You are on page 1of 13

The Effect of the Initial Microstructure on Recrystallization

and Austenite Formation in a DP600 Steel


M. KULAKOV, W.J. POOLE, and M. MILITZER

The effects of initial microstructure and thermal cycle on recrystallization, austenite formation,
and their interaction were studied for intercritical annealing of a low-carbon steel that is suitable
for industrial production of DP600 grade. The initial microstructures included 50 pct cold-
rolled ferrite–pearlite, ferrite–bainite–pearlite and martensite. The latter two materials recrys-
tallized at similar rates, while slower recrystallization was observed for ferrite–pearlite. If
heating to an intercritical temperature was sufficiently slow, then recrystallization was com-
pleted before austenite formation, otherwise austenite formed in a partially recrystallized
microstructure. The same trends as for recrystallization were found for the effect of initial
microstructure on kinetics of austenite formation. The recrystallization–austenite formation
interaction accelerated austenization in all the three starting microstructures by providing
additional nucleation sites and enhancing growth rates, and drastically altered morphology and
distribution of austenite. In particular, for ferrite–bainite–pearlite and martensite, the recrys-
tallization–austenite formation interaction resulted in substantial microstructural refinement.
Recrystallization and austenite formation from a fully recrystallized state were successfully
modeled using the Johnson–Mehl–Avrami–Kolmogorov approach.

DOI: 10.1007/s11661-013-1721-z
 The Minerals, Metals & Materials Society and ASM International 2013

I. INTRODUCTION austenite might transform back into ferrite, and the


remaining austenite is converted into martensite. The
DUAL-PHASE (DP) steels were originally devel- morphology and distribution of intercritical austenite
oped in the mid 1970s[1] but have become widely used determines the resulting martensite distribution in the
material for automotive manufacturers only in the last final microstructure and this affects the mechanical
10 years. The formability of high-strength low-alloy properties of intercritically annealed steels.[2–4]
(HSLA) steels is insufficient for the production of If the heating rate is sufficiently low, then recrystal-
complex shapes of automotive parts and thus limits lization is completed before austenite starts to form.
their applications. These limitations can be resolved by Austenite formation in as-heat-treated or fully recrys-
using DP steels with a composite microstructure of hard tallized materials has been studied extensively for a
martensite embedded in a soft ferrite matrix. Compared range of initial microstructures.[5–11] Microstructures
with HSLA steels at the same level of tensile strength, with a uniform cementite distribution and fine ferrite
DP steels possess lower yield strength, no yield point grain size, e.g., tempered martensite, result in faster
elongation, higher work hardening rate, and higher austenization rates and more homogeneous austenite
ductility. distribution as opposed to coarse-grained microstruc-
DP steels for automotive sheets are primarily cold- tures with clusters of cementite, e.g., in pearlite.[8,9]
rolled and coated products. In this case, the DP Modern industrial annealing lines permit heating at
structure is produced through intercritical annealing rates sufficiently high to postpone recrystallization such
after cold rolling. Depending on cooling conditions on that it overlaps with austenite formation. The interac-
the run-out table in the hot mill, various microstructures tion between recrystallization and austenite formation
can be produced as an input for intercritical annealing in has not received attention until recently. These recent
a continuous annealing or galvanizing line. During studies showed acceleration of austenite formation from
heating and holding at an intercritical temperature, the a nonrecrystallized microstructure and drastic changes
cold-rolled microstructure recrystallizes and partially in austenite morphology and distribution.[12–18]
transforms into austenite. Upon cooling, a fraction of The purpose of the current study is to determine the
effect of initial cold-rolled microstructures on recrystal-
lization, austenite formation, and their interaction in a
low-carbon steel with a typical DP600 grade chemistry.
Three initial microstructures are included in this inves-
M. KULAKOV, Ph.D. Student, W.J. POOLE, and
M. MILITZER, Professors, are with the Centre for Metallurgical
tigation, i.e., (1) ferrite–pearlite, (2) ferrite–bainite–
Process Engineering, The University of British Columbia, 309-6350 pearlite. and (3) martensite, to cover the full range of
Stores Road, Vancouver, BC V6T 1Z4, Canada. Contact e-mail: structures which could be potentially used for cold
nkulako@gmail.com rolling and subsequent annealing.
Manuscript submitted October 23, 2012.

METALLURGICAL AND MATERIALS TRANSACTIONS A


II. MATERIALS AND EXPERIMENTAL and 15 seconds dwell time. An average of five hardness
PROCEDURES measurements is reported. Austenite formation was
studied during continuous heating at 1, 10 and 100 K/s
The chemical composition of the investigated steel is and heating–holding tests with 1200 seconds holding at
shown in Table I. The industrially processed steel was 1043 K (770 C) after heating either at 1 or 100 K/s. A
supplied by ArcelorMittal Dofasco Inc. (Hamilton, ON, dilatometer was mounted on samples either transversely
Canada) from the same heat in form of 3.6 mm thick (Gleeble tests) or longitudinally (Bähr experiments) to
hot-rolled and 1.8 mm thick 50 pct cold-rolled sheets. measure dimensional changes during austenite forma-
The hot-rolled sheets were used to create additional tion with respect to the same rolling direction. At least
starting microstructures for the intercritical annealing two tests were conducted for each thermal path.
studies. For this purpose, the hot-rolled material was Austenite volume fractions were calculated from the
first re-austenized in a box furnace at 1173 K (900 C) dilatometer data using the lever rule (ASTM A1033-10).
for 1800 seconds and then either furnace-cooled or To validate the dilation results and obtain further
water-quenched to produce two different microstruc- insight into the austenite formation mechanisms, par-
tures. Subsequently, these materials were cold rolled in a tially austenized samples were water-quenched at cool-
single stand laboratory mill having 150 mm diameter ing rates of 1000 K/s to convert all intercritical
rolls. The cold-rolling schedule consisted of ten passes austenite into martensite at room temperature. The
with 5 pct reduction per pass to obtain a total reduction water-quenched samples were etched with LePera’s
of 50 pct as in the industrially cold rolled sheets. Rolling reagent to reveal martensite.[20] The volume fraction of
conditions in the laboratory rolling mill were selected martensite was measured using automatic image anal-
such as to obtain a homogeneous strain distribution ysis according to ASTM E1245-03. Martensite mor-
through the sample thickness similar to that commonly phology and distribution were investigated with
obtained in industrially cold-rolled sheets. Thus, labo- scanning electron microscopy on samples etched with
ratory rolling was conducted with the ratio of the mean 2 pct Nital. Further, fully austenized samples were water
sample thickness to the length of the sample-roll contact quenched to determine the austenite grain size resulting
being always below unity.[19] from austenite formation. For these fully martensitic
Subsequent heat treatments were conducted under samples, the prior austenite grain boundaries were
high vacuum of 0.26 Pa using Gleeble 3500 thermo- revealed by etching at 363 K (90 C) for 75 seconds in
mechanical simulator equipped with a dilatometer a saturated water based solution of picric acid, 1.3 g of
(Dynamic Systems Inc., Poestenkill, NY). From the sodium dodecylbenzene sulfonate and 1 mL of concen-
as-cold-rolled materials, 10 9 60 9 1.8 mm sheet sam- trated hydrochloric acid per 100 mL solution. Equiva-
ples were machined with the longitudinal axis coinciding lent area diameter of ferrite grains before and after
with the rolling direction. Temperature was controlled recrystallization, as well as austenite grains in fully
through a K-type thermocouple spot welded at the austenized samples were determined on at least 500
center of each sample. One test series, i.e., the heating– grains according to ASTM E1382-97 using the grain
holding transformation tests with 100 K/s heating rate, area measurement technique.
was conducted with a Bähr 805A/D dilatometer using
10 9 5 9 1 mm sheet samples with the longitudinal axis
aligned with the transverse rolling direction. Equal
amounts were ground off both sides to reduce their III. EXPERIMENTAL RESULTS
thickness from 1.8 to 1 mm. The microstructure was A. Characterization of the Initial Microstructures
characterized using a Nikon Epiphot 300 optical micro-
scope with image analysis software Clemex (Clemex Three materials which had been cold rolled by 50 pct
Technologies Inc., Longueuil, QB, Canada) and Hitachi were used for the recrystallization and austenite forma-
S2300 and S3000 scanning electron microscopes. tion experiments in the current study. Their microstruc-
Recrystallization in all starting microstructures was tures before and after cold rolling are shown in Figure 1
studied during isothermal holds at 873 K, 898 K, and and can be described as follows:
923 K (600 C, 625 C, and 650 C); the heating rate to (a) After reaustenizing at 1173 K (900 C) for
the holding temperature was 50 K/s. Further, recrystal- 1800 seconds and slow furnace cooling of the as-
lization was measured during continuous heating tests at received hot-rolled material, its microstructure con-
1 K/s. Partially recrystallized samples were helium- sisted of 85 vol pct of ferrite and 15 vol pct of
quenched. Recrystallized volume fraction was measured pearlite (F–P), Figure 1(a). The average ferrite
directly on secondary electron images using the point grain size in the as-heat-treated condition was
counting method (ASTM E569-89). Vickers micro- 11 lm. Pearlite bands were parallel to the rolling
hardness measurements were conducted at 1 kg load direction with an average band spacing of 14 lm
after 50 pct cold rolling, Figure 1(d). The pearlite
Table I. Major Alloying Elements in the Investigated Steel band thickness characterized using the intercept
(Weight Percent) method with the test lines being aligned perpendic-
ular to the bands was equal to 5.6 and 3.4 lm
C Mn Si Cr before and after cold rolling, respectively. The rela-
0.105 1.858 0.157 0.340 tive change in pearlite band thickness of 40 pct
reduction is less than the macroscopic strain of

METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 1—Initial microstructures before (a) to (c) and after 50 pct cold rolling (d) to (f): (a, d) furnace cooled (ferrite–pearlite), (b, e) as-received
(ferrite–bainite–pearlite), (c, f) water-quenched (martensite) (P: pearlite, F: ferrite).

50 pct. The strain partitioning is a common feature Hereafter, the initial microstructures will be desig-
observed during cold rolling of ferrite–pearlite nated by their abbreviations, i.e., F–P, F–B–P, and M.
structures.[21] After the heat treatment the hardness
was 135HV and increased by 93HV to 228HV
B. Recrystallization
after cold rolling.
(b) The microstructure of the as-received material con- Hardness of F–P and F–B–P decreased gradually in
sisted of 5 vol pct of pearlite, with the rest being the course of isothermal annealing, while M softened in
either ferrite or bainite (F–B–P), Figure 1(b). No two stages, Figure 2. The first, rapid softening stage,
attempt was made to differentiate the latter two took place in less than 10 seconds, where the micro-
constituents because of their similar appearance in structure did not show any evidence of recrystallization.
the micrographs. Grain size of the ferrite/bainite The second stage led to a more gradual hardness
constituents before cold rolling was equal to 3 lm. decrease. When recrystallization was complete after
The hardness values were equal to 189 and 272HV long annealing times, a hardness of 145HV was mea-
before and after cold rolling, respectively, corre- sured for all the initial microstructures. Figure 3 shows
sponding to a 83HV gain in hardness due to cold the metallographically measured values for recrystalli-
rolling, Figure 1(e). zation fraction as a function of holding time at 923 K
(c) After reaustenizing at 1173 K (900 C) for (650 C). The F–B–P initial microstructure had the
1800 seconds and water quenching of the as-re- highest rate of recrystallization, the F–P material
ceived hot-rolled material, a martensitic micro- required the longest times for recrystallization to be
structure was formed, Figure 1(c), with a hardness completed, while M recrystallized at intermediate rates.
of 378HV. This hardness falls within the range Similar trends were observed for the other investigated
from 340 to 380HV expected for martensite with holding temperatures as well as for recrystallization
0.1 wt pct carbon[22] indicating that this third start- during continuous heating at 1 K/s, as shown in
ing microstructure can indeed be considered as lath Figure 4. After recrystallization, all the microstructures
martensite (M). The martensite hardness increased consisted of various arrays of carbide particles in a
by 63HV to 444HV in the course of cold rolling, ferrite matrix. For the F–P, starting material pearlite
Figure 1(f). partially spheroidized in the course of annealing, and

METALLURGICAL AND MATERIALS TRANSACTIONS A


450 Figure 5(a). Relatively coarse carbide particles were
Ferrite-Pearlite either clustered or arranged into linear arrays in the fully
400 Ferrite-Bainite-Pearlite
Martensite
recrystallized state resulting from the F–B–P initial
structure, see Figure 5(b). The clusters were formed
Vickers hardness

350
presumably at the former bainite or pearlite areas, while
300 the linear arrays could be a result of grain boundary
carbides in the cold-deformed structure. A finer and
250 more homogeneous distribution of carbides was found
200 after recrystallization from the M initial structure,
Figure 5(c). The ferrite grain size after recrystallization
150 completion was found to be independent of the anneal-
ing temperature and equal to 7.6, 5.1, and 6.5 lm for
100 F–P, F–B–P and M, respectively. It is noteworthy that
1
0 10 100 1000
recrystallization in F–B–P and M led to a coarser grain
Time, s
size, while in F–P the final grain size was lower than the
one before recrystallization.
Fig. 2—Hardness evolution during annealing at 923 K (650 C).

C. Austenite Formation
100
Ferrite-Pearlite The fraction of austenite formed as a function of
Percentage recrystallized, pct

90 Ferrite-Bainite-Pearlite temperature for the F–B–P initial microstructure is


80 Martensite shown in Figure 6(a) for continuous heating rates of 1,
70 10, and 100 K/s. The ortho- and para-equilibrium
60 austenite fractions are also shown as calculated with
Thermo-Calc using the FE-2000 database (Thermo-Calc
50
Software, Stockholm, Sweden). The austenite fraction as
40
a function of temperature appeared to be rather
30 independent of heating rate even though in none of
20 the cases equilibrium was reached for any given tem-
10 perature. Thus, the austenite formation rates increased
0
with heating rate, and this acceleration of austenite
0
1 10 100 1000 formation scaled in a first approximation with heating
Time, s rate. There was even a slight tendency that higher
heating rates can lower the transformation tempera-
Fig. 3—Recrystallization kinetics at 923 K (650 C) (Symbols corre- tures. Similar trends were observed for the other two
spond to experimental results, error bars represent 95 pct confidence initial microstructures: Temperatures needed to achieve
interval, lines are the JMAK models). 50 pct of austenite during continuous heating were
similar for all heating rates and in some cases (M, F–B–
P) appeared to be lower by about 10 K for higher
Temperature, K heating rates, Figure 6(b). Further, the austenite forma-
913 933 953 973 993 1013 1033
100 tion start temperature was for all continuous heating
Percentage recrystallized, pct

90 cases approximately 1003 K (730 C), e.g., Figure 6(a).


80 During holding at 1043 K (770 C) after slow heating
70
at 1 K/s, the rate of austenite formation was virtually
60
the same for F–B–P and M, while the F–P starting
microstructure took longer holdtimes to achieve the
50
same austenite volume fractions, Figure 7(a). As shown
40
in Figure 7(b), the rate of austenite formation was
30
accelerated in all the three starting microstructures when
20 Ferrite-Pearlite the heating rate to the intercritical temperature was
10 Ferrite-Bainite-Pearlite
Martensite
increased from 1 to 100 K/s. After heating at a rate of
0 100 K/s, the austenite volume fraction approached the
640 660 680 700 720 740 760 paraequilibrium austenite content after 1200 seconds
0
Temperature, C long holding.
During heating at 1 K/s, recrystallization was com-
Fig. 4—Recrystallization kinetics during continuous heating at 1 K/s pleted below 993 K (720 C) for all the three starting
(Symbols correspond to experimental results, error bars represent 95
pct confidence interval, lines are the JMAK models).
microstructures (see Figure 4), i.e., before austenite
formation starts. This was confirmed by direct metallo-
graphic observations of the early stages of austenite
clusters of cementite particles in the recrystallized formation in F–B–P, Figure 8(a). Austenite nucleated at
microstructure inherited the distribution and morphol- the boundaries of recrystallized ferrite grains in this case.
ogy of the former pearlite bands, as can be observed in A faster heating rate of 10 K/s shifted recrystallization

METALLURGICAL AND MATERIALS TRANSACTIONS A


of austenite morphologies and distributions. Figure 9(a)
illustrates that for the fully recrystallized F–P condition,
both pearlite bands and the boundaries of recrystallized
ferrite grains provided nucleation sites for austenite
while predominantly banded austenite was found when
austenite formed in unrecrystallized F–P, Figure 9(d).
The morphology and distribution of austenite formed in
F–B–P and M was similar. After recrystallization was
completed, austenite appeared along ferrite grain
boundaries and formed a continuous boundary net-
work, Figures 9(b) and (c). In unrecrystallized F–B–P
and M, a uniform distribution of fine nearly equiaxed
austenite particles was formed, Figures 9(e) and (f). The
resulting austenite grain size decreased with increasing
heating rate, see Table II.
To provide a comparison basis for analyzing the effect
of incomplete recrystallization on austenite formation,
step-heating experiments were conducted. Here, the
samples were first heated at 1 K/s to 973 K (720 C),
which is 10 K lower than the austenite formation start
temperature (as shown in Figure 6(a)), to provide
sufficient time for recrystallization to be completed and
then the heating rate was increased to either 10 or
100 K/s. This procedure permitted to study austenite
formation from a completely recrystallized microstruc-
ture for these higher heating rates. As an example,
Figure 10(a) shows the effect of heating rate on the
austenite formation kinetics from the fully recrystallized
condition for the F–B–P starting microstructure. In
contrast to the continuous heating experiments,
Figure 6(a), the transformation curves were shifted to
higher temperatures as heating rate increased as one
might have expected. Interestingly, the austenite forma-
tion rates as measured by the temperature required to
reach an austenite fraction of 10, 50 and 90 pct were
very similar for all the three starting microstructures, see
Figure 10(b). An increase of heating rate from 1 to
100 K/s resulted in an increase of transformation
temperature by 40 K. Moreover, the final austenite
grain size was found to decrease with heating rate but
the magnitude of this decrease was lower for the step-
heating cases than for the continuous heating experi-
ments, Table II.
Fig. 5—Fully recrystallized microstructures after annealing at 873 K
(600 C): (a) ferrite–pearlite after 4020 s, (b) ferrite–bainite–pearlite
after 3000 s, (c) martensite after 3000 s (P: intact pearlite, SP:
spheroidized pearlite).
IV. MODELING
A. Recrystallization Model
completion into the intercritical temperature region and The Johnson–Mehl–Avrami-Kolmogorov (JMAK)
resulted in concurrent processes of recrystallization and model[23–25] was employed to describe the recrystalliza-
austenite formation. It was observed that unrecrystallized tion kinetics, i.e., for the isothermal case:
regions were found to provide additional nucleation sites
for austenite, Figure 8(b). During heating at 100 K/s X ¼ 1  expð0:693  ðt=t50 Þn Þ ½1
austenite formation was found to initiate before the onset
of recrystallization. As a result, a dense and relatively where X is the recrystallized volume fraction, t is time,
uniform distribution of fine martensite particles (i.e., the t50 is time needed to achieve 50 pct of recrystallization
former austenite grains) was observed in an unrecrystal- and n is the JMAK exponent.
lized ferrite matrix, as shown in Figure 8(c). The temperature dependence of the parameter t50 is
The different starting microstructures and the differ- expressed in the following form:
ent scenarios for the interaction between ferrite recrys-
tallization and austenite formation resulted in a variety t50 ¼ ðt50 Þ0 expðQ=R  TÞ ½2

METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 6—Austenite formation kinetics during continuous heating: (a) in ferrite–pearlite–bainite, (b) temperature-heating rate map for 50 pct auste-
nization.

70 70
Paraequilibrium Paraequilibrium
Austenite content, pct

Austenite content, pct

60 60
Orthoequilibrium Orthoequilibrium

50 50

40 40

30 30

Ferrite-Pearlite Ferrite-Pearlite
20 20
Ferrite-Bainite-Pearlite Ferrite-Bainite-Pearlite
Martensite Martensite
10 10
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Time, s Time, s
(a) (b)

Fig. 7—Austenite formation kinetics at 1043 K (770 C) after heating at (a) 1 K/s and (b) 100 K/s (symbols correspond to experimental results,
lines—to the JMAK model).

are in close agreement with the experimentally measured


with the pre-exponential factor, (t50)0, the effective activa-
recrystallization kinetics for continuous heating at 1 K/s,
tion energy, Q, while R and T have their usual meaning.
however for the F–B–P material the model overesti-
There are three adjustable parameters n, (t50)0 and Q
mates recrystallization rates at the initial stages of the
in this model. Based on the isothermal recrystallization
process. The experimental and modeling results, Fig-
kinetics at 873 K, 898 K, and 923 K (600 C, 625 C,
ure 4, are within ±7 K at all times and the models are,
and 650 C) separate sets of the fitting parameters were
thus, considered as a good representation of the
obtained for each of the three starting microstructures,
recrystallization process for all initial microstructures.
Table III. The quality of the fit is shown in Figure 3.
The overall conformity of the experimental and model-
The JMAK model in its differential form, i.e.,
ing results is consistent with the additive nature of the
 n1 recrystallization process. The recrystallization models
n  ð1  XÞ ln ð1=1  XÞ n were subsequently applied to predict temperatures of
dX=dt ¼ 0:693  ½3 recrystallization start and finish (here indicated by 10
t50 0:693
and 90 pct of recrystallization, respectively) for contin-
and the additivity principle[26,27] were then used to uous heating at rates between 0.1 and 100 K/s to create a
predict the recrystallization kinetics during continuous temperature-heating rate recrystallization–austenite
heating at 1 K/s. As shown in Figure 4, the predictions formation map as seen in Figure 11. As heating
of the model for the F–P and M initial microstructures rate increases, the completion of recrystallization is

METALLURGICAL AND MATERIALS TRANSACTIONS A


relative accuracy in determination of the critical heating
rates in Figure 11.

B. Austenite Formation Model


Austenite formation is a complex multistage process
involving austenite nucleation, dissolution of individual
carbide particles or pearlite colonies and ferrite-to-
austenite transformation. Moreover, the potential inter-
action of recrystallization and austenite formation adds
further complexity as the starting microstructure also
affects austenite formation when the two processes
overlap as discussed above. In the current study,
austenite formation was modeled for the fully recrystal-
lized microstructures by fitting the JMAK model to the
results of the step-heating experiments assuming addi-
tivity. Real austenite volume fractions were normalized
by the orthoequilibrium fractions, Eq. [4], for the
intercritical temperature region between 963 K and
1085 K (690 C and 812 C), i.e.,
feq
c ¼ 50  0:1033  T þ 0:0000536  T
2
½4

where T is temperature in K; the coefficients were


obtained from fitting to the Thermo-Calc predictions
shown in Figures 6(a) and 10(a).
The austenite formation start temperature (defined as
when fc/feq
c = 0.05 is reached) was expressed as a
function of heating rate using:
Ts ¼ A  qB ½5
where q is the heating rate; A and B are empirical fitting
parameters, Table IV.
The JMAK model parameters, Table IV, were
deduced from the results of the step-heating experiments
using the Rios method.[28] The JMAK model fit is in
excellent agreement with the results of the step-heating
experiments at 10 and 100 K/s, Figure 10(a), while for
heating at 1 K/s the model provides a good description
for the first 50 pct transformed but then overestimates
the austenite content for higher fractions. However, it is
worth noting that for intercritical annealing, the initial
portion of austenite formation (i.e., up to 50 pct) is
Fig. 8—Early stages of austenite formation in ferrite–bainite–pearlite primarily of interest. The model predictions for austen-
during continuous heating to 1013 K (740 C) at (a) 1, (b) 10, (c)
100 K/s (M: martensite, NF: nonrecrystallized ferrite).
ite formation after heating at 1 K/s and holding at 1043
K (770 C) for the three starting microstructures had
similar trends as the experimental results: austenite
postponed to higher temperatures. The rate of recrys- formed at higher rates in F–B–P and M, than in F–P,
tallization is fastest in F–B–P and slowest in F–P, while Figure 7(a). However, the model overestimates the
M recrystallizes at intermediate rates. The intercept of austenite content for the F–B–P and underestimates it
the austenite formation start temperature of 1003 K for the F–P initial microstructures, respectively. Never-
(730 C) during continuous heating and temperatures theless, the predicted austenite content is at all holding
required for recrystallization to be completed gives a times within ±5 pct of that measured in the experiments.
critical heating rate below which austenite will form in a
fully recrystallized microstructure whereas for heating
rates above this critical value, recrystallization and V. DISCUSSION
austenite formation will, at least partially, take place
simultaneously. The critical heating rate is 3 K/s for F–P The three starting microstructures have different grain
and 7 K/s for F–B–P and M. The aforementioned sizes and different distributions of carbon. In F–P, carbon is
agreement between experimental and modeling results clustered in pearlite bands, and for M, carbon is uni-
within the margin of ±7 K translates into ~20 pct formly distributed, while the F–B–P initial microstructure

METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 9—Austenite morphology after continuous heating to ~1033 K (760 C) at (a) to (c) 1 and (d) to (f) 100 K/s in (a, d) ferrite–pearlite, (b, e)
ferrite–bainite–pearlite, (c, f) martensite (M: martensite, F: ferrite).

Table II. Austenite Grain Sizes (lm)

Continuous Heating (K/s) Step Heating (K/s)

Material 1 10 100 1 to 10 1 to 100


Ferrite–Pearlite 9.2 5.8 5.5 7.1 5.4
Ferrite–Bainite–Pearlite 6.8 4.2 3.4 6.8 6.2
Martensite 8.3 6.3 3.5 6.4 5.4

represents an intermediate case in terms of carbon distri- during recrystallization due to the time- or spatial
bution. These three materials cover the full spectrum of dependence of the nucleation and growth rates, see
structures which could be potentially used for cold rolling Reference 31. However, the exponent for M is the
and subsequent annealing. highest among the three microstructures studied here
For the ideal site-saturated nucleation and constant and approaches the ideal value of 3; this indicates a
three-dimensional growth conditions, the exponent n in more homogeneous structure of M compared with the
the JMAK model is equal to 3. The exponents found in other two initial microstructures. The activation energy,
this study for recrystallization of the F–P and F–B–P Q, representing in a first approximation the average
initial microstructures are equal to 1.4 and 1.7, respec- activation energy for the migration of high-angle grain
tively, Table III, and close to the values typically boundaries, Table III, was found to be nearly constant
obtained for recrystallization of cold-deformed iron for the three starting microstructures with the average of
and steels.[14,29,30] The difference in the ideal and 375 kJ/mol being similar to 330 kJ/mol found for
measured exponents is because the simplifying assump- another steel with a typical DP 600 chemistry (Fe-
tions of the JMAK model (homogeneous nuclei distri- 0.06C-1.86Mn-0.16Mo).[14]
bution, site saturation or constant nucleation rate, The following simple physical model is proposed to
constant and isotropic growth rate) are not fulfilled explain the observed recrystallization rates in different

METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 10—Austenite formation kinetics during the step-heating experiments (a) in ferrite–bainite–pearlite (symbols correspond to experimental re-
sults, lines—to the JMAK model), (b) temperature-heating rate austenization map (dashed lines are trendlines).

Table III. Recrystallization Model Parameters  1


p  d3
Microstructure n (t50)0 (s) Q (kJ/mol) N¼ ½7
6
18
Ferrite–Pearlite 1.7 6.7 9 10 342
Ferrite–Bainite–Pearlite 1.4 2.9 9 1020 377 where d is the average ferrite grain size after recrystal-
Martensite 2.5 2.7 9 1021 398 lization completion.
The growth rate of a recrystallizing grain is equal to
the product of grain boundary mobility, M, and driving
pressure, P:
V¼MP ½8

The mobility of high-angle grain boundaries can be


assumed to be the same for all the three initial
microstructures. While the driving pressure for recrys-
tallization is the stored energy having two components:
the energy stored during phase transformations before
cold rolling, PTransformation, and the stored energy due to
the deformation, PDeformation:
P ¼ PTransformation þ PDeformation ½9

To estimate the driving pressure, the relative softening


during annealing in all the three initial microstructures
was considered:
Fig. 11—Temperature-heating rate recrystallization–austenite forma-
tion map. H0  H
S¼ ½10
H0  HRex
starting microstructures. Recrystallization proceeds where H0, H, and HRex are the microhardness values of
through nucleation and growth. Assuming site satura- the as-deformed, partially recrystallized, and fully
tion conditions and three-dimensional growth, the recrystallized materials, respectively.
recrystallized volume fraction is equal to: For the F–B–P initial microstructure, the relative
X ¼ 1  expðN  ðV  tÞ3 Þ ½6 softening was in a close agreement with the recrystal-
lization percentage as shown in Figure 12. Therefore,
where N is the nuclei density and V is the growth rate. the total hardness change in the course of annealing can
Since annealing temperatures are comparatively low be assumed to be due to the dislocation density decrease,
[£923 K (650 C)], no substantial ferrite grain coarsen- Dqdis, neglecting all other possible factors affecting
ing is expected such that the nuclei density can be hardness, e.g., grain size, different inherent hardness of
assumed to be equal to: different microconstituents, etc.:

METALLURGICAL AND MATERIALS TRANSACTIONS A


Table IV. Austenite Formation Model Parameters

Transformation Start
Temperature (K) (Eq. [5]) JMAK Parameters
1B B
Microstructure A (K Æs ) B n (t50)0 (s) Q (kJ/mol)
56
Ferrite–Pearlite 1008 0.0087 0.29 1.4 9 10 1155
Ferrite–Bainite–Pearlite 1008 0.0104 0.44 1.3 9 1041 843
Martensite 1015 0.0090 0.42 2.8 9 1045 921

100 fragmentation of cementite lamellas thereby promoting


pearlite spheroidization upon subsequent anneal-
90
ing.[13,17] Even in the fully recrystallized microstructure
80 of F–P, unspheroidized areas can still be found, as
70
shown in Figure 5(a). Spheroidization took place con-
currently with recrystallization, and therefore, it is
Softening, pct

60 rather challenging to discriminate their individual effects


50 on hardness changes. However, for 0.1C (wt pct) plain
carbon steel complete spheroidization of fine pearlite
40
leads to ~20HV hardness decrease,[32] while total soft-
30 ening for the F–P initial microstructure during isother-
20
mal annealing was 82HV in the current study.
Ferrite-Pearlite
Therefore, the hardness changes are primarily due to
10 Ferrite-Bainite-Pearlite
recrystallization, and Eq. [14] can still be applied for an
Martensite
0 order of magnitude estimate of the stored energy for
0 10 20 30 40 50 60 70 80 90 100 F–P. Finally, turning to the case of M, during cold
Recrystallized percentage, pct deformation of a low-carbon martensite, laths are
replaced with dislocation cell structures.[33,34] Upon
Fig. 12—Comparison of relative softening estimated using hardness subsequent annealing, the hardness of M dropped by
changes and recrystallized percentage quantified through metallo- ~150HV within the first 10 seconds without any evi-
graphic measurements.
dence of recrystallization, Figure 2. This initial soften-
ing was ascribed to the martensite tempering. Greater
annealing times result in a gradual hardness decrease
P  0:5  G  b2  Dqdis ½11 through recrystallization of tempered M. Typically high
temperature anneals, such as those used for the recrys-
where G is the shear modulus of iron, equal to 83 GPa; b
tallization experiments in the current study, lead to
is the magnitude of Burger’s vector, equal to 0.25 nm.
extremely rapid tempering. For instance, Caron and
The dislocation density decrease is related to the yield
Krauss[35] found, in Fe-0.2C martensite, a hardness drop
strength change, Dry:
of 50 pct due to carbide precipitation after a 0.26-second
pffiffiffiffiffiffiffiffiffiffiffi
Dry  0:5  G  b Dqdis ½12 heat treatment at 923 K (650 C). Thus, tempering was
assumed to take place first, followed by recrystallization
for the M initial microstructure. The hardness achieved
In a first approximation, the yield strength change (in after the first 10 seconds of annealing is assumed to be
MPa) is equal to one third of the Vickers hardness drop, the initial hardness for the estimates of the stored energy
DH (in MPa): to separate this from the tempering effect. Based on
Dry  DH=3 ½13 these arguments, the estimated values for the nuclei
density and stored energy are summarized in Table V.
They are the lowest for the F–P initial microstructure.
The stored energy is thus proportional to the square The M microstructure possessed the highest stored
of the hardness change for the F–B–P initial micro- energy, while the highest number of nuclei formed
structure: during recrystallization of F–B–P.
The average nuclei density and stored energy repre-
2
P DH2 ½14 senting an average microstructural compartment were
9G used to estimate ratios of characteristic times required
for 50 pct recrystallization, t50, for different initial
Only a fraction of the softening for F–P and M can be microstructures using Eqs. [6], [7], [14]:
attributed to recrystallization, as other processes also
lowering hardness took place in these materials, Fig- d
ure 12. In F–P, the additional softening mechanism is t50 / ½15
pearlite spheroidization. Cold-deformation leads to DH2

METALLURGICAL AND MATERIALS TRANSACTIONS A


Table V. Nuclei Density and Stored Energy for Recrystalli- the nuclei density at the beginning of the transforma-
zation tion, more nuclei were formed during faster heating in
the step-heating experiments, Table II. Nucleation and
Nuclei growth are competing phenomena. Higher heating rates
Density Stored Energy
Initial (1015 m3) (MPa)
lead to less growth of nuclei at any given temperature
Microstructure (Eq. [7]) (Eq. [14]) and promote additional nucleation at higher superheat-
ing.
Ferrite–Pearlite 4.4 1.8 As shown in Figure 11, for continuous heating
Ferrite–Bainite–Pearlite 14.4 4.2 experiments at a rate above the critical value, austenite
Martensite 7.0 5.3 forms in a partially recrystallized microstructure. Con-
sidering that during continuous heating at 1 and 100 K/s
austenite forms within the same temperature range of
1003 K to 1123 K (730 C to 850 C) and taking into
Table VI. Ranking of the Initial Microstructures Based on
Recrystallization Rates account the hundredfold difference in time available for
the transformation, Figure 6(a), the rate of austenite
Ratios of Times for formation in unrecrystallized material is accelerated by
50 Pct Recrystallization more than one hundred times. Comparison of austenite
formation kinetics during continuous and the step-
Based on heating experiments reveals that the presence of unre-
Hardness JMAK crystallized regions lowers the transformation start
Ratio of Change and Models temperatures during heating at 10 and 100 K/s for all
Initial Grain Size [T = 973 K
initial microstructures by approximately 20 and 40 K,
Microstructures (Eq. [15]) (700 C)]
respectively, e.g., compare Figures 6(a) and 10(b). The
M/FBP 1.0 1.3 recrystallization–austenite formation interaction also
M/FP 0.3 0.4 affects austenite nuclei distribution and density as can
FBP/FP 0.3 0.3 be seen in Figures 8 and 9. Employing the final austenite
grain size as a measure of the nuclei density, as much as
six times more nuclei were formed during the continuous
heating than during the step-heating experiments for a
given heating rate (see Table II). These observations are
The obtained ratios of annealing times needed to consistent with previous studies on the recrystallization–
reach 50 pct of recrystallization for different initial austenite formation interaction, where preferential
microstructures, Table VI, replicate the experimentally nucleation of austenite in unrecrystallized regions was
observed trends suggesting similar recrystallization rates reported.[13,15–17]
for F–B–P and M initial microstructures, and longer The higher austenite nuclei densities in partially
times required for F–P recrystallization. Moreover, the recrystallized microstructures can be explained with
ratios of times for 50 pct recrystallization are similar to the aid of the classical nucleation theory. The activation
those calculated using the JMAK models, i.e., Eq. [2] barrier for nucleation, DG*, is lowered by increasing the
and Table III. driving pressure for austenite formation, DgV, by the
To rationalize the observed trends for austenite magnitude of the stored energy, P, otherwise released in
formation, nucleation and growth also need to be the course of recrystallization:
considered. Two conditions must be met for the
nucleation to take place. First of all, according to the SðhÞ
DG / ½16
iron–carbon phase diagram, carbon solubility in aus- ðDgV þ PÞ2
tenite is much higher than in ferrite, and austenite nuclei
will form at carbon rich sites. Second, classical nucle- where S(h) is the shape factor (0 < S < 1) that is related
ation theory predicts preferred nucleation at high-energy to the benefits of heterogeneous nucleation sites in
nonequilibrium defects, which lower the activation lowering the activation barrier.
barrier for nucleation, i.e., heterogeneous nucleation. For instance, at the austenite formation temperature
For a distribution of cementites in ferrite matrix, such as of 1003 K (730 C), the chemical driving pressure for
fully recrystallized F–B–P and M, Figures 5(b) and (c), the nucleation is equal to 82 MPa under orthoequilib-
both conditions are met for cementite particles located rium conditions based on the Thermo-Calc prediction,
at ferrite grain boundaries as observed experimentally, while the stored energy varies between 1.7 MPa for the
Figure 8(a). In fully recrystallized F–P, preferable F–P initial microstructure and 5.3 MPa for M, Table V.
nucleation sites for austenite are F–P interfaces at the An additional contribution of 5 MPa is equivalent to a
beginning of the transformation, while ferrite grain 10 K temperature drop for the same magnitude of
boundaries become active nucleation sites at later driving pressure of 82 MPa for the nucleation. Local
transformation stages, this being observed in as- variations in the stored energy could also magnify the
quenched samples with different contents of martensite, effect of the stored energy on the austenite nucleation.
i.e., austenite at an intercritical temperature. Assuming Moreover, the quality of nucleation sites as quantified
that the density of austenite grains after the transfor- by the shape factor in Eq. [16] is also improved in
mation completion provides some indication of trends in deformed material thereby lowering the energy barrier

METALLURGICAL AND MATERIALS TRANSACTIONS A


for the nucleation even further. The combination of the and M). As a result, intercritical annealing of the F–P
lower activation energy for the nucleation and the initial microstructure leads to the development of the
higher density of potential nucleation sites increases the banded austenite, Figures 9(a) and (d). Primarily
nucleation rate in partially recrystallized or unrecrystal- because of its higher carbon concentration, pearlite
lized material. serves as a preferential site for austenite nucleation and
However, the enhanced nucleation alone cannot growth. In the F–B–P and M materials, carbon is more
account for the observed increase in the transformation uniformly distributed, and manganese segregation, in
rate, i.e., the growth rate changed dramatically as well. the current case, does not have any major effect on
Considering again the case of austenite formation austenite nucleation and morphology.
during continuous heating at 1 or 100 K/s between
1003 K and 1123 K (730 C and 850 C), Figure 6(a),
the ratio of the final austenite grain sizes was approx-
imately equal to two, see Table II. Taking into account VI. CONCLUSIONS
the hundredfold difference in the transformation time,
the average velocity of austenite–ferrite interface during It was found that the 50 pct cold-rolled ferrite–
the fast heating is estimated to be accelerated by at least bainite–pearlite (F–B–P) and martensite (M) initial
50 times. Further, austenite grain coarsening may have microstructures recrystallized at nearly the same rates,
taken place and its extent will be more pronounced while slower recrystallization was observed in ferrite–
during slow heating. Then, the growth rate increase pearlite (F–P) microstructure. The effect of initial
would be even greater than 50 times. One possible microstructure on austenite formation kinetics was
explanation for the fast austenite growth in non-recrys- nearly completely eliminated when recrystallization
tallized microstructures may be related to the presence was completed before austenite formation starts. Upon
of fast diffusion paths for substitutional alloying ele- heating to an intercritical temperature at a rate higher
ments in the deformed ferrite. The enhanced diffusion of than a critical value (3 to 7 K/s for the initial micro-
austenite stabilizing elements such as manganese toward structures examined here), the interaction of recrystal-
the ferrite–austenite interface may accelerate austenite lization and austenite formation was observed. Shorter
formation. The fast growth rate could be even further holding times are needed after faster heating (provided
amplified by reduced solute drag.[36–38] Solutes exert heating rate exceeds the critical value) to reach the
lower pressures on a boundary moving rapidly, this desired austenite content. Martensite morphology in DP
leading to its apparent mobility increase. steels can be controlled by variation of either the initial
Another consequence of the fast austenite formation microstructure or through the interaction of recrystal-
during rapid continuous heating is an inhomogeneous lization and austenite formation. Moreover, substantial
carbon distribution after complete austenization. For microstructural refinement can be achieved by forming
heating at a constant rate, the carbon diffusion distance austenite in unrecrystallized F–B–P and M initial
can be estimated from: microstructures.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Recrystallization and austenite formation in fully
T2  T1 recrystallized microstructures were successfully de-
L ¼ DcC  ½17 scribed by means of the JMAK model. Limitations of
q
this approach were found for conditions when recrys-
where T1 = 1003 K and T2 = 1223 K (730 C and tallization was not fully completed before austenite
850 C, respectively) are temperatures for the austenite formation and starting microstructures were essentially
formation initiation and completion during heating at a a function of heating rate, thereby affecting austenite
rate q = 1 or 100 K/s, DcC is the average carbon nucleation and growth rates. A more advanced model,
diffusion coefficient in austenite for this temperature for example, based on the phase-field approach, needs to
range.[39] The carbon diffusion distance in austenite be developed for an accurate description of the recrys-
decreases from 14 to 1.4 lm when the heating rate is tallization–austenite formation interaction and the
increased from 1 to 100 K/s, i.e., the diffusion distance observed variety of austenite morphologies.
in the latter case is smaller than the radius of the
austenite grain, Table II.
The through-thickness manganese distribution is not
uniform in the industrially produced steel sheets as ACKNOWLEDGMENTS
evidenced by the banded structure of pearlite in F–P,
Figure 1(a). The variation of manganese content is The authors acknowledge the financial support
typically inherited from casting and remains unaffected received from the Natural Sciences and Engineering
by the laboratory reaustenitization heat treatment at Research Council of Canada (NSERC) and Arcelor-
1173 K (900 C) for 1800 seconds in the current study, Mittal Dofasco Inc. (Hamilton, ON, Canada). The au-
as much higher soaking times and temperatures for the thors further wish to thank CanmetMATERIALS
homogenization treatment are typically employed. The (Hamilton, ON, Canada) and personally Fateh Fazeli
nonuniform distribution of manganese leads to pearlite for the opportunity to use the Bähr dilatometer. Stim-
banding in F–P, whereas there are no obvious signs of ulating discussions with Benqiang Zhu are greatly
banding in the other two initial microstructures (F–B–P, acknowledged.

METALLURGICAL AND MATERIALS TRANSACTIONS A


REFERENCES 19. W.F. Hosford and R.M. Caddell: Metal Forming: Mechanics and
Metallurgy, 3rd ed., Cambridge University Press, New York, NY,
1. M.S. Rashid: Science, 1980, vol. 208, pp. 862–69. 2007, pp. 163–64.
2. N.K. Balliger and T. Gladman: Met. Sci., 1981, vol. 15, pp. 95– 20. F.S. Lepera: J. Met., 1980, vol. 32, pp. 38–39.
108. 21. T. Gladman: Mater. Sci. Technol., 1990, vol. 6, pp. 1131–38.
3. M. Mazinani and W.J. Poole: Metall. Mater. Trans. A, 2007, 22. G. Krauss: Steels—Processing, Structure and Performance, 3rd ed.,
vol. 38A, pp. 328–39. ASM International, Materials Park, OH, 2005, p. 299.
4. G. Avramovic-Cingara, Y. Ososkov, M.K. Jain, and D.S. 23. A. Kolmogorov: Izv. Acad. Sci. USSR, Math. Ser., 1937, vol. 1,
Wilkinson: Mater. Sci. Eng. A, 2009, vol. 516, pp. 7–16. pp. 355–59.
5. G.R. Speich, A. Szirmae, and M.J. Richards: Trans. Soc. Min. 24. W.A. Johnson and R.F. Mehl: Trans. Am. Inst. Min. Metall. Pet.
Eng. AIME, 1969, vol. 245, pp. 1063–74. Eng., 1939, vol. 135, pp. 416–42.
6. C.I. Garcia and A.J. Deardo: Metall. Trans. A, 1981, vol. 12A, 25. M. Avrami: J. Chem. Phys., 1939, vol. 7, pp. 1103–12.
pp. 521–30. 26. J.W. Cahn: Acta Metall., 1956, vol. 4, pp. 572–75.
7. G.R. Speich, V.A. Demarest, and R.L. Miller: Metall. Trans. A, 27. M. Lusk and H.-J. Jou: Metall. Mater. Trans. A, 1997, vol. 28A,
1981, vol. 12A, pp. 1419–28. pp. 287–91.
8. X.-L. Cai, A.J. Garratt-Reed, and W.S. Owen: Metall. Trans. A, 28. P.R. Rios: Acta Mater., 2005, vol. 53, pp. 4893–901.
1985, vol. 16A, pp. 543–57. 29. W.C. Leslie, F.J. Plecity, and J.T. Michalak: Trans. Metall. Soc.
9. J.J. Yi, I.S. Kim, and H.S. Choi: Metall. Trans. A, 1985, vol. 16, AIME, 1961, vol. 221, pp. 691–700.
pp. 1237–45. 30. K. Mukunthan and E.B. Hawbolt: Metall. Mater. Trans. A, 1996,
10. V.I. Savran, Y. Van Leeuwen, D.N. Hanlon, C. Kwakernaak, vol. 27A, pp. 3410–23.
W.G. Sloof, and J. Sietsma: Metall. Mater. Trans. A, 2007, 31. M. Oyarzábal, A. Martı́nez-de-Guerenu, and I. Gutiérrez: Mater.
vol. 38A, pp. 946–55. Sci. Eng. A, 2008, vol. 485, pp. 200–09.
11. V.I. Savran, S.E. Offerman, and J. Sietsma: Metall. Mater. Trans. 32. W.D. Callister, Jr., and D.G. Rethwisch: Materials Science and
A, 2010, vol. 41, pp. 583–91. Engineering: an Introduction, 7th ed., Wiley, New York, NY, 2007,
12. M. Tokizane, N. Matsumura, K. Tsuzaki, T. Maki, and I. p. 341.
Tamura: Metall. Trans. A, 1982, vol. 13A, pp. 1379–88. 33. T. Swarr and G. Krauss: Metall. Trans. A, 1976, vol. 7A, pp. 41–
13. D.Z. Yang, E.L. Brown, D.K. Matlock, and G. Krauss: Metall. 48.
Trans. A, 1985, vol. 16A, pp. 1385–92. 34. S. Takaki, S. Iizuka, K. Tomimura, and Y. Tokunaga: Mater.
14. J. Huang, W.J. Poole, and M. Militzer: Metall. Mater. Trans. A, Trans., JIM, 1992, vol. 33, pp. 577–84.
2004, vol. 35A, pp. 3363–75. 35. R.N. Caron and G. Krauss: Metall. Trans., 1972, vol. 3A,
15. V. Andrade-Carozzo and P.J. Jacques: Mater. Sci. Forum, 2007, pp. 2381–89.
vols. 539–543, pp. 4649–54. 36. J.W. Cahn: Acta Metall., 1962, vol. 10, pp. 789–98.
16. T. Ogawa, N. Maruyama, N. Sugiura, and N. Yoshinaga: ISIJ 37. G.R. Purdy and Y.J.M. Brechet: Acta Metall. Mater., 1995,
Int., 2010, vol. 50, pp. 469–75. vol. 43, pp. 3763–74.
17. H. Azizi-Alizamini, M. Militzer, and W.J. Poole: Metall. Mater. 38. F. Fazeli and M. Militzer: Metall. Mater. Trans. A, 2005, vol. 36A,
Trans. A, 2011, vol. 42A, pp. 1544–57. pp. 1395–1405.
18. R.R. Mohanty, O.A. Girina, and N.M. Fonstein: Metall. Mater. 39. E.A. Brandes and G.B. Brook: Smithells Metals Reference Book,
Trans. A, 2011, vol. 42A, pp. 3680–90. 7th ed., Elsevier, Amsterdam, 1998, pp. 13–19.

METALLURGICAL AND MATERIALS TRANSACTIONS A

You might also like