Scripta Materialia 56 (2007) 777–780

www.actamat-journals.com

Nanometer martensite flakes in high-temperaturedeformation-induced ferrite grains of a low-carbon steel

Zhaoxia Liu, Dianzhong Li,* Guiwen Qiao and Yiyi Li

Laboratory for Special Environment Materials, Institute of Metal Research, Chinese Academy of Sciences,

72 Wenhua Road, Shenyang 110016, China

Received 20 November 2006; revised 3 January 2007; accepted 4 January 2007Available online 6 February 2007

High-temperature deformation-induced ferrite grains supersaturated with carbon offer superior properties such as higher elasticmodulus and microhardness compared with proeutectoid ferrite because of they contain nanometer martensite flakes. After anneal-ing at 700 �C, fine carbides precipitate due to the decomposition of these nanometer martensite flakes.� 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Nanometer martensite; High-temperature deformation-induced ferrite; Carbon supersaturation; Low-carbon steel; Carbon diffusion

Since the austenite–ferrite transformation is of greatimportance in refining ferrite grain size, it has beeninvestigated intensively and extensively over recentyears. The most frequently mentioned transformationis the deformation-induced ferrite (DIF) transformationbecause of its simplicity and efficiency in grain refine-ment [1–3]. Using this transformation, Priestner andco-workers observed local areas of very fine grains inrolled steels [4]. Yada and Matsumura claimed to haveproduced coils of hot-rolled strip with a ferrite grain sizeapproaching 1 lm [5].

A number of hypotheses regarding the formation ofDIF from austenite in steel have been suggested. Gener-ally speaking, the austenite–ferrite transformationduring undeformed austenite isothermal decompositionoccurs by diffusion-controlled nucleation and growthprocesses [6,7]. Some researchers [8] have proposed thatthe DIF transformation is also controlled by carbondiffusion. However, this has been questioned by otherresearchers [9–11] because of the high percentage ofthe DIF existing in specimens when the deformationhas taken place above Ar3 and at a high strain rate. Inthese conditions, carbon cannot diffuse fast enough tomaintain the carbon concentration equilibrium betweenaustenite and ferrite. Yang et al. [9,10] found that a highpercentage of the DIF was obtained when the strain rate

1359-6462/$ - see front matter � 2007 Acta Materialia Inc. Published by Eldoi:10.1016/j.scriptamat.2007.01.015

* Corresponding author. Tel.: +86 24 23971281; fax: +86 2423891320; e-mail: dzli@imr.ac.cn

was as high as 350 s�1. If the DIF transformation wascontrolled by long-range carbon diffusion, the carbonconcentration in the residual austenite would be highenough to decompose the austenite into pearlite andferrite. Thus, there would be deformed pearlite in thedeformed specimens. However, no pearlite was ob-served, even though the percentage of the DIF was upto 95%. Yang and co-workers therefore hypothesizedthat DIF transformation was not controlled by long-range carbon diffusion. Their hypotheses were con-firmed by Tong’s [12] simulation results of isothermalDIF transformation in a Fe–C binary system, in whichthere was no obvious directional carbon diffusion be-tween austenite and ferrite grains at a high strain rate.Thus, although there have been various hypothesesabout carbon diffusion, none of them have been directlyconfirmed by experiments. Moreover, because muchDIF transformation work by previous researchers wasundertaken below Ae3, the influence of supercoolinghas been a complicating factor.

In this work, we create conditions to induce high-temperature DIF. (In order to distinguish this materialfrom the DIF induced below Ae3, we name the DIFinduced above Ae3 as high-temperature DIF.) We willthen demonstrate experimentally the presence of high-temperature DIF grains with supersaturated carbon,and thereby reveal the storage location of the excess car-bon, hence clarifying the DIF transformationmechanisms.

The chemical composition of the low-carbon steel(Q235) used is 0.13 C–0.19 Si–0.49 Mn–0.012 P–0.013

sevier Ltd. All rights reserved.

778 Z. Liu et al. / Scripta Materialia 56 (2007) 777–780

S (wt.%). The austenite–ferrite equilibrium transforma-tion temperature Ae3, calculated by the Thermo-Calcsoftware package, is 848 �C. The hot compression spec-imens with diameter 8 mm and height 15 mm weremachined from a rough slab, normalized at 950 �C for1 h. The thermal simulation experiments were per-formed on a Gleeble 3500 machine.

Two routes were adopted here and specimens con-taining either DIF or proeutectoid ferrite were obtained.

The DIF specimens: After austenitization at 950 �Cfor 5 min, the specimens were cooled to 850 �C at a cool-ing rate of 1 �C s�1 and isothermally held for 30 s at850 �C. The specimens were then uniaxially compressedto a reduction of 80% at a strain rate of 20 s�1 at a targettemperature of 850 �C. However, from the recordedtemperature–time curve, the temperature increased to875 �C during this deformation due to the plastic work.Therefore, the temperature of the DIF transformation inthis study was much higher than the Ae3 temperature of848 �C. Quenching immediately after deformation wasconsidered to be the key to conserve the high-tempera-ture DIF microstructure, because, naturally, the high-temperature DIF can rapidly retransform to austeniteabove Ae3. The cooling rate is also important, becausethe cementite can precipitate from the high-temperatureDIF grains supersaturated with carbon, and proeutec-toid ferrite can form when the cooling rate is low [13].In this work, the DIF specimens were quenched immedi-ately after the compression finished not only by thewater sprayed from the standard nozzle of the Gleeble3500, but also by additional iced water poured onby hand. By this means a cooling rate of up to1500 �C s�1 was achieved.

DIF specimens were sectioned through the centerparallel to the compression axis. The sections wereground, polished and etched in a 3% nital solution.Under the microscope, the outer region of the DIFspecimens were seen to be composed of the high-temper-ature DIF and a little lath martensite, and the centerregion was composed totally of the high-temperatureDIF grains.

In the outer region of the DIF specimens (Fig. 1a) thecarbon concentration in the quenched martensite andthe high-temperature DIF grains was compared byelectron probe microanalysis (EPMA). To achieve opti-mal conditions for the measurement of light elementssuch as carbon, sensitivity was maximized by the useof a 10 kV accelerating voltage and 0.5 nA as reportedby Refs. [14,15]. The center part of the specimens(Fig. 1a), where only the high-temperature DIF grainsexisted, was studied by X-ray diffraction (XRD) mea-

Figure 1. The shape of (a) the DIF specimen, (b) the proeutectoidferrite specimen.

surements using a Rigaku D/Max-RC with a Cu target.Microstructural observations were carried out using aJEOL JEM 2010 transmission electron microscope(TEM). Nanoindentation tests were carried out with aNano Indenter XbTM (MTS, USA) with a Berkovichtip loading weight of 980 lN. The continuous stiffnessmeasurement (CSM) method was used. After the tests,the specimens were etched with 3% nital and then ob-served under a scanning electron microscope (SEM) toidentify the locations of the residual indentations.

The proeutectoid ferrite specimens: After austenitiza-tion at 950 �C for 5 min, the specimen was directlycooled to room temperature at a cooling rate of1 �C s�1. The specimens were sectioned perpendicularto their axis. The sections were ground, polished andetched in a 3% nital solution. Under the microscope,the specimens were seen to be composed of proeutectoidferrite and pearlite. The proeutectoid ferrite specimenswere studied by XRD and nanoindentation (Fig. 1b).

To compare the carbon concentration in the high-temperature DIF grains with the parent austenite(residual austenite transformed into martensite duringquenching), an EPMA line analysis of the carbon distri-bution across the section of the DIF specimen is shownin Figure 2 [16]. The concave areas denote the high-tem-perature DIF grains and are labeled a, c, d, g, h, i, j andk while the convex areas denote martensite and arelabeled b, e and f. Figure 2 illustrates that there is almostno difference between the carbon concentration of thehigh-temperature DIF grains and martensite grains, incontrast to expectations based on the carbon concentra-tion of ferrite which is two orders of magnitude lowerthan that of quenched martensite. The high-temperatureDIF grains are therefore supersaturated with carbon.

The XRD method involving step scan techniques wasalso used here to compare the parameters of the DIFgrains with that of proeutectoid ferrite. A software pro-gram (MDI Jade 5.0 Materials Data, Inc.) was used toobtain the lattice parameters for the high-temperatureDIF grains and proeutectoid ferrite. From Table 1 itwas found that the lattice parameter 0.286967 nm ofthe DIF was significantly larger than that of proeutec-toid ferrite at 0.286865 nm. Using the relationship pro-posed by Bhadeshia et al. [17] which relates the ferritelattice parameter to alloying element concentration,

Figure 2. EPMA line analysis of the carbon concentration in the high-temperature DIF (a, c, d, g, h, i, j, k) and the water-quenchedmartensite (b, e, f) in the border of the DIF specimen, showing verysimilar carbon concentrations in the high-temperature DIF and thequenched martensite.

Table 1. Ferrite lattice parameters and the residual stress of theexperimental steel

Ferrite aa (nm) Error Residualstress (MPa)

High-temperature 0.286967 ±0.00003 �232DIF grains 8Proeutectoid ferrite 0.286865 ±0.00003 �224

2

Z. Liu et al. / Scripta Materialia 56 (2007) 777–780 779

the carbon in the high-temperature DIF was quantita-tively estimated. The relation is

aDIF ¼ 0:28664þ ð3a2FeÞ�1

� bðaFe � 0:0279xacÞ

2ðaFe þ 0:2496xacÞ � a3

Fec� 0:003xa

Si þ 0:06xaMn þ 0:07xa

Ni þ 0:031xaMo

þ 0:05xaCr þ 0:096xa

V

ð1ÞIn the above equation, xa

i represents the mole fraction ofelements i in the ferrite. The lattice parameter of pureiron (aFe) is 0.28664 nm. Substitution of the appropriateconcentrations and aDIF = 0.286967 nm, the amount ofthe carbon in the high-temperature DIF is 0.104 wt.%,equivalent to xa

c ¼ 0:0046, in good agreement with theabove EPMA results (the mean carbon concentrationin this steel is 0.13 wt.%). The surface residual stresswas tested with an X-ray stress determinator (HandanAST Institute) with a Cr target on the {211} ferriteplanes. The two specimens have a value very close tothat of the surface compressive stress in the {21 1} fer-rite plane, with slightly lower lattice parameters, but inapproximately equal amounts. Therefore, the actual in-crease in the crystal lattice of the DIF grains very clearlyindicates the dilation caused by the excess carbon in theDIF grains.

Because the carbon concentration in ferrite is nor-mally two orders of magnitude lower than the parentaustenite, the precise form and location of this massiveexcess of carbon in the high-temperature DIF grainswas subjected to detailed study. TEM was employedto investigate the internal microstructures of the high-temperature DIF grains. The projected diameter of theselected area aperture was approximately 400 nm. Thisdiameter included some ferrite phase in addition to themartensite flakes. Figure 3a shows a typical TEM image

Figure 3. (a) TEM bright-field image of the microstructure in the DIFspecimen, showing the dark martensite flakes and the bright DIFphase; (b) SAD patterns of the area in (a): the arrows show martensiteflakes and the ferrite phase.

of one high-temperature DIF grain. It was found by en-ergy-dispersive X-ray spectroscopy (EDS) analysis thatthere were parallel dark flakes composed of Fe and C.No cementite or other carbides were found The corre-sponding selected area diffraction (SAD) pattern is dis-played in Figure 3b and the center location where thediffraction was obtained is shown by an arrow in Figure3a. The SAD analysis shows spots of martensite and fer-rite, and we can see splitting of some spots. This appearsto be the result of the extremely high-carbon concentra-tion in these martensite flakes. The DIF transformationis usually controlled by carbon diffusion, which takestime to complete. During these experiments, high strainand high strain rate were chosen, causing the DIF trans-formation to be incomplete, resulting in a very high-car-bon concentration in the residual austenite flakes. Theaustenite flakes subsequently transform into martensiteduring the rapid quenching. In this work the large defor-mation refines the grains, which in turn defines the max-imum possible length of the martensite flakes. The lackof time for diffusion in the high-temperature DIF grains(composed partly of ferrite and partly of martensite) ex-plains the higher carbon content, lattice dilation and thesplitting of TEM spots. For this high-carbon concentra-tion martensite, DIF transformation is controlled partlyby long-range carbon diffusion.

The high-temperature DIF grains with high elasticmodulus and hardness compared with proeutectoid fer-rite are shown in Figure 4. The high-temperature DIFgrains have a hardness distribution ranging from 2.33to 3.20 GPa with a mean hardness of 2.63 GPa, whichis significantly higher than that of proeutectoid ferriteat around 2.20 GPa. If the DIF transformation hadbeen completed, cementite would have formed at a suf-ficient volume per cent to increase the hardness valuessignificantly at the grain boundaries. However, in hard-ness surveys, the hardness values do not increase as theindentations approach the grain boundaries in thiswork. Furthermore, there is no observed precipitationin the grain boundaries. These facts strongly suggestthat the carbon is not stored at the grain boundaries.The elastic modulus of the DIF phase ranged from160 to 210 GPa with a substantially uniform distributionand was much higher than that of all proeutectoidferrite, which typically has an elastic modulus of 110–130 GPa. These results indicate that the nanometer

Figure 4. Hardness and elastic modulus values of the high-temperatureDIF and the proeutectoid ferrite.

Figure 5. Microstructure changes of the DIF specimen (a) beforeannealing, (b) after annealing at 700 �C for 1 h.

780 Z. Liu et al. / Scripta Materialia 56 (2007) 777–780

martensite flakes in the high-temperature DIF grains areresponsible for the increase in hardness.

During annealing, the high-temperature DIF grainswith martensite flakes are expected to decompose intoproeutectoid ferrite and cementite. Figure 5 shows theSEM microstructure of the high-temperature DIF spec-imens before [18] and after annealing. The polygonaland equiaxed ferrite grains are fine (with some less than1 lm) and occupy the greater part of the specimen.However, a light-etching phase identified as iron carbide(cementite) was precipitated around the tempered high-temperature DIF grains. Most of the carbides were lo-cated at the grain boundaries and triple points; only afew appear to exist within the ferrite grains. Naturally,the martensite is not stable during annealing, being eas-ily decomposed into cementite and ferrite. It is clear thatgrain growth would be expected to occur at relativelyhigh annealing temperatures. However, after a relativelymodest annealing treatment at 700 �C for 1 h the growthis limited, the mean grain size of the high-temperatureDIF only increases to 5 lm. Compared with the normalferrite–pearlite microstructure, the resulting uniformdispersion of fine cementite particles within the ferritegrains appears to be significantly more desirable forstrengthening the steel.

EPMA measurements confirm that the carbon con-centration of the high-temperature DIF grains is veryclose to that of quenched martensite. XRD analysis indi-cates that a larger lattice parameter of the high-temper-ature DIF grains is caused by the high supersaturationof carbon. The excess carbon in DIF grains is storedin nanometer martensite, leading to superior propertiessuch as high elastic modulus and hardness of the high-temperature DIF grains compared with proeutectoidferrite. Annealing at 700 �C precipitates fine carbidesaround the high-temperature DIF grains due to thedecomposition of nanometer martensite flakes.

The authors are indebted to Dr. Wei Wang at Insti-tute of Metal Research, Chinese Academy of Sciencesand Professor John Campbell at the University ofBirmingham, UK. This work is supported by NationalNatural Science Foundation of China (No. 50471073).

[1] R. Priestner, P.D. Hodgson, Mater. Sci. Technol. 8 (1992)849.

[2] Y. Matsumura, H. Yada, Trans. ISIJ. 27 (1987) 492.[3] B. Mintz, J. Lewis, J. Jonas, Mater. Sci. Technol. 13

(1997) 379.[4] R. Priestner, L. Ali, Mater. Sci. Technol. 9 (1993) 135.[5] H. Yada, Y. Matsumura, in: I. Tamura (Ed.), Proc. Int.

Conf. Physical Metallurgy of Thermomechanical Process-ing of Steel and Other Materials, ISIJ, 1988, p. 200.

[6] S.E. Offerman, N.H. Dijk, J. Sietsma, S. Grigull, E.M.Lauridsen, L. Margulies, H.F. Poulse, M.T. Rekveldt, S.Zwaag, Science 298 (2002) 1003.

[7] J. Speer, D.K. Matlock, B.C. De Cooman, J.G. Schroth,Acta Mater. 51 (2003) 2611.

[8] L.X. Du, X.H. Liu, G.D. Wang, Acta Metall. Sin. 38(2002) 196.

[9] Z.M. Yang, Y. Zhao, R.Z. Wang, Y.W. Ma, Y.M. Che,Acta Metall. Sin. 36 (2000) 818.

[10] Z.M. Yang, Y. Zhao, R.Z. Wang, Y.M. Che, Y.W. Ma,Q.A. Chen, Acta Metall. Sin. 36 (2000) 1061.

[11] C.M. Li, H. Yada, H. Yamagata, Scripta Mater. 39(1998) 963.

[12] M.M. Tong, J. Ni, Y.T. Zhang, D.Z. Li, Y.Y. Li, ScriptaMater. 50 (2004) 909.

[13] Y. Choi, W.Y. Choo, D. Kwoi, Scripta Mater. 45 (2001)1401.

[14] C.-U. Ro, J. Osan, R. Van Grieken, Anal. Chem. 71(1999) 1521.

[15] J. Osan, I. Szaloki, C.-U. Ro, R. Van Grieken, Mikr-ochemica Acta. 132 (2000) 349.

[16] Z.X. Liu, M.M. Tong, C.J. Huang, D.Z. Li, Acta Metall.Sin. 40 (2004) 930.

[17] H.K.D.H. Bhadeshia, S.A. David, J.M. Viteck, R.W.Reed, Mater. Sci. Technol. 7 (1991) 686.

[18] Z.X. Liu, D.Z. Li, S.P. Lu, G.W. Qiao, ISIJ. Int. 47(2007) 289.