Microstructure and cleavage resistance of
low-carbon bainitic steels .
P. Brozzo, G. Buzzichelli, A. Mascanzoni, and M. Mirabile
Charpy- V and three-point bending tests, coupled with
uniaxial tensile tests on unnotched specimens, were
carried out to investigate the relationship between
microstructure and cleavage resistance in low-carbon
2Mn-3Cr bainitic steels (C < 0·05%). While the
influence of microstructure on yield strength has to be
viewed mainly in terms of the mean subgrainsize, the
structural unit controlling cleavage fracture can be
identified as the covariant (bainitic) packet whose size
dB is very close to the average unit crack path
measured on fractured specimens. Although
apparently still dependent on C and S content, the ..
impact transition temperature FA TT decreases at the
rate of -14 degC mml/2 as the dBl/2parameter varies
in the 9·1-14·4 mm-1/2 range. The cleavage stress,
instead, increases linearly from 1700 to 2700 MPa
regardless of C content and the volume fraction of
sulphides. The critical stage in the fracture process
appears to be the propagation of a Griffith crack from
one bainitic packet to another, and the resistance
offered by high-angle boundaries is approximately the
same as that of steel with a polygonal ferrite
microstructure.
Paper No. MS 577. Manuscript received 21 May 1976; in final
form 2 December 1976. The authors are with the Centro
Sperimentale Metallurgico S.p.a., Rome, Italy.
Considerable efforts have been made during the past few
years to quantify the structure/property relationship of low-
carbon acicular ferrite steels, since this type of steel
possesses strength, toughness, and ductility in a particularly
satisfactory combination. Compared to traditional ferrite-
pearlite structures, low-C .acicular ferrite can give the
following higher performances:I-12 (i) a 200-300 MPa
increase in yield stress; (ii) a substantial improvement of
shelf energy in the Charpy test; (iii) maintenance of the
ductile --+ brittle transition temperature within acceptable
limits.
Initial attempts to rationalize the experimental data
relative to acicular ferrite steels of different compositions
followed the same lines adopted for ferrite-pearlite steels.
Therefore, in interpreting variations of properties such as
mechanical strength and cleavage resistance, particular
attention has been devoted to the role of the bainitic or
martensitic packet, i.e. the microstructural unit of acicular
ferrite equivalent to the polygonal ferrite grain of mild steels.
Alternatively, in some cases, the prior austenite grain has
been considered, on the assumption that the size of the
packet varied in accordance with it.
As far as cleavage resistance is concerned, some published
results appear to indicate that the role of the packet is similar
to that (already well-known) of the polygonal ferrite grain in
mild steels or, more generally, in iron alloYS.13Irvine and
Pickering,14 in fact, had already reported, for 0·12%C
bainitic steels, that increasingly improved 50% FA TT values
could be obtained by progressively refining the prior
austenite grain size. Roberts15 reported variations of FATT
with martensitic packet size in Fe-Mn alloys. Dolby and
Knott16 gave metallographic evidence of packet-size control
in HY80 HAZ microstructures. Finally Gladman, Dulieu,
and McIvor17 also propose a dependence of FATT in
bainitic structures on the mean spacing between high-angle
intersections d, with a value for the ~F ATT/ ~et-l/2
coefficient identical to that measured for ferrite, Le.
-11·5 degC mm1/2.*
Other investigations 18-23have related FA TT to para-
meters other than packet size measured on metallographic
specimens: unit crack path (UCP),18 or effective grain
size,20-22these being measured on fractured surfaces. The
UCP is directly related to the size of the covariant
packet but appears to be systematically larger, 19,22,23
because, as is generally accepted, packets generated
from the austenite according to different variants of the
Kurdjumov-Sachs orientation relationship may present a
continuity of their {100}cleavage planes within a misorienta-
tion of ~ 100.
A similar concept is expressed in the work of Naylor and
Krahe24 in which, however, the correspondence between
packet diameter and fracture facet is explained by a rough
continuity which various sets of cleavage planes could find
across groups of laths. In any case it could be concluded that
the mean straight path of a propagated cleavage crack would
provide a measure of the frequency of 'effective' high-angle
misorientations between sets of cleavage planes.
Nevertheless, even on. the basis of these results, the
marked structural differences distinguishing low-C bainitic
steels (sub-boundaries, high dislocation density, occasional
absence of carbides) from mild steels (low dislocation
density, presence of carbides) do not allow direct ex-
trapolations with regard to the fracture mechanisms actually
operating. Recent experiments, indeed, tend to throw doubt
upon the findings of some authorsl5,25-27 who claim that
packet size has a marked influence on yield stress, and
therefore any theory leading to a calculation of critical
fracture stress based on Petch's law would not be directly
applicable. In fact, Brownrigg28 has found that the prior
* However, it should be noted that the size of the packet determined
by metallographic methods does not necessarily coincide with the
dimension of the covariant packet (i.e., the a region transformed by
adopting the same variant of the Kurdjumov-Sachs orientation
relationship).
Metal Science April 1977 123
124 Brozzo et al. Microstructure and cleavage resistance

austenite grain size has a very weak influence

(K =0·14-0·19 MN m-3/2) on aO•2 in martensite with low-C
content and with 4-6%Mn. An analogous conclusion about
the absence of a Petch effect due to packet size was reached
by Buzzichelli and Mascanzoni,12 for bainitic and
martensitic 2%Mn-3%Cr low-C steels subjected to various
heat-treatments (supporting evidence for this conclusion will
be given later in this paper), and by Gladman, Dulieu, and
McIvor for bainites.17
It may be concluded that the mechanical strength and
cleavage resistance of low-C bainites appear to be controlled
by different structural parameters and not, as in the case of
polygonal ferrite steels, by the same structural unit.
The purpose of the present research work was to
investigate in greater detail the relationship between the
microstructure and the cleavage resistance of low-C bainitic
steels. Fracture stress has been chosen as a parameter for the
following reasons: (i) fracture stress is more significant,
from a physical point of view, than impact transition
temperature, since it is generally accepted that cleavage
fracture is regulated by a critical stress criterion; (ii) since
numerous data on critical cleavage stress of mild steels are
available in the literature,29-39 a direct comparison can be
made between the cleavage resistance of these two classes of
steels, with the further possibility of pointing out the
differences in the fracture mechanisms. A certain number of
Charpy- V transition curves were also determined, in order to
achieve a more complete characterization of the materials
tested.

MATERIALS AND EXPERIMENTAL TECHNIQUES

The basic composition of the steel investigated was
2%Mn-3%Cr, with two carbon and sulphur contents 1 Typical bainite microstructure (SEM).
(-0·025%. and ,....,0·050%C; ,....,0·005% and ,....,0·014%S)
(Table I). The steels were vacuum-melted and the ingots, microscopy observations and by SEM investigations with x-
each ,....,
100 kg, were hot-rolled to ,....,
12 mm thick plates. The ray energy microanalysis equipment.
hot-rolling operation was programmed so as to obtain a total Critical cleavage stress values were determined by three-
thickness reduction of ,....,90%at temperatures < 1050°C and point bending tests on Charpy V-notch specimens (taken in
finishing temperatures of ,....,950°C. After rolling the plates the transverse direction) using an Instron machine (rate of
were air-cooled to room temperature. Subsequently some of crosshead movement: 17 x 10-3 mm S-I) and by uniaxial
the materials were heat-treated in a Data Track simulator tensile tests on unnotched specimens with a strain rate of
furnace, with a wide range of controlled heating and cooling ,....,10-4 S-1 in the temperature range -196°C~T~ + 23°C.
rates, so as to vary the size of the austenite grain produced The load-deflection diagrams were used to obtain the
by the hot-rolling programme. After austenitizing, the general yield loadst (i.e. the load corresponding to the point
cooling rate was programmed to ensure for each type of steel where the load-deflection curve departed from linearity) and
tested an entirely 'acicular' microstructure corresponding to the fracture loads. The temperature at which the general
that of the as-rolled condition. yield curve intercepts the fracture load curve was taken as
The prior austenite grains of the bainitic structures were the nil ductility temperature (NDT) for the steel being tested.
revealed by means of Bechet-Beaujard etch ant. Scanning Ductile -- brittle transition curves were also obtained for
electron microscopy was used on specimens etched with 2% three of the steels investigated, using a second series of
Nital for measuring the size of the bainitic grains (or Charpy- V specimens taken in the transverse direction and
covariant packets). The results obtained by this techni- subjected to the same treatment mentioned above.
que-which proved extremely useful in the case of materials
with a very fine austenitic grain (dA ~dO fJ.m)-were
checked against those given by transmission electron t General yield in a notched specimen is defined as the state reached
microscopy integrated with selected-area electron diffraction when the minimum cross-sectional area. of the specimen has yielded
patterns. The latter technique was used extensively for sub- completely. The general yield load for a Charpy V-notch testpiece is
structural investigations. the load required to propagate the hinges from the root of the notch
The inclusional state was determined by optical to the side of the specimen opposite to the notch.

Table I Chemical composition of steel

Material C Si Mn P S Cr N 0 Al (tot.)

MC25 0·025 0·04. 2·04 0·015 0·006 3·0 0·005 0·007 0·005
MCS25 0·028 0·04 2·05 0·015 0·012 3·10 0·005 0·01 0·005
MC50 0·050 0·04 2·07 0·015 0·003 3·06 0·005 0·007 0·005
MCS50 0·050 0·04 2·05 0·010 0·015 3·0 0·006 0·003 0·005

Metal Science April 1977

 Brozzo et al. Microstructure and cleavage resistance i25

2 Substructure of low-C bainitic steels deriving from 3 Cell structure in low-C bainitic steels deriving from
transformation of coarse-grained austenite (£1A~ 50 ,urn). transformation of fine-grained austenite(dA~10 ,urn).
Steel MC25. Steel MC25.

The fracture surfaces of the specimens were protected
with a nickel coating several flm thick and then etched with
5% Nital for -- 20 s. Two sections were then cut from the
fractured testpiece, one parallel to and the other at
right angles to the fracture plane, and the unit crack paths
(UCP) were determined by optical microscopy.
EXPERIMENTAL RESULTS AND DISCUSSION
Microstructure and yield strength
The hot-rolling programme adopted ensured that the
austenite was completely recrystallized at the time of phase
transformation with an average grain size of -- 50 flm.
Fig. 1 shows the typical microstructure of the tested
materials: the SEM clearly reveals the individual grains of
the bainite structure, consisted mainly of' packets of laths,
very thin in proportion to their length (Fig. 2). The sub-
structure, therefore, was morphologically very similar to that
oflow-C martensitic steels (C < 0·2%).40-43
This transformation product has been tentatively defined
as 'bainite', following a kinetic definition,44 as it can be.
derived from the characteristic B s temperature shown by
CCT diagrams.12 No significant variation of lath width with
C content (--500°C~Bs~--600°C) was recorded.
When the austenitic grain size had been reduced to
--10 flm by thermal cycling between 300 and 900°C, it was-
observed that the laths had been replaced in a considerable
number of packets by roughly polygonal subgrains, with a
resulting substructure similar to that developed in cold-
worked steels (Fig. 3). Furthermore, with the assistance of
SEM, the following empirical relationship was worked out
between the average size dB of the bainitic packet and
the corresponding average size d A of the austenitic grain:
dB = 1.4dg·sS
in which both dB and d A are expressed in flm and are
calculated by the mean linear intercept method. In many cases
the correctness of the bainitic packet size determined by SEM
was checked by microdiffraction on thin foils, observed under
the electron microscope.12
found possible to write 0'0.2 as the sum of three separate con-
tributions: a friction stress, O'i' similar to that of ferrite; a
term ~aC+N expressing solid solution (C + N) strengthening
by the same form .as in the case of low-carbon martensites;
and, thirdly, a term ~O's for hardening due to the sub-
structure, expressing an inverse proportionality of a to the
transverse dimensions of laths (I). This was concluded after
a study12 of YS variations as a function of C content
(--0·01 ~C%~--o·05), packet size (--3·5 ~da<flm)~25)
and subgrain size (bainitic or martensitic state), which
allowed the following observations.
Firstly, the presence of high-angle boundaries delimiting
the covariant packets appeared to have no significant effect.
The small increase in value obtained for 0'0.2 by refining the
austenitic grain size (~0'0./~dl/2~0·13 MN m-3/2) (Fig. 4)
was attributed to the higher strengthening effect of the
substructure of the bainite derived from very fine austenitic
grain (Fig. 3), as more recent results seem to confirm.4s
Secondly, at constant packet size (austenite grain size
~40 flm) and constant lath size (0·55 flm for bainite and
0·30 flm for martensite), the yield stress for both martensite
and bainite follows a (C + N)l/2 law. The slope of the
diagram when expressed as ~ (shear stress)1 ~(atomic
fraction)1/2 proved to be --1/18 of the shear modulus, in
agreement with accepted theories of the solid-solution
strengthening of low-C martensite.46
-1/2 -1/2
dB ,mm
o 5 10 15
_/::,.
700
o MC25
600
/::,. MC50
o
/::"/::"J&--r
Do 4 MPa mm1/2
--===::I:

Small carbon-enriched areas were found to be fairly o

600 _00-
o
o
common in all the steels tested, located along the boundaries o
of the laths or of the bainitic grains and containing residual 500
austenite or martensite in accordance with the cooling rate o 5 10 15
-1/2 -1/2
from the y phase. No carbides of any kind were observed in dA ' mm
the steels with 0·025%C, but occasional particles of Fe)C
were noted in those with 0·05%C. 4 0·2% proof stress as function of austenite grain size and
In interpreting the yield strength of these steels, it was corresponding bainite packet size.

Metal Science April 1977

126 Brozzo et al. Microstructure and cleavage resistance

Thirdly, the use of the expression Kt-I proposed by

Langford and Cohen47 for substructural strengthening of
200
cold-drawn iron, corroborated by the work of Young and
Sherby,48 not only fitted the 00.2 value for bainite, but could
account particularly well for the difference of yield stress
existing in the same steel between bainite (i",,0·55 11m) and
martensite ([",,0,30 11m).Following Langford and Cohen, the 100
term KI-I takes also into account dislocation density inside
the subgrains.
To summarize, the following expression was then used:
00'2 (MPa) = 0i + L\oc+ N + L\os -50
°C
4
= 30 + 1900 {(C + N)%}t + 1'22 x 10- (/)-]
5 Effect of bainitic packet refinement on Charpy-V
where the interstitial concentration is in wt.-% and 7 is the transition curves for three steels. Three sets of curves
average transverse dimension of the laths in metres. refer to packet sizes respectively given in Table II.
Finally, analysis of the inclusional state showed that the
inclusions were mainly chromium oxide (very many of small
size) in the two low-S steels, but, as expected, a prevalence of the calculation developed by Ewing :50

manganese sulphide in the two steels with higher S content
was observed. An estimate of the total ,volume fractions of
inclusions, based on chemical analysis and on the nature of
the inclusions observed, gave two distinct levels, Le. 0·06%
for low-S steels and O· 1% for steels with higher S content. A
check of these values by means of the quantitative optical
microscope was not altogether satisfactory, since the values
thus obtained were systematically lower than quoted above,
possibly owing to the presence of numerous inclusions
< 2 11m in size which were beyond the resolving power of
the instrument.
Fracture tests
It is known that when fracture coincides with the general
yield (say, at nil ductility temperature (NDT», the critical
cleavage stress can be calculated from the uniaxial yield
stress a y by means of the following simplified formula of
Green and Hundy,49 valid in pure bending (four-point
bending):
O~yax /(2 7s) ~ 1'95
The critical cleavage stresses in this work have been
calculated using the above ratio and applying the Von Mises
yield criterion, so that we have:
ocr = O~yax = 2 7s x 1'95 = 2'24 oy(NDT)
The aCf values obtained for the steels tested are given in
Table II, together with the corresponding values of the NDT,
the yield stress at room temperature aO•2' and the parameter
dB1/2. Analysis of the tabulated data leads to two
conclusions: (1) increasing up to two times the volume
fraction of sulphides has no effect whatsoever on cleavage
resistance; (2) the critical cleavage stresses measured are
considerably higher than those usually reported for mild
steels or iron alloys.37
. Charpy test results for the MC25, MCS25, and MC50
materials are given in Fig. 5 and Table II. The graph Fig. 5
gives the energy absorbed at the various test temperatures,

(7T1+2"-2" e)
while Table II gives the 50% FA TT values recorded and the
,Ocf=Oyy
max
=27s values of the dBI/2 parameter.
Fig, 6 shows the influence of dp,1/2 on FATT, which
where rs is the yield stress in pure shear and () is the notch- increases with strength level and inclusion content. The
aperture angle. For a 45 ° V-notch, and adopting either the average value measured for the ~FA TT / ~(dBI/2) coefficient
Von Mises (rs=oy/y3) or the Tresca (rs=oy/2). yield is ",,-14 degC mml/2, intermediate between the -11 degC
criterion, we have mml/2 reported by Gladman, Dulieu, and McIvor for
polygonal ferrite and bainite in general,17 and the -25 degC
0cf= 2·51 Oy (Von Mises) mm 1/2calculated using the data of Ohtani et al.18, 23
0cf= 2·18 0y (Tresca)
For a Charpy- V specimen and three-point bending, the Unit crack path measurements
stress-intensification factor drops to 1· 95, as is also shown in Following Matsuda et al.,20, 21the unit crack path (UCP) has

Table II Results of variable-temperature mechanical tests

Material Slow bending tests Charpy tests Uniaxial tensile test

(room temperature)
d;:,1/2
B , ocr' NDT, d;:,l/2
B , FATT, °0.2'
mm-I/2 MPa °C mm-If2 °C MPa

MC25 9,2 1850 -166 8·9 15 550

11·2 2370 -198 12·4 -50 620

MCS25 9·1 1690 -140 9·1 40 550

12·9 2410 -205 12·1 - 5 580

MC50 10 1850 ;
- 60 9·1 75 '700
14·4 2680 -196 16·9 0 750

MCS50 11·2 2240 -170

14·4 2670 -190

Metal Science April 1977

 Brozzo et al. Microstructure and cleavage resistance 121

100

"
50

u
0

~
I-
<{
u..
0
\~
-50 MC25
\
0
0 MCS25
/:::,. MC50

-100
o 5 10 15
-1/2 -1/2
dB ,mm
8 Secondary microcracks below principal fracture surface
6 Effect of bainite packet size on FAIT (from Fig. 5 and of Steel MC25.
Table II).

FA TT parameter, acf is virtually independent of C content

been defined as the region within which the crack propagates
in a nearly straight fashion. UCP's were first identified by
examining a total length of a few millimetres along a crack
visible in the micrograph, and were subsequently grouped
into dimensional classes in order to work out their average
value. The average UCP values were then plotted against the
size of the bainitic packets, obtaining the linear relation:
<UCP) = 1·3 dB (see Fig. 7). This result is consistent both
with the ratio <UCP)/dp~ 1·4 reported by Tanaka et al.22
for a polygonal ferrite structure and with the value '" 1· 5
found by Ohmori et al.21 for lower bainite. As mentioned
above, the result presumably depends on the fact that
packets generated by different variants of the y/ a orientation
relationship may have cleavage planes within 100 mis-
orientation.
Moreover, metallographic analysis of the crack profile has
shown that secondary microcracks exist below the fracture
surface with lengths approximately twice the diameter of the
bainitic packets (Fig ..8).
Thus the experimental results on crack paths, taken
collectively, lead to the positive identification of the bainitic
covariant packet as the microstructural unit in the cleavage-
fracture mechanism of low-C bainites.
and of the inclusional level and depends only upon the
average size of the bainitic packet. This fact might explain
the differences in FA TT recorded when C content increases
from ",0·025 to ",0·050%. In fact, since acr will not vary if
dB is kept constant, any increase in yield stress due to an
increase of C percentage will automatically cause an
increase of the NDT and presumably the FA TT as well.
There appears to be no valid explanation instead for the
differences in FA TT recorded for Steels MC25 and MCS25
with approximately the same packet size and C content.
Fig. 9 also shows that the test values are all clustered
around a straight line passing through the origin of the axes
and having a slope equal to ~acrl~dBl/2~ 190 MPa mm1/2.
Since the steels tested for this study can be considered
virtually as single-phase materials, unlike the greater part of
those studied to date by other authors,29-39 there are two
existing theories which may account for this behaviour of acf"
The first of these two theories, developed by Cottrell, 53
assumes that the critical stage of the fracture process is the
growth of a microcrack within each microstructural unit of
size d. This assumption leads to the following relation

Fracture mechanisms I
2500 • MC25
The diagram Fig. 9 shows the critical cleavage stress • MCS25
obtained for the steels investigated as a function of dB1/2, A MC50
together with published data referring to polygonal ferrite 2000 • MCS50
microstructures.31, 33-37,39 The figure shows that, unlike the
d
Q..
~ 1500
b'U <> De Kazinczy
o MC25 1000 f::,. Oates
20 0 MCS25 <J Wilshaw et al.
MC50 a Knott et at.

/6
/:::,.
fV 500 D Almond et al.
~ 15 + Green and Knot t
a:
u 0
:::> 10 0 5 10 15
d-1/2, mm -1/2
5 ,/

For equation and parameters generating curve drawn through literature data,
0 see text
0 5 10 15 9 Critical fracture stress as function of size of
dB,l-lm microstructural units separated by high-angle boundaries
in steels with different microstructure: polygonal ferrite +
7 Relationship between unit crack path and bainite packet carbides (literature data) and low-C bainitic steels
size. Points. and T from Ref. 51 refer respectively to (present work). A point by Green and Knott52 for a bainitic
2%Mn-3%Cr steels with 0·045% and 0·03%C. A533B steel is also included.

Metal Science April 1977

128 Brozzo et al. Microstructure and cleavage resistance
between (J cf and d-I12:
. 4G
0cf = K;- d "1'1 -1/2
where G is the shear modulus, Y'I is the effective surface
energy, and Ky is Petch's constant.
The second theory, developed by Griffith-Orowan,54
assumes instead that the critical stage consists in the
propagation of a microcrack from one microstructural unit
to another, leading to the relation
4£ ,y'll -1/2
°cf = (1- v2) . d
where E is Young's modulus, v is Poisson's ratio, and Y'll is
the effective surface energy, necessarily different from Y'I in
formula (1).
Since Cottrell's theory relies' on Petch's model for
interpreting the yield mechanism, it would not appear to be
applicable to the present case, where the physical entity
controlling the yield stress is not the microstructural unit
delimited by high-angle boundaries. Furthermore,
assuming for Ky the value obtainable from the diagram in
even
Fig. 4 (Le. "",0·13 MN m-3/2), Cottrell's formula gives a value
for Y'I which is far too low (: j" m-2, Le. barely equal to the
true surface energy of ferrite). It can therefore be concluded
that Cottrell's theory is not applicable to the results of this
in vestigation.
The Griffith-Orowan theory seems to be the only one
capable of inte,rpreting the experimental results and in fact
the value of 1201 m-2 worked out for Y'll using formula (2) is
in fair agreement with the values 90-1901 m-2 found by
Hahn et al.29 for low-Mn steels with polygonal ferrite
microstructure, where the critical fracture stage was
definitely the crack propagation. It may be noted also that
the point by Green and Knott52 included in Fig. 9 for the
A533B bainitic steel seems to be in line with the present
results, obtained in similar microstructures.·
The observed presence of stationary microcracks beneath
the fracture surface would be yet another proof supporting
the hypothesis of a mechanism acting according to Griffith's
theory. Nevertheless, it still has to be explained why, in the
steels investigated, with bainitic structure and practically free
of carbides, the formation of a cleavage nucleus and its
growth into a microcrack equal in length to the size of the
covariant packet is more likely to occur than the
propagation of a micro crack from one packet to another.
A dislocation mechanism consistent with the above
situation has not yet been identified, but it is not in-
conceivable that, in a substructure with a high dislocation
density and characterized by high values of (JO.2' the stress
ma~ locally reach values which are sufficiently high to open
a mlCrocrack.
Fig. 9 shows also a series of (JO.2 values (open symbols),
taken from a recent research study by Ritchie and Knott,37
as well as from other studies by Wilshaw et al.,33, 34Oates 35
Almond et al.,36 Groom and Knott,39 and De Kazinczy a~d
Backofen.31 They refer exclusively to static tests carried out
on notched specimens of steels with ferrite + carbide
microstructures, and ar'e calculated by means of the Von
Mises yield criterion. The two points by Wilshaw et al.,33, 34
* Sin~e the submission of this paper, further evidence has been
prOVIded by Naylor and Blondeau55 for the concept of packet-size-
c,ontrolled .cl~avage fracture in acicular structures. In. a study of
FATT vanatIOns of a low-C Mn-Ni~Cr-Mo steel these authors
propose a subsidiary dependence of the fracture str~ss also on lath
width, which in the present work is constant.
Metal Science April 1977
originally calculated by the Tresca criterion, have been
multiplied by the factor 2/Y3. Data relating to the De
(1)
'Kazinczy-B-ackofen results were, - however, obtained by
multiplying the yield stress values reported by these authors
by the factor 2·24, since more accurate information was not
available on the notch' geometry adopted for their tests;
nevertheless, it is believed that eventual corrections to be
made do not materially affect the following argument.
For these materials (where the rupture of carbides is the
key event which determines fracture and where the Griffith
mechanism would not appear to be active) Knott38 has
suggested that: (i) the fracture mechanism should be that
(2) proposed by Smith;56 and (ii) the observed dependence of (J f
on d~1/2 is ~ctually a con~equence o~the dependence of (JCfO~
carbIde thIckness (a thIckness whIch is expected to vary
according to the treatment adopted for grain refinement),
even though the original Smith's formula does take into
account both carbide thickness and grain size. The same
trend of (Jcf vs. cr
I/2 could also be interpreted by the
following relationship worked out by Almond et al.:36
a = (K~ d + 4G)" 1)1/2 _ K yd 12
/
cf 4 t2 t 2t ,
where Ky is Petch's constant, t is the carbide thickness and
, .
Y 1represents the effective surface energy for the growth of a
'
microcrack opened up by the fracture of an intergranular
carbide. For example, the curve in Fig. 9, similar to that
according to which Ritchie and Knott have interpolated the
experimental points in Ref. 37, can be worked out by taking
y' 1=201 m-2 and t= 1·5Ilm, withKy=0·47 MN m-3/2•
If therefore the available data relative to polygonal ferrite
steels are extrapolated to high values of d-I12 (> 9 mm-1/2) by
the Almond et al. or the Smith relationship, the critical
cleavage stress values obtained will be far lower than those
measured in this study for low-C bainitic steels, even taking
into account the considerable scatter of those data. The low
rate of increase of (Jcf with cr I/2 would show up even more
clearly if the groups of results obtained by each author were
~xamined separately, since this would eliminate the possible
mfluence of their experimental conditions not being entirely
identical.
Finally, Fig. 9 as a whole indicates the presence of two
fracture mechanisms (carbide-controlled, grain-size-
controlled) operating respectively in polygonal and acicular
structures. Nevertheless the results of Hahn et al.29 clearly
point out that the Griffith mechanism can also operate in
carbid~~containing polygonal ferrite, and McMahon and
Cohen32 have demonstrated that microcracks initiated -at
intergranular carbides can be arrested at high-angle
boundaries. This might suggest that the two mechanisms are
competitive over the entire range of crI/2 values in ferritic
steels. At relatively low values of cr 1/2 the literature seems to
establish that the first mechanism (Smith-Almond et al.)
predominates, but for higher values of d-I/2 the Griffith
mechanism could be operating in the majority of cases. A
recent discussion by Knott38 points out that for some of the
finer grain sizes at least, control of cleavage fracture is
exerted exclusively by high-angle boundaries. The determina-
tion of critical cleavage stresses in polygonal ferrite steels
with d-I/2> 8 mm-I/2 should, therefore, prove an interesting
subject for future research.
CONCLUSIONS
T~e covariant (bainitic) packet appears to be the
mIcrostructural unit controlling the Cleavage r,esistance of
2Mn-3%Cr low-carbon bainitic steels; on the other hand,
the mechanical strength is controlled by the substructure
present inside the covariant packet.
 Brozzo et al. Microstructure and cleavage resistance 129

When the bainitic packet is refined by reducing the size of
the prior austenite grain, the FA TT value for the Charpy test
decreases at the rate of ~F ATT/~dB ~ -14 degC mm 1/2 (in
agreement with the values reported by other authors for
steels with polygonal, or acicular microstructures) and the
critical cleavage stress aCf increases at the rate -of
~a crl ~dB -1/2 ~ 190 MPa mm 1/2• However, while the FA TT
values seem to depend on carbon content and inclusional
state, the critical cleavage stress acf is controlled solely by the 21. S.MATSUDA eta[: Trans. iron Steel Inst. Japan, 1972: 12,325.
fineness of the bainitic structure. Its values (1700 < acf
(MPa) < 2700) are systematically greater than those which
can be obtained (for corresponding values of d-1/2) by ex- "
trapolating the data available in the literature for mild steels,
which refer to carbide-controlled cleavage fracture.
Linear regression of aCf on the parameter d;1/2 through the
origin, and the observed presence of stationary microcracks
beneath the fracture surfaces, both point to the fact that, for
the steels investigated, the critical stage of the cleavage-
fracture process consists in the propagation of a Griffith
crack from one bainitic packet to an adjacent one. The
effective surface energy which can be calculated
(",,120 J m-2) seems also to be consistent with the values
reported by other authors for this type of fracture
mechanism.
ACKNOWLEDGEMENTS
The authors wish to thank Dr M. Castagna for his
contribution in the early part of this research programme
and Dr G. Baldi for fruitful discussions. The work was partly
supported by a financial contribution from the EEC.
REFERENCES
1. A. J. McEVILY et al.: 'Transformation and harden ability in steel',
179. 1967: Ann Arbor (Climax Molybdenum).
2. H. MARTENSSON: Scand. J. Metallurgy, 1972, 1,319.
3. T. TERAZAWA et al.: 'Toward improved ductility and
toughness', 101. 1971: Kyoto (Japan Inst. Metals and
Climax Molybdenum).
4. J. M. GRAY: 14th Mech. Work. Steel Process. Conf., Chicago,
1972, p. 174.
5. J. B. BALLANCE et al.: Int. Symp. Proc. Prop. Low-C Steels (ed.
J. M. Gray), AIME, 1972, p. 251.
6. B. L. BRAMFITT and A. R. MARDER: ibid., p. 191.
7. A. P. COLDREN et al.: ibid., p. 163.
8. J. H. WOODHEAD and J. A. WHITEMAN: ibid., p. f45.
9. R. R. PRESTON and c. R. MACKENZIE: paper presented at
Noranda Conf., Copenhagen, April 1973.
10. W. E. LAUPRECHT, H. IMGRUND, and- A. P. COLDREN: Stahl u.
Eisen, 1973, 93~1041.
11. D. B. McCUTCHEON, T.W. TRUMPER, and J. D. EMBURY: paper
presented at 'Journee internationale de Siderurgie', Paris, 4
October 1974; Rev. Met., 1976, 73, 143.
12. G. BUZZICHELLI and A. MASCANZONI: paper presented at
'Journees d'Automne' of the SFM, Paris, 30 September-3
October 1974; Mem. sci. Rev. Met., 1976, 73, 335.
13. J. E. ROBERTS: paper presented at Rosenhain Centenary Conf.,
Teddington, London, 22-24 September 1975.
14. K. J. IRVINE and F. B. PICKERING: J. Iron Steel Inst., 1963,201,
518.
15. M.J. ROBERTS: Metall. Trans., 1970, 1,3287.
16. R. E. DOLBY and J. F. KNOTT: J. Iron Steel Inst., 1972,210,857.
17. T. GLADMAN, D. DULIEU, and I. D. MciVOR: paper presented at
'Microalloying 1975' Conf., Washington, 1-3 October
1975.
18. H. OHTANI, F. TERASAKI, and T. KUNITAKE: Trans. Iron Steel
Inst. Japan, 1972, 12, 118.
19. T.INouEetal.: Trans. Japan Inst. Metals, 1970,11,36.
20. s. MATSUDA, T. INOUE, and M. OGASA WARA: ibid., 1968,9,343.
2~. S. TANAKA,S. TANI, and C.OUCHI: ibid., 1975, 15, 19.
23. Y. OHMORI, H. OHTANI, and T. KUNITAKE: Metal Sci., 1974, 8,
357.
24. J. P. NAYLOR and P. R. KRAHE: Metall. Trans., 1974,5, 1699.
25. R. A. GRANGE: Trans. ASM, 1966,59,26.
26. L.-A. NORSTROM: Scand.J. Metallurgy, 1976,5, 165.
27. A. R. MARDER and C. KRAUSS: Proc. 2nd Int. Conf. Strength of
Metals and Alloys, Asilomar, ASH 1970, vol. III, p. 822.
28. A. BROWNRIGG: Scr. Metall. 1973, 7, 1039.
29. G. T. HAHN et al.: 'Fracture' (ed. B. L. Averbach et al.), 91.
1959: New York (Wiley).
30. F. De KAZINCZY and w. A. BACKOFEN: ibid., p. 65.
31. F. De KAZINCZY and W.A. BACKOFEN: Trans. AIME, 1961,221,
640.
32. C.J.McMAHON and M. COHEN: Acta Met., 1965,13,591.
33. T. R. WILSHAW: J. Iron Steel Inst., 1966,204,936.
34. T. R. WILSHAW, C. A. RAU, and A. S. TETELMAN: Eng. Fracture
Mech., 1968, 1, 191.
35. G. OATES: J. Iron Steel Inst., 1969,207,353.
36. E. A. ALMOND, D. H. TIMBRES, and J. D. EMBURY: 'Fracture 1969'
(ed. P. L. Pratt), 253. 1969: London (Chapman and Hall).
37. R. O. RITCHIE and J. F. KNOTT: Mater. Sci. Eng., 1974, 14, 7.
38. J. F. KNOTT: paper presented at 'Mechanics and physics of
fracture' Conf., Cambridge, 6-8 January 1975.
39. J. D. D. GROOM and J. F. KNOTT: Metal Sci., 1975,9,390.
40. G. KRAUSS and A. R. MARDER: Metall. Trans., 1971,2,2343.
41. G. R. SPEICH and P. R. SWANN: J. Iron Steel Inst., 1965, 203,
480.
42. G. R. SPEICH and H. WARLIMONT: ibid., 1968, 206, 385.
43. J. M. CHILTON, G. J. BARTON, and G. R. SPEICH: ibid., 1970,208,
184.
44. R. F. HEHEMANN, K. R. KINSMAN, and H. I. AARONSON: Metall.
Trans., 1972,3, 1077.
45. G. BALDI, P. BROZZO, G. BUZZICHELLI, and A. MASCANZONI:
paper presented at 4th Int. Conf. Strength of Metals and
Alloys, Nancy, 30 Aug.-3 Sept. 1976.
46. J. W. CHRISTIAN: 'Strengthening methods in crystals' (ed. A.
Kelly and R. B. Nicholson), 261. 1971: Amsterdam
(Elsevier).
47. G. LANGFORD and M. COHEN: Trans. ASM, 1969,62,623.
48. c. M. YOUNG and o. D. SHERBY: J. Iron Steel Inst., 1973, 211,
640.
49. A. P. GREEN and B. B. HUNDY: J. Mechanics and Physics Solids,
1956,4, 128.
50. D. J. F. EWING: ibid., 1968, 16, 265.
51. G. BUZZICHELLI et al.: EEC Convention No. 6210/48/401, 1st
Report CSM, Rome, September 1974.
52. A. P. GREEN and J. F. KNOTT, quoted by J. F. Knott, 1976, private
communication.
53. A. H. COTTRELL: Trans. AIME, 1958,212, 192.
54. E. OROWAN: Rep. Progress Physics, 1948, 12, 185.
55. J. P. NAYLOR and R. BLONDEAU: Metall Trans., 1976, 7A, 891.
56. E. SMITH: Proc. Conf. 'Physical Basis of Yield and Fracture',
1966, p. 36 (LO.P. and Phys. Soc.).

© THE METALS SOCIETY 1977

Metal Science April 1977

 Polish Academy of Sciences

ELEVENTH GENERAL ASSEMBLY AND

INTERNATIONAL CONGR'ESS OF
CRYSTALLOGRAPHY

Warsaw, Poland 3-12August 1978

The arrangement of the scientific programme will in general be similar to that of the Tenth Congress held in
Amsterdam in 1975. There will be general lectures, scientific sessions on topics of interest for today's
crystallography, chemistry and solid state physics, poster sessions, open sessions of Commissions of the
Union, and ad hoc meetings. Participants will be invited to submit abstracts of recent work on
crystallographic subjects. Accepted contributions will be printed in the book of abstracts of the Congress.
For oral presentation in the formal scientific sessions or for the poster sessions, a selection will be made from
the papers lying within the range of the Congress topics. The ad hoc meetings are intended mainly to
encourage free discussion.

An exhibition of non-commercial equipment and of photographs and drawings of crystallographic interest

will be held during the Congress. A commercial exhibition will be organized in which manufacturers and
distributors 'of equipment related to crystallographic research may display their products. Further
information about both exhibitions may be obtained from the Organizing Committee. For the accompanying
members, a choice of tours will be offered. The details of this programme and the social programme will be
published in the First Circular. Travel information can be obtained from the offices of Polish Airlines LOT
and Polish Travel Office ORBIS.

Professor J. Auleytner is Chairman of the Organizing Committee, other members of which are Dr J.
Leciejewicz (Vice-Chairman), Dr Z. Galdecki, and Dr T. Warminski (Secretary). Correspondence should be
addressed to:

The Secretariat,
Eleventh International Congress of Crystallography,
Polish Academy of Sciences,
Institute of Physics,
IF PAN SL-1,
AI. Lotnikow 32/46, Pawilon 9,
02-668 WARSZA WA, Poland.

Telephone: Warszawa 43 6034

Telex: 81 2464

Further details of the Congress, including arrangements for registration, accommodation, scientific visits,
etc., will be published in the First Circular, which will be distributed in the second half of 1977. Persons
interested in receiving the First Circular are requested to complete an application card and return it to the
Organizing Committee before 1 June 1977. Application cards may be obtained from the Secretaries of the
National Committees for Crystallography or from the Organizing Committee.

Meta1 -Science April 1977