You are on page 1of 12

Current Opinion in Solid State and Materials Science 8 (2004) 267278

The mechanism of acicular ferrite in weld deposits


Sudarsanam Suresh Babu

Metals and Ceramics Division, Bldg 4508, Mail Stop 6096, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States

Abstract
Research has shown that the acicular ferrite microstructure in steel weld metal, which provides an optimum combination of
strength and toughness, is indeed intragranularly nucleated bainite. It is possible to maximize the content of acicular ferrite by
increasing the intragranular nucleation sites while maintaining a critical weld metal cooling rate and the steel hardenability. This
paper highlights recent research related to nucleation and growth of acicular ferrite during decomposition of austenite.
Published by Elsevier Ltd.
Keywords: Steels; Weld microstructure; Acicular ferrite; Bainite and phase transformations

1. Introduction
In the late 1970s, the importance of an acicular ferrite
microstructure towards optimizing strength and toughness was identied [1]. Since then, extensive research
has been done on acicular ferrite formation and its relation to oxide inclusions, weld metal hardenability and
cooling conditions [**2,**3]. An overview of the hightemperature phase changes that occur in steel weld metal
is presented (see Fig. 1) to describe acicular ferrite formation in the context of the overall microstructure evolution. The weld metal microstructure is aected by
melting, gas dissolution, solidication and solid-state
transformations. Since the weld pool region is heated
to temperatures as high as 2500 K, the liquid steel dissolves oxygen. The extent of oxygen dissolution depends
upon the thermodynamic properties of liquid metal, gas,
and slag phases [4]. As the liquid weld metal cools from
this temperature (Fig. 1a), in the temperature range
20001700 C, the dissolved oxygen and deoxidizing
elements in liquid steel react to form complex oxide
inclusions in the range of 0.11 lm size range. In the

Tel.: +1 865 574 4806; fax: +1 865 574 4928.


E-mail address: babuss@ornl.gov

1359-0286/$ - see front matter Published by Elsevier Ltd.


doi:10.1016/j.cossms.2004.10.001

temperature range 17001600 C (Fig. 1b), solidication


to d-ferrite (body centered cubic phase) starts and envelops these oxide inclusions; and this d-ferrite transforms
to austenite (c-face centered cubic phase). In the temperature range from 1600 to 800 C (Fig. 1c), austenite
grain growth may occur. In the temperature range
800300 C (Fig. 1dg), the austenite decomposes to different ferrite (body centered cubic phase) morphologies.
The austenite to ferrite decomposition (Fig. 1d and e)
starts with the formation of allotriomorphic ferrite (a)
at prior cc boundaries and eventual coverage of these
boundaries. With continued cooling (Fig. 1f), the Widmanstatten (aW) ferrite nucleates at the a/c boundaries
and extends into the untransformed austenitegrain
interiors. Further cooling to low temperatures, (Fig.
1g) the acicular ferrite (aa) would nucleate on the inclusion. If there are no potent inclusions, bainitic (ab) ferrite might form instead of acicular ferrite, from the
remaining austenite. On further cooling to temperatures
close to room temperature, any remaining austenite may
completely or partially transform to martensite. This
mixture of martensiteaustenite phases is referred to as
MA constituent. These phase transformation sequences are important while discussing the mechanism
of acicular ferrite formation because each of the above
reactions modies the nucleation and growth kinetics

268

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

Fig. 1. Schematic illustration of a weld metal cooling curve and a


hypothetical continuous cooling transformation diagram shows dierent phase transformations that may occur as the weld metal cools to
room temperature.

of acicular ferrite. The importance of considering the


overall microstructure evolution was recently stressed
even for identication of microstructural constituents
[*5].
A typical weld microstructure containing acicular ferrite (aa) is shown in Fig. 2 [6]. This morphology is usually referred as a basket-weave structure [7]. The
optical microstructure (see Fig. 2a) shows the interlocking nature of acicular ferrite (aa) co-existing with other
ferrite morphologies such as Widmanstatten ferrite
(aW) and allotriomorphic ferrite (a). The acicular ferrite
plates have an aspect ratio of 0.10.2 in any random
cross sectional area. High-magnication analyses of
the same microstructure in a transmission electron
microscope show primary plates (marked as P) that
nucleate on inclusions and the secondary ferrite pates
(marked as S) that nucleate on preexisting ferrite plates
(see Fig. 2b). This chaotic arrangement of lenticular ferrite plates within an austenite grain is considered to be
one of the mechanisms by which the toughness is improved in the welds [**8]. This arrangement is promoted
by the nucleation of acicular ferrite on oxide inclusions.
The concept of intragranular nucleation on inclusions
has also been extended to structural steels also by inten-

Fig. 2. Typical acicular ferrite (aa) microstructure in a low alloy steel


(Fe0.06C0.51Si1.11Mn0.48Crwt.%) weld metal with small additions of aluminum, titanium and oxygen: (a) optical micrograph shows
the presence of grain boundary (a) and Widmanstatten ferrite (aW)
coexisting with acicular ferrite, (b) transmission electron micrograph
shows the apparent nucleation of primary acicular ferrite plate
(marked as P) on an oxide inclusion and a secondary acicular ferrite
plate (marked as S) on prexisting ferrite plate.

tionally adding titanium oxide during steelmaking before any other thermomechanical processing [*9,*10].
In this review, recent progress in understanding the
details of nucleation and growth of acicular ferrite, the
parameters that aect the formation of acicular ferrite
during weld cooling, overall transformation kinetics of
acicular ferrite formation and the transitions from acicular ferrite to bainite microstructure are reviewed. The
concepts presented are also applicable to acicular ferrite
that forms in oxide-inoculated structural steels.

2. Mechanisms of austenite to acicular ferrite


transformation
Past research has revealed that acicular ferrite plates
are crystallographically related to the parent austenite
grain approximately either by the NishiyamaWasserman (NW) orientation relationship or by Kurdjumov
Sachs (KS) orientation relationship [**8,**11,12]. In
addition, the acicular ferrite microstructure exhibits surface relief phenomena [13] indicating the formation of
these ferrite plates from austenite involves an invariant

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

plane strain with a large shear component. Elemental


analyses across acicular ferrite and austenite phases
have also revealed no substitutional alloying element
(i.e., Si, Mn, Cr, etc.) partitioning [14,6]. In addition,
atom probe eld ion microscopy indicated that interstitial alloying elements (i.e., carbon) could partition
between austenite and ferrite [15]. The austenite to acicular ferrite transformation kinetic data exhibit incomplete reaction phenomena, i.e., the transformation
stops when the carbon concentration in austenite
reaches a critical value (T0 condition) above which the
diusionless transformation is thermodynamically
impossible [**8,1214]. Furthermore, the basketweave morphology of acicular ferrite plates can be
changed to a non-random orientation by applying
elastic stresses during transformation [16]. During
transformation under stress, evidence for transformation-induced plasticity has also been observed. Based
on the above-mentioned research, one can conclude that
the transformation mechanisms of acicular ferrite and
bainite are similar, except for the requirement of potent
inclusions within the austenite grain for acicular ferrite
formation. Although, the bainite and acicular ferrite formation mechanisms are similar, due to the presence of
many crystallographic variants of plates in acicular ferrite and resulting chaotic microstructure, the toughness
of the acicular ferrite microstructure in welds is better
than the toughness of a bainitic microstructure [17,*10].

269

2.1. Nucleation of acicular ferrite


To understand the mechanism of acicular ferrite formation, one needs to understand why it nucleates on
inclusions? Based on the published literature, four major
mechanisms can be put forward to explain the
nucleation of acicular ferrite [*18,1922,*23,*24,25
27,*28,29,*30,**31,*32,33,34,*35,36]. These mechanisms are illustrated schematically in Fig. 3. (1) In the
rst mechanism (see Fig. 3a), the inclusions act as inert
surfaces leading to a reduction of activation energy and
ferrite nucleation is promoted. The calculations show
that the ratio of the activation energy for nucleation
on inclusions to that for homogeneous nucleation will
decrease with an increase in inclusion diameter. (2)
Good lattice matching between inclusions and ferrite
can also reduce the activation energy for nucleation.
Due to the constraint of a reproducible orientation relationship between austenite and ferrite, the probability of
achieving suitable orientation relationships between
inclusion and ferrite and also between ferrite and austenite may be dicult. (3) Inclusions may deplete the elements such as carbon, manganese and silicon from the
austenite. This depletion may lead to a local increase
in the driving force for the nucleation of ferrite from
austenite at the inclusion surface [*28]. (4) Due to a difference in thermal expansion coecients (De) of austenite and inclusions, thermal strains may develop near the

Fig. 3. Schematic illustrations of dierent mechanisms for nucleation of acicular ferrite on inclusions: (a) Inclusion surface acts as an inert surface for
nucleation and therefore large inclusions are potent sites. (b) Matching between ferrite and inclusion lattice will decrease the interfacial energy
between inclusion and ferrite in comparison to inclusionaustenite interface. (c) Local depletion of hardening elements such as carbon and/or
manganese which might lead to an increase in driving force for nucleation of ferrite from austenite. (d) Strain energy near the inclusion may increase
due to dierences in thermal expansion between austenite and inclusion, which might reduce the activation energy for nucleation.

270

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

inclusion austenite interface, which may reduce the activation energy for the formation of a ferrite nucleus [36].
In this section, the above mechanisms are evaluated
based on two methodologies. In the rst methodology,
the formation of acicular ferrite involves thermal activation. Based on this assumption, one can dene the nucleation rate per unit volume per second (IV) by the
following equation [**8]:


DG
I V m exp 
;
1
kT
where m is the attempt frequency, k is the Boltzmann
constant, T is the temperature in Kelvin and DG* is
the activation energy for nucleation per unit volume of
the matrix. Classical textbook analyses have shown that
DG* is inversely proportional to the square of the chemical driving force. Recent research has focused on relating the activation energy for nucleation on inclusion to
dierent metallurgical variables. In a classical work on
acicular ferrite nucleation, Ricks et al. [*18] used classical nucleation theory to describe the nucleation potency
of inclusion (see Fig. 3a). By assuming a spherical geometry, the activation energy for nucleation on an inclusion
was related to the driving force for the austenite to ferrite transformation (DGc!a
max jvol ), interfacial energies of
austenite-inclusion (rc/inc), ferriteinclusion (ra/inc) and
austeniteferrite (rc/a) [see Fig. 3a]. As per these calculations, the nucleation potency of inclusions will increase
with an increase in their radius, i.e. activation energy decreases as inclusion diameter increases. Furthermore,
the calculations also showed that the activation energy
nucleation of ferrite at the grain boundary is always
lower than at inclusions. It is also known that not all
inclusions provide preferential nucleation sites. This
may be related to possible changes in the interfacial ferriteinclusion (ra/inc) with dierent inclusion types. Recent publications have used the model by Ricks et al.
has been used to explain acicular ferrite transformation
both in welds and structural steels by calculating a
reduction in activation energy for nucleation due to
one or more phenomena listed: (i) increase in the inclusion radius [21]; (ii) reduction of interfacial energy by
lattice matching (see Fig. 3b) [*23,*24], (iii) increase in
the DGc!a
max jvol by local depletion of substitutional alloying elements including manganese [37] or (iv) local
depletion of carbon [27,*28] near the inclusion.
In the alternative methodology, the nucleation of
acicular ferrite can be understood by a theory based
on bainite nucleation [**8,*38]. In this theoretical treatment, the activation energy is assumed to be proportional to the driving force for transformation, as given
by the following equation:
DG bT ;

where b is assumed to be an constant. This theory is


based on martensite nucleation theory and assumes that

the ferrite nuclei are forming by displacive mechanisms


with the partitioning of carbon to the austenite. As per
this theory, the austenite already contains preexisting
ferrite embryos and the nucleation of ferrite occurs as
soon as the chemical driving force increases above the
barrier energy for interface motion. In addition, the theory also requires the presence of a glissile interface between the ferrite and austenite. Rees and Bhadeshia
[*38] found experimental evidence for Eq. (2) by analyzing the acicular ferrite start temperature as a function of
weld metal composition. With this concept, one can derive an expression for nucleation rate of acicular ferrite
[**8,39,40],

c!a 
C 4 C 4 DGmax

I V C 3 exp 
;
3
RT
C 2 RT
where C3 is related to the number of nucleation sites and
C2 and C4 are constants based on experimental measurements. For acicular ferrite transformation, C3 would be
related to number density of potent inclusions and for
bainite transformation C3 is related to grain boundary
nucleation sites.
In the following paragraphs, possible mechanisms for
the modication of parameters in Eq. (3) by the presence
of inclusions within the austenite will be discussed. The
nucleation rate of acicular ferrite can be associated with
c!a
a change in the magnitude of C2, C3, C4 or DGmax
. Since
C2 is associated with the slope of the universal nucleation function for acicular ferrite or bainite, one can
assume that this does not change with the inclusions
type. The presence of inclusions can increase the C3 by
providing an inert surface [**41] within the austenite
and thereby promoting the nucleation of acicular ferrite.
However, this assumes that the interface structure of
inclusionaustenite is more or less similar to that of an
austeniteaustenite grain boundary. Therefore, it is not
clear how changing the inclusionaustenite crystallographic orientations will lead to a dierent interface dislocation structure, which might inuence the nucleation
of acicular ferrite.
In the original derivation of Eq. (3), the C4 is also related to number of array dislocations in the fault which
acts as embryo for nucleation, shear resistance of the lattice to dislocations, magnitude of Burgers vector, energy barrier between adjacent equilibrium positions of
dislocations and the interfacial energy between austenite
and ferrite. Rees and Bhadeshia [*38] speculated that
embryos might exist at the dislocation debris created
near the inclusions due to thermal strains induced by
dierences in the thermal expansion between inclusions
and austenite. Furthermore, it is suggested that in case
of large inclusions, these stresses cannot be accommodated and may lead to the generation of a dislocation array [**8] by dislocation motion. In this mechanism, it is
important to note that the stress build up is proportional
to both De and the degree of undercooling (DT) below

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

the austenite to ferrite transformation temperature


[33,34,35*,36]. These treatments indicate that a plastic
zone may develop near the inclusion interface. Pan
et al. [36] extended the treatment of Brooksbank and
Andrews [*35] and indicated that the strain energy value
might not be very high in comparison to the chemical
driving force because of the plastic deformation of austenite close to the inclusion. The plastic deformation
may help in nucleating the acicular ferrite by creating
dislocation defect structures that would be suitable for
displacive nucleation of ferrite. At the same time, excessive plastic deformation may not work in favor of acicular ferrite growth, which will be described in the next
section. In real weld cooling conditions, the stresses near
the inclusionaustenite interface can also be relaxed
due to stress relaxation phenomena. This implies that
the acicular ferrite formation at the inclusion will be
favored only under certain thermomechanical cooling
conditions.
The depletion of alloying elements near the inclusion
interface may increase the driving force for transformation (DGc!a
max ), which will lead to an increase in the nucleation rate. It is noteworthy that the concentration of
alloying elements such as manganese [*32,37] may not
be homogenized after repeated thermal cycling between
the austenite and ferrite phase regions during welding
due to the low diusivity in austenite. At the same time,
the manganese-depleted regions are not always observed
in welds [21]. Decarburization of austenite close to the
inclusionaustenite interface may also promote nucleation, as proposed by Gregg and Bhadeshia [27,*28].
However, several questions exist with regard to this
mechanism. How can this carbon depletion be present
during repeated thermal cycling? Will the inclusion potency decrease with repeated thermal cycling between
austenite and ferrite phase eld? Will the inclusion act
as a carbon pump, i.e., deplete the carbon while cooling
and reject it back to the austenite while heating to high
temperature? Further work is needed to elucidate this
mechanism by careful experimentation with repeated
thermal cycling.
Recently, He and Edmonds [42] speculated that the
formation of FeV clusters could act as nucleation sites
for acicular ferrite, in addition to inclusions. In recent
elegant work, Furahara et al. [**31] showed that addition of V and N to CMn steel containing MnS
inclusions led to the formation of intragranular idiomorphic ferrite. Although, this ferrite forms above the bainitic start temperature for this alloy, i.e. at and above
873 K, the discussions in this work are very relevant to
the overall consideration of nucleation potency. Furahara et al. [**31] concluded that the reduced interfacial
energy between the VC and/or V(CN) and ferrite might
contribute to an increase in the nucleation rate of intragranular ferrite. Furahara et al. [**31] also found that
the strains around the inclusion might promote acicular

271

ferrite at 823 K, a lower transformation temperature.


Previous to this work, Guo et al. [*43] showed that the
addition of vanadium could lead to precipitation of
VC on preexisting MnS sulde inclusions, which may
lead to the nucleation of pearlite colonies in a hypereutectoid FeMnC alloy. In addition, they also found
that there is no preferred crystallographic orientation
relationship between MnS and VC, as well as between
VC and austenite. Based on this observation, the
authors concluded that the kinetics of transformation
is accelerated, possibly due to depletion of carbon in
the austenite as a result of precipitation of carbide.
Based on these results, the nucleation of acicular ferrite
on FeV clusters during weld cooling suggested by He
and Edmonds [42] appears to be improbable. The nucleation very well could have been accelerated due to the
precipitation of VC or V(CN). A simple thermodynamic
calculation of their alloy system shows that is possible to
form V(CN) in their steels. However, the evidence for
the precipitation of VC or V(CN) was not present, because the precipitates were very small. While the
above-mentioned models, in many ways, assume simplistic and uniform properties for inclusions, the real
inclusions in the weld are very heterogeneous in composition and often exhibit amorphous phases [*32,44].
Therefore, a critical experiment is needed to add wellcontrolled particles that induce only one of the mechanisms in the austenite grain. In summary, for displacive
acicular ferrite nucleation on inclusions, the possibility
of thermal strains and local depletion of carbon and/
or manganese appears to be the most plausible
mechanisms.
In this section, the nucleation potency of inclusions
for nucleation of a single ferrite plate was discussed.
Experimental observations show that there is no oneto-one correspondence of inclusion with every acicular
ferrite plate because acicular ferrite is also promoted
by autocatalytic nucleation on a pre-existing ferrite plate
(see Fig. 2). Recent work using electron back scattered
diraction methods have related morphology to the
crystallographic misorientation and showed the importance of autocatalytic nucleation of secondary ferrite
plates on pre-existing ferrite plates [*10,**11]. The
importance of autocatalysis will be discussed later in
the context of overall transformation kinetics of acicular
ferrite.
2.2. Growth of acicular ferrite
Acicular ferrite plates grow by repeated nucleation of
ferrite subunits by displacive transformation with no
partitioning of carbon [**8]. Subsequent to the formation of one subunit, the carbon escapes to austenite.
With continued growth of acicular ferrite, the carbon
in the austenite enriches and reaches a critical limit beyond which the displacive formation of acicular ferrite

272

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

becomes impossible. This critical carbon concentration


corresponds to a condition (T 00 condition) where the free
energy of the austenite equals that of ferrite with a
stored energy of 400 J/mole. This cessation of the transformation at this point is referred as incomplete reaction. Experimental evidence for such incomplete
reaction during acicular ferrite transformation in welds
has been published [**8,45]. As mentioned earlier, the
acicular ferrite growth also leads to invariant plane
strain surface relief [**8]. The formation of sheaves similar to bainite are stied due to hard impingement of one
set of plates with another set of plates nucleated from
nearby inclusions. However, if the number density of
potent nucleation sites is less, it may be possible to form
sheaves even within the austenite grain due to autocatalytic nucleation on primary acicular ferrite plates. Similar to bainite, by applying an elastic stress, the selection
of a particular growth variant among the 12 dierent
NW- or 24 dierent KS orientational relationships can
be forced, leading to alignment of acicular ferrite plates
(see Fig. 4). Interestingly, this result was observed in

conjunction with transformation-induced plasticity also


[16]. It is important to note that this eect was observed
with the presence of an elastic stress.
An interesting question arises: what will happen if the
austenite deforms plastically and leads to extensive dislocation network? As per the displacive transformation
theory, the formation of stable dislocation networks
must retard the glissile interface and therefore must slow
down the acicular ferrite transformation kinetics. This
phenomenon is referred to as mechanical stabilization
and it occurs in bainitic steels [**8]. A denitive proof
for this phenomenon during acicular ferrite transformation was obtained by Lee et al. [*46]. In their work, the
austenite grain boundaries were made ineective by decorating them with allotriomorphic ferrite [45,47] thereby
forcing the decomposition of austenite to acicular ferrite. The austenite was then strained to 10%, 25% and
45% before isothermally heat-treating at 390 C and
360 C. The transformation kinetics of austenite to acicular ferrite were measured at both temperatures. The
measurements from 390 C showed the retardation of
acicular ferrite transformation kinetics with an increase
in percentage strain, which is in agreement with the earlier hypothesis. Interestingly, the measurements at
360 C failed to show any dierence between strained
samples. This was attributed to an increase in chemical
driving force for transformation due to the increase in
undercooling. This result shows that the growth of acicular ferrite in welds will be aected by complex thermo
mechanicalchemical conditions that may prevail within
the austenite during weld cooling.
At this juncture, it is noteworthy for certain weld metal compositions under certain cooling conditions it may
be possible to form intragranularly nucleated Widmanstatten ferrite above the Bs temperature [*48]. This
intragranular Widmanstatten ferrite will be coarser than
acicular ferrite that forms below the bainitic start temperature. All the mechanisms we have discussed will be
applicable except for the reduction in undercooling for
this transformation since it occurs at a higher
temperature.
2.3. Transition from bainite to acicular ferrite

Fig. 4. Optical micrographs from FeCrC weld metal samples


subjected to isothermal acicular ferrite transformation with an applied
elastic stress of (a) 12 MPa and (b) 174 MPa during isothermal
transformation of austenite to acicular ferrite [6].

The discussion above shows that the acicular ferrite


and bainite are similar in nature. An obvious question
arises: is it possible to switch the microstructure in the
same steel or weld by modifying the ratio of intragranular to intergranular nucleation sites? Based on the published literature the answer is yes. It is possible to
switch between bainite and acicular ferrite microstructures in three ways, as schematically shown in Fig. 5.
The rst method relies on increasing the austenite
grain size, which in turn reduces the cc grain boundary
area per unit volume. Since bainite nucleation initiates
at the cc grain boundaries, coarse austenite grains re-

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

273

Fig. 5. Schematic illustrations of dierent mechanisms by which bainite transformation can be stied to obtain predominantly acicular ferrite: (a)
Austenite grain size eect: with large grain size, the cc grain boundary area per unit volume decreases for a given inclusion density. (b) Poisoned
austenite grain boundary: by forming a thin layer of allotriomorphic ferrite, the cc grain boundary can be made ineective due to reduced area of
favorably oriented a/c interface, as well as, carbon enrichment near the interface (c) Presence of potent inclusions: for a given austenite grain if we
increase the number of potent inclusions, then acicular ferrite nucleation will be favored [6].

duce the bainite kinetics and favor the intragranular


nucleation sites. This condition will lead to a predominant acicular ferrite [49,**8] microstructure.
The next method relies on the removal of active cc
grain boundary by decorating it with ne allotriomorphic ferrite [45,47]. It is noteworthy that a favorably oriented ac interface may also develop bainite sheaves
[50]; however, it can be stied by the carbon enrichment
near the ac interface. This was proven by reducing the
carbon enrichment prole ahead of the ac interface,
which leads to the emanation of bainitic sheaves from
preexisting the ac interfaces [6].
The third method is related to the presence of potent
inclusions within the austenite grain. If there are no
inclusions, acicular ferrite formation will not occur,
and this was proved by Harrison and Farrar in their elegant experiment [51]. In addition, even if the inclusions
are present, if they are ineective, bainite formation will
be favored once again. This was recently observed in
many cases by varying the titanium addition to the weld
metal [5254]. At very low levels of titanium, the inclusions are predominantly manganese silicates, which are
ineective acicular ferrite nucleants, and the weld metal
microstructure is predominantly bainitic. However, on
addition of small amounts of titanium, the inclusion
characteristics are modied and the microstructure is
predominantly acicular ferrite. This illustrates the
importance of inclusion engineering during the design

of weld metal consumable and inoculated structural


steel development.
Elimination of cc grain boundaries as potential
nucleation sites for bainite by boron segregation is also
been postulated by Peng et al. [55]. However, grain
boundary allotriomorphic ferrite can be seen in their
micrograph and therefore the mechanism for acicular
ferrite enhancement may be similar to the one shown
in Fig. 5. Snieder and Kerr have observed more than
90% acicular ferrite without any grain boundary allotriomorphs by adding titanium and boron [56]. These
researchers argued that the boron segregates to the austenite grain boundary and sties the nucleation of allotriomorphic ferrite at cc grain boundary and the
titanium forms potent inclusions within the austenite
grain to promote acicular ferrite. These researchers did
not report any presence of bainitic sheaves originating
from the cc grain boundary. This suggests that the
boron may be retarding the bainitic nucleation too.
Another interesting suggestion for making the cc
grain boundaries ineective nucleation sites for bainite
has been attributed to possible segregation at cc grain
boundaries [57]. These authors found that, in a Fe-CMn structural steel inoculated with titanium oxide, the
acicular ferrite microstructure can be promoted by
increasing the manganese concentrations from
1.4 wt.% to 2.46 wt.%. At high concentrations of manganese, the authors observed no allotriomorphic ferrite

274

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

and still observed enhanced acicular ferrite formation.


Even with the presence of free cc grain boundaries,
there was no evidence for bainitic sheave formation.
Therefore, these authors, based on the work of Liu
and Zhang [58], suggested that the Mn segregates to
cc grain boundaries and reduces the driving force for
bainite nucleation at the interface. Liu and Zhang [58],
concluded that this segregation is of a nonequilibrium
type and will depend upon the austenitization temperature. Although, such segregation can occur even in alloy
welds, changes in the austenite-to-ferrite driving force
under paraequilibrium conditions needs to be quantied
to evaluate the relative eects of segregation.
Finally, based on the above discussions, one can wonder, if acicular ferrite will form all the time in welds. Past
research and recent work [54,59] shows that, if the weld
cooling rates are too fast, even with the presence of potent inclusions a mixed bainite and acicular ferrite
microstructure will be obtained. For example, two weld
metal samples with dierent titanium contents were
austenitized at 1200 C for 10 min and were cooled at
a rate of 80 c/s. The micrograph in Fig. 6a is of a weld
metal sample with no potent inclusions and therefore
exhibits predominantly bainite and martensite. The
micrograph shown in Fig. 6b is of a weld metal sample
with potent inclusions exhibits grain boundary allotriomorphic ferrite, predominant acicular ferrite microstructure and small fractions of bainitic microstructure.
Detailed microstructural analyses showed that the
allotriomorphic ferrite was not continuous along the
austeniteaustenite grain boundaries due to rapid cooling conditions and as a result, the formation of bainite
sheaves (ab) was been favored in some regions. This
mixed microstructure formation is also related to the
mechanisms presented in Fig. 5.
2.4. Overall transformation kinetics of acicular ferrite
If the acicular ferrite formation is similar to that of
bainite, is it possible to apply already existing bainite kinetic models to acicular ferrite? Such an exercise was
carried out recently by Babu and David [54]. They applied the overall transformation kinetic model developed by Chester and Bhadeshia given by the following
equation [*38,39,40]:


dn
uK 1

1  n1 bhn
dt
h
 

 
K2
DG0m
K 2 DG0m  GN
n ;
1
 exp

rRT
RT
r
4
where K1 is a parameter related to austenite grain
boundary surface area per unit volume, u is the ferrite
subunit volume, n is the extent of transformation, h is
the maximum bainite volume fraction that can form at

Fig. 6. Optical micrographs of weld samples continuously cooled from


1200 C at a rate of 80 K/s. (a) A predominantly bainite and martensite
microstructure was observed due to lack of titanium containing potent
inclusions. (b) Predominantly acicular ferrite microstructure was
obtained due to presence of titanium containing inclusions. However,
a small fraction of bainite was also found due to lack of continuous
allotriomorphic ferrite coverage along the austenite grain boundary
[54].

that temperature, K2 is a constant, R is the gas constant,


T is temperature in Kelvin, DG0m is the maximum driving
force for the nucleation of ferrite, r is another constant
given by 2540 J/mole, GN is the universal nucleation
function given by GN (J/mole) = 3.6375T  r, b is the
autocatalysis factor. The autocatalysis factor is given
by the following equation:
b k1 1  k2x;

where x is the bulk carbon concentration in the units of


mole fraction and k1 and k2 are constants. Using isothermal kinetic data for bainite one can derive the constants
K1, K2, k1 and k2 [*38,39] and the variation of subunit
volume with temperature. The K1 parameter represents
the nucleation rate of bainite at any boundary. If the
austenite grain boundaries are free, K1 for bainite formation will be related to grain boundary area per unit
volume (S cc
V ) [40]. If the austenite grain boundaries
are decorated by the allotriomorphic ferrite, it will be related to the ac interfacial area per unit volume S Vca

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

275

Fig. 7. Comparisons of overall transformation kinetic model (black lines) predictions with the experimentally measured isothermal bainite and
acicular ferrite kinetics in FeCMn steel welds (open circles) with two dierent inclusion characteristics during continuous cooling at 50 K/s. (a)
Data from a sample with no intentional titanium addition shows sluggish transformation kinetics compared to (b) rapid kinetics measured from the
sample with 17 wt. PPM titanium addition. (c) Corresponding optical micrograph from the sample with kinetic data shown in (a) shows
predominantly bainite microstructure. (d) Optical micrograph from the samples with kinetic data shown in (b) shows predominantly acicular ferrite
microstructure [54].

that is capable of nucleating bainite [50]. In the case of


acicular ferrite, the K1 parameter is related to the inclusionaustenite interfacial area (S Vcinc ) that is capable of
nucleating acicular ferrite subunits [**41,*28,*24]. Babu
and David [54] evaluated experimental measurements of
volume fraction as a function of temperature during
continuous cooling conditions (see Fig. 7a and b) for
two weld samples using Eq. (5). In the weld containing
no titanium addition, the austenite transforms predominantly to bainite and in the weld containing titanium
the austenite transforms predominantly to acicular ferrite (see Fig. 7c and d). The tted K1 value for bainitic
data was (1.0 1017) higher than the one (4.4 1016) obtained for the acicular ferrite data. The autocatalytic
factor (k1) for acicular ferrite (34) was found to be higher than that of bainite (4). It is important to note that
these values are achieved by purely tting the equation
to experimentally observed kinetic data and are not derived from fundamental knowledge of nucleation potency. Nevertheless, it is interesting to compare the
tted parameters.
The increase in k1 for acicular ferrite may be possibly
related to the exibility to choose many variants from
the inclusion, which in turn may allow for many more
variants due to autocatalysis eect. Moreover, in the
case of bainitic microstructure, the maximum experimental ferrite fraction achieved at 737 K is 0.81. Below
737 K at 680 K, the transformation rate of austenite is

accelerated further due to formation of martensite. In


contrast, the maximum experimental ferrite fraction
achieved at 737 K is 0.97 for predominant acicular ferrite microstructure. This result shows that the rate of
transformation to acicular ferrite higher than bainitic
transformation. This change may be related to larger
carbon trapping between the acicular ferrite subunits
than that between the bainitic ferrite subunits [39]. This
discussion shows it is possible to describe the acicular
ferrite transformation kinetics by considering changes
in S Vcc (austeniteaustenite grain boundary area per unit
volume), S Vca (austeniteferrite interface boundary area
per unit volume), and S Vcinc (inclusionaustenite interface boundary area per unit volume) on K1 and k1 as a
function of weld metal composition and austenite grain
size. Further work is needed to evaluate these values
with large number of experimental data set on acicular
ferrite transformation.

3. Future work
In the beginning of this review, it was stressed that
acicular ferrite formation depends on various high-temperature phase transformation sequences (see Fig. 1).
Therefore, these reactions must follow a particular
direction to allow for acicular ferrite formation at
low temperature. This points to the obvious need for

276

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

an integrated model that describes inclusion formation,


primary solidication to d-ferrite, d-ferrite to austenite
transformation, austenite grain growth, and austenite
decomposition to dierent ferrite morphologies. There
are recent developments to describe inclusion formation
during solidication [37,60,61*,62,63*,6471]. There exist dendrite-tip growth and diusion-controlled-growth
models that can be used to describe both phase selection
and kinetics of solidication, respectively. These models
have been extended to weld solidication [72,73*,74].
Diusion controlled growth models can be used to describe peritectic solidication [75]. The same models
can be used to describe diusion-controlled growth of
austenite into d-ferrite [74,75]. With further development, these models can be coupled with Monte Carlo
type grain growth models [76] to describe the nal state
of an austenite grain before the onset of ferrite formation. Finally, the allotriomorphic-, Widmanstatten-,
bainitic- and acicular-ferrite kinetic models [**77] must
be linked through a simultaneous transformation kinetic
framework [78,79].

4. Conclusions
High-temperature transformation sequences in welds
that aect acicular ferrite formation were highlighted.
Evidence for the similarity between bainite and acicular
ferrite transformations is overwhelming. Using nucleation on inclusions and autocatalytic nucleation on preexisting ferrite plates, the nucleation of acicular ferrite
can be described. The most probable mechanisms for
nucleation on inclusions are either due to the generation
of thermal strains in the austenite or the formation of
solute-depleted regions. The experimental measurements
of acicular ferrite growth under externally imposed elastic and plastic strains illustrated the complex interplay
between thermalmechanicalchemicalcrystallographic
parameters. Now, it is possible to model the acicular ferrite formation by using bainitic kinetic models. The
importance of computational models that describe the
overall microstructure evolution in the context of predicting acicular ferrite formation is stressed.

Acknowledgments
Research sponsored by the Division of Materials Sciences and Engineering, and Assistant Secretary for Energy Eciency and Renewable Energy, Industrial
Technologies Program, Industrial Materials for the Future, US Department of Energy, under contract DEAC05-00OR22725 with UT-Battelle, LLC. The author
also thanks Dr. Z. Feng of ORNL for translating one
of the research paper published in Chinese language.
The author thanks Drs. J. M. Vitek and R. L. Klueh

of Oak Ridge National Laboratory for helpful discussions on this paper.

References
The papers of particular interest have been highlighted as:
* of special interest;
** of very special interest.
[1] Ito Y, Nakanishi M. Study on Charpy impact properties of
weld metals with submerged arc welding. Sumitomo Search
1976;15:4262.
[**2] Abson DJ, Pargeter RJ. Factors inuencing as-deposited
strength, microstructure, and toughness of manual metal arc
welds suitable for CMn steel fabrications. Int Met Rev
1986;31:14194.
[**3] Grong O, Matlock DK. Microstructural development in mild
and low alloy steel weld metals. Int Met Rev 1986;31:2748.
[4] David SA, DebRoy T. Current issues and problems in welding
science. Science 1992;257:497502.
[*5] Thewlis G. Classication and quantication of microstructures
in steels. Mater Sci Technol 2004;20:14360.
[6] Babu SS. Acicular ferrite and bainite in FeCrC weld deposits.
Ph.D. Thesis. University of Cambridge, UK; 1992.
[7] Liu S, Olson DL. The role of inclusions in controlling HSLA
steel weld microstructures. Weld J 1986;65:139s49s.
[**8] Bhadeshia HKDH. Bainite in steels. 2nd ed. Carlton House
Terrace, London: IOM Communications Limited; 2001.
[*9] Homma H, Ohkita S, Matsuda S, Yamamoto K. Improvement
of HAZ toughness in HSAL steel by introducing nely
dispersed Ti-oxide. Weld J 1987;66:301s9s.
[*10] Diaz-Fuentes M, Iza-Mendia A, Gutierrez I. Analysis of
dierent acicular ferrite microstructures in low-carbon steels
by electron backscattered diraction: study of their toughness
behavior. Metall Mater Trans A 2003;34A:250516.
[**11] Gourgues AF, Flower HM, Lindley TC. Electron backscattering diraction study of acicular ferrite, bainite, and martensite
steel microstructures. Mater Sci Technol 2000;16:2640.
[12] Yang JR, Bhadeshia HKDH. Orientation relationships between
adjacent plates of acicular ferrite in steel weld deposits. Mater
Sci Technol 1989;5:937.
[13] Strangwood M, Bhadeshia HKDH. The mechanism of acicular
ferrite transformation in steel weld deposits. In: David SA,
editor. Proceedings of the conference on advances in welding
science and technology. Ohio, USA: ASM Intenational; 1987.
p. 18791.
[14] Yang JR, Bhadeshia HKDH. Acicular ferrite transformations
in alloy-steel weld metals. J Mater Sci 1991;26:83945.
[15] Chandrasekharaiah MN, Duben G, Kolster BH. An atom
probe study of retained austenite in ferritic weld metal. Weld J
1992;71:247s57s.
[16] Babu SS, Bhadeshia HKDH. Stress and the acicular ferrite
transformation. Mater Sci Eng A 1992;156:19.
[17] Evans GM, Bailey N. Metallurgy of basic weld metal. Abington, Cambridge, England: Abington Publishing; 1997.
[*18] Ricks RA, Howell PR, Barritte GS. The nature of acicular
ferrite in HSLA steel weld metals. J Mater Sci 1982;17:73240.
[19] Dowling JM, Corbett JM, Kerr HW. Inclusion phases and the
nucleation of acicular ferrite in submerged arc welds in high
strength low alloy steels. Metall Trans A 1986;17:161123.
[20] Zhang Z, Farrar RA. Role of non-metallic inclusions in
formation of acicular ferrite in low alloy weld metals. Mater
Sci Technol 1996;12:23760.

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278
[21] Lee T-K, Kim HJ, Yang BY, Hwang SK. Eect of inclusion size
on the nucleation of acicular ferrite in welds. ISIJ Int
2000;40:12608.
[22] Zhang S, Hattori N, Enomoto M, Tarui T. Ferrite nucleation at
ceramic/austenite interfaces. ISIJ Int 1996;36:13019.
[*23] Madariaga I, Romero JL, Gutierrez I. Upper acicular ferrite
formation in a medium-carbon microalloyed steel by isothermal
transformation: nucleation enhancement by CuS. Metall Mater
Trans A 1998;29:100315.
[*24] Madariaga I, Gutierrez I. Role of the particlematrix interface
on the nucleation of acicular ferrite in a medium carbon
microalloyed steel. Acta Mater 1999;47:95160.
[25] Mills AR, Thewlis G, Whiteman JA. Nature of inclusions in
steel weld metals and their inuence on formation of acicular
ferrite. Mater Sci Technol 1987;3:105161.
[26] St-Laurent S, LEsperance G. Eects of chemistry, density, and
size distribution of inclusions on the nucleation of acicular
ferrite of CMn steel shielded-metal-arc welding weldments.
Mater Sci Eng A 1992;149:20316.
[27] Gregg JM, Bhadeshia HKDH. Titanium-rich mineral phases
and the nucleation of bainite. Metallurgical and Materials
Transactions A 1994;25A:160311.
[*28] Gregg JM, Bhadeshia HKDH. Bainite nucleation from mineral
surfaces. Acta Metall Mater 1994;42:332130.
[29] Kim HS, Lee H-G, Oh K-S. MnS precipitation in association
with manganese silicate inclusions in Si/Mn deoxidized steel.
Metall Mater Trans A 2001;32:151925.
[*30] Shim J-H, Oh Y-J, Suh J-Y, Cho YW, Shim J-D, Byun J-S.
Ferrite nucleation potency of nonmetallic inclusions in medium
carbon steels. Acta Mater 2001;49:211522.
[**31] Furuhara T, Yamaguchi J, Sugita N, Miyamoto N, Maki T.
Nucleation of proeutectoid ferrite on complex precipitates in
Austenite. ISIJ Int 2003;43:16309.
[*32] Byun J-S, Shim J-H, Cho Y-W, Lee DN. Non-metallic inclusion
and intragranular nucleation of ferrite in Ti-killed CMn steel.
Acta Mater 2003;51:1593606.
[33] Brooksbank D, Andrews KW. Stress elds around inclusions
and their relation to mechanical properties. J Iron Steel Inst
1972;210:24655.
[34] Pagounis E, Lindroos VK. The role of internal stresses on the
phase transformation of iron alloys. Scripta Mater
1997;37:659.
[*35] Brooksbank D, Andrews KW. Tesselated stresses associated
with some inclusions in steels. J Iron Steel Inst 1970;208:4959.
[36] Pan T, Yang ZG, Bai BZ, Fang HS. Study of thermal stress and
strain energy in c-Fe matrix around inclusion caused by thermal
coecient dierence. Acta Metall Sinica 2003;39:103742. [in
Chinese].
[37] van der Eijk C, Grong O, Walmsley J. Mechanisms of inclusion
formation in low alloy steels deoxidized with titanium. Mater
Sci Technol 2000;16:5564.
[*38] Rees GI, Bhadeshia HKDH. Thermodynamics of acicular
ferrite nucleation. Mater Sci Technol 1994;10:3538.
[39] Rees GI, Bhadeshia HKDH. Bainite transformation kinetics. 1
Modied-model. Mater Sci Technol 1992;8:98593.
[40] Chester NA, Bhadeshia HKDH. Mathematical modeling of
bainite transformation kinetics. J Phys IV 1997;7(C5):416.
[**41] Lee J-L. Evaluation of the nucleation potential of intragranular
acicular ferrite in steel weldments. Acta Metall Mater
1994;42:32918.
[42] He K, Edmonds DV. Formation of acicular ferrite and
inuence of vanadium alloying. Mater Sci Technol
2002;18:28996.
[*43] Guo Z, Kimura N, Tagashira S, Furuhara T, Maki T. Kinetics
and crystallography of intragranular pearlite transformation
nucleated at (MnS + VC) complex precipitates in hypereutectoid FeMnC alloys. ISIJ Int 2002;42:103341.

277

[44] Barbaro FJ, Krauklis P, Easterling KE. Formation of acicular


ferrite at oxide particles in steels. Mater Sci Technol 1989;5:
105768.
[45] Babu SS, Bhadeshia HKDH. Transition from bainite to
acicular ferrite in reheated FeCrC weld deposits. Mater Sci
Technol 1990;6:100520.
[*46] Lee CH, Bhadeshia HKDH, Lee H-C. Eect of plastic
deformation on the formation of acicular ferrite. Mater Sci
Eng A 2003;360:24957.
[47] Babu SS, Bhadeshia HKDH. Mechanism of the transition from
bainite to acicular ferrite. Mater Trans 1991;32:67988.
[*48] Thewlis G, Whiteman JA, Senogles DJ. Dynamics of austenite
to ferrite phase transformation in ferrous weld metals. Mater
Sci Technol 1997;13:25774.
[49] Yang JR, Bhadeshia HKDH. Thermodynamics of acicular
ferrite transformation in alloy steel weld deposits. In: David SA,
editor. Proceedings of the conference on advances in welding
science and technology. Ohio: ASM International; 1987. p.
20913.
[50] Babu SS, Bhadeshia HKDH. A direct study of allotriomorphic
ferrite crystallography. Mater Sci Eng 1991;A142:20919.
[51] Harrison PL, Farrar RA. Inuence of oxygen-rich inclusions on
the c to a transformation in high strength low alloy (HSLA)
steel weld metals. J Mater Sci 1981;16:221826.
[52] Blais C, LEsperance G, Evans GM. Characterization of
inclusions found in CMn steel welds containing titanium. Sci
Technol Weld Joining 1999;4:14350.
[53] Rowe MD, Liu S, Reynolds TJ. The eect of ferro-alloy
additions and depth on the quality of underwater wet welds.
Weld J 2002;81:156s66s.
[54] Babu SS, David SA. Inclusion formation and microstructure
evolution in low alloy steel welds. ISIJ Int 2002;42:134453.
[55] Peng Y, Chen W, Xu Z. Study of high toughness ferrite for
submerged arc welding of pipeline steel. Mater Charact 2001;47:
6773.
[56] Snieder G, Kerr HW. Eects of chromium additions and ux
type on the structure and properties of HSLA steel submerged
arc weld metal. Can Metall Q 1984;23:31525.
[57] Byun J-S, Shim J-H, Cho YW. Inuence of Mn on microstructure evolution in Ti-killed CMn steel. Scripta Mater
2003;48:44954.
[58] Liu SK, Zhang J. The inuence of the Si and Mn concentrations
on the kinetics of the bainite transformation in FeCSiMn
alloys. Metall Trans A 1990;21:151725.
[59] Babu SS, Evans GM, David SA. Unpublished research, Oak
Ridge National Laboratory, 2004.
[60] Hong T, DebRoy T. Nonisothermal growth and dissolution of
inclusions in liquid steels. Metall Mater Trans BProc Metall
Mater Process Sci 2003;34:2679.
[*61] Hong T, Debroy T. Eects of time, temperature, and steel
composition on growth and dissolution of inclusions in liquid
steels. Ironmaking Steelmaking 2001;28:4504.
[62] Kluken AO, Grong O. Mechanisms of inclusion formation in
AlTiSiMn deoxidized steel weld metals. Metall Trans A
1989;20:133549.
[*63] Lehmann J, Rocabois P, Gaye H. Kinetic model of non-metallic
inclusions precipitation during steel solidication. J Non-Cryst
Solids 2001;282:6171.
[64] Kim H-S, Lee H-G, Oh K-S. Evolution of size composition and
morphology of primary and secondary inclusions in Si/Mn and
Si/Mn/Ti deoxidized steels. ISIJ Int 2002;42:140411.
[65] Babu SS, David SA, Vitek JM, Mundra K, DebRoy T. Development of macro- and microstructures of CMn low alloy steel
weldsinclusion formation. Mater Sci Technol 1999;11:18699.
[66] Babu SS, David SA, Vitek JM, Mundra K, DebRoy T. Model
for inclusion formation in low alloy steel weld metals. Sci
Technol Weld Joining 1999;4:27684.

278

S.S. Babu / Current Opinion in Solid State and Materials Science 8 (2004) 267278

[67] Wintz M, Bobadilla M, Lehmann J, Gaye H. Experimental


study and modeling of the precipitation of non metallic
inclusions during solidication of steel. ISIJ Int 1995;35:
71522.
[68] Quintana MA, McLane J, Babu SS, David SA. Inclusion
formation in self shielded ux-cored arc welds. Weld J
2001;80:98s105s.
[69] Ichikawa K, Koseki T, Fuji M. Thermodynamic estimation of
inclusion characteristics in low alloy steel weld metals. Sci
Technol Weld Joining 1997;2:2315.
[70] Koseki T, Ohkita S, Yurioka N. Thermodynamic study of
inclusion formation in low alloy steel weld metals. Sci Technol
Weld Joining 1997;2:659.
[71] Hsieh KC, Babu SS, Vitek JM, David SA. Calculation of
inclusion formation in low alloy steel welds. Mater Sci Eng A
1996;215:8491.
[72] Pavlyk V, Dilthey U. Simulation of weld solidication microstructure and its coupling to the macroscopic heat and uid ow
modeling. Model Simul Mater Sci Eng 2004;12:S3S45.

[*73] Fukumoto S, Kurz W. Solidication phase and microstructure


selection maps for FeCrNi alloys. ISIJ Int 1999;39:12709.
[74] Babu SS, David SA, Quintana MA. Modeling microstructure
evolution in self-shielded ux cored arc welds. Weld J
2001;80:91s7s.
[75] Hillert M, Hoglund L. Simulation of the peritectic reaction in
FeC alloys. Mater Trans JIM 1999;40:5646.
[76] Radhakrishnan B, Zacharia T. On the Monte Carlo simulation
of curvature driven grain growth. Modeli Simul Mater Sci Eng
2002;10:22736.
[**77] Bhadeshia HKDH, Svensson L-E, Gretoft B. A model for the
development of microstructure in low-alloy steel deposits. Acta
Metall 1985;33:127183.
[78] Jones SJ, Bhadeshia HKDH. Kinetics of the simultaneous
decomposition of austenite into several transformation products. Acta Mater 1997;45:291120.
[79] Thewlis G. The nature of acicular ferrite in ferrous weld metals
and the challenges for microstructure modeling. Mater Sci
Forum 2003;426432:401926.

You might also like