain con-
ditions, the b.c.c, a phase forms from the f.c.c.
phase by the process of composition-invariant
massive transformation. The massive transfor-
mation product phase has been observed in
alloys containing up to 8.7 at.% Ni. The na-
ture of the massive phase has been confirmed
by both metallographic features, such as cross-
ing of prior V phase boundaries, as well as by
microprobe studies which confirm composition
invariance, even in the two-phase region b e y o n d
the retrograde boundary.
2. Because the diffusion rates in the Fe - Ni
system are very slow, a massive transformation
is observable only if the rate of cooling is slow
enough to allow some decomposition by a
massive transformation prior to a martensitic
transformation. In low Ni alloys (approximate-
ly 2 at.% Ni) the massive transformation rate
is relatively faster and, even at the cooling rates
of a b o u t 3000 deg C/sec, a completely massive
product can be obtained. However, in alloys
containing a b o u t 8 at.% Ni, very slow air cool-
ing is needed to produce even a small amount
of the massive phase. Even slower rates would
be needed to produce the equilibrium ~ +
segregation.
3. The occurrence of the massive transfor-
mation in the two-phase field suggests, in
agreement with previous such studies in non-
ferrous systems, that the massive transforma-
tion takes place without local equilibrium at
the transformation interface.
ACKNOWLEDGEMENTS
The authors are grateful to Dr. A.J. Perkins
for helpful discussions during the period of this
research.
This study was supported in part by a grant
from the U.S. Army Research Office, Durham,
N.C., which is gratefully acknowledged.
REFERENCES
1 J.I.Goldstein and R.E. Ogilvie,Trans. AIME,
233 (1965) 2083.
2 G.R. Speich and P.R. Swann, J. Iron Steel Inst.
(London), 203 (1965) 480.
3 T.B. Massalski, Phase Transformations, Am. Soc.
Metals, Metals Park, Ohio, 1970.
4 R.H. Goodenow and R.F. Hehemann, Trans. AIME,
233 (1965) 1777.
5 W.D. Swanson and J.G. Parr, J. Iron Steel Inst.
(London), 202 (1964) 104.
6 L. Kaufman and M. Cohen, Trans. AIME, 206
(1956) 1393.
7 A. Gilbert and W.S. Owen, Acta Met., 10 (1962)
45.
8 T.B. Massalski,A.J. Perkins and J. Jaklovsky,
Met. Trans., 3 (1972) 687.
9 K. Hirano, M. Cohen and B.L. Averbach, Acta
Met., 9 (1961) 440.
10 S.K. Bhattacharyya, J.H. Perepezko and T.B.
Massalski, Acta Met., 22 (1974) 879.
11 D.A. Karlyn, J.W. Cahn and M. Cohen, Trans.
AIME, 245 (1969) 197.
12 L.S. Darken and R.W. Gurry, Physical Chemistry
of Metals, McGraw-Hill, New York, 1953.