Coarsening and dendritic instability of spheroidal graphite in high silicon cast iron under thermal cycling in the ferritic domain

OATAO is an open access repository that collects the work of Toulouse

researchers and makes it freely available over the web where possible

Any correspondence concerning this service should be sent

to the repository administrator:

tech-oatao@listes-diff.inp-toulouse.fr

This is an author’s version published in:

http://oatao.univ-toulouse.fr/25153

To cite this version:

Ebel, André

and Pegoraro, Mariana Alves

and Malard, Benoît

and

Tenailleau, Christophe

and Lacaze, Jacques

Coarsening and dendritic

instability of spheroidal graphite in high silicon cast iron under thermal cycling in

the ferritic domain. (2020) Scripta Materialia, 178. 86-89. ISSN 1359-6462

Coarsening

and

dendritic

instability

of

spheroidal

graphite

in

high

silicon

cast

iron

under

thermal

cycling

in

the

ferritic

domain

André Ebel

a ,b

_,

_Mariana

_Alves

_Pegoraro

a

_,

_Benoit

_Malard

a

_,

_Christophe

_Tenailleau

a

_,

Jacques

Lacaze

a ,∗

a CIRIMAT, ENSIACET, Université de Toulouse, CNRS, BP 44362, 31030 Toulouse Cedex 4, France b LCTS, Université de Bordeaux, France

a

r

t

i

c

l

e

i

n

f

o

Keywords:

Cyclic heat treatment Dendritic growth Graphite

Spheroidal graphite cast iron

a

b

s

t

r

a

c

t

Highsiliconferriticspheroidalgraphitecastironshavebeendevelopedforhightemperatureservice,in particularunderthermalcyclingconditions.Thetheoreticalmaximumservicetemperatureisdefinedas theupperlimitofthetwo-phaseferrite+graphitedomain,whichincreaseswiththealloysiliconcontent. Whileisothermalheattreatmentclosetothistemperatureshowedlittleevolutionofthegraphite distri-bution,thermalcyclingledtoasignificantcoarseningofthegraphiteparticlesassociatedwithdendritic overgrowthofthelargegraphiteparticles.Thisunexpectedbehaviourishereobservedanddescribedfor thefirsttime.

Studies devoted to hightemperaturebehaviour of ferritic cast ironshavebeenconcernedwithoxidationresistanceand mechan-ical properties.Because siliconis known to increase both oxida-tion resistance andthe upper temperature of the domain where ferrite is stable, high-silicon spheroidal graphite cast irons (SGI) havebeendevelopedsincethe1980s[1 ,2] .Thesealloysareusedin manyapplicationsincludingloadingunderthermalcycling.Studies abouttheeffectofthermalcyclingontensileproperties[3 ,4] and about the thermomechanical behaviour [5 ,6] of high silicon SGI havebeen carried out withmaximum temperature up to 800°C whichwas within the ferritic domain forthe investigated alloys. These studies involved a limited number of cycles and the only changesinmicrostructurethatwerereportedconcernedoxidation ofinclusionsassociatedwiththe last-to-solidifyareasandpartial recrystallizationofferrite forthehighestmaximumtemperatures (750°Cand800°C).

Investigating the behaviour of high-silicon SGI when submit-tedto a much higher number ofthermal cycles isof interest as thismaybe morerepresentative oftheactual lifecycleof indus-trialcomponents.Inorderto doso,a material containingmainly 3.10wt.%C,4.45wt.%Si,0.25wt.%Mnandalso0.0037wt.%Sb (bal-ance Fe) was selected. Before casting, the melt was treated for graphite spheroidisation and inoculated as described previously

[7] . Metallographic observation showed a fully ferritic matrix in

∗ _{Corresponding author.}

E-mail address: Jacques.lacaze@ensiacet.fr (J. Lacaze).

the as-caststate, with an evendistribution of graphitespheroids asseen in Fig. 1 a. The surfacenumber densityof spheroids was 580particlespermm2 _{andthegraphitefractionwasabout0.1.}

Twotypes of experiments were run on 10_{× 20× 2}mm3

sam-ples:isothermalheattreatmentandcyclicheattreatment. Isother-malheattreatmentwasperformedinamufflefurnacefor50hat 800°C.Thermalcyclingwasperformedbetween100°Cand800°C ina dedicatedfacility.Inthisfacility, 10sampleswere hungto a holderwhich wascyclically introduced intoa furnacepre-heated to 800°C, and then withdrawn from the furnace andfan-cooled beforebeingheatedup again. Oneofthe sampleshada thermo-coupleweldedonit forcontinuouscontrol ofthethermalcycles. The whole thermal cycle duration was 480s, including 360s of heating,60s ofholdingat800_{± 10}°Cand60sofcooling.Partof thesamples waswithdrawnforanalysisafter 1000,2000 and fi-nally3000cycles.Thislatternumberofcycleswasselectedsothat thetimenominallyspentat800°Cshouldbe50h.

Metallographic observations and2D image analysis were car-ried out on the bulk of the heat treated specimens which was notaffectedbysurfaceoxidationanddecarburization.Image anal-ysiswascarriedouttoquantifytheevolutionofthegraphite vol-umefraction,particles numberdensityandsize distribution. Par-ticlesize wasevaluated usingan equivalent diameterDequal to 4_·A

P,whereAandParerespectivelythe areaandperimeterofa

particle.Measurementswerecarriedouton10micrographstaken alongthecentreofthesamplesectionandthenaveraged.

Fig. 1 bshows that isothermal heat treatment may resultin a veryslightcoarseningofthegraphitespheroidswhichwasbetter

Fig. 1. Optical micrographs of the alloy in the as-cast state (a), after isothermal treatment of 50 h at 800 °C (b), and after 30 0 0 thermal cycles (c).

Fig. 2. Variation of the normalised number distribution of 2D equivalent diameters of the graphite particles after isothermal holding at 800 °C and after 10 0 0, 20 0 0 and 30 0 0 thermal cycles, as compared to the distribution in the as-cast material. The insert corresponds to a zoom in the domain of particles larger than 30 μm in equivalent diameter. The maximum observed standard deviation was ± 10% of the value shown for any of the classes.

evidencedbythenumberdensityofgraphiteparticles(seebelow). In contrast, Fig. 1 c shows that the microstructure of the sample after 3000 cycles presents a marked coarsening of the graphite particles whichisassociatedwithmorphologicalinstabilityofthe largestgrowingparticles.

It wasobserved that thenumber densityofgraphite particles decreased from580 mm−2 for the as-cast sample to 520 mm−2 forthesampleisothermallytreatedandto470–500mm−2 forthe thermally cycledsamples. These valuesconfirm a slight coarsen-ing process duringisothermal treatment, andastronger effectof thermalcycling.

Thenormalizeddistributionsofgraphiteparticlessizeobtained afterisothermalholdingandafter1000,2000and3000cyclesare comparedtotheinitialdistributionintheas-castmaterialinFig. 2 . Therelativedistributionsfortheheat-treatedsamplesareall sim-ilar,showingan increaseinthenumberof0–10μmand20–30μm particles, and a decrease in the number of the 10–20μm ones when compared to the as-cast sample distribution.More signifi-cantly,it isseenthat thermalcyclingleadstothe increaseinthe numberofparticleswithequivalentdiameterhigherthan30μm.It

isalsoclearlyseenthat thesizeofthe largestparticlesincreased from30–40μmintheas-caststatetomorethan60μmafter3000 cycles.

Coarseningofgraphite particleshasbeen reportedinthecase ofheat-treatment ata temperature whereausteniteis stablebut neverwhenthematrixremains ferritic.Incontrasttothecaseof austenite,thecarboncontentinferriteisassumedsolowthatsuch aphenomenonhasbeengenerallyconsideredasunexpected. How-ever,inthepresentwork,itwasobservedthat theapparent frac-tionofgraphite measuredon 2D metallographicsectionsdid sig-nificantlyincreasewithcyclingfrom0.1intheas-caststateto0.17 after 3000 cycles. Measurements of the sample density showed a correlated decrease, from 6.9g/cm3 _{for the as-cast sample to}

6.8, 6.6 and finally 6.4g/cm3 _{respectively after 1000, 2000 and}

3000 cycles. This correlation means that cavities developed dur-ingcyclingwhichwereconsidered asgraphiteduringimage anal-ysis because they had the same contrast on light optical micro-graphs.In other words, graphiteparticles did partlydissolve and carbon atoms were transferred to other locations in relation to coarsening.

In thepresent case, these cavities certainly developed first at the interface between graphite precipitates and the matrix, but thenevolvedintoholesleftbythesmallspheroidsbeingdissolved duringthe coarseningprocess. Unfortunately, it was not possible to identify unambiguously these cavities on metallographic sec-tions asgraphiteparticles mayhavealso spalled off during sam-ple preparation. Toevaluate the actual possibility for graphite to dissolveduringheat-treatmentintheferriticdomain,theFe-C iso-plethsectionoftheFe-C-Si-Mnphasediagramat4.45wt.%Siand 0.25wt.%Mnwasdrawn(Fig. 3 ).Itisseenthatthecarboncontent inferriteincreasesabove700°C,beingabout0.012wt.%at800°C whileitisvirtually zeroat700°Candbelow.Itcanthus be con-jecturedthatthischangeinferritecarboncontentbetween700°C and800°Cisinstrumentalinthecoarseningprocess.

Fig. 3 shows also that the lowest temperature of the fer-rite/austenite/graphitethree-phasefieldis860°C,i.e.,thata heat-treatment with a maximum temperature at 800°C should not lead to the appearance of austenite. However, because micro-segregation of silicon that has developed during solidification couldhavetriggered partialtransformationofferrite toaustenite, itappearednecessarytoverifythat therewasnophasechangein thestudiedmaterialduringthermalcyclingwithanupper temper-atureof800°C.Differentialthermalanalyseswerethusrunat2,5, 10and20°C/min(seesupplementary information).Themeasured temperaturesforthestart oftransformationupon heating,Tstart-C,

andfortheendoftransformationuponcooling,Tpeak-R,were

plot-ted versus the scanning rate. The values extrapolated to a zero scanning rate are 860°C upon heating and840°C upon cooling, thusconfirmingthat thereisnophase changeduringthermal cy-clingwhenthemaximumtemperatureissetat800°C.

Themostintriguingobservationmadeduringthepresentwork was the characteristics of the protuberances developing on the

Fig. 3. Isopleth Fe-C section of the Fe-C-Si-Mn phase diagram at 4.45 wt.% Si and 0.25 wt.% Mn showing that ferrite (BCC_A2) is stable up to a temperature of 860 °C where austenite (FCC_A1) appears. The calculations were made with the Thermocalc software and using the TCFE8 database [8] .

Fig. 4. Optical micrograph (polarised light) of dendritic protuberances observed on the sample having been submitted to 20 0 0 thermal cycles. The white dashed circle superimposed on the graphite particle to the left defines the central compact part which is considered as the original nodule. The extensions out of this circle are thus the protuberances grown during thermal cycling.

largestspheroids.Fig. 4 showsthatafter2000cyclestheyassume a dendritic shape protruding fromthe initial compact spheroids. Thelength oftheseprotuberancesgoesupto 30–40μm inFig. 4 . After 1000 cycles, it was observed that the protuberances have juststartedtodevelopwhileafter3000cyclesthecoarsening pro-cessledtoathickeningofthedendriticprotuberanceswhichoften joinedeach other.In thislattercase, areasof matrixwere found isolated by graphite "arches" very similar to those described by Monchouxetal.[9] .Notethatthislatterstudyinvolvedheatingin theausteniticdomain whilethe presentresultsdemonstratethat graphitedissolutionmaytake placewithouttheferriteto austen-itetransformationofthematrix.

Some hints for understanding the growthphenomenon could be gained by evaluating the extent of graphite dissolution upon heatingfromroomtemperature(RT)to800°C.Thecarbonbalance canbewrittenas:

ρ

iron_{· w} 0

C=

ρ

g· w g

C· g g+

ρ

α· w Cα· g α (1)

where

ρ

iron_andw0

Carethedensityandcarboncontentofthealloy,

ρ

φ_,gφ _andwφ

C are,respectively, thedensity, thevolume fraction

andthe carbon content ofphase

ϕ

(g: graphite;

α

: ferrite). This massbalancemaybedifferentiatedand,afterrearrangementusing

dgα=−dgg_,onegets: dgg= −w g C· g g· d

ρ

g+

(

1 − g g

)

·



ρ

α_{· dw} α C + wCα· d

ρ

α



ρ

g_{· w} g C−

ρ

α· w Cα (2)

High temperature X-rays were conducted to evaluate the change in lattice parameter of ferrite in the temperature range fromRT to 800°C. The expansion coefficient wasfound equal to 1.5_•10−5°C−1.Usingthedensityofacastironcontaining4.25wt.% Si[10] ,

ρ

α wasevaluatedat7570kg.m−3atRT.Withtheabove ex-pansioncoefficient,theferritedensityisevaluatedas7300kg.m−3 at800°C,anditsaverage value

ρ

¯α betweenRT and800°C could be set to 7435kg.m−3. The density of graphite

ρ

g _{varies from}

2262kg.m−3 atRT to 2221kg.m−3 at 800°C [11] and its average value

ρ

¯g _{wassetto2240kg.m}−3_{.From allthesevalues,itcan be}

concluded that the first andlast terms of the numerator in the right-hand side ofEq. (2) are negligible.The averagechange



gg

betweenRTand800°Cmaythusbeapproximatedas:



gg_{≈ −}

(

1 − g g

)

· ¯

ρ

α·



wαC

¯

ρ

g_{· w} g

C− ¯

ρ

α· ¯w αC

(3)

According to Fig. 3 ,



wα_C was set at 0.012wt.% and its aver-age value w¯_Cα at half of it. With these values and the graphite fraction gg _{set to 0.1, the change ingraphite fraction upon}

heat-ingfromRTto800°Cisabout



gg _=-3.6•10−4 _(-3.6•10−2_%).This

changeingraphitefractionmaybeconvertedinto thesizeofthe gapformedinbetweenthegraphitespheroidsandthesurrounding matrixuponheatingfromRTto800°C.Forsodoing,the2D nod-ulecount,NA,givenabove,maybeconvertedtothevolumenodule

countbyNV= _π2 ·N_DA

2 [12] ,where

¯

D2istheaveragediameterofthe

spheroidsasmeasuredonthe2Dsection.Withgg_=N A·π·(

¯

D₂)2

4 ,D¯2

couldbesetto15μmwhichgivesNV≈ 24,600mm−3.Thechange

ingraphitefractioncanalsobewrittenas



gg_=N

V· 4·

π

· ¯r2·



¯r,

inwhich ¯rand



¯rare,respectively, theaverage radiusand aver-agechangeinradius ofthegraphitespheroids. Thischangein ra-dius gives thesize of the gapformed betweenthe outer surface ofgraphitespheroidsandthesurroundingmatrixwhentheupper

temperature of 800°C is reached. By inserting the value of



gg

evaluatedabove,onegetsagapsizeof2.1•10−2μm.

Though the above gap appears very small, it was readily re-alised that if thisvalue could accumulate from one cycle to an-other it would sum up to 42μm after 2000 cycles. As a matter offact, thislattervalue appearsvery similar tothelength ofthe protuberancesseeninFig. 4 .Thisresultthussuggeststhatthe car-bondissolvedduringheatingandholdingat800°Cre-precipitated in preferred locations during cooling and not uniformly around thelargeparticles.Onthelargestparticles,reiteratingthisprocess leadstotheprotuberancesillustratedonFig. 4 .Thermalcyclingis thusessentialforthedevelopmentofthesemorphological instabil-ities, andthisexplainswhythey couldnot haveappeared during isothermalheat-treatment.

Thermalcyclingintheferritic domainhasbeenshownto dra-maticallydecreasemechanicalpropertiesofspheroidalhigh-silicon castirons[3 ,4 ,6] .However,noneofthesepreviousworksreported a microstructure evolution as observed in the present work. In the case of the works by Lin et al. [3 ,4] this was certainly due to the maximum temperature they considered, either 700°C or 750°C,which mayhavebeen too low forgraphite dissolutionto take place. It appears more surprising that Cheng et al. [5] and Averyetal.[6] didnot observemicrostructurechanges similarto those detailedhereas thehighesttemperature that they investi-gatedwasalso800°C.Thisisprobablybecausetheseauthorswere focused onthe areas damaged by thermomechanicalcycling, but thiscouldaswellberelatedtotheveryrapidcyclesthattheyused whichmaynothavegivenenoughtimeforgraphitedissolutionin theuppertemperaturerangetooccur.

Understandingthedetailedmechanismleadingtothe morpho-logicalinstabilitiesdescribedhereiscertainlyworthyoffurther in-vestigation,aswellasevaluatingiftheymayplayaroleinthe de-creaseofthemechanicalpropertiesofhighsiliconSGIundercyclic thermalloadings.

Funding None.

DeclarationofCompetingInterest

Theauthorsdeclarethattheyhavenoknowncompeting finan-cialinterestsorpersonalrelationshipthatcould haveappeared to influencetheworkreportedinthispaper.

Supplementarymaterials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.scriptamat.2019.11. 001 .

References

[1]W. Fairhurst , K. Röhrig , Foundry Trade J. 146 (1979) 657–681 .

[2] L. Delin, R. Logan, G. Burger, G. Liao, D. McFarlan, B. Black, R. Williams, SAE Transactions, J. Mater. Manuf. 116, 2007, 530-538.

[3]H.-M. Lin , T.-S. Lui , L.-H. Chen , Mater. Trans. 44 (2003) 173–180 .

[4]H.-M. Lin , T.-S. Lui , L.-H. Chen , Mater. Trans. 44 (2003) 1209–1218 .

[5]C.P. Cheng , T.S. Lui , L.H. Chen , Metall. Mater. Trans. A 30A (1999) 1549–1558 .

[6] K. Avery, J. Pan, C. Engler-Pinto, SAE Technical Paper 2015-01-0557, 2015, doi: 10.4271/2015- 01- 0557 .

[7]R. González-Martínez , U. de la Torre , J. Lacaze , J. Sertucha , Mater. Sci. Eng. A 712 (2018) 794–802 .

[8]J.O. Andersson , T. Helander , L. Höglund , P.F. Shi , B. Sundman , Calphad 26 (2002) 273–312 .

[9]J.P. Monchoux , C. Verdu , G. Thollet , R. Fougères , A. Reynaud , Acta Mater. 49 (2001) 4355–4362 .

[10]T. Matsushita , E. Ghassemali , A. Gomez Saro , L. Elmquist , A.E.W. Jarfors , Metals 5 (2015) 10 0 0–1019 .

[11]P. Dietrich , G. Lesoult , in: State of the Art of Computer Simulation of Cast- ing and Solidification Processes, Les éditions de Physique, E-MRS, 1986, pp. 225–235 .