KINETICS OF THE PROEUTECTOID FERRITE REACTION A STUDY OF THE INFLUENCE OF Mn AND Ni ON THE IN STEELS
Kawasaki Stel Corporation Technical Reseach Laboratories Mizushima Kurashiki 712-8511
K. OI C. LUX and G. R. PURDY
Japan IRSIDGroue USINOR BP30320 57283MaiziereslMetz Frane and Dparmet fMaterials Science and Engineering McMaster University Hamilton Ontario Canada L8S 4L7
Receled 2 December J9; accepted 26 Jarry 2000)
AbstraetThe kinetic transition between partitioned and unpartitioned growth of proeutectoid ferrte has been stuadied for high-purity FeCMn and FeCNi alloys and for temperatures just above the eutectoid.These resuls (and crain rsults of previous stigatins) are cmpared with cmptd paaquilbrim and equilibrium temary phase diagrams and it is shown that the transition ocurs well within the para- p s jo s p so d qulibeium o-phase regions but signifcantly otside the lmit pdicted by the lcal quibrium analsisforce is due to substitutional solute diffusion within the moving interfaoe. The oquilbrium binding energies inferred that both Mn and Ni exert a drag on the moving ferritelaustenite interfaces and that this dragof each of the substitutional solutes to the boundary are expected to be of order RT. 20 Ach Mef fwrgica Inc Publishend by Elsevier Science Lid. Aff righrs reserred.
Keyworad: Phase transformations; Steels; Nucleation growth
1. INTRODUCTION
the moving z/ interface is assumed and the diffu-solved for the interface velocity. The choice of sion equations for solute transport in austenite areassumption concerning the behavior of the substitu- tional alloying element is then critical in distinguish-ing among the different approaches.If the substitutional solute (X) is presumed also to be inlocal equilibrium the large difference in diffusion coefficients of the two solutes requires either thatthe far-field concentration of X in austenite be thesame as that in the product ferrite or that the rate of ferrite growth be controlled by the diffusion ofponent X. This divides the isothermal section of the ternary diagram into two regions as noledby Kirkaldy and co-workers [1 4] with the region of higher supersaturation characterized by rapidpartitioned growth. The boundary between the two unpartitioned growth and a second region by slowdzjo pn q ssaustenite-stabilizing substitutional alloying element tition" [4] as shown schematically in Fig 1 for ansuch as Mn or Ni.
The kinetics of the formation of ferrite from auste- nite is one of the most important determinants ofthe microstructure of stees. It has been clear for many years that the velocity of the ×/ interface isstrongly influenced (most often retarded) by thepresence of alloying elements such as Mn and Ni. However although this problem has been the objectof numerous experimental and theoretical studies [112] a quantitative explanation of alloying el-ement effects has proven somewhat elusive.
The approach most monly employed hasbeen to pare the results of isothermal growthexperiments with theories which are ternary or mul- tiponent extensions of the binary local-equili-brium analysis first presented by Zener [13]. In these treatments a (local) equilibrium for carbon at
A consequence of the local equilibrium treatmentof non-partitioned growth (LE-NP) would be the existence of a thin “spike" of solute X in the auste-been observed however (e.g. Ref. [11]) that the nite just ahead of the interface. It has frequently
Fig. 1. Schematic relation among the isothermal equib- brium paraequilibrium and local equilibrium non-parti-tioning boundaries for the system FeCX where X is an austenite-stabilizing suhstitutional alloying element.
thickness of this profile as estimated from volumediffusion coefficients in austenite would be less thanatomic dimensions for most cases investigated.
An alternative approach first postulated byassumes that the substitutional solute is lefl essem- Hultgren [2] and developed in detail by Hillert [3] tially undisturbed by the passage of the interface and that tbe interstitial carbon is in a form of con-strained local equilibrium in which the chemical po- tential of the mobile solute is continuous throughthis constraint gives rise to a type of equilibrium the interface. Minimizing the free energy subject totermed paraequilibrium (PE); it is as accessible toputation as is full equilibrium [5 14]. Under the assumption of paraequilibrium the region wherethan expected from the LE-NP approach unpartitioned growth is expected is somewhat larger sesuggested by Fig 1.
Several reports of experimental investigations offerrite growth from FeCMn austenites (e.g. Refs
Fig. 2. A schematic representation of the solute drag forceas a function of interface velocity [12]. The local chemical force for ferite grouth is also shown; a solutien isobtained at the intersection of the two functions.
[4 5]) have indicated that the transformation at low supersaturation generally proceeds at rates fasterthan those predicted by LE-P theory. However observed rates are slower than those predicted byXouedsp sig u suonesudns soqq 1e 3dhas been accounted for in part by Enomoto and Aaronson who invoke fast interfacial transport asa mechanism for transporting Mn from the growing ferrite (the rejecter plate mechanism [8]). Growth atparaequilibrium model has been attributed by rates considerably less than that predicted by theeffect (SDLE) as originally proposed by Kinsman Bradley and Aaronson [9] to a solute-drag-likeand Aaronson [10].
of the solute-drag model as formulated by Cahn Purdy and Brechet [12] considered the applicationtutional element through the transformation inter- [15] to the problem of the distribution of the substi-face.This efeet is represented schematicall in
Fig. 3. Substitutionl solute profles through the inteface fo an stenite-stabilizing alloying elmentcorresponding to: (a) a low relative velocity (VA/D 1).
Table 1. Compositioe of alloys stodied (mass%)
Alloy Mn N C2.08 0.095B D C 389 2.17 3.93 0.120 0.237 0.217266 2.42 0.196 620
Fig. 2 in which the dependence of the solute dragforce with interface velocity is pared with the local chemical driving force (also shown as a func-tion of interface velocity). Their intent was to inves- tigate theoretically a continuum of possible kineticmodynamic limits defined by the LE-NP and PE states (for unpartitioned growth) between the ther-models. This approach suggests that the kinetic boundary between fast and slow transformationshould be located between the PE and LE-NP limits; it is not a simple matter to predict the preciselocation of this experimental boundary however.Neither the interaction energies of the solutes with the interphase boundary nor the cross-boundarydiffusion coefficients for the substitutional solutes are well established. Figure 3 represents a series ofputed solute profiles through a transformation interface for a solute (like Mn) that partitions tothe parent phase. At higher dimensionless velocities [Fig. 3(c)] the profile flattens as the interfaceapproaches a paraequilibrium state. Enomoto [16] has extended this model to include an energy of in-teraction between substitutional and interstitialsolutes in the vicinity of the boundary an effect dis- cussed by Purdy and Brechet [12] but not explicilytaken into account in their analytical approach.
In a recent contribution [17] the PurdyBrechetapproach has been applied in semi-quantitative aSupnxqSudqwith the maximum solute drag force. If the avail- able chemical driving force is less than the maxi-will be dominated by the solute drag term. mum solute drag it is likely that the transformationApplication of this concept yielded reasonablesoluteboundary interaction energies for the system Fe-CMo [17].
precisely as possible the experimental boundary The purpose of the present work is to define asbetweenpartitioned growth regimes in high purity FeCMn and Fe- and unpartitionedferriteCNi alloys at temperatures near the eutectoid and to pare these results with puted equilibriumand paraequilibrium isothermal phase diagrams.Both equilibrium and paraequilibrium diagrams were puted [14] with the aid of a moderm data-buse (FEDAT) obtained from ThermoCale AB. Manganese was chosen as one of the most -monly used alloying elements in steels. Nickel which has an austenite stabilizing effect similar tothat of Mn but which has an opposite sign of inter-
Fig. 4. The location of the alloy positions relative to an isothermal section (700°C) for the Fe-CMn system.The puted paraequilibrium and LE-NP boundary (broken line) are also shown.
action with carbon in solution was chosen for - parison.
2. EXPERIMENTAL
A number of experimental termary alloys wereprepared by are-melting zone refined iron and high- purity alloying elements to yield the positionsshown in Table 1. The alloys were analyzed afterfound to contain < 0.005 wt% N and < 0.002 wt%sections of the ternary FeCMn and FeCNi O. The alloy positions are shown on isothermalequilibrium diagrams in Figs 4 and 5. The ingots were sealed in silica capsules and homogenized formore than a week at 1000°C checked for homogen- eity using electron microprobe analysis then cutinto small specimens (~ 10 × 5 × 2 mm?) austeni- Fig. 5. A puted 700°C isothemm for the Fe-CNi sys-tem. Table 2. Ferite growth data Estimated ferie fctin afr I h (%) Grovth rate at 700C (m/s2)() du] 740 730 720 710 700 670A 3 20廿 65 40 × 10-0 3 0 < 1 18 4.8 × 10- 3. RESULTS tized and isothermally reacted in temperature-stabil- ized salt baths. Other samples were analyzed usingdifferential thermal analysis under an argon atmos-phere employing continuous cooling rates ranging from 10 to 20°C/s. The resulting microstructureswere analyzed using optical metallographic tech- niques. structures and thermal histories in Figs 68. The kinetic results are presented as typical micro-Estimated growth rates for grain-boundary ferritemates of the fraction of ferrite formed after 1 h at allotriomorphs are given in Table 2 along with esti- Fig. 6. Microstructures and thermal histories of typical FeCMn alloys The isothermal sections areshown for the isothermal reaction temperatures: (a) alloy B 700°C; (b) alloy B 670C. Fig. 7. Microstructures and themmal histories for Fe-CNi alloys: (a) alloy F 720°C; (b) alloy F.700°C. temperature. A clear kinetic boundary between par-titioned and unpartitioned growth is evident. Figure 8 shows a result characteristic of doubleheat treatments for temperatures above and below the partitioning transition. A well-defined exper-imental boundary between partitioned and unparti- tioned growth is shown in Figs 9 and 10 whichpare the experimentally determined boundarywith puted LE-NP and PE limits for FeCMn and FeCNi alloys. The results of the continuous-cooling differential thermal analyses are also shown in Figs 9 and 10. 4.DISCUSSION Our main observation is this the experimental envelope of unpartitioned growth is seen to lie sig-nificantly beyond the puted LE-NP region but well within the PE region in each case. Other simi-lar results obtained by Aaronson and co-workers [7 9] are consistent with this result. For example fora 0.11 w1% C3.28 w1% Ni alloy Aaronson andbetween 700 and 717°C; for a stee containing Domian [7] determined that the transition lay0.37 wt% C and 3.04 wt% Ni the boundary lay between 684 and 700°C. In each of these cases theexperimental transition lies outside the puted LE-NP region. This kinetic transition is quite well defined asof growth. The double heat treatments (Fig. 8) indicated by the mierostructures and estimated ratesfurther demonstrated that the envelope is in some
