Materials Transactions, Vol. 45, No. 12 (2004) pp.

3272 to 3281
#2004 The Japan Institute of Metals

Time-Temperature-Transformation Diagram
within the Bainitic Temperature Range in a Medium Carbon Steel
Francisca G. Caballero, Maria Jesus Santomia* , Carlos Garca-Mateo
and Carlos Garca de Andres
Department of Physical Metallurgy, Centro Nacional de Investigaciones Metalurgicas (CENIM),
Consejo Superior de Investigaciones Cientcas (CSIC), Avda. Gregorio del Amo, 8. 28040 Madrid, Spain

The time-temperature-transformation (TTT) diagram within the medium temperature range of medium carbon steel has been determined.
A single type of C-curve is found within the bainite temperature range for the studied steel. Distinct reaction C-curves were not observed for both
types of microstructure, upper bainite and lower bainite in the TTT diagram. Experimental results on the kinetics of the isothermal formation of
bainite at dierent temperature have demonstrated that both type of microstructure, upper and lower bainite, possesses similar overall
transformation kinetics. Some applications of phase transformation theory towards the formation of bainitic microstructures are discussed, with
particular emphasis on the bainite start temperature, transition temperature from upper to lower bainite, martensite start temperature and the
thickness of bainitic plates.

(Received July 20, 2004; Accepted October 19, 2004)

Keywords: bainite, transformation kinetics, steel

1. Introduction In lower bainite formation, the slower diusion associated

with the reduced transformation temperature provides an
Bainite can be regarded as a non-lamellar mixture of ferrite opportunity for some of the carbon to precipitate in the
and carbides, but within this broad description, it is possible supersaturated bainitic ferrite. A ne dispersion of plate-like
to identify two classical morphologies, traditionally named carbides is then found inside the ferrite plates,6) with a single
upper and lower bainite. Both upper and lower bainite consist crystallographic variant within a given ferrite plate7) although
of plates of ferrite, known as sub-units, separated by it is possible to observe more than one variant of carbide
cementite. The plates of ferrite grow in aggregates called precipitation in a lower bainite sub-unit.8)
sheaves of bainite. Within each sheaf, the plates of ferrite are Takahashi and Bhadeshia9) developed a model to estimate
parallel and share a common crystallographic orientation. the transition temperature from upper to lower bainite based
The essential dierence between upper and lower bainite is on the hypothesis that bainitic ferrite grows with a super-
with respect to the carbide precipitation. In upper bainite, the saturation of carbon in solid solution. In this model the lower
bainitic ferrite is free of precipitation; carbides grow from the bainite start temperature (LBs ) is calculated comparing the
regions of carbon enriched austenite, which are trapped time required to decarburise supersaturated ferrite and the
between the sub-units of ferrite. By contrast lower bainitic time required obtaining a detectable degree of cementite
ferrite contains a ne dispersion of plate-like carbides within precipitation in the bainitic ferrite. Thus, the time required to
the bainitic ferrite plates. decarburise a supersaturated bainitic ferrite plate of thickness
Both upper and lower bainite are formed by the propaga- w is given by,
tion of displacive ferrite sub-units followed by the partition of w2 x

x
 2
the excess carbon into the residual austenite. Cementite can td 1
then precipitate from the enriched austenite between the 16D
x

x
2
ferrite plates. The growth of the bainitic sub-units is where x
is the average carbon concentration in the steel as a
accompanied by a change in the shape of the transformed whole, x
 and x
are the carbon concentrations in ferrite
region, which can be described as an invariant-plane strain and austenite, respectively, when both phases are in para-
with a large shear component. The parent austenite cannot equilibrium, and D
is the weighted average diusivity of
accommodate the large shape deformation elastically and carbon in austenite. Likewise, the time interval t necessary
relaxes by plastic deformation in the region adjacent to the to obtain an amount of cementite precipitation in the ferrite
bainite. This plastic deformation sties the growth of bainite can be determined from the Avrami relationship, which
sub-units before they hit any obstacle, and the transformation expresses the change in volume fraction of cementite
proceeds with the formation of a new sub-unit.1) The precipitation as a function of time and temperature. The
mechanism of transformation continues until the carbon parameters of this relationship were calculated from temper-
concentration of the residual austenite reach the To curve, at ing experimental data of plain carbon steels.9)
which ferrite and austenite with identical chemical compo- The bainite reaction has its own C-curve centred between
sition have the same free energy25) (or To0 curve, if the strain the pearlite and the martensite ranges on a time-temperature-
energy due to the displacements associated with the trans- transformation (TTT) diagram, whether upper bainite and
formation is considered). lower bainite have the same C-curve10,11) or their own
C-curves12) are not clear. Likewise, the displacive C-curve
*Graduate Student, Universidad de Cordoba. presents a characteristic at top at a temperature Th which is
 Isothermal Bainite Formation in a Medium-Carbon Steel 3273

the highest temperature at which ferrite can form by is focussed on the specimen by means of a bi-elliptical
displacive transformation.13) The transformation product at reector. The temperature is measured with a 0.1 mm
Th can be Widmanstatten ferrite or bainite. If the driving diameter Chromel-Alumel (Type K) thermocouple welded
force at Th is sucient to account for diusionless growth, to the specimen. Cooling is performed by blowing a jet of
then Th is identied with the bainite-start temperature (Bs ) helium gas directly onto the specimen surface. These devices
and Widmanstatten ferrite does not form at all. This theory ensure an excellent eciency in controlling the temperature
provides a thermodynamic method for the calculation of Bs .2) and holding time of isothermal treatments and as well as fast
There also exist several empirical formulas for the prediction cooling in quenching processes.
of Bs as a function of the chemical composition,1417) which Cylindrical dilatometric test pieces of 2 mm in diameter
contain dierent alloying elements and ranges of use. and 12 mm in length were austenitised for 60 seconds at
The martensite start temperature (Ms ) generally features on 1200 C, and then isothermally transformed at temperatures
a TTT diagram as a horizontal line parallel to the time axis as ranging from 300 to 550 C for dierent times before
a lower limit to the bainite range. As occurs in the case of the quenching to room temperature. A simple lever rule based
Bs calculation, there are some empirical formulas for Ms as a on the consideration that there is no carbon enrichment of the
function of the chemical composition14,1823) with dierent austenite during transformation, which causes a variation of
grades of accuracy depending on the allowed range of the lattice parameter of the austenite, allows us estimating the
alloying elements. An alternative approach is to use evolution of the volume fraction of bainite, Vb , with time
thermodynamic theory, which has the advantage that any from the recording of the change in length of the specimen
solute combination can be considered on condition that the during transformation,
thermodynamic data are available. In the thermodynamic Lt
Lo
approach, martensite is said to be triggered when the free Vb Vb max  2
Lf
Lo
energy change accompanying the transformation of austenite
to martensite of the same composition achieves some critical where Vb max is the maximum volume fraction of bainite
value at Ms . Ghosh and Olson24) explicitly dened the critical formed at the end of the isothermal reaction assessed by
driving force needed to start martensite formation in terms of metallography, Lt, Lo and Lf are the lengths of the specimen
solute concentration by using solid solution theory. The at t, before and after the completion of the transformation,
inputs of Ghosh and Olsons model are the chemical respectively. This simple procedure was conrmed by
composition and the free energy change for the transforma- comparing the bainite volume fraction assessed by metal-
tion of austenite to ferrite of the same composition, G
. lography and dilatometry using eq. (2).
More recently, Capdevila et al.2527) developed a model using Optical and scanning electron microscopes (SEM) were
articial neural networks for the determination of Ms , used to examine the etched microstructures. Specimens were
including the eect of chemical composition and the prior ground and polished using standard techniques and etched in
austenite grain size (PAGS) in calculations. 2% nital solution and examined using a Jeol JSM-6500F
In the present work, the partial TTT diagram within the Field Emission Scanning Electron Microscope operating at
bainitic temperature range of medium carbon steel has been 7 kV. The volume fraction of bainite was estimated by a
determined. Experimental results on the kinetics of the systematic manual point-counting procedure on scanning
isothermal formation of bainite at dierent temperatures have electron micrographs.29) A grid superimposed on the micro-
allowed studying separately the overall transformation structure provides, after a suitable number of placements, an
kinetics of upper and lower bainite. Moreover, some unbiased statistical estimate of Vb . Stereological errors for Vb
applications of phase transformation theory towards the values correspond to the standard deviation of the measure-
formation of bainitic microstructures are discussed, with ment. Moreover, the true plate-thickness of bainitic ferrite, w,
particular emphasis on the bainite start temperature, tran- was estimated on scanning electron micrographs of fully
sition temperature from upper to lower bainite, martensite transformed specimens by measuring the mean lineal
start temperature and the thickness of bainitic plates. intercept, L
T w=2, in a direction normal to the plate
length. The thickness w can be also related to the mean lineal
2. Materials and Experimental Procedures intercept measured using randomly oriented test lines by the
relation L
 2w,30) but L
T is easier to measure since it is rare
The chemical composition of the studied steel is given in in scanning electron micrographs that entire plates of bainite
Table 1. The material was supplied as 30 mm hot rolled can be imaged.
square bar. Dilatometric analysis and heat treatments were The Ms temperature of the steel was measured by
carried out using an Adamel Lhomargy DT1000 high- dilatometry. Cylindrical specimens 12 mm in length and
resolution dilatometer described elsewhere,28) The heating 2 mm in diameter were heated to 1200 C and then rapidly
device consists of a very low thermal inertia radiation cooled. Each dilatometric test was performed twice. The
furnace. The power radiated by two tungsten lament lamps formation of martensite during cooling was detected by
monitoring the fractional change in dilatation with temper-
ature.
Table 1 Chemical compositions of the studied steel in mass%. Vickers hardness tests were carried out on heat treated
C Mn Cr Ni Si Mo V S samples. The results correspond to the average of at least ve
tests.
0.31 1.22 0.14 0.10 0.25 0.03 0.004 0.01
Cylindrical samples 3 mm in diameter and 12 mm in length
3274 F. G. Caballero, M. J. Santomia, C. Garca-Mateo and C. Garca de Andres

were used to reveal grain boundaries by the thermal etching decreasing decomposition temperature from 550 C to 525 C,
method.31) For this purpose, a surface 2 mm in width was microstructure markedly changes. At 550 C, austenite trans-
generated along the longitudinal axis of samples by polishing forms into a mixture of proeutectoid ferrite and pearlite,
and nishing with 1 mm diamond paste. Later on, samples whereas upper bainite consisting of ferrite laths forming in
were austenitised in vacuum (>1 Pa) at 1200 C for 60 parallel becomes dominant at 525 C. The morphology of
seconds. Subsequently samples were cool down to room proeutectoid ferrite is not polygonal at 550 C, side-plates in
temperature at 1 C/s. These samples do not require metal- addition to allotriomorphs are observed at this temperature.
lographic preparation after heat treatment; the prior austenite On the other hand, as increasing time from 900 to 1800 s at
grain boundaries are revealed without chemical etching. The 525 C degenerated pearlite characterised by the polygonal
average austenite grain size was measured using an image gross microstructure involving non-lamellar carbide is seen
analyser. together with upper bainite in the microstructure. However,
the microstructure becomes fully bainitic and degenerate
3. Results and Discussion pearlite is not longer seen at temperatures below 500 C. All
these metallographic results suggest that Bs temperature of
3.1 Determination of Ms , Bs and LBs temperatures the steel is around 525 C. As decreasing the decomposition
The Ms temperature of the steel was measured by temperature the carbide density in the bainitic microstructure
dilatometry monitoring the related change in dilatation with becomes higher as shown in Fig. 3. This gure shows
temperature. A value of Ms 342 C was recorded on microstructures formed at temperatures between 500 and
dilatometric curves. Figure 1 shows representative micro- 375 C at relatively late stages of transformation.
structures at relatively late stages of transformation at There is no fundamental dierence in transformation
temperature just above and below the Ms temperature mechanism for upper and lower bainite.9) The bainitic ferrite
assessed by dilatometry. Micrographs suggest that martensite grows supersaturated with carbon. The excess carbon may
was formed at 340 C, whereas the microstructure obtained at then partition into the residual austenite or precipitate in the
355 C consists of lower bainite. Likewise, an increase in ferrite in the form of carbides. If the latter process is
hardness of 40 HV is detected when decomposition temper- dominant, lower bainite is obtained. Upper bainite is obtained
ature decreases from 355 to 340 C. These results thus only when carbon partitions relatively rapidly into the
conrm the Ms temperature initially assessed by dilatometry. residual austenite, before the carbides have an opportunity
Figure 2 shows optical and scanning electron micrographs to precipitate. According to this model, it is also possible to
of samples isothermally transformed at 550 C and 525 C. As form a mixed microstructure of upper and lower bainite in a
narrow temperature range around LBs temperature. This is
because the carbon enrichment of the austenite caused by
upper bainite transformation can result in a subsequent
formation of lower bainite.9)
In the studied steel, upper bainite is formed at temperatures
ranging from 525 to 475 C (See Figs. 2(d) and 3(a)(b)),
whereas lower bainite is present in microstructures formed at
temperatures between 425 C and 355 C (See Figs. 3(d)(f)
and Fig. 1(b)). A mixed microstructure of both forms of
bainite is observed at 450 C as shown in Fig. 3(c). It is noted
that a lower bainite plate contains one major and some minor
variants of carbides precipitates within bainitic ferrite in the
right-hand side of the micrograph. However, in the left-hand
side, upper bainite sheaves can be found. It can be then
concluded that the transition from upper to lower bainite
occurs at a temperature about 450 C.
Table 2 lists the Ms temperature of the studied steel
calculated from dierent empirical formulas,14,1823) thermo-
dynamic theory,24) and neural network analysis,2527) in
comparison with the Ms experimental value. The free energy
change for the transformation of austenite to ferrite of the
same composition, G
, input in Ghosh and Olson model24)
was calculated using MTDATA.32) Likewise, the articial
neural network method27) requires the PAGS as an input
(39 mm in the studied steel). Calculated values in Table 2
suggest that thermodynamic theory and articial neural
network analysis are accurate methods to estimate Ms
temperature. Likewise, empirical formula developed by
Fig. 1 Micrographs of microstructures obtained by isothermal transforma- Carapella, and Rowland and Lyle predict for the studied
tion for 900 s at (a) 340 C, light optical microscopy (LOM), (b) 355 C, steel a value of Ms quite close to the experimental value.
SEM. On the other hand, Table 3 lists the Bs temperature of the
 Isothermal Bainite Formation in a Medium-Carbon Steel 3275

Fig. 2 Micrographs of microstructures obtained by isothermal transformation at 550 C for 1800 s: (a) LOM and (b) SEM; at 525 C for
900 s: (c) LOM and (d) SEM; and at 525 C for 1800 s: (e) LOM and (f) SEM. PF is proeutectoid ferrite; P is pearlite; UB is upper bainite;
and DP is degenerated pearlite.

studied steel calculated from dierent empirical formu- empirical formulas tested fail in the prediction of Bs
las,14,15,17) and bainite theory.33) If the process of bainite temperature.
reaction is consider diusionless, the highest temperature at Finally, Takahashi and Bhadeshia model9) has been used to
which bainite can grow should be below the To line. The To estimate the LBs temperature of the studied steel. In this
concept was rst introduced by Zener13) and modied by sense, the time required to decarburise supersaturated ferrite,
Aaronson et al.34,35) A further calculation was carried out by td , and the necessary time to obtain a detectable amount of
Bhadeshia and Edmonds36) allowing for the eect of elastic cementite precipitation in the ferrite, t , as a function of
strain due to transformation, giving a strain energy of about decomposition temperature were calculated. There is no
400 J mol
1 (To0 line). A way of predicting Bs temperature is reliable method for the calculation of t as a function of
to introduce the universal nucleation function (GN ), which temperature. Here, data obtained by Takahashi and
was determined experimentally by Ali and Bhadeshia,37) in Bhadeshia9) corresponding to a Fe-0.3 mass%C plain carbon
addition to the growth criterion (To0 ) described previously. steel were used. On the other hand, eq. (1) was used for the
Nucleation of bainite is possible only when the maximum determination of td . In this equation, the diusivity of carbon
free energy change accompanying the nucleation (GM ) is in austenite, D
, was calculated using Babu and Bhadeshia
larger than GN in absolute value. Thus, the Bs temperature for model,38) and the carbon concentrations in ferrite and
a given alloy can be predicted to be the lower value of its austenite, x
 and x
, respectively, were determined using
corresponding To0 temperature and the temperature at which MTDATA in paraequilibrium conditions.32) Likewise, the
jGM j > jGN j. Bs calculated values in Table 3 suggest that bainitic ferrite plate thickness, w, was measured for the
Bhadeshia method based on bainite theory is an accurate studied steel at dierent transformation temperature on
method to estimate Bs temperature. In this case, all the scanning electron micrographs. An average w value of
3276 F. G. Caballero, M. J. Santomia, C. Garca-Mateo and C. Garca de Andres

Fig. 3 Scanning electron micrographs of microstructures obtained by isothermal transformation for 900 s at: (a) 500 C, (b) 475 C,
(c) 450 C, (d) 425 C, (e) 400 C and (f) 375 C.

Table 2 Calculation of Ms temperature. Table 3 Calculation of Bs temperature.

Martensite start temperature, Bainite start temperature

Empirical formulas or Model Empirical formulas or Model
Ms / C Bs / C
Experimental value 342 Experimental value 525
Empirical Formulas: Empirical Formulas:
Payson and Savage18 355 Steven and Haynes14 621
19
Carapella 348 Zhao et al.15 559
Rowland and Lyle20 351 Bodnar et al.17 570
Grange and Stewart21 376 Bainite Theory:
Nehrenberg22 361 Bhadeshia33 515
14
Steven and Haynes 369
Andrews23 363
Thermodynamic Theory:
Ghosh and Olson24 349 the maximum temperature at which t < td is, approximately,
Neural Network Analysis: 370 C. This temperature would correspond to the transition
Capdevila et al.27 350 temperature from upper to lower bainite. The calculated LBs
temperature is quite low in comparison to the experimental
value, 450 C. A more precise determination of t , including
0.18 mm was estimated. Results of td along with t values are the inuence of alloying elements in calculation, is indis-
shown in Fig. 4. pensable for the estimation of LBs temperature. Moreover,
Comparing both td and t values, it can be concluded that the reactions of partition of carbon into the residual austenite
 Isothermal Bainite Formation in a Medium-Carbon Steel 3277

1000 0.35

Bainite PlateThickness, w /m
Lower Bainite Upper Bainite
100 0.30
td
0.25
10
Time, t /s

t 0.20
1 Ms 0.15
0.1 0.10
LB s calculated Experimental values
0.01 0.05
LB s experimental Bs According to Ref. [40]
0.00
0.001
350 400 450 500 550
0 100 200 300 400 500 600 Temperature, T /C
Temperature, T /C
Fig. 5 Thickness of bainite plates as a function of transformation temper-
Fig. 4 Calculated curves of td and t times. Square symbols correspond to ature.
experimental Ms , Bs and LBs temperatures.

Bhadeshia.40) The model accurately ts the experimental

and the precipitation of carbides in the bainite should interact. results even out of its range of validity. As expected, w
Partitioning reduces the amount of carbon available for increases with the isothermal transformation temperature.
precipitation and vice versa. This eect should be taken into Thus, results suggest that the origin of this trend is in the
account in a more reliable theory for the transition from upper growth mechanisms of bainitic ferrite sub-units and is
to lower bainite. independent of carbide precipitation processes. There is a
plastic relaxation in the austenite adjacent to the formed
3.2 Scale of microstructure bainite plates and the dislocations generated resist the
In the absence of precipitation, the thickness of bainite advance of the bainite/austenite interface, the resistance
plates depends primarily on the strength of the austenite at the being greatest for strong austenite. As the yield strength of
transformation temperature, and the chemical free energy the austenite decreases with increasing temperatures, it is
change accompanying transformation.39) This applies specif- expected that the plates become thicker. Signicant dif-
ically when the shape deformation causes plastic strain in the ferences in this behaviour have not been detected between
adjacent austenite. Strong austenite or a large driving force upper and lower bainite as Fig. 5 shows.
results in ner plates, the former because there is a larger Two main factors contribute to the hardness of a fully
resistance to interface motion and the latter because an bainitic microstructure: the plate thickness of bainitic ferrite
increased nucleation rate leads to microstructural renement. and the presence of cementite particles inside or between
Temperature by itself does not have much of an eect. In this ferrite plates. Figure 6 shows the hardness of specimens fully
sense, Singh and Bhadeshia40) developed a model using transformed to bainite as a function of isothermal trans-
articial neural network for the determination of the thick- formation temperature. As expected, the lower the trans-
ness of bainite plates, including the eect of temperature, formation temperatures the higher the hardness of bainitic
austenite strength and driving force for nucleation of bainite. microstructure. This trend is explained by the decrease in the
The validity range of temperature for the model is 260
460 C.
In order to nd the role that cementite precipitation might 450
play in the nal thickness of bainitic plates, Singh and Lower Bainite Upper Bainite
Bhadeshia model was used to calculate the thickness of
400
Vickers Hardness

bainite plates in the studied steel beside cementite precip-

itation occurs during bainite formation. The driving force for
bainite nucleation was determined using thermodynamic 350
theory33) and the austenite strength was calculated using
Young and Bhadeshia empirical model,41) where the yield 300
strength of austenite is expressed as a function of the
transformation temperature and the chemical composition of 250
the alloy.
The measured plate thickness w is plotted as a function of 200
transformation temperature in Fig. 5 and compared with the
350 400 450 500 550
results of the model above described. Experimental error bars Temperature, T /C
corresponds to the standard deviation on the thickness of
400 bainite plates. Moreover, an average error of 0.03 mm Fig. 6 Vickers hardness of specimens fully transformed to bainite as a
can be considered for calculations according to Singh and function of isothermal transformation temperature.
3278 F. G. Caballero, M. J. Santomia, C. Garca-Mateo and C. Garca de Andres

1 600 1 600
Vb dilatometry Vb dilatometry

Vickers Hardness
0.8 Vb metallography 0.8

Vickers Hardness
Vb metallography
Bainite Fraction

Bainite Fraction
HV 500 HV
500

0.6 (a) 0.6 (b)

400 400
0.4 0.4

300 300
0.2 0.2

0 200 0 200
1 10 100 1000 1 10 100 1000
Time, t /s Time, t /s
1 600 1 600
Vb dilatometry Vb dilatometry

Vickers Hardness

Vickers Hardness
0.8 0.8
Bainite Fraction

Bainite Fraction
Vb metallography Vb metallography
500 HV
500
HV
0.6 (c) 0.6 (d)
400 400
0.4 0.4

300 300
0.2 0.2

0 200 0 200
1 10 100 1000 1 10 100 1000
Time, t /s Time, t /s
1 600 1 600

Vb dilatometry Vb dilatometry

Vickers Hardness
Vickers Hardness

0.8 0.8
Bainite Fraction

Vb metallography
Bainite Fraction

Vb metallography
500 HV
500
HV
0.6 (e) 0.6 (f)
400 400
0.4 0.4

300 300
0.2 0.2

0 200 0 200
1 10 100 1000 1 10 100 1000
Time, t /s Time, t /s
1 600
Vb dilatometry
Vickers Hardness

0.8
Bainite Fraction

Vb metallography
HV 500

0.6 (g)
400
0.4

300
0.2

0 200 Fig. 7 Isothermal transformation kinetics for the studied steel at

1 10 100 1000 (a) 525 C, (b) 500 C, (c) 475 C, (d) 450 C, (e) 425 C, (f) 400 C
and (g) 375 C.
Time, t /s

plate thickness of bainitic ferrite (Fig. 5) and the increase in ning electron micrographs, whereas solid line represent the
the density of carbides present in the microstructure. Results evolution of the volume fraction of bainite with time
also suggest a signicant increase in hardness when transition determined from dilatometric curves using eq. (2). Both
from upper to lower bainite occurs due to the precipitation of procedures essentially lead to the same results. Fully bainitic
ner carbides within bainitic ferrite plates. microstructures are obtained at all the tested temperature.
Likewise, squares in Fig. 7 represent the evolution of
3.3 Overall transformation kinetics hardness as a function of time. Hardness decreases as bainite
Isothermal kinetics for the studied steel in the temperature transformation proceeds since the volume fraction of mar-
range 525375 C is presented in Fig. 7. Close cycles in this tensite is reduced in the microstructure.
gure correspond to the volume fraction of bainite measured As an example of the evolution of the volume fraction of
by a systematic manual point-counting procedure on scan- bainite with time, the optical micrographs corresponding to
 Isothermal Bainite Formation in a Medium-Carbon Steel 3279

Fig. 8 Evolution of the volume fraction of bainite with time at 375 C: (a) 3 s, (b) 6 s, (c) 15 s, (d) 30 s, (e) 60 s and (f) 300 s.

the isothermal treatments carried out at 375 C are shown in Therefore the value of n at the dierent tested temperatures is
Fig. 8. After 3 s of holding time at this temperature, bainite given by the slope of lnln1=1
 vs. ln t plots of the
transformation has not started and a fully martensitic micro- overall transformation data shown in Fig. 7. In this analysis,
structure is obtained by quenching (Fig. 8(a)). The rst the data were limited to those having a value of lnln1=1

sheaves of bainite nucleated on prior austenite grain  less than zero, that is  < 0:6, where linearity between
boundaries are observed after 6 s of holding time (Fig. 8(b)). lnln1=1
 and ln t is well preserved.42) Figure 9 repre-
Figs. 8(c)(e) show intermediate stages in the transforma- sents the values of n as a function of the transformation
tion. Transformation is completed after 300 s of holding time temperature. It is clear that isothermal kinetics exhibit a
when a fully bainitic microstructure is obtained (Fig. 8(f)). constant Avrami exponent close to 2 in the whole inves-
The conventional Avrami equation is generally used to tigated range. This value indicates that bainitic reaction can
describe the overall transformation kinetics under isothermal be described by a linear growth of ferrite plates nucleated at
conditions: austenite grain surfaces with a constant rate.43) Growth is
indeed considered to occur mainly in one dimension assum-
 1
expf
btn g 3
ing that thickening of the plates can be neglected compared to
where  is the volume fraction of austenite transformed their lengthening. Dierences in the n values corresponding
within holding time t, n is the Avrami exponent indicative of to upper and lower bainite have not been found.
the transformation mechanism, and b is a temperature About the temperature dependence of the isothermal
dependent factor. Equation (3) can be rewritten as: bainite transformation rate, it has been reported44) that there
   exists an overall activation kinetics given by:
1
ln ln n ln t ln b 4
1

3280 F. G. Caballero, M. J. Santomia, C. Garca-Mateo and C. Garca de Andres

6 550
Lower Bainite Upper Bainite
5
500 Bs
Avrami exponent, n

Temperature, T /C
4

3 450
LB s
2
400
1 5% bainite
350 50% bainite
0 Ms
350 400 450 500 550
Temperature, T /C 300
0.1 1.0 10.0 100.0 1000.0
Fig. 9 Value of Avrami exponent n as a function of the transformation Time, t /s
temperature.
Fig. 10 5 and 50% bainite TTT curves for the studied steel.

Table 4 Overall activation energy Q0:5 for 50% reaction.

Overall activation energy slight acceleration of transformation to lower bainite as Ms

Microstructure for 50% reaction, temperature is approached.8)
Q0:5 /kJmol
1
Upper bainite (525475 C) 17 4. Conclusions
Lower bainite (425375 C) 16
A single type of C-curve has been found in the TTT
diagram within the bainite temperature range for the studied
  steel. Distinct reaction C-curves were not observed for both
@ ln t
Q R 5 types of microstructure, upper bainite and lower bainite in the
@1=T 
TTT diagram. Experimental results on the kinetics of the
where t is the time to reach a certain transformed fraction , T isothermal formation of bainite at dierent temperature have
is the isothermal holding temperature and R is the gas demonstrated that both type of microstructure, upper and
constant. The term  is usually taken as 0.5 in most lower bainite, possesses similar overall transformation
literature12) and the value of Q0:5 is calculated with the slope kinetics. Finally, it has been shown that phase transformation
of the straight line ln t vs. 1=T. Table 4 lists Q0:5 values theory can be used to calculate with a very good agreement
determined for upper and lower bainite in the studied steel. the martensite and bainite start temperatures, and the thick-
Overall kinetics data at 450 C were not considered in the ness of bainitic ferrite plates, but a more precise determi-
calculations since a mixed microstructure was obtained at nation of the transition temperature from upper to lower
this temperature. Similar Q0:5 values for upper and lower bainite is required.
bainite were found, but lower than those reported else-
where.12,44) This might indicated that the rate of overall Acknowledgements
transformation in the steel studied is higher. It is clear from n
and Q0:5 results (Fig. 9 and Table 4) that upper and lower The authors acknowledge nancial support from Spanish
bainite present similar overall transformation kinetics. Ministerio de Ciencia y Tecnologa ((project-MAT 2001-
Figure 10 shows the 5 and 50% bainite TTT curves 1617)). F.G. Caballero would like to thank Spanish Minister-
determined by dilatometry for the studied steel. Solid lines in io de Ciencia y Tecnologa for the nancial support in the
this graph mark the tendency of experimental values (circles form of a Ramon y Cajal contract (Programa RyC 2002). M.J.
symbols) for the bainite curves taking into account the error Santofmia would also like to express her gratitude to
in time (2 seconds) associated to the determination of the Consejo Superior de Investigaciones Cientcas (CSIC) for
volume fraction of bainite using dilatometric analysis the nancial support in the form of a PhD Research Grant
(dashed lines). By contrast to previous reports,8,45) TTT (I3P Program).
curves appear to exhibit a single C-curve within the bainite
temperature range for the studied steel for the upper and
REFERENCES
lower bainite. Distinct reaction C-curves were not clearly
observed for both types of microstructure, upper bainite and 1) H. K. D. H. Bhadeshia and J. W. Christian: Metall. Trans. A 21 (1990)
lower bainite in the TTT diagram. This is consistent with n 76797.
and Q0:5 results found for upper and lower bainite. A 2) H. K. D. H. Bhadeshia: Acta Metall. 29 (1981) 11171130.
signicant acceleration of lower bainite overall transforma- 3) H. K. D. H. Bhadeshia and A. R. Waugh: Acta Metall. 30 (1982) 775
784.
tion kinetics near the LBs temperature compared with upper 4) L. C. Chang and H. K. D. H. Bhadeshia: Mater. Sci. Eng. A 184 (1994)
bainite formed just above LBs temperature is not observed. L17L20.
However, transformation curves indicate that there may be a 5) I. Stark, G. D. W. Smith and H. K. D. H. Bhadeshia: Solid-Solid Phase
 Isothermal Bainite Formation in a Medium-Carbon Steel 3281

Transformations, ed. by G. W. Lorimer (Institute of Metals, London, lvarez:

28) C. Garca de Andres, F. G. Caballero, C. Capdevila and L. F. A
1988) pp. 211215. Mater. Charact. 48 (2002) 101111.
6) S. J. Matas and R. F. Hehemann: Trans. Met. Soc. AIME 221 (1961) 29) G. F. Vander Voort: Metallography. Principles and Practice,
179185. (McGraw-Hill Book Company, New York, U.S.A., 1984) pp. 426428.
7) H. K. D. H. Bhadeshia: Acta Metall. 28 (1980) 11031114. 30) D. B. Wiles and R. A. Young: J. Appl. Cryst. 14 (1981) 149151.
8) L. C. Chang: Mat. Sci. Eng. A 368 (2004) 175182. 31) C. Garca de Andres, M. J. Bartolome, C. Capdevila, D. San Martn,
9) M. Takahashi and H. K. D. H. Bhadeshia: Mater. Sci. Technol. 6 (1990) F. G. Caballero and V. Lopez: Mater. Charact. 46 (2001) 389398.
592603. 32) MTDATA: Phase Diagram Calculation Software, National Physical
10) B. L. Bramtt and J. G. Speer: Metall. Trans. A 21 (1990) 817829. Laboratory, Teddington, UK, 2004.
11) H. Ohtani, S. Okaguchi, Y. Fujishiro and Y. Ohmori: Metall. Trans. A 33) H. K. D. H. Bhadeshia: Materials Algorithms Project, www.msm.
21 (1990) 877888. cam.ac.uk/map/steel/programs/mucg46-b.html.
12) M. K. Kang, D. M. Chen, S. P. Yang and G. L. Hu: Metall. Trans. A 23 34) H. I. Aaronson, H. A. Domain and G. M. Pound: Trans. Metall. Soc.
(1992) 785795. AIME 236 (1966) 753767.
13) C. Zener: Trans. AIME 167 (1946) 550595. 35) H. I. Aaronson, H. A. Domain and G. M. Pound: Trans. Metall. Soc.
14) W. Steven and A. G. Haynes: J. Iron Steel Inst. 183 (1956) 349359. AIME 236 (1966) 768780.
15) Z. Zhao, C. Liu, Y. Liu and D. O. Northwood: J. Mater. Sci. 36 (2001) 36) H. K. D. H. Bhadeshia and D. V. Edmonds: Acta Metall. 28 (1980)
50455056. 12651273.
16) L. C. Chang: Metall. Mater. Trans. A 30 (1999) 909915. 37) A. Ali and H. K. D. H. Bhadeshia: Mater. Sci. Technol. 6 (1990) 781
17) R. L. Bodnar, T. Ohhashi and R. I. Jaee: Metall. Trans. A 20 (1989) 784.
14451460. 38) S. S. Babu and H. K. D. H. Bhadeshia: J. Mater. Sci. Lett. 14 (1995)
18) P. Payson and C. H. Savage: Trans. ASM 33 (1944) 261280. 314316.
19) L. A. Carapella: Metal. Progress 46 (1944) 108118. 39) S. B. Singh and H. K. D. H. Bhadeshia: Mater. Sci. Eng. A 245 (1998)
20) E. S. Rowlan and S. R. Lyle: Trans. ASM 37 (1946) 2747. 7279.
21) R. A. Grange and H. M. Stewart: Trans. AIME 167 (1946) 467490. 40) S. B. Singh and H. K. D. H. Bhadeshia: Materials Algorithms Project,
22) A. E. Nehrenberg: Trans. AIME 167 (1946) 494498. www.msm.cam.ac.uk/map/neural/programs/plate-b.html.
23) K. W. Andrews: J. Iron Steel Inst. 203 (1965) 721727. 41) C. H. Young and H. K. D. H. Bhadeshia: Mater. Sci. Technol. 10 (1994)
24) G. Ghosh and G. B. Olson: Acta Metall. Mater. 42 (1994) 33613370. 209214.
25) C. Capdevila, F. G. Caballero and C. Garca de Andres: Mater. Sci. 42) A. Matsuzaki and H. K. D. H. Bhadeshia: Mater. Sci. Technol. 15
Technol. 19 (2003) 581586. (1999) 518522.
26) C. Capdevila, F. G. Caballero and C. Garca de Andres: ISIJ Int. 42 43) D. Quidort and Y. J. M Brechet: ISIJ Int. 42 (2002) 10101017.
(2002) 894902. 44) M. Umemoto, K. Horiuchi and I. Tamura: Trans. Iron Steel Inst. Jpn. 22
27) C. Capdevila, F. G. Caballero and C. Garca de Andres: Materials (1982) 854861.
Algorithms Project, www.msm.cam.ac.uk/map/steel/programs/MS. 45) H. K. D. H. Bhadeshia and D. V. Edmonds: Metall. Trans. A 10 (1979)
neural.htm. 895907.