MATERIALS CHARACTERIZATION 44:291–299 (2000)© Elsevier Science Inc., 2000. All rights reserved. 1044-5803/00/$–see front matter655 Avenue of the Americas, New York, NY 10010 PII S1044-5803(99)00060-1

291

M

icrostructural Characterization of Bainitic Steel Submitted to Torsion Testing and Interrupted Accelerated Cooling

A. B. Cota* and D. B. Santos

†

*Federal University of Ouro Preto, Department of Physics, Ouro Preto, M.G., Brazil; and

†

Federal University of Minas Gerais, Department of Metallurgical and Materials Engineering,Belo Horizonte, M.G. Brazil

HSLA low-carbon bainitic steel containing B was submitted to torsion tests to simulate con-trolled rolling, followed by interrupted accelerated cooling. Microstructural characteristicsand the mechanisms for the refinement of structure were evaluated using light microscopy,scanning electron microscopy, transmission electron microscopy, and Vickers hardness test-ing. The final microstructure was found to contain complex mixture of granular bainite,small islands of MA constituent, bainitic ferrite, and polygonal ferrite. Increasing the cool-ing rate or decreasing the finish cooling temperature resulted in a decrease in the volumefraction and average size of the MA islands and the polygonal ferrite. A finish cooling tem-perature of 400

8

C produced a microstructure consisting of fine laths of bainitic ferrite withan interlath MA constituent. A quantitative relationship between the accelerated coolingvariables and the ferrite grain size was developed. © Elsevier Science Inc., 2000. All rights

reserved.

INTRODUCTION

In the last 2 decades there have been con-siderable advances in the theory and prac-tice of accelerated cooling after controlledrolling. Several investigations have beenconducted with the objective of studyingthe influence of cooling variables on the mi-crostructure and mechanical properties ofHigh-Strength Low-Alloy (HSLA) low-car-bon bainitic steels [1–6]. The study of themicrostructural characteristics of these steelshas been carried out using the classificationsystem developed for bainitic and ferriticmicrostrutures by Krauss and Thompson[7]. Most of the bainitic microstructures ob-served in the low-carbon microalloyedsteels can be well described as bainitic fer-rite (acicular ferrite) and granular bainite[with discreet islands of MA (martensiteand

/

or residual austenite)].

The MA islands formed in granular bain-ite during the continuous cooling of steelshave a strong effect on the mechanicalproperties [8]. This constituent leads to in-creases in the tensile strength but a deterio-ration in the impact toughness. It also has asmall effect on the yield strength. For thesereasons, it is very important to understandthe influence of cooling variables on theformation of MA islands in the granularbainite.

In the present work, an alternativemethod, based on hot torsion testing tosimulate controlled rolling and on gas-accel-erated cooling, was used to study the influ-ence of cooling rate (

CR

) and finish coolingtemperature (

T

FC

) on the microstructure ofHSLA low-carbon bainitic steel recently de-veloped by the metallurgical industry. Themicrostructure refinement mechanism andthe MA islands (granular or equiaxed mor-

292

A. B. Cota and D. B. Santos

phology) formation mechanism are dis-cussed. A quantitative relationship betweenthese cooling variables and the ferrite grainsize was obtained by multiple regressionanalysis.

EXPERIMENTAL PROCEDURE

The chemical composition of the steel usedin this investigation is given in Table 1. Tu-bular torsion specimens, having a gaugelength of 16.5mm, 6.5mm outer, and 2mminner diameter, were used to study micro-structural characteristics after interruptedaccelerated cooling. A thermocouple, con-nected to an interface card installed in apersonal computer, was placed inside thespecimen, allowing control of the thermalprofiles during thermomechanical process-ing and data collection of the coolingcurves.

The specimens were deformed using anine-pass torsion test. They were reheatedto 1200

8

C for 15 min and, while coolingdown at a rate of 1

8

C

/

s, they were submit-ted to five passes at temperatures above thenonrecrystallization temperature (

T

n

r

5

970

8

C), followed by four passes below thistemperature. A strain per pass of

e

5

0.2and a strain rate of

5

2 s

2

1

were used. De-tails of the temperature of each pass duringof the hot torsion experiments are given inFig. 1.

The continuous cooling transformation(CCT) diagram of this steel (Fig. 2) was de-termined by thermal analysis of the coolingcurves from deformed austenite accordingto the nine-pass torsion test schedule de-scribed above [9]. This diagram was uti-lized in the selection of the cooling ratesand finish cooling temperatures used afterthe thermomechanical processing, and are

e •

also indicated in Fig. 1. The cooling ratesmeasured correspond to average values forthe temperature range of 800 to 500

8

C.The cooling rate and the finish cooling

temperature after deformation were con-trolled by means of a cooling device usinghelium and a control panel, which permittedclose regulation of the helium flow rate [10].

Samples for metallographic observationand Vickers hardness (VH) measurements(4.905N load) were sectioned longitudi-nally, producing a plane tangential to thegauge length to the torsion specimens. Twopercent nital and LePera’s etchants [11]were used to examine the microstructures.The samples were observed by light mi-croscopy (LM) and scanning electron mi-croscopy (SEM). Volume fractions and di-mensions analyses of the constituents wereperformed using computerized image anal-ysis, IMAGE PRO-PLUS

TM

, from light mi-croscopy. More detailed microstructuralcharacterization, for selected cooling ratesand finish cooling temperatures, was con-ducted using transmission electron micros-copy (TEM) of thin foils in a bright-fieldmode.

RESULTS AND DISCUSSION

MICROSTRUCTURE

Figure 3 shows the microstructure corre-sponding to a specimen quenched after thelast pass of the nine-pass test. Elongatedaustenite grains are observed, revealingthat accumulation of deformation has takenplace during hot working in the low-tem-perature range. The elongated grains have,in the plane of observation, an approximateaspect ration of 4:1, given an average valueof 40

m

m.

Table 1

Chemical Composition of the Steel (wt. %)

C Mn Si P Al Nb V Ti Ni B S N

0.08 1.70 0.25 0.021 0.029 0.033 0.058 0.026 0.17 0.0024 0.002 0.0048

Bainitic Steel

293

The combined effects of cooling rate andfinish cooling temperature on the micro-structure are illustrated in Figs. 4 (LM pho-tomicrographs) and 5 (SEM photomicro-graphs). It can be seen from Figs. 4(a) and5(a) that the microstructure associated with

the highest finish cooling temperature(

T

FC

5

650

8

C) exhibits a mixture of finepolygonal ferrite (with a volume fraction

<

12%), granular bainite, and small islands ofthe MA constituent. The MA constituenthas a granular or equiaxed morphology(volume fraction

<

4.5% and with islands

FIG. 1. Schematic representation of the controlled rolling and accelerated cooling simulation.

FIG. 2. CCT diagram of the HSLA bainitic steel deter-mined via thermal analysis of the cooling curves fromdeformed austenite using nine-pass torsion testing [9].

FIG. 3. Light photomicrograph of the austenitic micro-structure of sample quenched after the last pass of thenine-pass test.

294

A. B. Cota and D. B. Santos

of average size less the 2

m

m). By contrast,the microstructures associated with the nextlower finish cooling temperature (

T

FC

5

500

8

C) [Figs. 4(b) and 5(b)] consist of granu-lar bainite and bainitic ferrite. A smallamount of polygonal ferrite is also in evi-dence in these microstructures.

In the case of the two higher finish cool-ing temperatures (

T

FC

5

650

8

C and T

FC

5

600

8

C), the MA islands are distributedpractically uniformly throughout the bain-ite matrix (Fig. 6). This figure, a compila-tion of binary images in which the MA is-lands appear white, shows that the increasein cooling rate or decrease in finish coolingtemperature implies a decrease in the vol-ume fraction and the average size of theMA islands.

Figures 4(c)–(d) and 5(c)–(d) show that,for the lowest finish cooling temperature,

T

FC

5

400

8

C, the microstructure is essen-tially bainitic, with fine laths of bainitic fer-

rite and interlath MA constituent. It can beseen from Fig. 5(c),

CR

5

6.3

8

C

/

s, that themicrostructure contains a small amount ofpolygonal ferrite. Figure 7(a)–(d), TEM mi-crographs of samples cooled at rates of6.3

8

C

/

s and 33

8

C

/

s for

T

FC

5

400

8

C, showsthat the width of the bainitic ferrite lathsdecreases with increasing cooling rate, andthat the widths can be as narrow as 0.2

m

m.No evident interlath carbides were ob-served in this steel under present condi-tions. The refinement of bainitic ferrite byincreasing the cooling rate could be due tothe following: when the cooling rate in-creases, the reduction in the bainite trans-formation start temperature,

B

s

, leads to anincrease in the driving force (the differencein free energy between the austenite phaseand the ferrite phase) for the nucleationrate of subunits of ferrite and, conse-quently, to a decrease in the width of thebainitic ferrite laths. The increase in the

FIG. 4. Light photomicrographs of samples cooled at different rates and with different finish cooling tempera-tures: (a) 6.38C/s, 6508C; (b) 13.38C/s, 5008C; (c) 6.38C/s, 4008C; (d) 338C/s, 4008C.

Bainitic Steel

295

cooling rate leads to a continuous increaseinthe Vickers hardness from 262 to 320.

The analysis of the optical, SEM, andTEM photomicrographs reveals that thecooling variable of greater influence on thebainite morphology is the finish coolingtemperature. When the finish cooling tem-perature decreases from 650 to 400

8

C, it canbe observed that the ferrite crystals changefrom their granular morphology with thepresence MA islands (by default, these is-lands have granular or equiaxed shapes re-solvable in light micrographs) to finer fer-rite crystals with laths morphology (MAconstituent is retained between the ferritelaths with acicular morphology).

Figures 8(a)–(b) and 9(a)–(b) show the in-fluence of the cooling rate on the volumefraction and average size of the MA islandsand the polygonal ferrite, associated withthe various finish cooling temperatures, re-spectively. The results show that the in-

crease in the cooling rate or the decrease inthe finish temperature leads to a decreasein the volume fraction and the average sizeof the MA islands and the polygonal ferrite.For

T

FC

5

400

8

C no MA islands were de-tected.

FERRITE GRAIN SIZE

A quantitative relationship between thecooling variables and the ferrite grain sizewas obtained by means of multiple regres-sion analysis. The statistical treatment wascarried out by taking the cooling rate andthe finish cooling temperature as indepen-dent variables. The following equation wasobtained:

(1)

where

R

is the multiple correlation coeffi-cient,

d

a

is ferrite grain size. The values en-closed in the parentheses indicate 95% of

dα 0.12 CR( ) 0.27– TFC( )0.67 R 97.2%==

FIG. 5. SEM photomicrographs of samples cooled at different rates and with different finish cooling temperatures:(a) 6.38C/s, 6508C; (b) 13.38C/s, 5008C; (c) 6.38C/s, 4008C; (d) 338C/s, 4008C.

296 A. B. Cota and D. B. Santos

confidence level. Figure 10 shows that thecorrespondence between measured and pre-dicted [Eq. (1)] ferrite grain size is quite close,and discrepancies observed are very small.

The effectiveness of cooling rate on theferrite grain size appears to be reduced asthe finish cooling temperature is decreased.The grain size is reduced from 5.6 to 3.7mmfor an increase in the cooling rate from 6.3to 338C/s for TFC 5 6508C. Under the same

conditions for TFC 5 5008C, da is reducedfrom 4.6 to 3.2mm. From Eq. (1), when thecooling rate increases tenfold, the ferritegrain size is reduced to almost half (0.54).

MA CONSTITUENT

The formation of MA islands is a processcontrolled by carbon diffusion. Therefore, it

FIG. 6. Binary images of samples cooled at different rates and with different finish cooling temperatures: (a) 6.38C/s,6508C; (b) 13.08C/s, 6508C; (c) 338C/s, 6508C; (d) 6.38C/s, 5008C; (e) 13.38C/s, 5008C; (f) 338C/s, 5008C. LePera’setchant; MA constituent is white.

Bainitic Steel 297

was expected that both the cooling rate andfinish cooling temperature would influencethe average size and volume fraction of theMA islands. The displacive model has beenproposed and accepted [12–15] for the ki-netics of bainite transformation. In thismodel the nucleation of ferrite occurs by adiffusional mechanism, and longitudinal

growth takes place by a displacive mecha-nism. The growth of ferrite laths (or plates)is controlled by the repeated nucleation ofsubunits (individual ferrite laths), and itsnucleation rate strongly depends on thedriving force for the phase transformationreaction. Diffusionless growth requires thattransformation occurs at a temperature be-

FIG. 7. TEM photomicrographs of samples cooled at different rates but with the same finish cooling temperature:(a) 6.38C/s, 4008C; (b) 338C/s, 4008C.

FIG. 8. Influence of cooling rate on: (a) the volumefraction of the MA islands (fMA); (b) the average size ofthe MA islands (dMA).

FIG. 9. Influence of cooling rate on: (a) the volumefraction of the polygonal ferrite (fPF); (b) the averagesize of the polygonal ferrite (da).

298 A. B. Cota and D. B. Santos

low To, when the free energy of bainite be-comes less than that of austenite of thesame composition. The To curve definestemperatures at which austenite and ferriteof the same composition have equal freeenergy (Fig. 11). While the austenite–bain-ite transformation proceeds, any excess car-bon is rejected into the residual austenite.When carbon-enriched residual austenite iscontinuously cooled to a temperature be-low the martensite transformation starttemperature (Ms), it will transform to anMA constituent.

An increase in the cooling rate causes adecrease in the bainite transformation starttemperature (Bs) (Fig. 2) and, consequently,an increase in the difference of free energybetween austenite and bainite, as shown inFig. 11. The free energy of the austenite willrise from g1 to g2, and the To curve shiftsfrom To1 to To2. This results in an increasedamount of bainite and in a smaller amountof retained austenite, which subsequentlytransform into MA islands. This can ex-plain why the amount of the MA islandsdecreases when the cooling rate increases.

The finish cooling temperature also influ-ences the dimensions and volume fractionof MA islands. For a given cooling rate andfor a higher finish cooling temperatures(e.g., 6008C), bainite has not yet started to

form, or is only in the very initial stages ofdevelopment. During the subsequent slowcooling in air (18C/s), the sample remainsfor a long enough time at temperatureswhere diffusion of carbon atoms over longdistances is still possible. The bainite trans-formation, already initiated during acceler-ated cooling, can then continue until zonesof stable carbon-enriched austenite areformed, which will become lower tempera-ture MA islands. When the finish coolingtemperature is smaller, the formation ofbainite is already at an advanced stage atthe point of interruption of the acceleratedcooling, and is brought to completion dur-ing the subsequent air cooling. In this case,the sample remains a very small time attemperatures where diffusion of carbon at-oms over long distances is still possible.During the austenite–bainite transforma-tion the carbon is rejected into the adjacentresidual austenite, reducing the volumefraction and the size of the MA islands.

CONCLUSIONS

The effects of interrupted accelerated cool-ing on the microstructural characteristics ofthe HSLA low-carbon bainitic steel were

FIG. 10. Correlation between measured and predictedferrite grain size (da).

FIG. 11. Schematic dependence of carbon concentra-tion vs. free energy of austenite (g, and bainite (aB)and To temperature (after Honeycombe et al. [13]).

Bainitic Steel 299

studied, and the following conclusionswere reached.

For the conditions of accelerated coolingapplied in the present work, the final mi-crostructure consists of a complex mixtureof granular bainite, small MA islands, bai-nitic ferrite, and polygonal ferrite. For TFC 54008C, the microstructure is essentially bai-nitic, with fine laths of bainitic ferrite withan interlath MA constituent. No evident in-terlath carbides could be observed in thissteel.

The increase in the cooling rate or the de-crease in the finish cooling temperature re-sults in decreasing volume fraction andaverage size of the MA islands and the poly-gonal ferrite.

A quantitative relationship between thecooling variables and ferrite grain size isproposed. The empirical equation showsthat the cooling rate has a stronger influ-ence than the finish cooling temperature onthe ferrite grain size.

The authors are grateful for the financialsupport of CNPq and FAPEMIG. The authorsalso thank Usiminas S.A. for supplying thematerial used in this work.

References

1. P. Bufalini, M. Prontemolli, M. Chersi, A. Aprile,and C. Jannone: Optimizing the Microstructure andthe Strengthening Factors in High Strength PlatesProduced by Accelerated Cooling. Proc. Int. Conf. onAccelerated Cooling of Steels, Pittsburgh, PA, pp.387–400 (1985).

2. L. E. Collins, G. E. Ruddle, A. F. Crawley, and J. D.Boyd: Accelerated Cooling of a Nb-V MicroalloyedPlate Steel. Proc. Int. Conf. on Accelerated Coolingof Rolled Steels, Winnipeg, Canada, pp. 57–70(1987).

3. M. Pontremoli, P. Bufalini, A. Aprile, and C. Jan-none: Development of grade APIX80 pipeline steelplates produced by controlled rolling. Metals Tech-nol. 11:504–514 (1984).

4. P. C. Rodrigues, A. B. Cota, and D. B. Santos: Effect

of Cooling Conditions on the Microstructure and Me-chanical Properties of a HSLA Bainitic Steel. Proc. Int.Conf. on Thermomechanical Processing of Steeland Other Materials, THERMEC’97, Wollongong,Australia, pp. 787–792 (1997).

5. E. V. Perolama and P. D. Hodgson: MicrostructuralCharacterization of Strip Steel after Simulated Thermo-mechanical Processing and Accelerated Cooling. Proc.Int. Conf on Thermomechanical Processing ofSteel and Other Materials, THERMEC’97, Wollon-gong, Australia, pp. 803–809 (1997).

6. J. G. Willians, C. R. Killmore, P. D. Edwards, andP. G. Kelly: Thermomechanical Processing of Mo-NbHigh Strength Steels for Application to X70 and X80ERW Linepipe. Proc. Int. Conf. on Thermomechani-cal Processing of Steel and Other Materials,THERMEC’97, Wollongong, Australia, pp. 475–482 (1997).

7. G. Krauss and S. W. Thompson: Ferritic micro-structures in continuously cooled low and ultra-low carbon steels. ISIJ Int. 35:937–945 (1995).

8. S. W. Thompson and G. Krauss: Structure andProperties of Continuously Cooled Bainitic Ferrite–Austenite–Martensite Microstructures. Proc. Int.Conf. on Mechanical Working and Steel Process-ing, Chicago, IL, pp. 467–481 (October 1989).

9. A. B. Cota, P. J. Modenesi, R. Barbosa, and D. B.Santos: Determination of CCT diagrams by ther-mal analysis of a HSLA bainitic steel. ScriptaMater. 40:165–169 (1999).

10. A. B. Cota: Simulação da Laminação Controlada eResfriamento Acelerado em um Aço ARBL Bainit-ico de Baixo Carbono atrevés de Ensaios deTorção, Ph.D. Thesis, Federal University of MinasGerais, Brazil (1998).

11. F. S. LePera: Improved etching technique to em-phasize martensite and bainite in high-strengthdual-phase steel. J. Metals 38–39 (1980).

12. H. K. D. H. Bhadeshia and J. W. Christian: Bainiteof steels. Metall. Trans. A 21A:767–797 (1990).

13. R. W. K. Honeycombe and H. K. D. H. Bhadeshia:Steels: Microstructure and Properties. Edward Ar-nold Ltd., London, pp. 115–139 (1995).

14. E. Mazancová and K. Mazanec: Physical metal-lurgy characteristics of the MA constituent forma-tion in bainite granular. J. Mater. Process. Technol.64:287–292 (1997).

15. S. Yamamoto et al.: Effects of the austenite grainsize and deformation in the unrecrystallized aus-tenite region on bainite transformation behaviorand microstructure. ISIJ Int. 35:1020–1026 (1995).

Received April 1999; accepted July 1999.