Carbon partitioning into austenite after martensite ... › public ›...

Acta Materialia 51 (2003) 2611–2622www.actamat-journals.com

Carbon partitioning into austenite after martensitetransformation

J. Speera,∗, D.K. Matlocka, B.C. De Coomanb, J.G. Schrothc

a Department of Metallurgical and Materials Engineering, Colorado School of Mines, Hill Hall 274, Golden, CO 80401, USAb Department of Metallurgy and Materials Science, Ghent University, BE-9052 Ghent, Belgium

c Research and Development Center, General Motors Corporation, Warren, MI 48090, USA

Received 13 October 2002; received in revised form 23 January 2003; accepted 30 January 2003

Abstract

A model is developed to describe the endpoint of carbon partitioning between quenched martensite and retainedaustenite, in the absence of carbide formation. The model assumes a stationarya/g interface, and requires a uniformchemical potential for carbon, but not iron, in the two phases, leading to a metastable equilibrium condition identifiedhere as “constrained paraequilibrium” or CPE. The model is explained with example calculations showing the character-istics of the constrained paraequilibrium condition, and applications are discussed with respect to new microstructuresand processes, including a new “quenching and partitioning,” or Q&P process, to create mixtures of carbon-depletedmartensite, and carbon-enriched retained austenite. Important new implications with respect to fundamental elementsof the bainite transformation are also discussed. 2003 Published by Elsevier Science Ltd on behalf of Acta Materialia Inc.

1. Introduction

Carbon partitioning between ferrite and austeniteduring high temperature diffusional transform-ations is relatively well understood. These reac-tions are frequently referred to as reconstructivetransformations, because of the short-range dif-fusional movements of iron (and substitutional)atoms that accomplish a change in crystal structurebetween bcc and fcc [1]. In contrast, the detailsof carbon partitioning during or after displacive ormartensitic transformations are somewhat more

∗ Corresponding author. Fax:+1-303-273-3016.E-mail address: [email protected] (J.G. Speer).

1359-6454/03/$30.00 2003 Published by Elsevier Science Ltd on behalf of Acta Materialia Inc.doi:10.1016/S1359-6454(03)00059-4

controversial, particularly with respect to thegrowth of bainite. In martensite, the displacivetransformation (i.e. a process that involves coordi-nated rather than diffusional movements of the ironatoms) is usually believed to occur without dif-fusion of carbon or other interstitials [2], and thusthe body-centered martensite phase can be substan-tially supersaturated with carbon. Subsequent car-bon partitioning between martensite and anyretained austenite is usually not considered,because the temperature is too low for substantialamounts of diffusion to occur after quenching, andalso because carbon supersaturation is usuallyeliminated by competing processes, e.g. carbideprecipitation during tempering. There is, however,evidence that carbon partitioning from martensite

2612 J.G. Speer et al. / Acta Materialia 51 (2003) 2611–2622

to (retained) austenite does occur, to thin interlathfilms during cooling [3] or by isothermal holdingin a Si-containing steel after transformation [4].Carbide-free bainite microstructures may also formby diffusionless martensitic growth, followed by,or along with, carbon partitioning into austenite[1]. (It should be noted that the bainite transform-ation is not universally accepted to involve a shearmechanism, and also that it has been suggested thatthe term “bainite” should not be applied to carbide-free microstructures [5]). Carbon partitioning isone means of stabilizing austenite against furthertransformation at lower temperatures, and is likelyto be especially important in these steels containingalloying additions (e.g. silicon) that suppress for-mation of iron carbides.

Carbon migration after martensite transform-ation is most often considered in the context of car-bide precipitation reactions during tempering.Consequently, the thermodynamics of carbon par-titioning between martensite and retained austenitehave not been examined completely. Thus, thepresent work is focused on developing a model toaddress carbon partitioning from as-quenched mar-tensite into austenite, under conditions where com-peting reactions are suppressed. (Such reactionsinclude cementite or transition carbide formation,and decomposition of retained austenite by otherprocesses such as bainite transformation.) Themodel does not address partitioning kinetics, butrather predicts the “endpoint” where partitioning iscomplete at a given partitioning temperature. Thestarting microstructure (i.e. initial fractions of mar-tensite and retained austenite) is an input to thecarbon partitioning model developed below, andmay be controlled, for example, by the martensitetransformation behavior (particularly the Ms tem-perature of the austenite in relation to the quench-ing temperature during heat treatment).

2. Thermodynamics of carbon partitioning

Under equilibrium conditions, mixtures ofquenched martensite plus retained austenite inbinary Fe-C alloys are expected to decompose toferrite and iron carbide [6]. The phase compo-sitions at equilibrium are given by the phase

boundaries on the usual Fe-C phase diagram, andthe phase fractions may be determined from thelever rule. In the presence of substitutional (“X” )alloying additions, such as in a ternary Fe-C-X sys-tem, long-range diffusional processes at low tem-perature are limited primarily to carbon atommovements, and a “paraequilibrium” condition(following Hultgren [7,8]) is used to describe con-ditions where substitutional atoms do not partitionbetween the phases during transformation, and thusthe Fe/X atom ratios are unchanged in the differentphases. It should be recognized that the usual para-equilibrium condition does not have meaning in theabsence of substitutional alloying additions, i.e. inthe Fe-C binary system. Thermodynamically, theparaequilibrium condition represents a minimumfree energy condition, with the additional con-straint that the Fe/X atom ratios are unchanged. Aparaequilibrium transformation, such as proeutec-toid ferrite precipitation from austenite at highundercooling, may proceed by interface migrationinvolving short-range diffusion of the iron and sub-stitutional atoms to accomplish the necessary struc-ture change, along with long-range carbon dif-fusion to accomplish the necessary compositionchange. At paraequilibrium, the phase fractionsand phase compositions are uniquely fixed, rep-resenting the minimum free energy condition in theabsence of substitutional partitioning.

Under conditions where iron carbide precipi-tation is suppressed, then the phase diagram is oflittle use in predicting the metastable equilibriumcondition that applies to heating of mixtures ofmartensite and retained austenite at temperatureswell below the eutectoid temperature. The meta-stable equilibrium condition between martensite(i.e. ferrite) and austenite, identified by Hultgren[7,8], must therefore be considered. In the case ofan Fe-C binary alloy, if the minimum free energycondition (“orthoequilibrium” ) applies, then the aand g phase fractions and compositions wouldadjust themselves until the metastable a/g orthoe-quilibrium was achieved, where the chemicalpotentials (or partial molar free energies) of eachcomponent (Fe and C) would be equal in both theferrite and austenite. In the presence of slow-diffusing substitutionals, the free energy minimiz-ation would be modified to represent a paraequilib-

2613J.G. Speer et al. / Acta Materialia 51 (2003) 2611–2622

rium condition in which the Fe/X ratios are fixedand the carbon potentials are equal in the phases.In either case, it is critical to recognize that theapproach to metastability (i.e. to the appropriateminimum free energy condition) necessarilyinvolves interface migration to adjust the phasefractions appropriately. When the position of themartensite/austenite interface is effectively fixed,as we consider here when examining carbon par-titioning after martensite transformation is com-plete, then clearly it is not possible for theferrite/austenite mixture to reach either orthoequili-brium (in Fe-C) or paraequilibrium (in Fe-C-X),and a new model is required.

The metastable condition representing the com-pletion of carbon partitioning between martensiteand austenite is developed below, and will bereferred to as “constrained paraequilibrium,” orCPE. The constrained paraequilibrium condition isthus defined by complete absence of iron or substi-tutional atom movements, along with freedom ofthe carbon (or interstitial) atoms to migrate asneeded. Like paraequilibrium, constrained parae-quilibrium presumes that only carbon atoms moveover distances greater than the unit cell dimen-sions. Unlike paraequilibrium, under constrainedparaequilibrium conditions short-range move-ments of the iron and substitutionals do not occurat the interface and interface migration is thus pre-cluded, hence the “constraint” . Constrained parae-quilibrium also differs from paraequilibrium in thatit applies to binary Fe-C alloys as well as higher-order systems containing substitutional alloyingadditions.

Constrained paraequilibrium can be readilyunderstood through consideration of two key con-ditions that define its unique thermodynamic andmatter balance constraints. First, carbon diffusionis completed under constrained paraequilibriumconditions when the chemical potential of carbonis equal in the ferrite and austenite phases, or equi-valently, when the carbon activity is equal in thetwo phases (with reference to a common standardstate). A similar condition has been invoked pre-viously by Hillert and coworkers to describe apossible local boundary condition at the a/g inter-face during bainite formation by a coupleddiffusional/displacive process [9]. This thermodyn-

amic condition also represents a minimum freeenergy condition for the constrained system, butdiffers from (unconstrained) orthoequilibrium,which also requires that the chemical potential ofiron must be equal in both phases. Second, thenumber of iron atoms must be conserved in eachphase during the approach to constrained paraequi-librium. Such a matter balance is consistent withthe presumption of a stationary a/g interface aftercompletion of martensite growth. While alloyingelements such as silicon are important in promot-ing conditions where constrained paraequilibriumapplies (i.e. where carbide formation is suppressedduring tempering), the fundamentals of the con-strained parequilibrium condition apply equallywell to the Fe-C binary system, which is used inthe following discussion to derive and illustrate theproperties of constrained paraequilibrium.

A schematic Gibbs molar free energy vs. com-position diagram is shown in Fig. 1 for ferrite andaustenite in the iron-carbon system at a given tem-perature. The chemical potential of a componentin a phase (mphase

component) is given by the tangent inter-cept with the respective composition axis. Thecommon tangent construction in Fig. 1 thus illus-trates the unique orthoequilibrium conditionbetween ferrite and austenite at this temperature,showing the equilibrium phase compositions,XaEQ and XgEQ. The same schematic free energy vs.composition curves for ferrite and austenite arereproduced in Fig. 2, along with two arbitrarilyselected examples (I and II) satisfying the thermo-

Fig. 1. Schematic Gibbs free energy vs. composition diagramshowing the common-tangent construction representing orthoe-quilibrium between ferrite and austenite.


Fig. 2. Schematic Gibbs free energy vs. composition diagramshowing two possible ferrite and austenite compositionssatisfying the CPE requirement that the chemical potential ofcarbon is equal in the two phases.

dynamic condition for constrained paraequilibriumat this temperature. In each example, the tangentsto the ferrite and austenite free energy curves inter-sect the carbon axis at a single point, indicatingequal carbon potentials in a and g, but the chemicalpotential of iron is clearly different in each phase,indicating that orthoequilibrium is not satisfied.(The orthoequilibrium condition does represent apossible constrained paraequilibrium condition,however.) This figure shows that there is not aunique ferrite and austenite composition that satis-fies the general thermodynamic condition for con-strained paraequilibrium at this temperature, butrather there is an infinite set of possible ferrite andaustenite phase compositions. This is an importantproperty of constrained paraequilibrium, where thematter balance constraint is also needed touniquely determine the applicable phase compo-sitions.

3. The constrained paraequilibrium conditionin Fe-C alloys

The constrained paraequilibrium condition canbe calculated for Fe-C alloys using published ther-modynamic data, and knowledge of the as-quenched microstructure. Carbon activities in fer-rite and austenite have been determined by Loboand Geiger [10,11], with reference to graphite as

the standard state. For the purposes here, it is con-venient to use the data in the simplified form [12]:

RT ln�aC�gC

� 76,789�43.8T�(169,105 (1)

�120.4T)XgC

where �gC and �gC represent the Henrian activitycoefficients for carbon in ferrite and austenite, T isabsolute temperature, R is the universal gas con-stant, and XgC is the austenite composition (molefraction of carbon). Since the activity of carbon ineach phase is given by the product of the activitycoefficient and the mole fraction of carbon, thecondition whereby carbon activity is equal in thephases is satisfied when:

XgC � XaC·e76,789�43.8T�(169,105�120.4T)XgC

RT (2)

where XaC and XgC represent the mole fractions ofcarbon in ferrite and austenite. The relationship inEq. (2) describes a locus of phase compositions atany temperature; an example is shown in Fig. 3that illustrates the appropriate ferrite and austenitecompositions at 400 °C. The ferrite and austenitecompositions in Fig. 3 are quite similar to previousresults [9] determined using a different source of

Fig. 3. Calculated locus of ferrite and austenite compositionshaving equal carbon activities at 400 °C.


free energy data [13]. In the CPE calculationspresented below, temperatures between 200 and600 °C are considered. The possible combinationsof ferrite and austenite compositions over thisrange of temperatures are shown in Fig. 4. Theresults in Fig. 4 are indicative of reduced carbonsolubility in ferrite at lower temperatures in thisrange.

Calculation of the constrained paraequilibriumcondition for a particular steel requires twoadditional inputs to address the matter balance con-straint: 1) the total carbon concentration, and 2) thefractions of martensite and retained austenite in theas-quenched condition. The phase fractions aftermartensite transformation are generally controlledby the degree of undercooling below the marten-site-start (Ms) temperature of the austenite, accord-ing for example, to the Koistinen and Marburgerequation [6]. The matter balance conditionsaccount for changes in the phase compositions ascarbon partitions between the phases, and the spe-cial CPE requirement that iron atoms are not trans-ferred across the interface during partitioning, i.e.the number of iron atoms is conserved in eachphase. For an initial condition (prior to any carbon

Fig. 4. Calculated loci of ferrite and austenite compositionshaving equal carbon activities at temperatures between 200 and600 °C.

partitioning) where the martensite and austenitehave identical compositions (i.e. the same as theoverall carbon content of the steel, Xalloy

C ) and theinitial mole fractions of each phase are given byfai and fgi , respectively, then the matter balance foriron is given by:

fgCPE(1�XgCCPE) � fgi(1�Xalloy

C ) (3)

where fgCPE and XgCCPErepresent the austenite phase

fraction and composition after partitioning, and werecognize that the iron concentration is alwaysgiven by XgFe = 1�XgC for binary alloys. It is per-haps counterintuitive that the phase fractionsshould change (i.e. fgCPE � fgi) under conditionswhere the a/g interface is presumed to be con-strained fully, but it should be clear that movementof carbon atoms between the phases necessarilyinvolves a small adjustment in the total number ofatoms in each phase, and thus the mole fractionsof the phases. It also follows that the densities andlattice parameters of the phases may be changedslightly by carbon partitioning, which could leadto some elastic strain considerations that are notaddressed here, and may be associated with someinteresting effects that warrant investigation. Thematter balance for carbon is straightforward, wher-eby the total carbon is given by the sum of theamounts in each of the phases, according to:

faCPEXaCCPE� fgCPEXgCCPE

� XalloyC (4)

The constrained paraequilibrium condition isrepresented by the solution of four equations ((2),(3), (4) above, and (5) below), in four unknownsXaCCPE

, XgCCPE, faCPE, and fgCPE, where the relationship

between the phase fractions is simply:

faCPE � fgCPE � 1 (5)

In this solution, the ferrite and austenite compo-sitions in Eq. (2) take on their unique values,XaCCPE

and XgCCPE, representing a specific con-

strained paraequilibrium condition.The formulation illustrated here is strictly appli-

cable to binary Fe-C alloys, but remains approxi-mately correct for low alloy steels. Rigorous modi-fication of the approach to account for alloyingeffects on the carbon activities would simplyrequire adjustment of the activity coefficients inEq. (1) using appropriate interaction parameters for


the elements of interest. Dislocation trapping couldinfluence the distribution of carbon atoms [14],although effects on carbon solubility of the defectpopulation in as-quenched martensite are not incor-porated in the treatment here. A Zener orderingcontribution to the carbon activity in ferrite is notrequired once partitioning has begun, and is thusnot included here. Finally, the constrained parae-quilibrium condition could potentially apply insome instances where the initial compositions ofthe ferrite and austenite phases are not identical,for example after completion of bainitic ferritegrowth in carbide-free bainite microstructures. Amore general expression for the iron matter balancethen becomes:

fgCPE(1�XgCCPE) � fgi(1�XgCi

) (6)

where XgCirepresents the carbon concentration in

the austenite, prior to final carbon partitioning inthe presence of a fully constrained interface.

4. Example calculations of constrainedparaequilibrium

Determination of the CPE condition using theformulation presented in Eqs. (2)–(5) requires onlythe steel composition (i.e. carbon content), and theinitial microstructure (mole fractions of martensiteand retained austenite) to be specified. Forexample, we can consider an iron-carbon alloycontaining 0.5 wt. pct. carbon, quenched to aninitial condition where 75% (molar basis) marten-site is present, along with 25% retained austenite.At a partitioning temperature of 400 °C, the calcu-lations yield a solution for the CPE condition rep-resented by:

WaCCPE� 0.0015%, WgCCPE

� 1.97%,

faCPE � 0.267, and fgCPE � 0.733

where the CPE phase compositions in this instance(WaCCPE

and WgCCPE) are reported in wt. pct. The

results illustrate substantial carbon partitioningbetween the phases, along with slight adjustmentin the phase fractions due to carbon partitioning.

The constrained paraequilibrium phase compo-sitions were calculated for a variety of Fe-C alloy

compositions having different mole fractions ofmartensite (or correspondingly, retained austenite)prior to a hypothetical CPE partitioning treatment.As mentioned previously, such differences in mar-tensite fraction would be expected to arise fromquenching to different temperatures below Ms, forexample. Figs. 5, 6 and 7 present the calculatedCPE phase compositions at temperatures between200 and 600 °C for steels containing 0.25, 0.5, and1.0% carbon by weight, respectively. Initial mar-tensite (mole) fractions of 25, 50, 75, and 90% areshown to illustrate the results. Suppression ofcementite formation becomes increasingly difficultat elevated temperature, [4] and thus temperaturesabove 600 °C were not considered. Furthermore,at elevated temperatures, the mobility of iron andsubstitutionals is eventually expected to becomesufficient to allow short-range lattice reconstructivediffusion, at which time the fundamental basis ofconstrained paraequilibrium is violated, and a tra-ditional paraequilibrium model (or orthoequilib-rium in binary alloys) would be more applicable.

The results in Figs. 5–7 are of practical use forestimating the endpoint of carbon partitioning, butare also useful in illustrating important character-istics of constrained paraequilibrium. It is note-worthy that the CPE phase compositions are influ-enced by the alloy composition and the initialphase fractions, in addition to the temperature. Incontrast, orthoequilibrium in the Fe-C binary sys-tem is represented by a single tie line at any tem-perature, with the phase compositions independentof the alloy composition. (Also, the phase fractionsat orthoequilibrium adjust themselves to satisfy thelever rule for the appropriate alloy composition andphase compositions given by the tie-line, whereasthe CPE phase fractions are only slightly changedduring partitioning to account for carbon atomtransfer across the constrained interface.) Theresults show that the CPE carbon concentration inferrite increases with increasing temperature for agiven initial condition, whereas the CPE carbonconcentration in austenite decreases slightly withincreasing temperature. The relative temperaturedependence of the austenite composition is smaller,because the ferrite has a greater compositiondependence of the carbon activity and quite lowcalculated solute carbon levels, and the austenite


Fig. 5. Constrained paraequilibrium calculations for an Fe-0.25C (wt. pct.) alloy, showing the dependence of CPE ferrite and austen-ite compositions on temperature and initial martensite fraction (molar basis).

Fig. 6. Constrained paraequilibrium calculations for an Fe-0.5C (wt. pct.) alloy, showing the dependence of CPE ferrite and austenitecompositions on temperature and initial martensite fraction (molar basis).


Fig. 7. Constrained paraequilibrium calculations for an Fe-1.0C (wt. pct.) alloy, showing the dependence of CPE ferrite and austenitecompositions on temperature and initial martensite fraction (molar basis).

inherits most of the carbon in the steel at con-strained paraequilibrium. The CPE condition isassociated with phase compositions that can beeither carbon-rich, or carbon-depleted, relative toorthoequilibrium. When the austenite fraction isgreater than would be associated with orthoequilib-rium, then the CPE phase compositions have car-bon levels less than their orthoequilibrium values;and at austenite fractions less than the orthoequilib-rium values, the CPE phase fractions have highercarbon levels. For a given initial martensite frac-tion, higher carbon steels have CPE phase compo-sitions with greater carbon levels. Interestingly, thecarbon levels in the retained austenite after CPEpartitioning can be quite high, exceeding 2% evenin the lowest carbon steel considered here (0.25%carbon), at a retained austenite fraction of 10%.Some caution is perhaps warranted when consider-ing high predicted levels of carbon enrichment inthe austenite. First, it should be recognized that thecarbon activity data used for the calculations weredetermined in austenite compositions well below1% carbon, and so extrapolation to much highercarbon levels should be only approximate. Second,the highest CPE austenite compositions calculated

here (for a 1% carbon steel containing 10%retained austenite) are greater than 6.7 wt. pct., andtherefore exceed the carbon concentration ofcementite! Consequently, it is implausible, or dif-ficult, to envision conditions under which compet-ing reactions such as carbide formation could besuppressed sufficiently to achieve constrainedparaequilibrium in such instances. It is useful torecognize that an opportunity apparently exists toexplore even greater silicon/aluminum levels thanare typical of carbide-free bainitic steels, or todevelop alternative alloying approaches to furthersuppress carbide precipitation, thereby enhancingthe potential for employing new microstructurescontaining highly enriched austenite. Nonetheless,the overall conclusion of this model is clear: sig-nificant carbon enrichment of retained austeniteshould be possible by deliberate control of par-titioning after martensite transformation.

5. Application of CPE partitioning

It is important to consider applications of CPEcarbon partitioning and implications of the model


results, although some further alloy and processdesign considerations are also reported elsewhere[15]. It is envisioned that CPE partitioning will bemost applicable for processing conditions whereaustenite is formed at elevated temperature (eitherduring full austenitization or intercritical heattreatment), followed by cooling to a temperaturecarefully selected to control the fractions of mar-tensite and retained austenite, and finally by a ther-mal treatment that accomplishes the desired extentof carbon partitioning. We refer to this processingsequence as Q&P (quenching and partitioning), todistinguish it mechanistically from conventionalquenching and tempering (Q&T) of martensite,where carbide precipitation and decompositon ofretained austenite are typical, rather than chemicalstabilization of austenite by carbon partitioning.

Steels of greatest interest for Q&P processingwill make use of controlled amounts of retainedaustenite, and include TRIP (transformationinduced plasticity) sheet steels and high-strengthcarbide-free bainitic grades [16,17]. Alloyingadditions such as Si, Al or P are already presentin these steels to suppress carbide formation [4], somajor changes in alloy composition are not initiallyneeded to apply the new process. The Q&P processcan also be considered as a possible replacementfor bainitic treatment in austempered ductile iron.In each example, carbon enrichment after marten-site transformation is considered to replace bainiteformation (where carbon partitioning to austeniteis intimately coupled with ferrite growth). Decoup-ling of carbon partitioning and microstructuredevelopment (i.e. control of phase fractions andmorphologies through athermal martensitetransformation) represents a dramatic and funda-mental departure from conventional approaches tomicrostructure control. There are widespreadimplications of this difference, including:

1. Greater amounts of carbon enrichment in aus-tenite become possible when compared to con-ventional carbide-free bainites. It has been sug-gested that transformation to bainite by a shearmechanism effectively ceases when the austen-ite composition is enriched to To, the compo-sition where the free energies of ferrite and aus-tenite are equal. In contrast, much higher carbon

levels in austenite are possible when partitioningoccurs under CPE conditions reported above,given suitable control of the phase fractionsafter quenching. The CPE model also indicatesthat the phase compositions after partitioning ina given steel are identical in martensite/austenitemixtures compared to bainite/austenite mix-tures, if the fraction of martensite is equal to thefraction of bainitic ferrite.

2. The temperature dependence of carbon enrich-ment is different for bainitic and Q&P pro-cessing. For bainite, the extent of partitioning iscontrolled by the temperature dependence of To,rather than the CPE conditions developedabove.

3. The processes used to develop the phase frac-tions and morphologies in the final microstruc-ture are completely separate from the carbonpartitioning treatment, allowing complete andindependent control during Q&P, whereas bain-itic ferrite growth and carbon partitioning areeffectively linked in the production of conven-tional carbide-free bainites. Different quenchedmicrostructures can also be designed prior topartitioning, such as lath or twinned martensite,providing additional opportunity to influencefinal properties in new ways. For example, sub-stantial amounts of carbon-depleted lath marten-site could be used in TRIP sheet steels to achi-eve higher strength levels. Somewhat elevatedcarbon levels in ferrite might also be achievableby carefully controlling CPE conditions,increasing the strength of the ferritic matrix.

4. Microstructure evolution occurs by an athermalmechanism in the Q&P process, whereas bain-itic transformation kinetics are thermally acti-vated. The thermal history needed to obtain con-ventional carbide-free bainitic microstructure isinfluenced by many factors that control the TTTor CCT kinetics, whereas the Ms and quenchtemperatures largely control transformation dur-ing Q&P [15]. During the subsequent partition-ing treatment, only carbon diffusion is needed,rather than carbon diffusion plus ferritenucleation or growth. Decoupling of these pro-cesses likely leads to a different time depen-dence of the carbon partitioning, and allowsgreater flexibility in the intermediate tempera-


ture process when compared to bainite. Forexample, alloying additions made to increasehardenability or suppress ferrite formation dur-ing initial cooling from the intercritical or aus-tenitizing temperature may substantially influ-ence bainitic transformation kinetics, while onlyslightly influencing the Q&P process throughsmall effects on the Ms temperature. While notof primary importance, it must be recognizedthat hardenability is still relevant in Q&P pro-cessing, influencing the response of the austen-ite during cooling or quenching, and also thepropensity for competing microstructurechanges during carbon partitioning.

5. Carbon enrichment of austenite can beemployed in steels with microstructures that aretraditionally martensitic, rather than just bain-itic. For example, higher carbon austenites typi-cal of moderately alloyed bearing steels or car-burized gear steels are quenched and onlytempered lightly, such that high hardness is pre-served and significant quantities of austenitemay be retained. The austenite can provide ben-eficial properties such as pitting resistance inrolling contact fatigue [18], and thus alloyingmodifications with suitable partitioning treat-ment could be used to alter the austenite stab-ility, and thus the behavior in service. Similarly,in other alloys where substantial levels of sili-con are currently applicable, such as ductile castirons or martensitic steels (e.g. for temperingresistance in spring steels), Q&P treatments toinfluence the amount and character of theretained austenite may also be considered.

Finally, it is worthwhile to consider possibleimplications of the CPE partitioning model withrespect to bainite transformation via a coupleddiffusional/displacive [9,19,20] mechanism. In thiscontext, the CPE model developed here predictsthe endpoint of carbon partitioning in the presenceof constrained interface boundaries, rather thanprocesses occurring during bainitic transformation(i.e. when the boundaries are still moving). It fol-lows that the final bainitic ferrite/retained austenitemixture in a carbide-free bainite should achieveconstrained paraequilibrium when carbon partition-ing is taken to completion. In such microstructures,

the final austenite composition has been reportedby different studies [1,21] to approximate To, andtherefore to provide support for the displacivegrowth mechanism. The carbon concentration ofbainitic ferrite is on the order of 0.03 wt. pct. [22]or greater [23], and is believed to be influencedsignificantly by trapping of solute carbon atoms bydislocations in the ferrite [24]. Most importantly,the steel is either at constrained paraequilibrium atthe end of the bainite transformation, or elseadditional carbon partitioning must occur to achi-eve constrained paraequilibrium. While possible[23], such partitioning after completion of the bain-ite transformation is not emphasized in the litera-ture, and indeed the remaining carbon available forpartitioning from ferrite is rather small. Thus, itseems reasonable to conclude that the bainite trans-formation may be completed under near-CPE con-ditions, where the To composition represents theCPE austenite composition.

The ferrite and austenite compositions at theinterface during transformation are also of funda-mental interest. Among the instantaneous interfacecompositions that have been considered forcoupled diffusional/displacive growth of bainiteare paraequilibrium [20] and “ local equilibrium forcarbon” [9]. Both the paraequilibrium and CPEcondition are effectively subsets of this latter “ localequilibrium” condition, which is effectively rep-resented by any of the phase compositionssatisfying Eq. (2) above. It is appealing to specu-late that the interface compositions during growthof bainite could perhaps be given by the particularCPE phase compositions having the austenite com-position of To (where the appropriate free energiesare adjusted for elastic strain energy, Zenerordering, and dislocation/solute interactions). Thatis, the transformation is not only completed underCPE conditions, but also may proceed under near-CPE conditions at the interface. Under such con-ditions, bainite growth might be considered toinvolve a diffusionless martensitic mechanism ofinterface migration, along with carbon diffusion-controlled growth kinetics with essentially steadystate boundary conditions that are far from(para)equilibrium, and yet satisfy very reasonablethermodynamic constraints. In this theory, growthof a bainitic subunit is eventually inhibited both


by transformation strains and by the diminishingcarbon concentration gradient in the austenite. Thetransition from bainitic to martensitic transform-ation is associated with the greater driving forceand reduced carbon mobility at lower temperatures,both which allow growth to proceed quicklyenough to avoid carbon buildup at the interface.While further work is needed to establish whetherthe kinetics and other characteristics of bainite for-mation are fully consistent with this explanation,the proposed behavior provides a model for thebainite transformation mechanism that is, at thesame time, both “ fully” diffusional and “ fully”martensitic.

6. Conclusions

Consideration of carbon partitioning fromquenched martensite to retained austenite in thepresence of a stationary a/g interface, under con-ditions where carbide precipitation is precluded,indicates that the metastable orthoequilibrium con-dition (or paraequilibrium in multicomponentalloys) between ferrite and austenite cannot be ach-ieved. Consequently, a “constrained paraequilib-rium” or CPE model was developed here to predictthe endpoint of carbon partitioning in the presenceof a stationary a/g interface, i.e. applicable to thecarbon redistribution processes after growth of themicrostructural constituents is complete. It isshown that the CPE condition involves specialmatter balance constraints (such that the numberof iron atoms is conserved in each phase duringpartitioning), while satisfying the thermodynamiccondition where the chemical potential of carbon isequal in austenite and ferrite (i.e. carbon depletedmartensite). Specific implications of the CPEmodel results include:

1. The ferrite and austenite compositions at con-strained paraequilibrium are dependent on thesteel composition and phase fractions, inaddition to temperature, in contrast to orthoequi-librium in Fe-C alloys where the tie-line isdependent only on temperature.

2. The CPE carbon concentration in ferriteincreases with increasing temperature for a

given initial condition, whereas the carbon con-centration in austenite decreases with increasingtemperature. The CPE phase compositions canbe either carbon-rich, or carbon-depleted, rela-tive to orthoequilibrium.

3. The austenite inherits most of the carbon in thesteel at constrained paraequilibrium, and theretained austenite can be very enriched with car-bon in some instances.

4. Applications of CPE partitioning may be con-sidered in steels where carbide formation is sup-pressed (e.g. with Si, Al, P, or even Niadditions). The potential for a new “quenchingand partitioning” process, or Q&P, is proposed,where the resulting martensite/austenite mix-tures may be substituted for more conventionalcarbide-free bainitic microstructures such ashigh-strength TRIP sheet steels or even aus-tempered ductile cast iron.

5. It is suggested that constrained paraequilibriumshould also apply when partitioning is completeafter transformation to carbide-free bainitemicrostructures. Austenite having a compositiongiven by the To curve is shown to represent apossible CPE condition, and it is speculated thatbainite growth involving a martensitic mech-anism of interface migration can occur with dif-fusion-controlled kinetics involving steady stateboundary conditions corresponding to CPEcompositions.

Acknowledgements

Two authors (Speer and Matlock) gratefullyacknowledge the support of the sponsors of theAdvanced Steel Processing and Products ResearchCenter at the Colorado School of Mines. Prof.H.K.D.H. Bhadeshia, University of Cambridge, isgratefully acknowledged for helpful discussions.

References

[1] Bhadeshia HKDH. Bainite in steels. London: The Instituteof Materials, 2001.

[2] Bhadeshia HKDH. Carbon content of retained austenite inquenched steels. Metal Science 1983;17(March):151–2.


[3] Sarikaya M, Thomas G, Steeds JW, Barnard SJ, SmithGDW. Solute element partitioning and austenite stabiliz-ation in steels. In: Aaronson HI, editor. Proceedings of anInternational Conference on Solid to Solid Phase Trans-formations. Warrendale, PA: The Metallurgical Society;1982. p. 1421–5.

[4] Gallagher MF, Speer JG, Matlock DK, Fonstein NM.Microstructure development in TRIP-sheet steels contain-ing Si, Al, and P. In: Proceedings of the 44th MechanicalWorking and Steel Processing Conference. Warrendale,PA: Iron and Steel Society; 2002. p. 153–72.

[5] Hillert M, Purdy GR. On the misuse of the term bainite.Scripta Materialia 2000;43(9):831–3.

[6] Krauss G. STEELS: Heat treatment and processing prin-ciples. Metals Park, OH: ASM International, 1990.

[7] Hultgren A. Isothermal transformation of austenite. ASMTransactions 1947;39:915–1005.

[8] Hultgren A. Isoterm omvandling av austenit. JernkontoretsAnnaler 1951;135:403–94.

[9] Hillert M, Hoglund L, Agren J. Escape of carbon fromferrite plates in austenite. Acta Metallurgica et Materialia1993;41(7):1951–7.

[10] Lobo JA, Geiger GH. Thermodynamics and solubility ofcarbon in ferrite and ferritic Fe-Mo alloys. MetallurgicalTransactions A 1976;7A(September):1347–57.

[11] Lobo JA, Geiger GH. Thermodynamics of carbon in aus-tenite and Fe-Mo austenite. Metallurgical Transactions A1976;7A(September):1359–64.

[12] Bhadeshia HKDH. Driving force for martensitic trans-formation in steels. Metal Science 1981;April:175–7.

[13] Gustafson P. A thermodynamic evaluation of the Fe-Csystem. Scandanavian Journal of Metallurgy1985;14:259–67.

[14] Kalish D, Cohen M. Structural changes and strengtheningin the strain tempering of martensite. Materials Scienceand Engineering 1970;6:156–66.

[15] Matlock DK, Brautigam VE, Speer JG. Application of thequenching and partitioning (Q&P) process to a medium-carbon, high-Si microalloyed bar steel, Proceedings ofTHERMEC’ 2003. Uetikon-Zurich, Switzerland: TransTech Publications, Inc.; 2003 (in press).

[16] Matlock DK, Krauss G, Speer JG. Microstructures andproperties of direct-cooled forging steels. Journal ofMaterials Processing Technology 2001;117:324–8.

[17] Caballero FG, Bhadeshia HKDH, Mawella KJA, JonesDG, Brown P. Design of novel high strength bainiticsteels, Parts I and II. Materials Science and Technology2001;17:512–22.

[18] Roache B, Pitsko R, Chidester A, Imundo J. The effect ofmicrostructure on debris dented rolling element bearingperformance, SAE Technical Paper #2002-01-1371. War-rendale, PA: Society of Automotive Engineers, 2002.

[19] Olson GB, Bhadeshia HKDH, Cohen M. Coupleddiffusional/displacive transformations. Acta Metallurgica1989;37(2):381–9.

[20] Mujahid SA, Bhadeshia HKDH. Coupleddiffusional/displacive transformations: Effect of carbonconcentration. Acta Metallurgica et Materialia1993;41(3):967–73.

[21] Matsuda H, Kitano F, Hasegawa K, Urabe T, Hosoya Y.Metallurgy of continuously annealed high strength TRIPsteel sheet. In: De Cooman BC, editor. International Con-ference on TRIP-Aided High Strength Ferrous Alloys—Proceedings. Antwerp: Technologisch Instituut VZW;2002. p. 113–9.

[22] Bhadeshia HKDH, Edmonds DV. The mechanism of bain-ite formation in steels. Acta Metallurgica 1980;28:1265–73.

[23] Bhadeshia HKDH, Waugh AR. Bainite: An atom probestudy of the incomplete reaction phenomenon. Acta Metal-lurgica 1982;30:775–84.

[24] Bhadeshia, HKDH. personal communication, Universityof Cambridge, UK; 2002.

Microstructural Evolution of a Low-Carbon Steel duringApplication of Quenching and Partitioning Heat Treatmentsafter Partial Austenitization

M.J. SANTOFIMIA, L. ZHAO, and J. SIETSMA

The ‘‘quenching and partitioning’’ (Q&P) process has been studied in a low-carbon steel con-taining 1.1 wt pct aluminum by heat treatments consisting of partial austenitization at 900 �Cand subsequent rapid cooling to a quenching temperature in the range between 125 �C and175 �C, followed by an isothermal treatment (partitioning step) at 250 �C and 350 �C for dif-ferent times. Characterization by means of optical and scanning electron microscopy, electronbackscattered diffraction (EBSD), magnetization measurements, and X-ray diffraction (XRD)has shown a multiphase microstructure formed by intercritical ferrite, epitaxial ferrite, retainedaustenite, bainite, and martensite after different stages of tempering. A considerable amount ofretained austenite has been obtained in the specimens partitioned at 350 �C for 100 seconds.Experimental results have been interpreted based on concepts of the martensite tempering,bainite transformation, and kinetics calculations of the carbon partitioning from martensite toaustenite.

DOI: 10.1007/s11661-008-9701-4� The Minerals, Metals & Materials Society and ASM International 2008

I. INTRODUCTION

TRANSFORMATION-INDUCED-PLASTICITY(TRIP) steels are usually produced via a thermome-chanical process of intercritical annealing followed byrapid cooling to a bainitic transformation regime, toobtain a microstructure of ferrite, bainite, and retainedaustenite. During the bainitic transformation, the for-mation of carbides is suppressed, due to the effect ofalloying elements such as silicon and aluminum; theaustenite is thus enriched with carbon and retained atroom temperature. Carbon-enriched metastable retainedaustenite is considered beneficial, because the TRIPphenomenon during the deformation can significantlycontribute to the formability and energy absorption ofthe material.

Recently, Speer et al.[1,2] proposed a novel heat-treatment concept, the so-called ‘‘quenching and parti-tioning’’ (Q&P) process, for the development ofmultiphase steels with considerable retention of austen-ite in the microstructure. The Q&P process consists of afirst quench (quenching step) to a temperature below themartensite-start (Ms) temperature but above the mar-tensite-finish (Mf) temperature, to form a mixture ofmartensite and austenite, and a subsequent isothermaltreatment (partitioning step) at the same temperature

(one-step treatment) or at a higher temperature (two-step treatment), in order to transfer the carbon from thesupersaturated martensite into the austenite. In this heattreatment, alloying elements such as silicon and alumi-num are also used, to avoid the carbide precipitationduring the partitioning step, because carbide precipita-tion acts as a sink of carbon that is no longer availablefor the stabilization of the austenite. Combining thisheat treatment with a previous partial austenitization, amicrostructure consisting of ferrite, carbon-depletedmartensite, and carbon-enriched retained austenite isobtained. This microstructure can lead to an interestingcombination of mechanical properties,[3,4] from a goodformability, as a result of the TRIP effect from theretained austenite, to a strength higher than that ofconventional TRIP steels, due to the presence ofmartensite instead of bainite.The design of adequate Q&P heat treatments requires

an understanding of all the phenomena that can affectthe final microstructure of the material during thisprocessing route. Most of the research on the Q&Pprocess has been focused on treatments that start withfull austenitization;[5,6] the use of partial austenitizationhas been studied less.[1] Therefore, there is insufficientexperimental evidence that the phenomena affect themicrostructure during the Q&P process in the case oftreatments starting with partial austenitization. In thiswork, the Q&P process is studied in a low-carbon steelcontaining 1.1 wt pct aluminum, by the application ofpartial austenitization followed by quenching to 125 �C,150 �C, or 175 �C and partitioning at 250 �C or 350 �Cfor different times. The use of a steel with a chemicalcomposition similar to the typical chemistry of aTRIP steel will help evaluate whether such compositionsare suitable for application of the Q&P process.

M.J. SANTOFIMIA, Postdoctoral Researcher, and L. ZHAO,Research Fellow, Department of Materials Science and Engineering,Delft University of Technology, are with the Materials InnovationInstitute (M2i), 2628 CD Delft, The Netherlands. J. SIETSMA,Associate Professor, is with the Department of Materials Science andEngineering, Delft University of Technology, 2628 CD Delft, TheNetherlands. Contact e-mail: [email protected]

Manuscript submitted March 26, 2008.Article published online November 14, 2008

46—VOLUME 40A, JANUARY 2009 METALLURGICAL AND MATERIALS TRANSACTIONS A

The resulting microstructures have been analyzed byoptical and scanning electron microscopy, magneticmeasurements, and X-ray diffraction (XRD). The inves-tigation pays attention to the evolution of the micro-structure along the whole process, but focuses on theaustenite retention.

II. EXPERIMENTAL PROCEDURE

The chemical composition of the material used in thiswork is shown in Table I. The as-received commercialmaterial was a 1.2-mm-thick steel sheet with a multi-phase microstructure consisting of ferrite, bainite, andretained austenite. After removing the galvanized layer,dilatometry specimens with dimensions of 1 9 5 910 mm3 were machined. Heat treatments were appliedto the specimens with a DT1000 high-resolution dila-tometer (Adamel Lhomargy SAS, Roissy en Brie,France). The thermal schedules applied to the specimensare displayed in Figure 1. Specimens were heated at5 �C/s, partially austenitized at 900 �C for 10 minutes,and cooled at 100 �C/s to 125 �C, 150 �C, or 175 �C toget a partial martensitic microstructure. Specimens werereheated and isothermally held at 250 �C or 350 �C(partitioning temperatures) for different times beforequenching to room temperature. In order to selectappropriate quenching temperatures, the Ms tempera-ture corresponding to the remaining intercriticalaustenite after the partial austenitization was measuredby dilatometry from the direct-quench treatment(Figure 2), leading to Ms = 260 �C. The formation ofa new phase was also observed. As will be explained inthe following Section III, this new phase is identified asepitaxial ferrite.

The specimens were ground and polished according tothe usual techniques. The etching procedure proposedby De et al.[7] was applied to the specimens to beobserved by light optical microscopy. This etching is atwo-step procedure that consists of a first etching in asolution of 5 pct picric acid with a few drops ofhydrochloric acid (20 seconds), followed by a secondetching in 10 pct aqueous sodium metabisulfide (8 sec-onds). Specimens were also analyzed after etching with5 pct nital with a JEOL* JSM-6500F field-emission gun

scanning electron microscope (FEG-SEM) operating at15 kV.

Some selected specimens were metallographicallyprepared for electron backscattered diffraction (EBSD)examination with a final polishing step of 0.5 mm, using

an oxide polishing solution (OPS) suspension. The lastspecimen-preparation step was electrolytic polishingwith an electrolyte consisting of 78 ml perchloric acid,90 ml distilled water, 730 ml ethanol, and 100 mlbutylglycol at 40 V for 10 seconds. The specimens wereanalyzed by an orientation imaging microscope attachedto a Nova 600 Nanolab dual-beam focused-ion-beammicroscope (FEI Company, Hillsboro, OR) equippedwith a FEG-SEM column. The analysis was performedunder the following conditions: acceleration voltage of20 kV, working distance of 7 mm, tilt angle of 70 deg,and step size of 50 nm. The EBSD measurements werecarried out in the cross section of the specimens, in aplane perpendicular to the normal direction of the sheet.The orientation data were postprocessed with Channel 5software provided by HKL Technology (Oxford Instru-ments, Abingdon, UK).Cubic specimens with an edge dimension of 1.0 mm for

magnetic measurements were machined from dilatome-try specimens using an electrodischarging machine. A7307 vibrating sample magnetometer (Lake Shore Cryo-tonics, Westerville, OH), calibrated with a standardNational Institute of Standards and Technology nickelspecimen, was used for the experiments. With thisequipment, magnetization curves at room temperaturewere measured by a stepwise change in the applied

Table I. Chemical Composition of the Studied Steel

(Weight Percent)

C Mn Si Al P

0.19 1.61 0.35 1.10 0.09

Fig. 1—Scheme of the Q&P treatments applied to the steel.

Fig. 2—Dilatometry curve corresponding to the cooling step of thespecimen obtained after direct quench at 100 �C/s from the intercriti-cal conditions.

*JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 40A, JANUARY 2009—47

magnetic field from 2.0 to -2.0 T. The saturationmagnetization values were obtained by fitting theapproach to the saturation of the experimentallyobtained magnetization curve.[8] The volume fraction ofretained austenite in every specimen, fc, is determined bycomparing the saturation magnetization values obtainedboth on the specimen with retained austenite, Msat(c),and on an austenite-free specimen, Msat(f), according to

fc ¼ 1� bMsat cð ÞMsat fð Þ

½1�

The coefficient b is obtained via the relation[8]

b ¼ 1� fh þ fhMh

sat

Masat

� �½2�

where fh is the volume fraction of cementite present inthe austenite-free specimen, Ma

sat is the saturationmagnetization of the ferrite,[9] and Mh

sat is the saturationmagnetization of the cementite.[10,11] The austenite-freespecimen was obtained by annealing one of the austen-ite-containing specimens at 600 �C for 1 hour.

In order to determine the average carbon content ofthe austenite, XRD experiments were performed on aBruker-type D8-Advance diffractometer (Bruker AXS,Karlsruhe, Germany) equipped with a Bruker VANTECposition sensitive detector. In the experiments, Co Ka

radiation was used; a 2h range from 30 to 135 deg,containing the (111), (200), (220), and (311) austenitereflections, was scanned using a step size of 0.05 deg.The austenite lattice parameter, ac, was determined fromthe peak position of each austenite reflection usingCohen’s method.[12] The carbon concentration xC of theaustenite was obtained using the relation[13]

ac ¼ 0:3556þ 0:00453xC þ 0:000095xMn þ 0:00056xAl

½3�

where xC, xMn, and xAl are the concentrations ofcarbon, manganese, and aluminum, respectively, inaustenite, in weight percent. The effects of silicon andphosphorus are not considered in Eq. [2], although the

effects of substitutional elements on the lattice param-eter are small compared with the influence of the carboncontent, as can also be observed from Eq. [3]. Theresults of this calculation indicate an average measure-ment of the carbon content in the austenite.

III. RESULTS

A. Optical Microscopy

Figure 3 shows the optical microscopy micrographsof the specimens quenched to 175 �C and partitioned at350 �C for 3 and 1000 seconds. These micrographs areused here to describe common microstructural detailsobserved for different quenching and partitioning tem-peratures. According to the work of De et al.,[7] theirproposed etching procedure allows the identification ofretained austenite in TRIP and dual-phase microstruc-tures by the revelation of white islands clearly differentfrom the ferrite phase. However, as has been analyzed ina previous study,[14] the distinction between martensiteand austenite from the metallographic study using thisagent is not clear in this material. In the present case,retained austenite and untempered martensite areobserved as light-colored islands in a ferritic matrix,whereas tempered martensite and bainite are observed ina dark color because of the finer microstructure and, insome cases, the presence of carbide precipitation.From the figures, the mixture of untempered mar-

tensite and retained austenite is observed to be moreabundant in the specimen partitioned for 3 seconds thanin the specimen partitioned for 1000 seconds. On theother hand, the combination of tempered martensite andbainite is more abundant in the specimen partitioned for1000 seconds. This observation is in agreement with theoccurrence, at longer partitioning times, of conventionalmartensite tempering processes such as carbide precip-itation in martensite and decomposition of austeniteinto bainite. This observation is analyzed in the follow-ing Section III–B.In Figures 3(a) and (b), ferrite is clearly observed as

the white matrix, but a detailed observation shows very

Fig. 3—Microstructure after Q&P heat treatments with quenching at 175 �C and partitioning at 350 �C for (a) 3 s and (b) 1000 s.


fine boundaries separating ferrite areas inside the samewell-defined ferrite grain. The ferritic areas that arecloser to the austenite/martensite grains correspond toferrite that has grown from the ferrite present after theintercritical treatment during the first cooling (the so-called epitaxial ferrite[15,16]). A detailed characterizationof the epitaxial ferrite formed during cooling in thismaterial can be found elsewhere.[14] Moreover, moredetails of the formation of epitaxial ferrite in thismaterial will be presented in the following Section III–B,III–C.

Optical microscopy observations give indicationsabout the microstructure present in the specimens forevery Q&P condition. However, they do not providemicrostructural details smaller than a few microns;therefore, scanning electron microscopy has been usedfor these purposes.

B. Scanning Electron Microscopy

Epitaxial ferrite can also be distinguished by scanningelectron microscopy. As an example, Figure 4 shows themicrostructure of the specimen quenched at 125 �C andtempered at 350 �C for 10 seconds, as observed byFEG-SEM. Epitaxial ferrite is clearly distinguishedfrom intercritical ferrite because of a different topogra-phy. This means that intercritical and epitaxial ferritehave a different response to the 5 pct nital etching.These differences are due to a distinct composition ofsubstitutional alloying elements in each type of ferrite,because epitaxial ferrite has grown during rapid coolingunder paraequilibrium conditions with the austen-ite,[15,17] whereas the partitioning of alloying elementswas possible to a certain extent during the intercriticalannealing.

From scanning electron microscopy observations, nosubstantial differences have been detected in the micro-structures observed after partitioning at 250 �C forevery quenching temperature. In particular, Figures 5(a)and (b) show the microstructures of the specimensquenched at 125 �C and partitioned at 250 �C for 3 and

100 seconds. A comparison of these figures shows thatthere has not been any substantial tempering of themartensitic microstructure during partitioning at 250 �Cfor less than 100 seconds. However, carbide precipita-tion is observed in specimens partitioned for 1000 sec-onds at this temperature, as shown in Figure 5(c) and(d) for the specimen quenched at 125 �C.The characteristic tempering phenomena occurring in

the martensite at 350 �C are different from thoseoccurring at 250 �C. In addition, the carbon partitioningfrom martensite to austenite is faster at 350 �C than at250 �C. These changes in the kinetics of the processesshould be reflected in the characteristics of the micro-structure of the specimens partitioned at 350 �C. In thisvein, Figure 6(a) shows the microstructure of thespecimen quenched at 125 �C and partitioned at350 �C for 10 seconds, in which blocks of temperedand untempered martensite in a ferrite matrix areobserved. One of the blocks of tempered martensite isenlarged in Figure 6(b), showing carbide precipitation.These two figures provide evidence that 10 seconds ofpartitioning at 350 �C is enough time to start thetempering of the microstructure. Similar behavior hasbeen observed after quenching at 150 �C and 175 �C.After partitioning for 1000 seconds at 350 �C, bainite

has been observed in all the specimens. The resultingbainitic microstructure, shown in Figure 7, is formed bybainitic ferrite plates separated by films of retainedaustenite or martensite formed during the final quench.Figure 7(a) displays the microstructure of the specimenquenched to 150 �C and partitioned at 350 �C for1000 seconds, showing a high fraction of bainite.Figure 7(b) contains an enlarged micrograph of thesquare area drawn in Figure 7(a), showing details ofthe decomposition of residual austenite in bainite.Figures 7(c) and (d) show the presence of bainite inthe specimens quenched to 125 �C and 175 �C, respec-tively, and partitioned at 350 �C for 1000 seconds. In allthe observed cases, bainitic ferrite plates are free ofinternal cementite particles, which is characteristic ofupper bainite.

Fig. 4—(a) and (b) FEG-SEM micrographs of a specimen quenched to 125 �C and partitioned at 350 �C for 10 s. The square in (a) is enlargedin (b). (EF is epitaxial ferrite, IF is intercritical ferrite, TM is tempered martensite, and RA/UM is retained austenite or untempered martensite).


C. Electron Backscattered Diffraction

In order to support some of the microstructuralfeatures mentioned, an EBSD analysis was performedon the specimens quenched to 150 �C and partitioned at350 �C for 10 and 1000 seconds.

Figure 8(a) shows a scanning electron microscopyimage of the EBSD scan displayed in Figure 8(b) of thespecimen quenched to 150 �C and partitioned at 350 �Cfor 10 seconds. In particular, Figure 8(b) is a combinedband-contrast (BC) map and color-coded phase map, in

Fig. 5—FEG-SEM micrographs of specimens quenched to 125 �C and partitioned at 250 �C for (a) 3 s, (b) 100 s, and (c) and (d) 1000 s. Thesquare in (c) is enlarged in (d), showing a block of tempered martensite containing carbide precipitation.

Fig. 6—(a) and (b) FEG-SEM micrographs of a specimen quenched to 125 �C and partitioned at 350 �C for 10 s. The square in (a) is enlargedin (b), showing a block of tempered martensite containing carbide precipitation.


which blue corresponds to bcc lattice, red corresponds tofcc lattice, and darker areas correspond to a very low BC,most probably indicating martensite. Retained austeniteis observed with an equiaxed morphology and is situatedvery close to martensite, near the ferrite boundaries.

A similar combination of figures is shown in Figure 9,for the specimen quenched to 150 �C and partitioned at350 �C for 1000 seconds. In this case, austenite isobserved with a filmy morphology and surrounded byferrite. This microstructure resembles, in morphologyand size, the one presented in Figure 7 and identified asbainite. This austenite morphology was not observedafter partitioning for 10 seconds and reinforces themicrostructural interpretation done in previous SectionIII–A, III–B.

The arrows in Figure 9(a) indicate the presence ofepitaxial ferrite. Figure 10(a) shows a combined BC andorientation image map of the microstructure analyzed inFigure 9. It is clear that epitaxial ferrite shares the grainorientation with the surrounding ferrite. Moreover, thepoint-to-origin misorientation profile along the blackstraight line shown in Figure 10(a) was analyzed, withpoint 1 being the origin of the analysis. The resultingmisorientation profile is shown in Figure 10(b). Themisorientation between the two neighboring ferrite

grains analyzed does not show significant changes whenthe boundaries between intercritical and epitaxial ferriteare reached. This observation confirms the formation ofepitaxial ferrite by the growth of the intercritical ferritein the absence of nucleation.

D. XRD and Magnetic Measurements

One of the desired goals in the Q&P process isobtaining multiphase microstructures with an adequatevolume fraction and adequate carbon enrichment of theretained austenite. Therefore, the characterization of theretained austenite in volume fraction and carbon con-tent is of pre-eminent importance in this investigation.Figures 11(a) and (b) show the volume fraction of

retained austenite (measured by the magnetizationmethod) and its carbon content (from XRD) in thespecimens thermally treated with a partitioning temper-ature of 250 �C. The horizontal solid line represents thecorresponding measurements in the specimen directlyquenched. The volume fraction of retained austenite forall the quenching temperatures and partitioning times israther close to the one measured in the specimen directlyquenched, leading to values between 0.03 and 0.06.Carbon content measurements are also close to the

Fig. 7—FEG-SEM micrographs of specimens partitioned at 350 �C for 1000 s: (a) microstructure of specimen quenched at 150 �C, (b) detail ofbainite corresponding to the square drawn in (a), (c) bainite in specimen quenched at 125 �C, and (d) bainite in specimen quenched at 175 �C.(UB is upper bainite, TM is tempered martensite, and RA/UM is retained austenite or untempered martensite).


value obtained in the direct-quenched specimen and areapproximately constant along the partitioning times,with a slight tendency to decrease in the case ofpartitioning for 1000 seconds.

The same measurements corresponding to the speci-mens treated with a partitioning temperature of 350 �Care presented in Figures 11(c) and (d). In this case, somecommon tendencies are observed for every quenching

Fig. 8—EBSD analysis of the sample quenched to 150 �C and partitioned at 350 �C for 10 s: (a) secondary electron image of the scan analysis;and (b) combined BC map and color-coded phase map corresponding to the scan shown in (a), in which blue corresponds to bcc lattice, red cor-responds to fcc lattice, and darker areas correspond to a very low BC, most probably indicating martensite.

Fig. 9—EBSD analysis of the sample quenched to 150 �C and partitioned at 350 �C for 1000 s: (a) secondary electron image of the scan analy-sis; and (b) combined BC map and color-coded phase map corresponding to the scan shown in (a), in which blue corresponds to bcc lattice, redcorresponds to fcc lattice, and darker areas correspond to a very low BC, most probably indicating martensite.


Fig. 10—(a) Combined BC and orientation image map of the microstructure analyzed in Fig. 9 and (b) misorientation measured with respect tothe first point along the line shown in (a).

Fig. 11—(a) Volume fraction and (b) carbon content of retained austenite, for Q&P heat treatments with partitioning at 250 �C; (c) volume frac-tion of retained austenite; and (d) carbon content for partitioning at 350 �C. The horizontal solid line represents the corresponding measurementin the direct-quenched specimen; shaded areas represent the estimated error of this measurement.


temperature studied: (1) after the partitioning for 3 and10 seconds, the volume fraction of austenite and thecarbon content are similar to the specimen directlyquenched, with a slight decrease in volume fraction at10 seconds, in comparison to the result at 3 seconds; (2)there is a substantial increase in the volume fraction ofaustenite after partitioning for 100 seconds, withoutchanges in the carbon content; and (3) at 1000 seconds ofpartitioning, there is a decrease in the volume fraction ofaustenite, with an obvious increase in its carbon content.

IV. DISCUSSION

A. Cooling from Intercritical Region

The calculation of the amount of ferrite remainingafter the partial austenitization at 900 �C using MTDA-TA (National Physical Laboratory, Teddington, UK)leads to a volume fraction equal to 0.34. However, theamount of ferrite found experimentally was 0.74 ± 0.05,which is almost two times greater than the amountexpected from this calculation. This important discrep-ancy originates from the formation of epitaxial ferrite.

The formation of epitaxial ferrite during the coolingstep of the heat treatments has important implications inthe subsequent microstructure before the partitioningstep. As schematically represented in Figure 12, epi-taxial ferrite formation introduces carbon gradients inthe remaining austenite during the first cooling step,leading to more carbon enrichment close to the austeniteboundaries and less carbon in areas well inside theaustenite grains. As a result, martensite formed at thequenching temperature will be more likely formed insidethe austenite grains, leaving the austenite that is closerto the boundaries untransformed. This morphology canbe observed in Figure 5(d), in which the martensite thatlooks tempered is situated in the center of the formeraustenite grain. This martensite is formed during thefirst quench of the Q&P process and tempered duringthe partitioning step, while the surrounding austenite

remains untransformed. After the second quench, thesurrounding austenite is either transformed to (untem-pered) martensite or retained, in both cases showing asmooth surface by scanning microscopy.A consequence of this morphology is that the initial

carbon content in the martensite and austenite beforethe partitioning step is not equal; it is higher in theaustenite than in the martensite. Therefore, the end ofthe carbon partitioning from martensite to austeniteduring the partitioning step will be reached at an earlierstage than is the case when the carbon content is thesame in both phases. Moreover, this additional carbonenrichment of the austenite is favorable for stabilizingthis phase.However, an excess of epitaxial ferrite would reduce

the amount of martensite formed at the quenchingtemperature to levels that could not be interesting interms of the strength of the steel. Therefore, it isimportant to control the formation of epitaxial ferriteduring the quenching step, in order to ensure anadequate volume fraction of austenite before the for-mation of martensite.The quenching temperature determines the amount of

martensite and austenite prior to the partitioning step.Since the experimental Ms temperature is known, theamount of martensite formed after quenching to tem-peratures of 125 �C and 175 �C is estimated by using theKoistinen and Marburger equation.[18] Table II showsthe calculated volume fractions of the phases. Theseestimations show that the volume fraction of austenite

Fig. 12—Scheme of morphology and carbon profiles during different stages of the Q&P process. (A is austenite, IF is intercritical ferrite, EF isepitaxial ferrite, and M is martensite.)

Table II. Estimation of Martensite and Austenite VolumeFractions at Every Quenching Temperature

QuenchingTemperature (�C)

Volume Fractionof Martensite

Volume Fractionof Austenite

125 0.20 0.06150 0.18 0.08175 0.16 0.10


available for further carbon partitioning is limited tovalues lower than 0.10 in the case of quenching to175 �C, and lower yet for quenching at 150 �C and125 �C, as a consequence of the formation of epitaxialferrite during cooling.

B. Partitioning at 250 �CIn a previous study,[19] the present authors have

simulated the kinetics of carbon partitioning frommartensite to austenite based on the constrained carbonequilibrium assumptions,[20,21] considering the alloystudied in this work. These calculations have been usedhere to make an interpretation of the experimentalresults observed for different partitioning temperaturesand times. In particular, calculations corresponding tothe case of quenching at 125 �C and 175 �C andpartitioning at 250 �C showed that the completion ofthe carbon partitioning and the homogenization ofcarbon in the austenite are reached after partitioning forapproximately 105 seconds. Therefore, after partitioningtimes of 3, 10, and 100 seconds, the carbon diffusion isnot substantial enough to get an additional carbonenrichment and, consequently, a volume fraction in-crease in the retained austenite, which is in agreementwith the experimental results. Given that, experimen-tally, the volume fraction and carbon content of theretained austenite is similar to those obtained in thespecimen directly quenched, the carbon enrichment ofthe austenite due to the formation of epitaxial ferrite isprobably the process responsible for the austeniteretention in these specimens.

As is shown in Figure 5, carbide precipitation wasobserved in specimens partitioned for 1000 seconds at250 �C. This carbide precipitation is detrimental to themicrostructural characteristics of the material, because itreduces the amount of carbon to be partitioned to theaustenite. Carbide precipitation explains the absence ofcarbon enrichment (and the constant volume fraction) ofthe austenite in specimens partitioned for 1000 seconds.In particular, this carbide precipitation is probablyresponsible for the decrease in the carbon content inthe austenite in specimens quenched at 125 �C and150 �C after 1000 seconds of partitioning (Figure 11(b)).

As has been shown in Table II, different quenchingtemperatures lead to slightly different amounts ofaustenite available for carbon partitioning during thepartitioning step. However, apart from small differencesin the carbon content of the retained austenite, whichare not significant if the error bars are taken intoaccount, no significant differences have been observedbetween the resulting microstructures formed underdifferent quenching temperatures.

C. Partitioning at 350 �CDuring partitioning at 350 �C, the volume fraction

and carbon content in the austenite show complexvariations (Figures 11(c) and (d)) that are independentof the quenching temperature used. The calculations ofcarbon partitioning kinetics at 350 �C presented inReference 19 showed that carbon partitioning from

martensite to austenite is in an advanced stage afterisothermal treatment for 100 seconds, whereas thecompletion of the process, including the homogeniza-tion of carbon in the austenite, takes place after1000 seconds. Partitioning during 3 or 10 seconds isnot long enough to get considerable carbon enrichmentof the austenite and, consequently, not long enough tofurther stabilize this phase. Therefore, the modelingresults can justify that the carbon content and volumefraction of austenite after 3 and 10 seconds of parti-tioning are similar to those obtained on the specimendirectly quenched; this is also true for the increase in thevolume fraction observed after 100 seconds of parti-tioning. Moreover, the initial stages of the carbonpartitioning process, from the relatively carbon-richmartensite to austenite, generate steep carbon gradientsin the austenite close to the austenite/martensite inter-face, which can be interpreted as creating possible sitesfor the formation of carbides. This considerationexplains the carbide precipitation in the martensiteobserved close to the ferrite boundary in the specimenquenched at 125 �C and partitioned 350 �C for 10 sec-onds (Figure 4(b)).After 1000 seconds of partitioning, all the specimens

show a decrease in the volume fraction of austenite anda significant increase in the carbon content of theretained austenite. The reason for this behavior is theobserved decomposition of the austenite in bainite(Figure 5). Bainite formation takes place through thesuccessive nucleation and diffusionless growth of bain-itic ferrite plates that soon afterward partition the excessof carbon to the surrounding austenite. The growth ofeach bainitic ferrite plate continues until it is stifled bythe strength of the residual austenite.[22,23] Cementitecould then precipitate from the enriched austenitebetween the bainitic ferrite plates. However, given thatthe steel is alloyed with a considerable amount ofaluminum, carbide precipitation between the bainiticferrite plates is likely to be inhibited. Therefore, thecarbon that is rejected from the bainitic ferrite enrichesthe residual austenite, leading to an additional carbonenrichment of this phase that explains the obviousincrease in the carbon content experimentally deter-mined after partitioning for 1000 seconds. Moreover,this process involves a reduction in the volume fractionof austenite, because part of the austenite transforms tobainitic ferrite plates.To understand the formation of bainite at this

temperature, the bainite start (Bs) temperature corre-sponding to the residual austenite just before themartensitic transformation (taking into account theformation of epitaxial ferrite) has been calculated usingthe procedure proposed by Bhadeshia,[24,25] whichincludes the effects of C, Si, Al, and Mn. Thiscalculation results in Bs = 360 �C, which means thata partitioning temperature of 350 �C is inside thetemperature range of the bainite formation and that,therefore, its formation would be possible during thepartitioning step, if the conditions for nucleation arefavorable.The thermodynamic method developed by Bhadeshia[26]

for the calculation of bainite incubation times, which is


based on Russell’s method,[27] has been used for theestimation of the incubation times for bainite at 250 �Cand 350 �C, considering the chemical composition of theaustenite just before the martensitic transformation.Following this method, the incubation time for theformation of bainite is given by

lnt DGmð Þp

Tz

�� ¼ Q0

RTþ C ½4�

where t is the incubation time; T is the temperature; R isthe gas constant; DGm is the available driving force fornucleation, calculated as a function of the activity ofcarbon in ferrite and austenite; and C, p, and z areconstants. Appropriated values for C, p, and z can befound in Reference 26. According to this calculation, theincubation time for bainite at 250 �C is 1330 seconds,which explains the absence of bainite in the specimenspartitioned at this temperature (250 �C is a temperaturebelow Ms, but bainite has been shown to form isother-mally at temperatures below Ms

[28]). On the other hand,the calculated incubation time for bainite at 350 �C isonly 20 seconds. This nucleation time reasonably agreeswith the observation of a significant nucleation andgrowth of bainite in the specimen partitioned for1000 seconds; this is in comparison with the specimenpartitioned for 100 seconds, in which bainite was notobserved.

During partitioning at 350 �C, the overall microstruc-ture evolution has been similar for every quenchingtemperature used, as is observed for partitioning at250 �C. Therefore, this indicates that, although theaustenite available for the partitioning step is slightlydifferent for every quenching temperature, variations ofapproximately 50 �C in the quenching temperature donot significantly affect the final microstructure.

D. Additional Comments

The application of Q&P heat treatments with partialaustenitization has led to a microstructure formed by alarge amount of ferrite (0.74), which limits the effective-ness of this process in producing a novel microstructure.

The microstructures after these heat treatments havealso shown martensite, retained austenite, and, in somecases, bainite. Carbide precipitation has been observedafter partitioning for 1000 seconds at 250 �C and for10 seconds at 350 �C. In general, the degree to whichthese processes overlap with the carbon partitioningfrom the martensite to the austenite, the key to the Q&Pprocess, is considerable.

For the reasons stated earlier, the application ofQ&P heat treatments to typical TRIP chemistries seemsdifficult, because these chemical compositions areoptimized for the promotion of bainite formation. Thisbainite, in TRIP steels, is formed by an isothermaltreatment at a relatively high temperature; it is under-standable, then, that the formation of epitaxial ferritedoes not overly affect the resulting microstructure.However, in the case of the Q&P process, the materialis first cooled to a temperature below the martens-itic start temperature. Therefore, in this case, the

formation of epitaxial ferrite takes place in a highertemperature range, which leads to a higher presence ofthis phase and a significant effect on the final micro-structure.

V. CONCLUSIONS

The Q&P process has been studied in a low-carbonsteel containing 1.1 wt pct aluminum. The main resultsare presented in the following points.

1. Epitaxial ferrite forms during the first cooling ofthe specimens, at a cooling rate of 100 �C/s. Theepitaxial ferrite introduces carbon-content gradientsin the remaining austenite before the formation ofmartensite, leading to higher carbon content in theareas closer to the borders of the austenite grains.Therefore, martensite is formed within the grainduring the quenching step, leaving untransformedthe austenite closer to the grain boundaries.

2. It is experimentally observed that partitioning at250 �C does not lead to an efficient carbon enrich-ment of the austenite. This might be due to the slowpartitioning kinetics and to the possibility of theformation of carbides.

3. The occurrence of carbide precipitation during thepartitioning step at 350 �C starts to be observed onsamples isothermally treated for 10 seconds, but thekinetics of carbon partitioning is a dominant pro-cess at 100 seconds of partitioning time, leading toa considerable increase in the austenite volume frac-tion, to levels of approximately 0.08.

4. At 1000 seconds of partitioning at 350 �C, thedecomposition of the austenite to upper bainiteleads not only to an obvious increase in the carboncontent of the austenite but also to a decrease inthe volume fraction, to values approximately 0.03.

5. The microstructure evolution during the partition-ing step has been observed to be independent of thequenching temperature used, indicating that,although the austenite available for the partitioningstep is slightly different for every quenching temper-ature, variations of approximately 50 �C in thequenching temperature do not significantly affectthe final microstructure in the cases analyzed in thisstudy.

ACKNOWLEDGMENTS

The authors thank Dr. Carlos Garcıa de Andres forproviding the dilatometry facilities at the NationalCentre for Metallurgical Research (CENIM-CSIC) inMadrid, Spain, and acknowledge Mr. Orlando Garcia-Leon and Dr. Roumen Petrov for their assistance andfor the provision of the EBSD facilities at the Univer-sity of Ghent, Belgium. The authors also thankDrs. Dave Hanlon and Theo Kop (Corus RD&T,IJmuiden, The Netherlands) for fruitful discussions,and acknowledge the support Corus RD&T providedfor this project.


OPEN ACCESS

This article is distributed under the terms of theCreative Commons Attribution Noncommercial Li-cense which permits any noncommercial use, distribu-tion, and reproduction in any medium, provided theoriginal author(s) and source are credited.

REFERENCES

1. J.G. Speer, A.M. Streicher, D.K. Matlock, F.C. Rizzo, and G.Krauss: inAustenite Formation andDecomposition, E.B. DammandM. Merwin, eds., TMS/ISS, Warrendale, PA, 2003, pp. 505–22.

2. J.G. Speer, F.C. Rizzo, D.K. Matlock, and D.V. Edmonds:Mater.Res., 2005, vol. 8, pp. 417–23.

3. J.G. Speer, D.V. Edmonds, F.C. Rizzo, and D.K. Matlock: Curr.Opin. Solid State Mater. Sci., 2004, vol. 8, pp. 219–37.

4. D.V. Edmonds, K. He, F.C. Rizzo, A.M.S. Clarke, D.K. Matlock,and J.G. Speer: 1st Int. Conf. Super High Strength Steels, Asso-ciazione Italiana Di Metallurgica, Rome, Italy, 2005, pp. 1–20.

5. D.K. Matlock, V.E. Brautigam, and J.G. Speer: Mater. Sci.Forum, 2003, vols. 426–432, pp. 1089–94.

6. D.V. Edmonds, K. He, M.K. Miller, F.C. Rizzo, A. Clarke, D.K.Matlock, and J.G. Speer: Mater. Sci. Forum, 2007, vols. 539–543,pp. 4819–25.

7. A.K. De, J.G. Speer, and D.K. Matlock: Adv. Mater. Processes,2003, pp. 27–30.

8. L. Zhao, N.H. van Dijk, E. Bruck, J. Sietsma, and S. van derZwaag: Mater. Sci. Eng., 2001, vol. 313A, pp. 145–52.

9. R.W. Cahn and P. Haasen: Physical Metallurgy, 3rd ed., Elsevier,Amsterdam, 1983, pp. 2558–60.

10. R.M. Bozorth: Ferromagnetism, D. Van Nostrad Company, Inc.,New York, NY, 1961, pp. 367–68.

11. F.C. Schwerer, C.E. Spangler, and J.F. Delly, Jr.: Acta Metall.,1978, vol. 26, pp. 579–89.

12. B.D. Cullity: Elements of X-Ray Diffraction, Addison-Wesley,Reading, MA, 1978, pp. 359–67.

13. N.H. van Dijk, A.M. Butt, L. Zhao, J. Sietsma, S.E. Offerman,J.P. Wright, and S. van der Zwaag: Acta Mater., 2005, vol. 53, pp.5439–47.

14. M.J. Santofimia, L. Zhao, R. Petrov, and J. Sietsma: Mater.Charact., 2008, doi:10.1016/j.matchar.2008.04.004.

15. B.M. Bronfin, M.I. Gol’dshteyn, A.A. Yemel’yanov, and A.Z.Shifman: Phys. Met. Metall., 1985, vol. 59, pp. 129–36.

16. S. Zaefferer, J. Ohlert, and W. Bleck: Acta Mater., 2004, vol. 53,pp. 2765–78.

17. G. Ghosh and G.B. Olson: Metall. Mater. Trans. A, 2001, vol.32A, pp. 455–66.

18. D.P. Koistinen and R.E. Marburger: Acta Metall., 1959, vol. 7,pp. 59–60.

19. M.J. Santofimia, L. Zhao, and J. Sietsma: Proc. 2nd Int. Conf.Simulation and Modelling of Metallurgical Processes in Steelmak-ing, Graz, Austria, Andreas Ludwig, ed., ASMET, The AustrianSociety for Metallurgy and Materials, 2007, pp. 368–73.

20. J.G. Speer, D.K. Matlock, B.C. De Cooman, and J.G. Schroth:Acta Mater., 2003, vol. 51, pp. 2611–22.

21. J.G. Speer, D.K. Matlock, B.C. De Cooman, and J.G. Schroth:Scripta Mater., 2005, vol. 52, pp. 83–85.

22. H.K.D.H. Bhadeshia and J.W. Christian: Metall. Trans. A, 1990,vol. 21A, pp. 767–97.

23. S.B. Singh and H.K.D.H. Bhadeshia: Mater. Sci. Eng., 1996, vol.12A, pp. 610–12.

24. H.K.D.H. Bhadeshia: Met. Sci., 1981, vol. 15, pp. 178–80.25. H.K.D.H. Bhadeshia: Bainite in Steels, The Institute of Materials,

London, 2001, pp. 132–34.26. H.K.D.H. Bhadeshia: Met. Sci., 1982, vol. 16, pp. 159–65.27. K.C. Russell: Acta Metall., 1969, vol. 17, pp. 1123–31.28. S.M.C. van Bohemem, M.J. Santofimia, and J. Sietsma: Scripta

Mater., 2008, vol. 58, pp. 488–91.


http://dx.doi.org/10.1016/j.matchar.2008.04.004

Materials Research, Vol. 8, No. 4, 417-423, 2005 © 2005

*e-mail: [email protected]

The “Quenching and Partitioning” Process: Background and Recent Progress

John G. Speera, Fernando C. Rizzo Assunçãob*, David K. Matlocka, David V. Edmondsc

aAdvanced Steel Processing and Products Research Center, Colorado School of Mines, Golden, CO 80401, USA

bDepartment of Materials Science and Metallurgy, Pontifícia Universidad Católica, 22453-900 Rio de Janeiro - RJ, Brazil

cSchool of Process, Environmental and Materials Engineering, University of Leeds, Leeds LS2 9JT, United Kingdom

Received: July 19, 2004; Revised: April 8, 2005

A new process concept, “quenching and partitioning” (Q&P) has been proposed recently for creating steel microstructures with retained austenite. The process involves quenching austenite below the martensite-start temperature, followed by a partitioning treatment to enrich the remaining austenite with carbon, thereby stabilizing it to room temperature. The process concept is reviewed here, along with the thermodynamic basis for the partitioning treatment, and a model for designing some of the relevant processing temperatures. These concepts are applied to silicon-containing steels that are currently being examined for low-carbon TRIP sheet steel applications, and medium-carbon bar steel applications, along with a silicon-containing ductile cast iron. Highlights of recent experimental studies on these materials are also presented, that indicate unique and attractive microstructure/property combinations may be obtained via Q&P. This work is being carried out through a collaborative arrangement sponsored by the NSF in the USA, CNPq in Brazil, and the EPSRC in the United Kingdom.

Keywords: carbon partitioning, retained austenite, martensite

1. Introduction

High strength ferrous alloys containing significant fractions of retained austenite have been developed in recent years, and have important commercial applications. In sheet steels, for example, carbon-enriched metastable retained austenite is considered beneficial because the TRIP phenomenon during deformation can contribute to formability and energy absorption. In gear and bearing surfaces, austenite is considered to provide damage tolerance in rolling/sliding contact fatigue applications. In thicker section structural applica-tions, retained austenite may provide enhanced resistance to fracture. Similarly, austempered ductile cast iron materials develop favorable property combinations through a microstructure of fine ferrite plates in combination with carbon-rich retained austenite.

Steels with substantial amounts of carbon-enriched retained austenite are typically produced by transforming at low tempera-tures, leading to a microstructure containing “carbide-free bainite” that consists of bainitic ferrite laths with interlath retained austenite. Alloying additions such as Si or Al are made to suppress cementite precipitation that usually accompanies bainite formation. Recently, an alternative processing concept, “quenching and partitioning (or Q&P), has been developed for the production of austenite-contain-ing steels, based on a new understanding of carbon partitioning hypothesized between martensite and retained austenite1. This paper reviews the fundamental elements of the process concept, and recent experimental investigations to examine the Q&P processing response of two commercial Si-containing steels and a commercial Si-contain-ing ductile cast iron.

2. Background and Q&P Fundamentals

2.1. Carbon partitioning concept

Carbon partitioning between martensite and retained austenite is usually ignored in quenched steels, because the temperature is normally too low for substantial amounts of carbon diffusion to occur after quenching, and because carbon supersaturation in martensite is ordinarily eliminated by a different mechanism, viz. carbide precipita-tion during tempering. Consequently, while carbon-enriched retained austenite has been identified in martensitic steels for some time2, the thermodynamics of carbon partitioning between martensite and retained austenite has been scarcely considered. Recently, a model has been developed to address carbon partitioning from as-quenched martensite into austenite, under conditions where competing reac-tions such as bainite, cementite or transition carbide precipitation are suppressed1. The model predicts the “endpoint” of partitioning, when martensite (i.e. ferrite) is in metastable equilibrium with austenite.

Metastable equilibrium between austenite and ferrite is not a new concept3, and equilibrium (e.g. orthoequilibrium) and paraequilibrium concepts are well understood at sub-critical temperatures for condi-tions where partitioning of slow-moving substitutional elements is ei-ther complete or absent, respectively. It must be recognized, however, that transformations occurring under equilibrium or paraequilibrium necessarily involve interface migration and thus require short range movements of iron and substitutional atoms, even when long-range substitutional diffusion is precluded as in the paraequilibrium case. When the position of the martensite/austenite interface is effectively

418 Speer et al. Materials Research

constrained, as we consider to apply for carbon partitioning between martensite and retained austenite at relatively low temperatures, then even short-range diffusional movements of iron and substitutionals are precluded, and it is not possible for a ferrite/austenite mixture to reach equilibrium in the Fe-C system (or paraequilibrium in multicomponent alloy systems). The metastable α/γ equilibrium in the case of an immobile or constrained interface, is therefore termed “constrained paraequilibrium” or CPE. Paraequilibrium and CPE derive fundamentally from the immobility of iron and substitutionals in comparison to carbon and other interstitials. Consequently, these two conditions are considered by the authors to be closely related, although this view is not held universally4 and remains the subject of discussion5.

Constrained paraequilibrium is essentially defined by one thermo-dynamic requirement, and one key matter balance constraint. First, carbon diffusion is completed under constrained paraequilibrium conditions when the chemical potential of carbon is equal in the fer-rite and austenite. Ignoring effects of alloying on carbon activity, this requirement may be represented using results of Lobo and Geiger6,7 for the Fe-C binary system as follows:

, . ( , . )x x e

RTT T x76 789 43 8 169 105 120 4

C CC$=

- - -c ac

(1)

where xαC and xγ

C represent the mole fractions of carbon in ferrite and

austenite. The relevant thermodynamics are embedded in Equation 1. This thermodynamic condition may be understood by comparing the schematic Gibbs molar free energy vs. composition diagram in Figure 1a representing metastable equilibrium in the Fe-C system, with constrained paraequilibrium in Figure 1b.

In (ortho) equilibrium, or paraequilibrium in higher order alloys, there are unique ferrite and austenite compositions (xα

EQ and xγ

EQ)

satisfying the common tangent construction whereby the chemical potentials of both carbon and iron are equal in both phases (mα

C = mγ

C

and mαFE

= mγFE

). (In paraequilibrium, the same construction would apply if the vertical axis at the composition of pure iron were replaced by the appropriate composition in multicomponent space representing the relative fractions of iron and substitutional elements in the alloy). In constrained paraequilibrium, the thermodynamic condition that the chemical potential of carbon is equal in both phases requires only that the tangents to the ferrite and austenite free energy curves must intersect the carbon axis at a single point. This condition can be satisfied by an infinite set of phase compositions8, and examples of two such conditions are given in Figure 1b, one which is associ-ated with phase compositions (xα

C -

P I

EI and xγ

C -

P I

EI

) having a higher carbon

concentration than the equilibrium phase compositions, and one as-sociated with phase compostions (xα

C -

P I

E and xγ

C -

P I

E ) having lower carbon

levels than equilibrium. The actual CPE phase compositions must also satisfy the unique matter balance constraint associated with the stationary α/γ interface. This second constraint requires that the number of iron (and substitutional) atoms is conserved in each phase during carbon partitioning. Mathematically, this matter balance for iron may be represented by:

f γCPE

(1 – xγCCPE

) = f γi (1 – x

Calloy) (2)

where xCalloy is the overall carbon content of the steel (in atom fraction,

recognizing also that in Fe-C binary alloys, 1 – xC = x

FE ), f γ

i is the

mole fraction of retained austenite before partitioning begins, and f γCPE

and xγ

CCPE represent the austenite amount and carbon concentration,

respectively, at constrained paraequilibrium when carbon partition-ing is complete. (A small change in austenite fraction is consistent with transfer of carbon atoms across the interface). Constrained paraequilibrium is achieved when Equations 1-2 above, and Equa-tions 3-4 below are satisfied, where the mass balance for carbon is represented by:

f αCPE

xαCCPE

+ f γCPE

xγCCPE

= xCalloy (3)

and the relationship between the phase fractions of α and γ is sim-ply:

f αCPE

+ f γCPE

= 1 (4)

Example CPE calculations have been reported previously1, where it was shown that most of the carbon in the steel is expected to parti-tion to the austenite, and quite high levels of carbon enrichment are possible. The dependence of the metastable CPE condition on alloy carbon content, temperature, and the as-quenched austenite and martensite phase fractions was also illustrated. While the detailed calculations are not difficult, it was found that the austenite composi-tion at constrained paraequilibrium can be closely approximated by assuming that virtually all of the carbon in the martensite partitions to the austenite, and then applying the appropriate carbon matter balance based on the amount of retained austenite present after quenching9.

The results of the constrained paraequilibrium model suggested a new process, whereby austenite is formed at high temperature (either by full austenitization or intercritical heat treatment), followed by cooling to a temperature carefully selected (between M

s and M

f) to

control the fractions of martensite and retained austenite, and finally by a thermal treatment that accomplishes the desired carbon partition-ing to enrich the austenite with carbon and stabilize some (or all) of it to room temperature. This process sequence and the corresponding microstructural changes are illustrated schematically in Figure 2 10. The process assumes that carbon supersaturation is relieved by dif-fusion into retained austenite, and is referred to as quenching and

(a)

(b)

Figure 1. Schematic molar Gibbs free energy vs. composition diagrams il-lustrating metastable equilibrium at a particular temperature between ferrite and austenite in the Fe-C binary system. a) equilibrium (EQ), and b) two possible constrained paraequilibrium conditions (I and II).

Vol. 8, No 4, 2005 The “Quenching and Partitioning” Process: Background and Recent Progress 419

partitioning, or Q&P, to distinguish it mechanistically from conven-tional quenching and tempering (Q&T) of martensite, where carbide precipitation and decomposition of retained austenite (to ferrite plus cementite) are typical. The example in Figure 2 indicates an initial full austenitization step, although intercritical annealing is also en-visioned for formable sheet products containing an equiaxed ferrite component in the microstructure. During intercritical annealing, a smaller initial fraction of austenite would be present with a higher initial carbon content.

The quenching and partitioning heat treatment was envisioned to have application to high-strength austenite containing TRIP sheet steel products, replacing an isothermal bainitic heat treatment of low-carbon steels containing substantial additions of Si, Al, or P to suppress carbide formation. Some suggested advantages of Q&P include the potential for greater carbon enrichment of austenite, decoupling of the (bainitic) ferrite growth kinetics from the carbon partitioning process, and increasing strength via formation of sub-stantial quantities of lath martensite in the microstructure. Other opportunities were identified to employ retained austenite through Q&P processing of higher strength bar steels or even austempered ductile cast iron. Finally, it was suggested that a specific CPE phase composition (where the austenite composition approximates T

o)

might even represent a viable steady state boundary condition at the α/γ interface during bainitic ferrite growth, providing a model for the bainite transformation mechanism that is both “fully” diffusional and “fully” martensitic1.

2.2. Importance of suppressing carbide precipitation

The absence of carbide formation is a fundamental element of the constrained paraequilibrium model, since the existence of metastable equilibrium between ferrite and austenite is precluded if the more stable ferrite plus iron carbide equilibrium can be achieved. Any carbide formation effectively “consumes” carbon, since these carbon atoms are no longer available to enrich the austenite. Thus, it is necessary to understand and control carbide precipitation proc-esses that may occur during any partitioning treatments associated with the Q&P process.

It is well known that cementite formation can be eliminated or suppressed through additions of silicon11,12, and also that aluminum and even phosphorus can produce a similar effect13. Such elements thus play a critical enabling role in the Q&P process. It is also well known in the martensite tempering literature that silicon suppresses cementite formation, or delays the transition from early-stage tem-pering (where ε or η carbides are present), to later-stage tempering (where θ-Fe

3C is present)14-16. In martensite, fine transition carbides

are usually not considered detrimental, whereas cementite can be of more concern. Thus, the greater emphasis has been on understanding when transition carbides are replaced by cementite formation16,17, rath-er than on the initiation of transition carbide precipitation. For Q&P processing, however, any transition carbide precipitation diminishes the potential for carbon enrichment of austenite, and it is necessary to develop a better understanding of the onset of transition carbide formation, including composition and processing effects9,18.

Precipitation of transition carbides within retained austenite during martensite tempering has not been documented. Since the chemical potential of carbon is much higher in as-quenched martensite than in the retained austenite, it is reasonable to conclude that carbide nucleation would be more likely in bcc ferrite than in austenite9,19. The α/γ interface is also a favored site for carbide formation. In the Q&P process, high carbon supersaturation of the martensite prior to partitioning could conceivably drive transition carbide formation to a greater extent than would be possible during bainite growth at the same temperature if bainitic ferrite grows with a much lower carbon content than the austenite. (In this context, it should be noted that the carbon supersaturation of bainitic ferrite during growth remains a subject of controversy). In any event, the extent to which carbide formation is suppressed will be a critical factor influencing the microstructures that are achievable using the Q&P process, and further studies are needed to establish more clearly the influences of alloying and processing on the carbide precipitation behavior and kinetics in these steels.

2.3. Process design (selection of quenching temperature)

A methodology for designing the quench temperature to achieve the maximum possible retained austenite fraction after Q&P process-ing, was developed in a recent publication9. The model ignores partitioning kinetics, and assumes that all of the carbon partitions from martensite to austenite, and that carbide precipitation is avoided completely. The model results are shown in Figure 3, for a 0.19%C,

Figure 2. Schematic illustration of the Q&P process for producing of austen-ite-containing microstructures. C

i, Cγ, Cm

represent the carbon concentrations in the initial alloy, austenite, and martensite, respectively. QT and PT are the quenching and partitioning temperatures10.

Figure 3. Predicted Q&P microstructure components for experimental steel containing 50% intercritical ferrite, vs. quench temperature, assuming full partitioning prior to final quenching to room temperature. The final austenite fraction at room temperature is given by the solid bold line. Dashed lines represent the austenite and martensite (M) present at the initial quench temperature, and the additional martensite formed during the final quench to room temperature. For this example, M

initial quench+ M

final quench+ γ

final = 0.5, and

the intercritical ferrite fraction is 0.5.


1.96%Al, 1.46%Mn, 0.02%Si (by weight) TRIP sheet steel composi-tion, assuming that intercritical annealing was conducted to achieve a microstructure containing 50% austenite and 50% ferrite prior to quenching. In this figure, the final austenite fraction after partitioning and cooling to room temperature is plotted (bold solid line) vs. the quenching temperature prior to partitioning. The austenite and mar-tensite fractions at the quench temperature are also plotted, along with the fraction of “fresh” martensite that forms during final cooling.

The model first estimates the fractions of austenite and martensite at the quench temperature (QT in Figure 1) based on the undercool-ing below M

s, according, for example, to the Koistinen-Marburger14

relationship:

fm = 1 – e – 1.1x10-2(Ms – QT) (5)

where fm is the fraction of austenite that transforms to martensite

upon quenching to a temperature QT below the Ms temperature, and

Ms for the applicable austenite composition can be estimated from

published correlations. (For processing where intercritical annealing is conducted rather than full austenitization, the initial carbon con-centration of the austenite is controlled by the intercritical annealing temperature, and may be estimated by assuming that nearly all of the carbon in the steel is contained in the austenite, since the carbon solubility in ferrite is very low). After completion of (full) partition-ing between martensite and austenite subsequent to quenching, the carbon concentration in the remaining austenite may be estimated, and the final phase fractions may be predicted after final cooling, again applying the Koistinen and Marburger relationship to the carbon-enriched austenite.

The model results indicate an “optimum” quenching temperature that yields a maximum amount of retained austenite. Above the peak temperature, substantial austenite fractions remain after the initial quenching step, but the austenite stability is too low during final quenching, and increasing amounts of fresh (M

final quench) martensite

are found at higher quench temperatures, reducing the final austenite fraction at room temperature. Below the peak temperature too much austenite is consumed during the initial quench prior to carbon par-titioning, and the carbon content of the retained austenite is greater than needed for stabilization at room temperature. The peak is found at the particular quench temperature where martensite formation is just precluded during the final quench, whereby the austenite has an M

s temperature of room temperature after full partitioning. This

methodology provides guidance for experimental processing design, and allows the effects of changes in a variety of processing variables to be explored and predicted. Partitioning kinetics are not predicted in this simple model, however, and development of a more sophisticated model will require further understanding of the length-scale of the microstructure over which partitioning occurs, and the kinetics of carbide precipitation processes that may occur.

3. Highlights of Recent Progress

3.1. Medium-carbon bar steels

Initial investigation of the Q&P processing concept verified the presence of significant amounts of carbon enriched austenite in a 0.35%C, 1.3%Mn, 0.74%Si (wt. pct.) microalloyed bar steel, despite the apparent formation of some transition carbides during the parti-tioning treatment10. More recently, the Q&P processing response of a 0.6%C, 2%Si (grade 9260) steel was examined by Gerdemann, and compared to the results of conventional austempering or quenching (to room temperature) and tempering20. Wafers (28.5 mm in diameter by 2.5 mm thick) were austenitized in molten salt for 15 minutes at 900 °C, quenched into a molten tin-bismuth bath at temperatures

ranging between 150 and 210 °C, and equilibrated for 120 seconds before partitioning at temperatures between 250 and 500 °C in mol-ten salt for times ranging between 10 and 3600 seconds, and finally, quenched to room temperature. The quenching temperatures were designed using the methodology described above.

The results showed that substantial levels of retained austenite could be achieved by Q&P processing of the 9260 alloy, approaching 30% by volume. The relationship between the amount of retained austenite and the quench temperature is reproduced here in Figure 4, for conditions involving a 10 seconds partitioning treatment at 500 °C. The figure shows that the amount of austenite measured by X-ray diffraction was in qualitative agreement with model calcula-tions, although the measured austenite fractions were lower than the maximum amounts predicted.

Partitioning at lower temperature (250 °C) led to partition-ing treatment times that would be more appropriate for industrial processing of bulk specimens (e.g. 45 to 60 minutes), whereas much shorter times were associated with the maximum austenite fractions at higher temperature (e.g. 10 seconds at 400 °C). Some encouraging property results were noted in this study, such as hardness levels in excess of HRC58 in combination with austenite fractions approaching 10%. In contrast, substantial austenite levels were not achievable by conventional quenching and tempering, and lower hardnesses were associated with bainitic processing (austempering). The combination of high hardness along with a significant retained austenite fraction is considered to be of possible interest for gear or bearing applications, where “damage tolerance” under pitting or contact fatigue conditions is enhanced by austenite that is present in the microstructure21.

Microstructure characterization is currently underway, and Figure 5 shows an example resulting from quenching to 190 °C, and holding in the bath for 120 seconds. Transmission electron microscopy (TEM) shows the martensite substructure in bright field (Figure 5a), along with finely dispersed retained austenite in dark field (light regions in Figure 5b). This heat-treatment condition is associated with much more retained austenite (> 6%) than is obtained by quenching directly to room temperature (< 2%), illustrating that partitioning has already begun during the 120 seconds equilibration at the quench temperature (190 °C)20.

Figure 4. Final volume fraction of retained austenite depending on the quench temperature at a partitioning temperature of 500 °C, and calculated austenite volume fraction over this quench temperature range20.


3.2. TRIP sheet steels

High strength sheet steels containing significant fractions of retained austenite have been developed in recent years, and are the subject of growing commercial interest23,24. Carbon-enriched meta-stable retained austenite is considered beneficial because the TRIP phenomenon during deformation can contribute to formability and energy absorbtion. These steels are typically produced by intercritical annealing followed by austempering, with additions of Si, Al, or P to suppress carbide formation that usually accompanies the bainite transformation. Initial studies on the Q&P processing response9 of TRIP sheet steel showed that substantial amounts of austenite could be obtained via Q&P processing, with measured retained austenite fractions similar to the predicted maximum of 15% in this 0.19%C, 1.96%Al, 1.46%Mn steel. Because of concerns related to uncertainty in the effects of aluminum on the M

s temperature, and

overlapping of the carbon partitioning and bainite transformation mechanisms owing to accelerated austenite decomposition kinetics associated with aluminum additions, more recent studies have been

conducted using a 0.19%C, 1.63%Si, 1.59%Mn TRIP sheet steel18. Transformation response and mechanical behavior are both being assessed, and initial results have been very encouraging. Variations in quenching temperature were examined, along with selected vari-ations in partitioning time and temperature, using either “1-step” or “2-step” Q&P processing. In 1-step processing, partitioning is carried out at the quenching temperature, while 2-step processing involves reheating to a selected partitioning temperature that differs from the quench temperature.

New microstructures that extend the strength levels of current TRIP steels resulted from Q&P processing, as shown in the results of Figure 6, comparing the measured strength and formability (ductility) combinations with current “state-of-the-art” sheet grades including dual-phase (ferrite-martensite), austempered TRIP (bainite), and martensitic steels. (The data used in this figure are discussed further in reference18.) Much additional opportunity remains to explore available property combinations, and optimize retained austenite fractions and austenite stability, as well as to understand the operative fundamental mechanisms and explore industrial processing capabilities. Scanning electron microscopy, as illustrated in the example of Figure 7, shows the presence of intercritical ferrite (dark featureless areas), along with a mixture of martensite and fine retained austenite. The fine sub-structure in the Q&P heat treated condition is apparently responsible for the elevated strength levels and distinguishes the resulting Q&P microstructure from bainite produced by conventional austempering at the same temperature as partitioning is accomplished in Q&P18.

3.3. Austempered ductile cast iron

Austempered ductile iron (ADI) contains substantial levels of silicon, and is usually processed by heating into the austenite-plus-graphite phase field, followed by austempering at a lower temperature to transform the austenite to “ausferrite,” which is essentially bainitic ferrite with carbon-enriched retained austenite. This microstructure provides ADI with high strength in combination with ductility and toughness that is sufficient for many applications. Because of the high-silicon levels and the importance of retained austenite, Q&P was considered to offer a potential heat treating alternative for ADI, and a team of 4th year undergraduate students at Colorado School of Mines

(a)

(b)

Figure 5. TEM bright field (a) and (002)γ dark field (b) images showing martensite and retained austenite in 9260 alloy quenched to 190 °C and equilibrated for 120 seconds before final cooling to room temperature22.

Figure 6. Total elongation vs. ultimate tensile strength for TRIP, Dual phase (DP), martensitic (M), and Q&P sheet steel products18.


has recently examined the response of a commercial 3.7%C, 2.5%Si, 0.34%Mn, 0.17%Cu ductile cast iron grade to Q&P processing, as part of a “capstone” materials-design course project25. A limited number of quenching (137, 157, and 172 °C) and partitioning temperatures (230, 270, 310, 350, and 390 °C) was assessed in this work, while the “austenitizing” temperature (850 °C) and partitioning time (1800 sec-onds) were held constant. The quenching temperatures were selected for austenite compositions predicted at 850 °C using Thermo-Calc software26 to estimate the austenite/graphite equilibrium condition.

The results again showed that substantial fractions of carbon-en-riched austenite could be retained via Q&P processing, although the austenite levels were found to be lower than obtained via austempering under the processing conditions evaluated. The strengths were greater in the Q&P condition, while the ductility and room-temperature impact properties were lower. Further experimentation and process optimization is needed to determine whether Q&P represents an advantageous alternative in this application.

4. Concluding Remarks

The Q&P process concept has been reviewed, along with the fun-damental thermodynamics of carbon partitioning between martensite (i.e. carbon-supersaturated ferrite) and austenite. A methodology for predicting the amount of retained austenite as a function of quench-ing temperature was described, based on the assumptions that full partitioning of carbon is achieved, and that carbide precipitation is fully suppressed. Recent experimental studies have confirmed that Q&P provides a viable means to create microstructures containing carbon-enriched retained austenite, and attractive property combina-tions have been achieved in a variety of materials. Carbon concentra-tions in the austenite vary with partitioning treatment parameters, and were found to reach levels in the 1.5% to 1.7% range for the materials examined here. In most cases, the measured austenite fractions and austenite carbon concentrations are less than values predicted from theory. Thus it appears that partitioning is generally interrupted before the hypothesized “constrained paraequilibrium” condition. Carbide precipitation is a likely means by which austenite enrichment and retention are reduced, especially in light of the substantial carbon supersaturation of martensite, and X-ray diffraction has confirmed the presence of carbides in some microstructures10,18. Further work is currently underway by the project team to characterize better the carbide precipitation behavior and other features of the microstructure

using high-resolution TEM techniques. Initial results have already confirmed the presence of ε-carbide in the 9260 bar steel after quench-ing to 150 °C and partitioning for 3600 seconds at 250 °C22. Clearly, substantial work remains to understand the fundamental transforma-tion characteristics of these alloys during Q&P processing, and to optimize the chemical compositions, microstructures and properties, although the results to date are of great encouragement.

Acknowledgments

The authors acknowledge the support of the Advanced Steel Processing and Products Research Center, a National Science Foundation Industry/University Cooperative Research Center at the Colorado School of Mines, and the Inter-American Materials Collaboration Program. Funding is acknowledged from NSF award 0303510 (USA), CNPq (Brazil), and EPSRC (United Kingdom). POSCO, Ispat-Inland Steel Company, MACSTEEL, North Star Steel, Applied Process, Inc., and Farrar Corporation are acknowledged for providing experimental materials.

References1. Speer JG, Matlock DK, De Cooman BC, and Schroth JG. Carbon parti-

tioning into austenite after martensite transformation. Acta Materialia. 2003; 51:2611-2622.

2. Sarikaya M, Thomas G, Steeds JW, Barnard SJ, Smith GDW. Solute ele-ment partitioning and austenite stabilization in steels. In: Proceedings of an International Conference on Solid to Solid Phase Transformations, ed. H.I Aaronson, Warrendale, PA: TMS; 1982. p. 1421-1425.

3. Hultgren A. Isothermal transformation of austenite. ASM Transactions. 1947; 39:915-1005.

4. Hillert M, Agren J. On the definitions of paraequilibrium and orthoequi-librium. Scripta Materialia. 2004; 50:697-699.

5. Speer JG, Matlock DK, De Cooman BC, Schroth JG. Comments on “On the Definitions of paraequilibrium and orthoequilibrium”. Scripta Materialia. 2005; 52:83-85.

6. Lobo JA, Geiger GH. Thermodynamics and solubility of carbon in ferrite and ferritic Fe-Mo alloys. Metallurgical Transactions A. 1976; 7A:1347-1357.

7. Lobo JA, Geiger GH. Thermodynamics of carbon in austenite and Fe-Mo austenite. Metallurgical Transactions A. 1976; 7A:1359-1364.

8. Hillert M, Hoglund L, Agren J. Escape of carbon from ferrite plates in austenite. Acta Metallurgica et Materialia. 1993; 41:1951-1957.

9. Speer JG, Streicher AM, Matlock DK, Rizzo FC, Krauss G. Quenching and partitioning: a fundamentally new process to create high strength TRIP sheet microstructures. In: Damm EB and Merwin M. Editors. Austenite Formation and Decomposition, Warrendale, PA: TMS/ISS; 2003. p. 505-522.

10. Matlock DK, Bräutigam VE, Speer JG. Application of the quenching and partitioning (Q&P) process to a medium-carbon, high-Si microalloyed bar steel. In: Proceedings of THERMEC’ 2003, Uetikon-Zurich, Switzerland: Trans. Tech. Publications, Inc.; 2003. p. 1089-1094.

11. Bhadeshia HKDH. Bainite in Steels, London: The Institute of Materials; 2001.

12. Bhadeshia HKDH, Edmonds DV. The bainitic transformation in a silicon steel. Metallurgical Transactions A. 1979; 10A:895-907.

13. Gallagher MF, Speer JG, Matlock DK, Fonstein NM. Microstructure development in TRIP-sheet steels containing Si, Al, and P. In: Proceed-ings of the 44th Mechanical Working and Steel Processing Conference, Warrendale, PA: ISS; 2002. p. 153-172.

14. Krauss G. STEELS: Heat Treatment and Processing Principles, Metals Park, OH: ASM International; 1990.

15. Honeycombe RWK, Bhadeshia HKDH. Steels, Microstructure and Properties, London: Edward Arnold; 1995.

16. Krauss G, Tempering and structural change in ferrous martensitic

Figure 7. Microstructure of Si-TRIP sheet steel containing 8.4% austenite, processed by intercritical annealing (75% γ + 25% α) followed by quenching to 200 °C and partitioning at 400 °C for 10 seconds. Etched in 2% nital.


structures. In: Proceedings of an International Conference on Phase Transformations in Ferrous Alloys, Warrendale, PA: The Metallurgical Society of AIME; 1984. p. 101-123.

17. Lee HC, Krauss G. Intralath carbide transitions in martensitic medium-carbon steels tempered between 200 and 300C. In: Gilbert R. Speich Symposium Proceedings: Fundamentals of Aging and Tempering in Bainitic and Martensitic Steel Products, Warrendale, PA: Iron and Steel Society; 1992. p. 39-43.

18. Streicher AM, Speer JG, Matlock DK, De Cooman BC. Quenching and partitioning response of a Si-added TRIP sheet steel. In: Speer JG Editor. Proceedings of the International Conference on Advanced High-Strength Sheet Steels for Automotive Applications, Warrendale, PA: AIST; 2004. p. 51-62.

19. Barbe LL, De Cooman BC, Conlon K. Characterization of the metastable austenite in low-alloy FeCMnSi TRIP-aided steel by neutron diffraction. Zeitschrift fur Metallkunde. 2002; 93:1217-1227.

20. Gerdemann FLH. Microstructure and Hardness of 9260 Steel Heat-Treated by the Quenching and Partitioning Process, Dipl. Thesis, Aachen University of Technology (RWTH), Germany; 2004.

21. Roache B, Pitsko R, Chidester A. Imundo J. SAE Technical Paper #2002-01-1371, Warrendale, PA: SAE; 2002.

22. He K. Unpublished research, University of Leeds, 2004.

23. Proceedings of the International Conference on Advanced High-Strength Sheet Steels for Automotive Applications, ed. J.G. Speer, Warrendale, PA: AIST; 2004.

24. Proceedings of the International Conference on TRIP-Aided High Strength Ferrous Alloys, ed. B.C. De Cooman, Aachen, Germany: Wissenschafts-verlag Mainz GmbH; 2002.

25. Unger S, Grahmann J, Shutts A, Wolf C, Gibbs J. Unpublished research, Colorado School of Mines; 2004.

26. Andersson JO, Helander T, Höglund L, Shi PF, Sundman B, Calphad. 2002; 26:273-312.

Carbon partitioning into austenite after martensite ... › public ›...

Documents

Transcript of Carbon partitioning into austenite after martensite ... › public ›...