You are on page 1of 8

Effed: of Applied Stress on t:he Mart:ensit:ic Transformation

by S. A. Kulin, Morris Cohen, and B. L. Averböch

The martensitic transformation can be initiated by elastic stresses


at temperatures above M. in a steel containing 20 pet Ni and 0.5 pet
C. Shear strains and normal tensile strains acting on a potential habit
plane promote the transformation but compressive strains oppose it.

T HEformation
strain sensitivity of the martensitic trans-
has long been recognized. is now It
and by the martensi tic re action has been depicted
by Scheil.' Basically, Scheil postulated that: I-at
well known that martensite formation can be in- temperatures hot too far above M. (where austenite
duced by plastic deformation at temperatures be- is less stable than martensite), a critical resolved
low, and not too far above, the M. point.' However, shear stress (within the elastic range) is required
the exact role of elastic vs. plastic strains and of the to promote the transformation to martensite; 2-at
concomitant stresses has proved to be a very elusive and below M" the austenite lattice becomes me-
matter. chanically, as well as thermodynamically, unstable
Many investigators'-O of the steel hardening re- and shears over spontaneously into martensite with-
action have made use of the stresses resulting from out the application of external stress; and 3-at
volume changes to account for such observations as temperatures sufficiently far above M" the critical
the existence of retained austenite and its variation resolved shear stress for martensite formation in-
across the section of a bar, the change in the amount creases to a level above that required for slip, and
of retained austenite with quenching rate, and the plastic flow then supersedes the transformation
self-stopping nature of the transformation when the when external stress is applied.
cooling is stopped. The effect of applied strain as an Scheil's concepts were tested by McReynolds· who
independent variable has been studied in 70 pct Fe- found no change in the elastic moduli (measured
30 pct Ni alloys by Scheir and McReynolds. 8 Some both statically and dynamically) of 71 pct Fe-29 pct
data are also available for lithium and Li-Mg al- Ni alloys in the vicinity of the M, temperature.
loys," Cu-Zn,'o Cu-Sn and Cu-Al alloys," and aus- Since the moduli did not approach zero or become
tenitic stainless steels. 12-'5 negative on cooling to M., it was concluded that the
It has been found that martensite formation can austenite lattice does not become mechanically un-
be induced isothermally, even at temperatures above stable at M" as postulated by Sc heil. McReynolds
M., by mechanical deformation of the parent phase. also reported that M, is not raised by elastic stresses.
The competitive nature of plastic yielding by slip Accordingly, plastic deformation was considered to
stimulate the transformation by generating mar-
S. A. KULlN, Student Associate AlME, is ossociated with West- tensite nuclei in the distorted regions of the parent
inghouse Atomic Power Div., Bettis Field, Pittsburgh; and phase.
M. COHEN, Member AlME, and B. L. AVERBACH, Juniar Member The above issues with regard to the role of ap-
AlME, are ossocioted with the Dept. of Metallurgy, Massachusetts
Institute of Technology, Cambridge, Mass.
plied stress require clarification, if a basic under-
Discussion on this paper, TP 3274E, may be sent, 2 copies, to standing of the martensitic transformation is to be
AlME by Dec. 1, 1952. Manuscript, Feb. 13, 1952. Philadelphia obtained. For example, in the reaction-path the-
Meeting,October 1952. ory,16,17 it is postulated that strain embryos exist
This paper is based on a thesis by S. A. Kulin submitted in in the austenite that provide part of the activation
partial fulfillment of the requirements for the degree of Doctor of energy for the nucleation process. These embryos
Science in Metallurgy at the Mossachusetts Institute of Technology, are visualized to be regions of localized strain in
June 1951. The investigation was sponsored by the Office of Naval which the atoms in the austenite lattice are dis-
Research under Contract N5-ori-07814, Task No. XIV, Designa-
placed part way along the path to their ultimate
tion No. NR-031- 142.
positions in the martensite. Consequently, it would

TRANSACTIONS AlME JUNE 1952, JOURNAL OF METALS-661


room temperature. The structures were eompletely
austenitie after the foregoing eycle, the M, tempera-
ture being about -37°C. * The stress-relieving
• In this aJloy the transformation starts with a burst," an~ hence
M. = Mb. Ho~ever, there are certain conditions (to be dlscussed
later) under which M. > Mb.

'"0: '6,ooo'j:-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _~-..s?O~'_"__(_I


treatment was introdueed to avoid the superimposi-
tion of quenehing stresses on the applied stresses
'"
w
rr
I-- 12000 .
during subsequent eooling. This faetor will be dis-
'" ' eussed later.
6,000
Prior to microexamination, the speeimens were
usually tempered at 315°C for 10 sec in order to
4,000 "darken" any martensite that might exist.
Tensile stresses were applied to austenitized and
°0 JOD. 200 300 400 500 600 700 800 900 1000
stress-relieved specimens having a gage length of
PlASTIC STRAlf'l x 10 6 2.5 in. and a gage diameter of 0.250 in. The loads
Fig. l-Stress VS. plastie strain at temperatures abave M,. were inereased in small inerements, with intermedi-
Horizontal diseontinuities indieate stress-indueed martensitie ate unloading. The permanent set, if any, was de-
transformation. 20 pet Ni-O.5 pet C steel. termined with two eleetrieal resistanee strain gag es,
mounted longitudinally on eaeh speeimen. For
be predicted that applied stress, even within the simultaneous transformation measurements, the
elastie range, might eontribute to the displacement ehanges in eleetrical resistanee of the specimens
in the embryos and thus enable them to aehieve the were observed by means of a Kelvin double bridge,
eritieal strain for nucleation at temperatures above using potential leads spotwelded near the fillets.
M Furthermore, beeause of the unique nature of
S'
The loading was carried out at various temperature
the atomie displacements that eonvert the austenite levels, maintained constant to ± 0.5°C. The sub-
lattiee into martensite,18 it ean be expeeted that the atmospheric temperatures were aehieved by adding
direetion, as weH as the magnitude, of the applied powdered CO, to the oil bath in whieh the speeimens
stress should be an impor.tant variable in the nu- were immersed during the testing. Other types of
cleation proeess. This paper deseribes a number of stress experiments will be described later in the
eritieal experiments that were designed to resolve paper.
these questions. Martensitic Transformation Under Applied Tensile
Experimental Details Stress
An alloy steel eontaining 20 pet Ni and 0.5 pet C The foregoing "load-unload" procedure was ap-
was chosen for this investigation beeause its normal plied to austenitized and stress-relieved speeimens
martensitie transformation oeeurs in a eonvenient of the 20 pet Ni-O.5 pet C steel at temperatures be-
subzero temperature range, and the extent of iso- tween 20° and -25°C. In eaeh instanee, stress vs.
thermal martensite formation","'·2l is negligibly plastic strain (permanent set) was plotted as shown
small, thereby minimizing any possible complieation in Figs. 1 and 2. The method was eapable of de-
from the time variable during the stress studies. teeting plastic strains down to ±2x10·'. Plastic
The eomplete analysis was 19.99 pet Ni, 0.52 pct C, strains up to 0.012 were studied here.
0.37 pet Mn, 0.47 pet Si, 0.010 pet P, 0.015 pet S, The sudden yielding denoted by the horizontal
0.04 pet Cr, balance Fe. lines in Figs. 1 and 2 eorresponds to martensite for-
The speeimens were austenitized at 1095°C for mation. This was established both by the aecom-
30 min in a purified nitrogen atmosphere, and panying decrease in eleetrieal resistanee and by
quenehed in oil at 20° to 22°C. They were then subsequent microexamination. At temperatures in
sealed in evaeuated Vyeor tubes and stress relieved the vicinity of -15° to -25°C, sharp clicks eould be
at 480°C for 30 min, followed by furnace eooling to heard as rather large amounts of martensite formed
in bursts, even though these temperatures were
definitely above the martensite range for the un-
40,000 stressed material.
·20'(
No transformation was detected in the specimens
36,000 ,--~ strained at 20°C and there was only a slight trace
20'(
at O°C; in these eases, the eleetrical resistanee in-
0'( creased progressively with the plastie deformation.
However, at -15°C and below, appreciable quan-
tities of martensite were produeed during the tests,
"
r and (for a given permanent set) inereased with de-
ereasing temperature. This trend is illustrated by
~ 20,000
w
er:
l
r the data in Table I.
0-
If)
16,000 -20'( ! Even more significant is the fact that the ineeption
of the transformation was indueed at temperatures
12'000l above M, or Mb by tensile stresses weH within the
8,000 elastic range. For example, at 20°C the elastie limit
(corresponding to a permanent set of 2x10·') was
4,000 about 17,000 psi and it would be expeeted to be at
- 2 5'(
least as high at -25°C. However, at the latter tem-
°OC=c:==::'::::j3==4=~::::..~6--:--:'8--:9'---1--::-0--:II-~,2 perature, the transformation was triggered by a
STRAIN x 10 3 tensile stress of only 1000 psi. It may be concluded
Fig. 2-Stress vs. plastic strain at temperatures above M,. 20 pet that the M, (and Mb) temperature ean be raised by
Ni-O.5 pet esteei. Compare with Fig. 1. elastic tensile stresses.

662-JOURNAL OF METALS, JUNE 1952 TRANSACTIONS AlME


From the results of these experiments the yield-
ing by martensite formation and plastic flow by slip .,
may be regarded as two competitive deformation
mechanisms. While it is likely that either may affect
the other, neither one is necessary for the other and
the martensitic transformation sets in befor~ the
slip process if the tensile stress is applied at tem-
peratures not too far above M,. The onset of the
transformation due to tensile stresses within the
elastic range results in a "premature" yielding, and
therefore the material may have an abnormaHy low
yield strength. However, it is evident from Fig. 2
that the formation of martensite under stress pro-
duces more effective "strain hardening" (cf. curves
for -15° and -20°C) than does the ordinary slip
mechanism (cf. curves for 20° to O°C), and hence
relatively high tensile strengths may be encountered
despite the low yield strengths.
It was observed that the austenite twin bound- Fig. 3-Mortensitic chevrons Iying along twin boundaries of 20 pct
aries are sometimes preferred regions for the nu- Ni-O.5 pct C steel ofter plastically straining ot O·C. Tempered at
cleation of martensite under the application of 315· C for 10 sec. Etched with 4 pct nital and 1 pct zephiran
stress. As shown in Fig. 3, the initial martensitic chloride. XSOO.
plat~s were nucleated in pairs, like chevrons, ex-
tendmg more or less symmetrically into both the was austenitized and stress relieved in the usual
twin and the mother crystal. This phenomenon is way. It was then loaded in uniform bending at 20°C
readily explained in terms of the transformation to a maximum fiber stress of 14,000 psi, this being
strain analysis of Machlin.18 The cooperative atomic below the elastic limit of the material at room tem-
dis placements involved in the formation of a mar- perature. While loaded in this way, thespeciinen was
tensite plate require considerable movement in the slowly cooled until a burst of transformation oc-
surrounding austenite, and the associated strain en- curred. The temperature at that instant was
er~y contributes materially to the energy of nucle-
-21.9°C, whieh was weH above the M, = Mb for the
~tlOn. However, since the twin boundary is a mirror
unstressed material. The specimen was immediately
lmag.e of the lattice on either side, the formation of heated to room temperature, unloaded, and tem-
a paIr of plates reduces the strain energy require-
pered to "darken" the martensite.
ments because both plates tend to move the inter- Metallographie examination revealed (Fig. 4)
vening austenite in ab out the same direction. Under that the transformation had taken plaee in the part
these conditions, the twin boundaries should be of the speeimen subjected to tensile stress, whereas
favored sites for nucleation, provided that the plates no transformation was observed in the part under
form cooperatively in pairs. In the specimen repre- compressive stress. Also, the neutral axis of the
sented by Fig. 3, only aminute amount of trans- bent bar had shifted toward the compressive side,
formati0l! occur~ed at O°C on plastically straining presumably beeause of the yielding on the tensile
to 0.012 In. per In., but most of it took place along
side due to the transformation there.
twin planes in the form of double plates. At the
lower straining temperatures where substantial
quantities of martensite appeared, the plates had a
more conventional distribution.
Martensitic Transformation Under Bending Stress
Because of the shear-like nature of the marten-
sitic reactions along the habit plane, it might be
assumed that the stimulating effect of elastic tensile
stress is caused solely by the resolved shear com-
ponent of the applied stress. If this were the case
uniaxial compressive stress should produee simila;
results. A simple bending experiment will suffiee
to prove that such a shear eriterion for the activa-
ti on of the transformation by stress is invalid.
A %x%x4 in. bar of the 20 pet Ni-O.5 pet C steel

Table I. Extent of Transformotion as a Function of Tensile-Test


Temperoture

Pet Martenslte
Temperature of Formed (For PlasUc
:renslle Strain of
Test, oe Approxlmately 0.012)

Fig. 4-longitudinal (vertical) section through bar of 20 pct Ni-O.5


20 o pet C steel cooled to Mb temperature while subjected to elastic
o Trace
-15 0.5 bending. Top is compression side, bottom is tension side. Tempered
-20 12.3 at 315°C for 10 sec. Etched with 4 pct nital and 1 pct zephiron
-25 26.0
chloride. X8.

TRANSACTIONS AlME JUNE 1952, JOURNAL OF METAL5--663


the case of the bending experiments, the trans~or­
o AS QuENCHtO mation takes place "prematurely" in the regl<;IllS
o STR ESS RELI(V(D
'0 20'\ No, O.'5-"C STEEL
sustaining tensile stresses, but not in those sus~aIn­
ing compressive stresses. The higher the resIdual
'0 - tensile stress in the core of the specimen, the sooner
does the burst occur on cooling and the more re-
'0 stricted it becomes. With little or no residual stress
present, the M. temperature is lower and the bur~t
10 takes place more uniformly throughout the specl-
men.
On the basis of this reasoning, it may be pre-
~~O---~'6~~-4~'--~-'~'--~-'~4--~-'~O~-~'6'-~-'~'--~-'I'
(Mb 1 BuRST TEMPERATuRE ·C dicted that the completely austenitic rim remaining
after the burst in a nonstress-relieved specimen,
Fig. 5-Percent martensite formed during the burst
ys. M. temperature. such as that of Fig. 6, should und ergo transforma-
tion on cooling below the M. temperature, since ~he
martensite formed in the core causes an expanSIOn
These findings demonstrate not only that elastic there which reduces the tensile stresses in the core
tensile stresses promote the transformation, but as weIl as the compressive stresses in the rim. This
also that elastic compressive stresses do not act in was confirmed: on cooling some 10° to 15 °e below
a corresponding manner. Inasmuch as the shear M., the transformation became fairly uniform across
stress on geometrically similar planes is the same in the transverse section of the specimen.
both the tensile and compressive regions of the bent To show the effect of the superimposition of re-
specimen, then it must be the sign of the normal sidual and applied stresses, a 1f4 in. diam x 2ljz in.
stress that is responsible for the striking difference rod was water quenched from the austenitizing tem-
in transformation behavior. Thus, it is now clear perature, placed in elastic bending along the central
that, although the component of shear stress along portion, and then cooled until a burst of transforma-
a habit plane undoubtedly plays an important part tion was noted by electrical resistance measure-
in the activation process, the component of stress ments. After tempering the specimen to "darken"
normal to this plane is likewise important. By the the martensite cross-sections were examined metal-
same token, a critical resolved shear stress or shear lographically in regions subjected to the bending
strain or shear-strain energy cannot provide a suit- and in regions extern al to the bending clamp. Fig.
able criterion for the initiation of the transformation, 7b illustrates the latter situation, with a symmetri-
because the contributions of the magnitude and di- cal core of transformation corresponding to the state
rection of the normal stresses are thereby neglected. of high tension in the interior produced by the water
Effect of Residual Stresses on Transformation
It is evident from the results presented thus far
that residual stresses may influence the transforma-
tion and, in fact, this was the reason for stress re-
lieving the austenitic specimens prior to the applied
stress studies. The role of residual stresses may be
illustrated in the following way: Two series of lJ4
in. diam x 1fz in. long specimens were austenitized,
one series was then stress relieved at 480°C for 30
min and the other was not. On subsequent cooling,
the stress-relieved specimens were found to have
M. temperatures between -34° and -40°C (av-
erage -37°C), while the specimens without stress
relief had M. temperatures between -20° and
-30 0 e (average -26°C). Furthermore, the amount
of transformation occurring during the burst de-
creased with increasing M. temperature. This cor-
relation was the same for both series of specimens
as shown in Fig. 5.
In the nonstress-relieved series, it was found that
the quantity of martensite formed during the burst
varied markedly across the seetion. The specimen
-
with the highest M. at -20 °C was entirely austenitic

,
around the rim, the transformation being confined
to the central portion. Fig. 6 shows the austenitic
rim on a transverse section, with some of the mar-
tensite formed toward the interior of the specimen.
l4
t ~
1*-
/ '
The other sampIes in this series with lower M. tem- I
peratures had progressively less of this rim effect, ~

and the stress-relieved specimens contained a rela-


tively uniform amount of transformation across the
''1' '
section, as in Fig. 7a.
Fig. 6-Martensite formation in 20 pet Ni-O.5 pet C steel austeni-
These results are attributable directly to the tized at 1095°C, quenched to room temperature, and subcooled to
residual stresses in the austenitized steeI. The M. (-20.1 °C). Tempered ot 315°C for 10 sec. Etched with 4 pet
quenching from the austenitizing temperature puts nital ond 1 pet zephiran chloride. Note eompletely austenitic rim.
the specimens into astate of tension in the interior Bottom of picture shows martensitic plates formed toword the in-
and astate of compression near the surface. As in terior of the specimen. X100.

664-JOURNAL OF METALS, JUNE 1952 TRANSACTIONS AlME


quenching. Fig. 7c shows how the core of trans-
formation was changed in shape and shifted to the
side of the specimen where the bending moment had
added tensile components to counterbalance the re-
sidual compressive stresses in the rim.
Criterion for Effectiveness of Applied Elostic Stress
It has been demonstrated that a proper criterion
for the initiation of the transformation by applied
elastic stress must take into ac count the normal, as
weIl as the shear, components. A feasible approach
to the problem is derived from the nature of the
transformation strains. The generation of a mar-
tensitic plate involves two displacements,18 the first
being a homogeneous strain that can be described
as the motion of planes of no rotation and no dis-
tortion lying parallel to the habit plane. The direc-
tion of atom motion does not lie in this plane; hence
the displacement is not one of simple shear. It may
be assumed that this first strain is the primary dis-
placement that occurs during the nucleation process.
The second strain, which finally yields the marten-
sitic structure, is homogeneous only over small re-
gions compared to the volume of the plate, and will
be regarded as playing no significant role in the
activation phenomenon. The planes of no rotation
and no distortion during the first strain slide past
one another, corresponding to a shear strain of about
0.2; but they also move apart, corresponding to a
normal (dilatation) strain of about 0.05.* . Thus, the
• These measurements were made on a 70 pet Fe-30 pet Ni alloy,
having a {259) habit," and are assumed to apply to the alloy under
study here. Both alloys transform in approximately the same range
of subzero temperatures and both exhibit burst phenomena.

effectiveness of an applied stress should depend


upon its ability to produce both shear and normal
strains of the correct magnitude with respect to a
potential habit orientation. Moreover, anormal com-
pressive component will oppose the dilatation strain
of the transformation, whereas anormal tensile
component will aid it.t According to this concept
t The sign of the shear eomponent is of little eonsequenee, sinee
the many habit orientations virtually permit shearing in either
direction.

the results of uniaxial tension and of uniaxial com-


pression are expected to be quite different.
The foregoing combined strain criterion can be
treated quantitatively through the use of a Mohr
two-dimensional strain circle. 23 Because of the sym-
metry involved only plane strains need be con-
sidered and the strain circles are representative of
any longitudinal plane passing through the tension
or compression axis.:j: In this method of depicting
:I: The plane strains are visualized to aet on a potential habit plane
whose normal lies in the plane of the strain elrele. Thus the latter
plane is perpendieular to the potential habit plane, and also eon-
tains the direetion of atom motion during the first transformation
strain.
Fig. 7-Martensite formation in 20 pet Ni-O.5 pet C steel austeni-
plane strains, the principal strains (E, and E2) are tized at 1095·C then treated as indieated. Finally tempered at
plotted along the x-axis from a common origin (Fig. 315·C for 10 sec and etched with 4 pet nital and 1 pet zephiran
chloride. XIO.
8) and a circle is constructed of radius
2 a-(top)-Oil quenched from austeniti:z:ing temperature, stress
relieved at 480·C, and subeooled to Mb (-37·C). Note relatively
with E, taken as the larger strain in the positive uniform distribution of martensite across the transverse seetion.
direction. The circle passes through E, and E2 and has
b-{centerl-Water quenched from austenitizing temperature and
E, + E2
=
subcooled to Mb (-20·C). Note completely austenitic rim with
its center at - - - - Letting v Poisson's ratio, transformation confined to the central portion of the section.
2
c-(bottom)-As in b, but with elastic bending moment superim-
then E2 = - VE, where E, is the principal (positive) posed on quenching stresses during the subeoaling to Mb (-20·Cl.
strain parallel to the tension axis and E2 is the princi- Note the shift of the transformed region toward the tension side
pal (negative) strain perpendicular to the tension (bottom of pieture) of the applied bending moment.

TRANSACTIONS AlME JUNE 1952, JOURNAL OF METAL5-665


of opposite sign in the lateral direction due to
Poisson's ratio.) However, since most of the orien-
tations in this instance sustain normal compressive
strains, a chain of reaction would tend to be in-
hibited, and Mb should be lowered even though M,
11 apeclmen may be raised by the applied compressive stress.
c
Thus, while M, = - 37°C = Mb for the unstressed
A axis
f:-lIf--
, I 'Ilcpplied alloy, it is predicted that M, > - 37°e > Mb for the
tensile
strain uniaxially compressed alloy. Of course, under uni-
axial tension both M, and Mb should be raised ac-
cording to the proposed criterion and this has been
demonstrated already (Figs. 1 and 2).
(al Correspondlng to 0 unloxiol tenstle stress
I To test the above prediction, austenitized and
stress-relieved specimens were elastically loaded at
room temperature, and electrical resistance changes
were measured during subcooling with the speci-
mens und er constant stress. Uniaxial compression of
15,000 psi was applied in one run, and uniaxial ten-
sion of the same magnitude was applied in another.
Unloaded specimens were also checked for compara-
tive purposes. The resulting curves in Fig. 9 show
opplled that M, and Mb were raised by 17°e in the tension
compreSSive straln experiment, while in the compression run, M, was
(bl CorrespondlnQ to 0 unloxtol compress1ve stress raised by 10 0 e and Mb was lowered by Boe.
Fig. 8-Two dimensional Mohr strain circle depicting The relative effectiveness of uni axial tension and
narmal and shear strains as 0 function of orientation. compression in inducing the transformation can be
calculated. Granting that the critical combined
strain can be defined by a shear component '}'. and a
axis. Similarly for uniaxial compression, E, = - 1'E" normal component E., eq 1 yields for uniaxial ten-
where E2 is the principal (negative) strain parallel sion:
to the compression axis, and E, is the principal (posi-
tive) strain perpendicular to the compression axis. -E.(1-1') +YE: (1 +1'),+1',),:
E, (positive) = [2]
These two cases are illustrated in Fig. Ba and b. 21'
A point on the strain circle lying at a radial angle and for uni axial compression:
2a from the x-axis corresponds to a plane whose
normal makes an angle IX with the E, direction. It can -E. (1-,,) -YE: (1 + ,,)2+1',},'-
be shown that abscissa of such a point denotes the E2 (negative) = --~--=-----.:._~-~-~ [3]
21'
normal strain (E) sustained by the plane in question
and the ordinate of the point denotes one-half the It is then assumed that the critical components ')'.
and E. stand in ab out the same ratio as the observed
shear strain ( ; ) sustained by this plane. The shear and normal components (0.2 and 0.05) of the

r
mathematical relationship is:
~

(_El_~_E2_ E, ~ E2 [1] 2 0
>-
....
:>
which is the equation of the strain circle. Because .... 20
(f)
of the random orientation of the potential habit (j)
W
planes and the symmetry of the applied strain Ir
40
system about the specimen axis,§ it is only necessary -'
<!
'::'
§ Also see the second (t) note on p. 665. Ir
....
0
w 60
to consider orientations of IX between 0 and 90°; this -'
w
corresponds to points lying along the semicirc1e ~
ABC in Fig. 8. Between A and B, E and '}' are positive w
80
and the orientations are such that transformation is (f)
<!
aided by the acting strains. On the other hand, be- W

tween Band C, € is negative and these particular ~


w
100
0
orientations sustain normal compressive strains that
oppose the dilatation component of the transforma- -'
~ 120
tion strain. 0
i=
From a comparison of Fig. Ba and b, it is evident 0
<!
that arc AB (tension case) is greater than arc AB Ir
CL
140 (b) (0) (c)

(compressive case), and hence there should be many 20 10 -40 - 50


more favorable orientations for transformation
under uni axial tension than und er uni axial com- Fig. 9-Fractional decrease in electrical resistivity vs. tem-
pression. Yet even in the latter case, the surprising perature far specimen of 20 pct Ni-O.5 pet C steel cooled
conclusion is reached that there should be some (al without applied stress, (bl with uniaxial tensile stress
orientations that are aided by the uniaxial compres- of 15,000 psi, and (cl with uniaxial' compressive stress of
sion. (Actually, this results from the normal strains 15,000 psi.

666-JOURNAL OF METALS, JUNE 1952 TRANSACTIONS AlME


a-Subcooled to -45 ° C with outer rim under triaxial eompressive b-Reeaoled to -45°C without applied stress. Note relative uni-
stress. Note lack of transformation there. formity of the transformation oe ross the section.

Fig. 10-Martensite formation in 20 pet Ni-OS pct C steel austenitixed at 1095°C, water quenched to room temperature, and subcooled
os indicated. Finally tempered at 315°C for 10 sec and etched in 4 pct nital and 1 pet xephiran chloride. Xl0.

first transformation strain.'8 Substituting in eqs 2 the specimen due to the tensile stress there. This
and 3 and letting v = 0.3, it follows that: was found to occur as a burst at -31°C. (It is evi-
E2 (compression) = -1.8 El (tension) [4] dent that the applied compressive stress partly
This indicates that the uniaxial stress in compression cancelled the residual tension in the core and there-
fore the transformation set in at a lower temperature
required to initiate the transformation under a given
than if the compressive load had not been applied.
set of conditions is 1.8 times the uni axial stress in
Without such loading, the water-quenched specimen
tension required to initiate the transformation under
would start to transform in the core at about
the same conditions.
- 26°C.) The aluminum has about the same thermal
A direct experiment to test this calculation has not
coefficient of contraction as the austenitic alloy, and
yet been made, but the fact that the raising of M. by
hence it resisted the expansion due to the trans-
15,000 psi tension is 1.7 times the raising achieved
formation in the core at -31°C. This generated com-
by 15,000 psi compression suggests that the theory
pressive stresses transverse to the specimen axis
stands a good chance of being quantitatively correct.
It would also give strong support to the re action- (estimated to be comparable to the elastic limit) and
path concept which provides the basis for the the applied axial compression completed the state of
triaxial compression. The lateral constraint pro-
'Yo 4 vided by the aluminum block was definitely opera-
assumption that - - ~ - . tive, as indicated by the fact that the applied axial
Eo 1
stress jumped from 15,000 to 16,000 psi simul-
Experiments with Triaxial and Hydrostatic Compressive taneously with the burst. However, the applied load
Stresses was quickly reduced to the original value in the
The positive normal strains that exist in a uni- testing machine before further cooling.
axially compressed specimen because of Poisson's The specimen, triaxially compressed as described,
ratio are eliminated in case of triaxial and hydro- was cooled to - 45°C, held there for 5 min, returned
static compression. Furthermore, in the latter case, to room temperature, and unloaded. The aluminum
the shear components are also avoided. Therefore, block was removed, and a portion of the specimen
the combined strain criterion indicates that both M. that had been surrounded by the block was tempered
and Mb should be lowered by tri axial and hydro- at 315°C to "darken" the martensite. From Fig. 10a,
static compression. it can be seen that the core of the rod had undergone
A 0.34 in. diam x 3 in. long specimen was austeni- transformation, corresponding to the burst at
tized and water quenched. Without being stress re- - 31°C, but the rim was austenitic even after the
lieved, the specimen was copper plated and pressed cooling to - 45°C. The small amount of transforma-
into a central hole in an aluminum block, 2 in. OD tion visible near the edge of the specimen was un-
x 1 in. long. The specimen was then loaded in uni- doubtedly an extraneous effect of the superficial dis-
axial compression to an elastic stress of 15,000 psi tortion that occurred when the rod was forced into
and cooled slowly below room temperature. It was the aluminum block. When the same specimen was
known from the experiments on residual stresses recooled to - 45°C, und er no external stress the
that the transformation would start in the core of martensitic transformation took place in the' pre-

TRANSACTIONS AlME JUNE 1952, JOURNAL OF METALS-667


viously untransformed rim, as shown in Fig. lOb. Martensite Reaction. Cold Working 01 Metals. A.S.M.
This experiment demonstrates that M. and M. ean (1949).
2 H. Scott: Dimensional Changes Accompanying the
be depressed by triaxial compressive stresses.
Through the eourtesy of Professor P. Bridgman Phenomena of Tempering and Aging Tool Steels. Trans.
at Harvard University, a few runs were carried out A.S.S.T. (1926) 9, p. 277.
under hydrostatie pressure. With austenitized and • H. Hanemann and A. Schrader: On Martensite.
stress-relieved speeimens, the burst temperature was Trans. A.S.S.T. (1926) 9, pp. 169-233, 364.
lowered from the customary -37°C to about -78°C • S. L. Hoyt: Stresses in Quenched and Tempered
Steel. Trans. A.S.S.T. (1927) 11, pp. 509-530, 658.
by apressure between 2000 and 4000 atm. The M. 5 K. Tamaru and S. Sekito: On the Quantitative De-
temperature was also lowered by the hydrostatic termination of Retained Austenite in Quenched Steels.
pressure, but the extent was not determined with Sci. Rep., Sendai, Series 1 (1930) 20, p. 377.
eertainty. Further experiments along these lines are 6 F. Lazlo: Tessellated Stresses, Part H. Journal Iron
underway. and Steel lnst. (1943) 148, p. 137.
7 E. Scheil: Uber die Umwandlung des Austenits in
Summary
Martensite in Eisen-Nickel-legierungen unter Belas-
The formation of martensite in a 20 pet Ni-0.5 pct tung. Ztsch. anorg. al/gem. Chemie (1932) 207, p. 21.
C steel und er the application of stress has been 8 A, W. McReynolds: Effects of Stress and Deforma-

studied as a function of the temperature of loading tion on the Martensite Transformation." Journal of
and the state of stress. The principal results may be Applied Physics (1949) 20, p. 896.
summarized as follows: 9 C. S. Barrett and O. R. Trautz: Low Temperature
1-The martensitic transformation ean be initiated Transformation in Lithium and Lithium-Magnesium
by elastic stresses at temperatures above M •. Hence Alloys. Trans. AlME (1948) 175, p. 579; METALS TECH-
the role of applied stress is not merely one of pro- NOLOGY (April 1948).
10 A. B. Greninger and V. G. Mooradian: Strain
ducing nuclei by plastic deformation.
2-It is suggested that the role"qf applied stress at Transformation in Metastable Beta Copper-Zinc and
temperatures just above M, is to supply critical Beta Copper-Tin Alloys. Trans. AlME (1938) 128, p.
strains to existing embryos. This makes it unneces- 337; discussion, p. 355.
11 A. B. Greninger: The Martensite Transformation in
sary to adopt the assumption that the austenite
lattice becomes meehanically unstable at MB' Beta Copper-Aluminum Alloys. Trans. AlME (1939)
3-The normal strain aeting on a potential habit 133, p. 204; discussion, p. 222.
12 K. Mathieu: Effect of Alloy-element Content and
plane has an important effect on the nucleation
process, and therefore a critical resolved shear stress Testing Conditions on Gamma to Alpha Transformation
or strain cannot be regarded as a valid criterion for in Cold-Deformation of Austenitic Steels. Archiv
Eisenhuttenwesen (December 1942) 16, p. 215.
inducing the transformation.
" C. B. Post and W. S. Eberly: Stability of Austenite
4-Shear strains and positive (tensile) normal in Stainless Steels. Trans. A.S.M. (1947) 39, p. 868;
strains acting on a potential habit plane promote discussion, p. 888.
the transformation, but negative (compressive),nor- 14 D. J. McAdam, Jr., W. Geil, and F. J. Cromwell:

mal strains oppose it. Hence, strain energy, being lnfluence of Low Temperatures on the Mechanical
independent of the sign of the normal strain, cannot Properties of 18: 8 Chromium-Nickel Steel. Journal of
provide a suitable criterion for the transformation. Research, Nat. Bur. Standards, Research Paper 1882
5-Based on the reaction-path theory, a combined (1948) 40, p. 375.
15 N. A. Ziegler and P. H. Brace: Hardening of Auste-
strain criterion is presented, which takes into ac-
count the shear and dilatation components of the nitic Stainless Steels by Mechanical Working at Sub-
Zero Temperatures. Proc. A.S.T.M. (1950) 50, p. 861.
atomic displaeements during the transformation, and 16 M. Cohen, E. S. Machlin, and V. G. Paranjpe:
which is consistent with the following observations: Thermodynamics of the Martensitic Transformation.
a-M, and M. are raised by uni axial tension. Thermodynamics in Physical Metallurgy. Trans. A.S.M.
b-M, is raised, but Mb is lowered, by uniaxial (1950) 42A, p. 242.
17 E. S. Machlin and M.. Cohen: The lsothermal Mode
compression.
of the Martensitic Transformation. JOURNAL OF METALS
c-M. and Mb are lowered by tri axial and (May 1952), Trans. AlME, p. 489.
hydrostatic compression. 18 E. S. Machlin and M. Cohen: Habit Phenomenon in

6-Applied and residual stresses appear to be the Martensitic Transformation. Trans. AlME (1951)
additive in their effect on the transformation. 191, p. 1019; JOURNAL OF METALS (November 1951).
19 G. V. Kurdjumow and O. P. Maksimova: Kinetics
7-Yielding by slip or by the stress-induced trans- of Austenite to Martensite Transformation at Low Tem-
formation may be considered as competitive de- peratures. Doklady Akademii Nauk SSSR (1948) 61,
forma.tion processes. Either may precede the other, No. 1, p. 83.
and eIther may influence the other. G. V. Kurdjumow and O. P. Maksimova: On the
Energy of Formation of Martensite NucleL Doklady
Acknowledgments Akademii Nauk SSSR (1950) 73, p. 95.
20 B. S. Lement and S. C. Das Gupta: lsothermal
The authors wish to express their appreciation to
Miriam E. Yoffa and to George Black of the Dept. Formation of Martensite at Sub-Zero Temperatures in
a High Chromium Steel. Trans. AlME (1951) 191, p.
of Metallurgy at the Massachusetts Institute of 727; JOURNAL OF METALS (September 1951).
Technology for assistance in various phases of this 21 S. A. Kulin and G. R. Speich: lsothermal Marten-
investigation. The authors are also indebted to the site Formation in an lron-Chromium-Nickel Alloy.
Office of Naval Research for sponsorship of a re- JOURNAL OF METALS (March 1952), Trans. AlME, p. 258.
search program of which this work was apart. 22 E. S. Machlin and Morris Cohen: The Burst Phe-

nomenon in the Martensitic Transformation. Trans.


References AlME (1951) 191, p. 746; JOURNAL OF METALS (Septem-
1 B. L. Averbach, S. A. Kulin, and M. Cohen: The
ber 1951).
Effect of Plastic Deformation on Solid Reactions: Part 2S R. Heindlhofer: Evaluation of Residual Stress.
II-The Effect of Applied Stress and Strain on the (1948) New York. McGraw-Hill Book Co.

668-JOURNAL OF METALS, JUNE 1952 TRANSACTIONS AlME

You might also like