Identification of the metallurgical parameters explaining the corrosion susceptibility in a 2050 aluminium alloy

(1)

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

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

makes it freely available over the web where possible.

This is an author-deposited version published in : http://oatao.univ-toulouse.fr/

Eprints ID : 14698

To link to this article : DOI :

10.1016/j.corsci.2015.10.020

URL :

http://dx.doi.org/10.1016/j.corsci.2015.10.020

To cite this version : Guérin, Mathilde and Alexis, Joël and Andrieu,

Eric and Laffont-Dantras, Lydia and Lefebvre, Williams and Odemer,

Grégory and Blanc, Christine Identification of the metallurgical

parameters explaining the corrosion susceptibility in a 2050 aluminium

alloy. Corrosion Science, Vol.102. pp.291-300. ISSN 0010-938X

Any correspondance concerning this service should be sent to the repository

administrator: [email protected]

(2)

Identification

of

the

metallurgical

parameters

explaining

the

corrosion

susceptibility

in

a

2050

aluminium

alloy

Mathilde

Guérin

a,b

_,

_Joël

_Alexis

b

_,

_Eric

_Andrieu

a

_,

_Lydia

_Laffont

a

_,

_Williams

_Lefebvre

c

_,

Grégory

Odemer

a

_,

_Christine

_Blanc

a,∗

a_Université_de_Toulouse,_Institut_CARNOT_CIRIMAT,_{UPS/CNRS/INPT,}_ENSIACET,₄_allée_Emile_Monso,₃₁₀₃₀_Toulouse,_France b_Université_de_Toulouse,_LGP,_ENIT/INPT,₄₇_Avenue_d’Azereix,_65016,_Tarbes,_France

c_Université_de_Rouen,_GPM—UFR_Sciences_et_Techniques,_Avenue_de_{l’Université—BP12,}_76801,_Saint_Etienne_du_Rouvray,_France

Keywords: Alloy Aluminium STEM Intergranularcorrosion Interfaces

a

b

s

t

r

a

c

t

Thecorrosionbehaviourofa2050aluminiumalloywasstudiedinaNaClsolution.Thestructureof precipitationdidnotfullyexplainthesusceptibilitytointergranular(inthe-T34state)and intragran-ularcorrosionfortheagedstate(the-T8state).Arelationshipbetweenthenatureofinterfaces,the grainscharacteristics(size,internalmisorientationandorientationaccordingtotheplaneexposedto theelectrolyte)ononehandandthecorrosionsusceptibilityofthealloyontheotherhandwasclearly established.Galvaniccouplingbetweengrainswithdifferentinternalmisorientationshelpedtoexplain theintergranularcorrosionsusceptibilityofthe-T34state.

1. Introduction

NewgenerationofAl–Cu–Li–Xalloysshowsremarkable combi-nationofdensity,mechanicalpropertiesandcorrosionresistance. Althoughotherphasescanbeobservedinthesealloys,the pre-cipitationofT1–Al2CuLiphaseiscommonlyconsideredasamajor

parameter to explain the corrosion behaviour of these alloys

[1–9].Thisintermetallicphasepossessesamorenegative

corro-sionpotentialthanthematrix[2,6].Whenthematerialisexposed toanaggressivemedium,agalvaniccouplingbetweentheT1phase

andthematrixoccursinfavourofthematrix.Inthe-T34 metallur-gicalstate,becauseofthepresenceofT1particlesonlyatthegrain

boundaries,galvaniccouplingleadstothedissolutionofthegrain boundarieswithamorenegativecorrosionpotentialthanforthe grains.Todesensitisethealloytointergranularcorrosion,anaging treatmentisappliedtothematerial.Thisleadstothe precipita-tionofT1particlesbothinthegrainsandatthegrainboundaries,

leavingtheagedstate(-T8).Fora-T8sample,becauseofthis struc-tureofprecipitation,thecorrosionpotentialsofthematrixandthe

∗ Correspondingauthorat:UniversitédeToulouse,CIRIMAT,UPS/CNRS/INPT,4 alléeEmileMonso,BP44362,31030ToulouseCedex4,France.

Fax:+330534323498.

E-mailaddress:[email protected](C.Blanc).

grainboundariesarequitesimilar.Consequently,galvaniccoupling doesnotoccurbetweenthegrainandthegrainboundariesandthe materialissusceptibletointragranularcorrosion[2–4].

Ourresultsfrompreviousworkconfirmedthesusceptibilityof anAA2050alloytointragranularcorrosionafteranageing treat-mentat155◦_C_for₃₀_h_and _correlated_this_to_the_homogeneous

distributionofT1 phaseparticles inthegrainsand atthegrain

boundaries[9].For theAA2050-T34 alloy,susceptible to inter-granularcorrosion,amajorpartofcorrodedgrainboundariesdid notevidencethepresenceofT1precipitates.Thisresultsuggested

thatthepresenceoftheT1 phasewasnotnecessarytoinducea

susceptibilitytointergranularcorrosioninanAA2050-T34alloy. Furthermore,forboth-T8and-T34alloys,preferentialdissolution ofsomegrains andgrainboundarieswasrespectivelyobserved while,for-T8alloyforexample,homogeneousdistributionofT1

precipitateswasobservedbothinthegrainsandatthegrain bound-aries.Theseresultssuggestedthatparametersdifferentthanthe structureofprecipitationshouldcontributetoexplainthe corro-sionbehaviouroftheAA2050alloy.Thisshouldbeinagreement withresultsfoundinliterature.Forexample,Luoetal.showedthat arelationshipexistsbetweenthedislocationdensityinagrainand theabilityofthegrainboundarytobecorroded[10].Kimetal. showedthatthegrainboundarycharacterdistribution(GBCD)has agreateffectonintergranularcorrosionsusceptibilityofaluminum inHCl[11].Ralstonetal.workedontheinfluenceofthegrainsizeon

(3)

Fig.1. MethodologyusedforstudyingtheimpactofdifferentmetallurgicalparametersatthepolycrystalscaleonthecorrosionbehaviourofAA2050.

Fig.2. Opticalmicroscopeobservationsof(a)intergranularcorrosioninthe-T34sampleand(c)intragranularcorrosioninthe-T8sample.Bright-fieldTEMimagesof(b) the-T34sampleand(d)the-T8sample.

thecorrosionsusceptibilityofamaterialandnoticedthatthe‘grain size—corrosionresistance’relationshipiscomplex.Theysuggested thatthiscomplexitycouldbeincreasedduetotheheterogeneityof thegrainsizeinasample[12].

Theaimofthisstudywastodeterminetheoriginofthe suscepti-bilitytointergranularandintragranularcorrosionforAA2050-T34

andAA2050-T8alloysrespectively.Thestructureofprecipitation wasconsidered.However,attentionwasalsopaidto microstruc-turalparametersatthepolycrystalscale:thismeansthatthenature oftheinterfaces(eithergrainboundariesorsubgrainboundaries correspondingtothelevelofmisorientationattheinterfaces),the internalmisorientationandsizeofthegrains,andtheirorientation

(4)

Fig.3.HAADFSTEMobservationofGPzonesalonga<110>axisoftheAlmatrixintheAA2050-T34alloy.

withrespecttotheplaneexposedtotheelectrolytewere consid-eredaspotentialcriticalmetallurgicalparametersinfluencingthe corrosionsusceptibilityoftheAA2050alloy.Tostudytheimpactof thesemicrostructuralparameters,electronbackscatterdiffraction (EBSD)analyseswereperformedonhealthyandcorrodedsamples. 2. Experimentalprocedure

2.1. Material

The material used for this study was an aluminium–copper–lithiumAA2050-T34alloy(Albase,3.86%Cu, 0.86%Li,wt.%)providedbyConstellium(France).Thematerialwas receivedasa50mmthickplateformedbyhotrollingandfollowed by solutionheat treatment, water quenching and stretchingto achieve the final -T34 metallurgical state.A piece of theplate wasaged and corresponds to a -T8 metallurgicalstate. Dueto therollingprocess,themicrostructureofboththe-T34and-T8 samplesexhibitedahighdegreeofanisotropy.Observationsusing opticalmicroscopy(OM)ofthethickplateshowedthatthegrains werehighlyelongatedintherollingdirection.Theaveragesizesin thetransverse(LT)andshorttransverse(ST)directionswere350 and60mm,respectively,withalargediscrepancyinsizefromone graintoanother.Inthelongitudinal(L)direction,thediscrepancy issuchthatthecalculationofameanvalueismeaningless.Further EBSDanalysisrevealedtwopopulationsofgrains:onepopulation oflargepolygonisedgrains(withsizesvarying from350mmto 5mmintheLdirection)andanotherpopulationofrecrystallised grainswithsmallerdimensions(between5mmand20mminthe Ldirection).

2.2. StudyofthestructureofprecipitationoftheAA2050alloy Inordertotakeintoaccountthestructureofprecipitationin thecorrosionsusceptibilityoftheAA2050alloy,atransmission electronmicroscope(TEM)inthescanningtransmissionelectron microscopy(STEM)mode,whichhasahighdegreeoflocal pre-cision,wasused.STEMobservationswererecordedusingaJEOL ARM200FequippedwithaSchottkyFEG(fieldemissiongun)and aprobeCs-correctoroperatingat200kV.TheHAADF-STEM(High

AngleAnnularDarkField)andLAADF-STEM(LowAngleAnnular DarkField)imageswererecordedwitha0.1nmFWHMelectron

probesizeof30mradsemi-convergenceangle.Thedetectorranges weresetto54–220mradforHAADFand24–54mradforLAADF.The HAADFandLAADFsignalsaredominatedbyZ-contrastand diffrac-tioncontrast,respectively.Forthesetechniques,thesampleswere preparedusingmechanicalpolishingtoobtaina100mm-thickslice followedbyelectrolyticpolishingtomakeaportionofthesample transparenttoelectrons.

2.3. Studyofthecriticalmicrostructuralparametersatthe polycrystalscale

EBSDanalyseswereperformedforthe-T34 and-T8samples usingaJEOL7000Ffieldemissionscanningelectronmicroscope equippedwithaNordlysIIF+cameraatanacceleratingvoltage of15kV.TheEBSDanalyseswerehelpfulinboth analysingthe microstructureofthealloyandcharacterisingthecorrosion dam-ageafterthetests.Allofthecorrosiontestswereperformedover 72handwereconductedinacontinuousimmersionatopencircuit potentialina0.7MsolutionofNaCl,whichwaspreparedby dis-solvingNormapurchemicalsaltsindistilledwater.Duringthetests, thesolutionwasstirredandopenedtotheair.Thetemperaturewas maintainedat25◦_C_using_a_Julabo_refrigerated_circulator._For_each

sample,EBSDmapswereperformedonthesame1.5mm×3mm areawithastepsizeof1–2mminthe(LT-ST)planebothbeforeand afterthecorrosiontests.Duringthecorrosiontests,onlythiszone wasexposedtotheelectrolyte(avarnishwaspaintedontoprotect therestofthesurface).Thesurfaceexposedtotheelectrolytewas largeenoughtoobtainrepresentativeresults.Overall,foreach sam-ple,morethan250grainswereanalysed.TheEBSDdatawerethen post-processedusingthecommercialorientationimagingsoftware packageOxfordChannel5.Tominimisemeasurementerrors,all grainscomprisinglessthan3pixelswereautomaticallyremoved fromthemapspriortodataanalysis.Inaddition,toeliminate spuri-ousboundariescausedbyorientationnoise,alowerlimitboundary misorientationcut-offof2◦_was_used._The_analyses_permitted_the

determinationoftheorientationofthegrains(seeFig.1,map1) andthenatureoftheinterfaces(seeFig.1,map2).Inthiswork, theword‘interfaces’referstothegrainboundariesindependentof theirlevelofmisorientation.A10◦_criterion_was_used_to

differenti-atelowanglegrainboundaries(i.e.subgrainboundaries)andhigh anglegrainboundaries(i.e.grainboundaries).Thismisorientation waschosenbasedondataavailableforpurealuminium[13].Two

(5)

Fig.4. HAADFSTEMimagesofsubgrainboundariesintheAA2050-T34alloy(Z-contrast).

Fig.5.(a)HAADFSTEM(i.e.Z-contrast)and(b)LAADFSTEM(i.e.diffractioncontrast)imagesofthesameregionintheAA2050-T34alloy.Theregiontotherightisinperfect <110>orientationwhereasthecrystalorientationslightlychangesacrossthesubgrainboundaryidentifiedbythealignmentofdislocationsin(a).Thisslightchangeofcrystal orientationisclearlyvisibleonthehighermagnificationHAADFSTEMimagesdisplayedin(c)and(d),whichshowprecipitatesindicatedbyawhitedottedsquarein(a).

categories of grainboundaries were distinguished: coincidence sitelatticegrainboundaries(CSL)andrandomgrainboundaries. Incoincidentsite latticegrainboundaries,thedegreeoffit (6) betweenthestructuresofthetwograinsisdescribedbythe recip-rocaloftheratioofcoincidencesitestothetotalnumberofsites orientedinoppositiontotherandomgrainboundarieswherethe structurewasobservedasbeingdisorganised.Thegrainsizewas quantifiedbymeasuringthegrainareaandcalculatingthe equiv-alentgraindiameter assumingeach grain asa circle. Afterthe corrosiontests,allthecorrodedinterfacesandgrainsinthesame

zonewereanalysed.Thisrepresentedapproximately50interfaces (forthe-T34metallurgicalstate)and15grains(forthe-T8 metal-lurgicalstate)ineachsample.Theanalysiswasreproducedtwice foreachmetallurgicalstateleadingtosimilarconclusions.Afterthe testing,thecorrosionproductswereremovedusinggentle mechan-icalpolishing.ThecorrodedzonewasanalysedusingSEM with theelectronbackscatterdetectortoeasilyidentifyanycorrosion defects(seeFig.1,map3a).Thecorrosiondefectswereisolated usingimageanalysis(ImageJsoftware,seeFig.1,map3b).Thislast mapwassuperimposedonmap1ormap2toobtainacorrosion

(6)

map(seeFig.1,map4).Themisorientationlevelforeachcorroded interfaceandthecharacteristicsoftheadjacentgrains(forthe-T34 metallurgicalstate)orofthecorrodedgrains(forthe-T8 metallur-gicalstate)wereextractedfrommap4.Thegraincharacteristics studiedoverthisareaincludedthesize,theinternalmisorientation andtheorientationrelativetotheplaneexposedtotheelectrolyte. 3. Experimentalresultsanddiscussion

Asalreadydiscussed,a previousstudyonthesamematerial revealedasusceptibilitytointergranularcorrosionformaterialin the-T34state(Fig.2a)andtointragranularcorrosioninthe-T8 state(Fig.2c)[9].TEMobservationsfor-T8samplesagreedwith theliterature;theT1phaseprecipitateswereobservedbothinthe

grainsandatthegrainboundaries,explainingthesusceptibilityto intragranularcorrosion(Fig.2d).Forthe-T34samples,noT1

pre-cipitatewasobservedinthemajorpartsofthegrainboundaries (Fig.2b).

3.1. CorrosionbehaviouroftheAA2050-T34alloy

The corrosion behaviour of the system investigated being stronglydependentonthestructureofprecipitationassuggested bytheliterature[1–9],it isnecessarytodeterminewhetherT1

and/oru′_{precipitates,}_that_are_also_expected_in_such_an_alloy,_can

beobservedintheT34state.BecausepreviousTEMobservations didnotallowthepresenceofT1precipitatestobeevidencedinthe

majorpartofthegrainboundaries,HAADFSTEMobservationswere performed.TheyrevealedfirstthatonlyCu-richGPzonesare iden-tifiedinthematrix(Fig.3)awayfromsubgrainorgrainboundaries, asexpected.ThelengthoftheseGPzonesmeasuredbyprojection alonga<110>axisisrangingbetween1and3nm.

Specialattentionwaspaidtothecharacterizationofsubgrain boundariesandHAADFSTEMimages(Fig.4)indicatedthat precip-itateswerealreadypresentalongtheseinterfaces.However,the densityofprecipitationwasfoundtostronglyvaryfromone sub-grainboundarytoanotheroralongthesameinterface.Inorder torevealelasticdistorsionsinducedbythedislocationsassociated totheseinterfaces,LAADFSTEMmodewasused.Suchamethod hasrevealeditsrelevancefortheimagingofdislocationsinSTEM

[14].In theAA 2050-T34alloy,thesameregionswere simulta-neouslyimagedinLAADF(Fig.5b)andHAADFSTEM(Fig.5a)to unambiguouslydissociatethecontrastassociatedwithasubgrain boundary(i.e.thealignmentofdislocationsacrosswhichthecrystal orientationisslightlytitled)totheZ-contrastrevealingthe even-tualpresenceofprecipitates.Fig.5showsasubgrainboundaryina nearly<110>orientation.Precipitateswereclearlyvisibleinsome regionsofthisinterfaceandweredisplayedinlargermagnification inFig.5candd.Acrossthesetwoimages,fromrighttoleft,the lossoflatticeresolutionintheAl-matrixwasduetoasmall dis-orientationofthelatticeacrossthesubgrainboundary.InFig.5c, theprecipitateinthelowerpartoftheimageexhibitedatypical contrastofaT1precipitatealong<110>zoneaxisofaluminium,as

reportedintheliterature[15,16].TheCu-richplanardefectslying along{111}planesofaluminiuminFig.5dcouldnotbeclearly identified.TherecentworkbyGaoetal.[17]hasreportedGP-T1

zones,whichstructuredidnotcorrespondtotheCu-richplanar defectsinFig.5.TheGP-T1zonesreportedbyGaoetal.actually

dis-playtwoCu-richplanesparallelto{111}Al,separatedbyaregion withacompositionthatcouldnotbedeterminedunambiguously thoughitscontrastseemstoindicatethatit consistsofasingle planeofaluminium.Ourobservationsweremoreconsistentwith thosemadebyAraullo-Petersetal.byatomprobetomography[18]. TheseauthorshavedemonstratedacomplexcouplingbetweenMg andCusegregationsalongdislocationswiththeformationofplanar

defectsenrichedinCu,identifiedasT1precursors.Nevertheless,the

mechanismofformationofT1precipitatesfallsoutsidethescope

ofthispaper,whichpurposeistorelatemicrostructural character-izationtothecorrosionbehaviouroftheAA2050-T34alloy.

Themainresultherewasthat,intheT34state,thealloyalready containedT1precipitatesandotherplanardefectsenrichedin

cop-peralongsubgrainboundaries.Lithiumcouldalsobepresentin theinterfaces but,due toits lowamount,it wasimpossible to detectthiselement.Despitethefactthatnou’precipitatecould beobserved,wecannotexcludethatsomeoftheseprecipitates werepresentintheAA2050-T34alloyalonginterfaces.The den-sityofprecipitatesand planardefectsstronglyvariedfromone subgraintoanotherandwemadetheassumptionthatthese varia-tionswereduetothenatureofdislocationspresentinthesubgrain boundaries.Assuggestedbyseveralauthors,agrainboundarywith ahighdegreeofatomicdisarrangementshouldreactmorewith impuritiesoralloyingelementsthanagrainboundarywithhigh coincidenceleveloreventhanagrainboundaryoflowangle mis-orientation[19–21].Thissuggestedthatthelevelofmisorientation ofinterfacescouldbecorrelatedtotheircorrosionsusceptibility.A detailedinvestigationwashencerequiredtoestimatethe possibil-ityofsuchacorrelation.Thisinvestigation,whichwasperformed usingEBSDanalyses,ispresentedhereafter.

TheresultsoftheEBSDanalysesperformedpriortocorrosion testsfortheAA2050-T34samplearepresentedinFig.6.The dis-tributionofinterfacesaccordingtotheirlevelofmisorientationis showninFig.6a.Thelevelofmisorientationofeachinterfacewas determinedbymeasuringthemisorientationbetweenthegrains situated onboth sidesof theinterface.Theresultsshowedthat 85% ofinterfaceswere subgrainboundariesand only15% were grainboundaries.Amongthegrainboundaries,20%wereCSLgrain boundariesand80%wererandomgrainboundaries.63interfaces representedthemajorityoftheCSL(35%,Fig.6b).Therefore,during analysisofthecorrodedsamples,onlythe63CSLgrainboundaries wereconsidered.

AftercompletingthecorrosiontestsintheAA2050-T34sample, thecorrodedinterfaceswerelocated,andthelevelsof misorien-tationwererecorded(Fig.1).Theanalysiswasperformedon50 corrosiondefects,andledtotheestablishmentofarelationship betweenthelevelofmisorientationofaninterfaceandits suscep-tibilitytocorrosion.Fig.6ashowsthecorrodedinterfacefrequency accordingtotheirmisorientationlevel.Thedistributionwas differ-entfromthedatatakenpriortothecorrosiontest.Atotalof70%of thecorrodedinterfacesweregrainboundarieswhereasonly30% ofthecorrodedinterfacesweresubgrainboundaries.Considering thehighproportionofsubgrainboundaries,theseresultssuggesta higherresistancetocorrosionofthesubgrainboundaries. More-over, amongthecorrodedgrainboundaries,theresultsshowed thatthemostmisorientedinterfacestendedtobemore suscepti-bletocorrosion.Thestrongreactivityofthegrainboundarieswith ahighlevelofmisorientationmaybelinkedtotheirdisarranged structureandcorrelatedwiththeirhighenergy[19].Onthe con-trary,theresistancetocorrosionofthesubgrainboundariesshould beassociatedwiththeirlowenergy.Theseresultswereconsistent withotherworkonvariousmetals[20,22–24].Itshouldbenoted thattheinterfacesusceptibilitytocorrosion,andinparticularof somesubgrainboundariescompared totheothers,mayalsobe explainedbytheirchemicalcompositionandstructureof precipi-tation,ashighlightedintheSTEM-HAADFobservations.Itwillbe interestingtostudytherelationshipbetweenthechemical com-position/structureofprecipitationofaninterfaceanditslevelof misorientationbutthisshouldbethesubjectoffuturestudy. Con-cerningthesusceptibilitytocorrosionoftheCSLgrainboundaries, onlyfourcorrosiondefectsat63grainboundarieswereidentified. Anintergranularcorrosionsusceptibilityindexwasdeterminedby theratiooftheproportionofcorrosiondefectscorrespondingto

(7)

Fig.6. AA2050-T34alloy—(a)Distributionofthelevelofmisorientationoftheinterfacesbeforeandafterthecorrosiontests;(b)distributionoftheCSLgrainboundaries beforecorrosiontests.

Fig.7. AA2050-T34alloy—(a)Distributionofthegrainsurfaceareasbeforeandafterthecorrosiontestsforcorrosionatsubgrainboundariesandatgrainboundaries;(b) distributionofthegrainsurfacearearatioofthegrainssituatedonbothsidesofthecorrosiondefectsatthegrainboundaries.

Table1

StatisticalanalysisoftheintergranularcorrosionintheAA2050-T34alloybasedonthenatureoftheinterfacesatthepolycrystalscale.

Natureoftheinterfaces Subgrainboundaries Randomgrainboundaries CSLgrainboundaries

%Innon-corrodedspecimen(a) 85 12 3

%Incorrodedspecimen(b) 30 62 8

(8)

Fig.8.AA2050-T34alloy—(a)Relationshipbetweenthegrainsurfaceareaandtheinternalgrainmisorientation;(b)distributionoftheinternalgrainmisorientationbefore andafterthecorrosiontestsforcorrosionatthesubgrainboundaries.

acertaintype ofinterface(subgrain boundaries,randomorCSL grainboundaries)dividedbytheproportionofinterfacesofthis sametypeinthesample.TheresultsshowedthattheCSLgrain boundariesweretwo timesless sensitivetocorrosionthanthe randomgrainboundaries,whichwasasexpectedfromprevious observations(Table1).Moreover,thecalculatedindicesshowed thatsubgrainboundariesweretheleastsensitiveinterfaces.This resultwaspartiallyconsistentwiththeliteratureevenifCSLgrain boundarieswereoftenconsideredmoreresistantthansubgrain boundaries[20].Thistrendmustbefurtherconfirmedduetothe lownumberof63studied.

Inlightofthepreviousresults,itwasimportanttodistinguish corrosionatgrainboundariesandatsubgrainboundariesforthe AA2050-T34alloy.Inbothcases,thecharacteristicsofthegrains situatedoneachsideofacorrosiondefect(inthecaseof corro-sionatthegrainboundaries)orcontainingcorrosiondefects(in thecase of corrosion atthe subgrainboundaries)should influ-encethereactivityoftheinterface.Therefore, thepropertiesof thegrainswerealsoinvestigated.Theanalysesperformedprior tothecorrosiontestsshowedthatmostofthegrainswereless than10,000mm2_in_area₍_Fig.₇_a)._More_precisely,_50%_of_the_grains

measuredlessthan100mm2_._After_the_corrosion_tests,_the_grains

containingcorrodedsubgrainboundariesintheAA2050-T34 sam-pleswereidentifiedandtheirareaswererecorded.Fig.7ashows thatthecorrosionatsubgrainboundariesoccurredingrainswith alargearea(>10,000mm2_,_and_more_precisely_>80,000_mm2_)._This

wasbestexplainedbythepresenceofmanysubgrainboundaries and dislocations insidethe largepolygonizedgrains, leadingto asignificantamountof storedenergyandconsequently,astate ofnon-thermodynamicequilibrium inthesegrains.Theanalysis of corrosion at thegrain boundaries of theAA 2050-T34 sam-pleincludedexaminingthegrainssituated onboth sidesofthe corrosion defects and recording theirareas. Fig. 7a shows the distributionof thegrains experiencing corrosion attheir inter-faces. Results highlighted that intergranular corrosion at grain boundariesoccurredattheinterfacebetweengrainsofvariable surfaceareas.Butbydrawingthedistributionofthearearatio, which representstheareaof thesmallestgraindivided bythe area of the largest grain (for the grains situated on each side of a corrosion defect), a dominant group of corrosion defects for a low area ratio (0–0.2, Fig. 7b) was observed. This result

meansthatintheAA 2050-T34sample,intergranularcorrosion developedmainlybetweengrainswithdisparateareas:agalvanic couplingbetweenasmallgrainandalargegrainshouldexplainthis resultbecausethereisastrongerreactivityattheirshared inter-face.Indeed,severalworkshaveshownthatthecorrosionpotential ofametalvariedwiththedislocationdensity[24,25]andthegrain

size[26,27].Itwaspossibletotransposethisresultonthe

poly-crystalscalebyconsideringthegalvaniccouplingbetweengrains ofdifferentsizesandpossessingdifferentdislocationdensities.

Therelationshipbetweenthegrainsizeanddislocationdensity isshowninFig.8aforanon-corrodedAA2050-T34samplewhere thedislocationdensityofagrainwasmeasuredthroughits inter-nalmisorientation.Fig.8ashowsthatthesmallestgrainshavethe smallestinternalgrainmisorientationinrelationtothelow dislo-cationdensity.Moreover,the-T34metallurgicalstateexhibiteda highproportionofgrainswithalowinternalmisorientation(<2◦_,

Fig.8b).Afterthecorrosiontests,thegrainscontainingcorroded subgrainboundariesfortheAA2050-T34samplewereidentified andtheirinternalmisorientationsrecorded.Asexpected, corro-sionatthesubgrainboundariesoccurredingrainscontainingahigh internalmisorientation(Fig.8b).Whenconsideringthecorrosionat thegrainboundariesfortheAA2050-T34sample,theratiobetween theinternal misorientation of each grainsituated oneach side ofthecorrodedgrainboundariesmustbeexamined.Theresults showedthatthecorrodedgrainboundariescorrelatedtoahigh graininternalmisorientationratio,suggestingagainthatagalvanic couplingbetweenthegrains withdifferentinternal misorienta-tionsledtocorrosionattheirsharedinterface.Fig.9illustrates thisphenomenonofgalvaniccoupling.Intergranularcorrosionwas observedattheinterfacebetweenagrainwithlowinternal misori-entation(ontherightofthecorrosiondefectn◦₁₎_and_a_grain_with

manysubgrainboundaries,whichassumedahighdegreeof inter-nalmisorientation(totheleftofthecorrosiondefectmarkedn◦_1).

ItnotedthatdespiteoftheaveragebehaviourshowninFig.8a,the lowinternalmisorientationwithinagrainisnotnecessarily asso-ciatedwiththesmallsizeofthisgrain(identifiedasthegraintothe leftofcorrosiondefectmarkedn◦_2)._Therefore,_to_a_greater_extent

thanthegrainsize,theinternalgrainmisorientationseemstobethe mostrelevantparametercapableofexplainingtheintergranular corrosionsusceptibilityoftheAA2050-T34alloy.

(9)

Fig.9. EBSD(IPF-Z0)micrographofthecorrosiondefectssituatedbetweengrainsshowinglargediscrepanciesofinternalmisorientationintheAA2050-T34alloyandthe correspondingSEMmicrograph.

3.2. CorrosionbehaviouroftheAA2050-T8alloy

ThesusceptibilitytointragranularcorrosionofAA2050-T8alloy waswell-correlatedtothehomogeneousdistributionofT1

pre-cipitatesbothinthegrainsandatthegrainboundaries.However, aspreviously noticed, sucha structure of precipitation didnot explainthepreferential dissolutionofsomegrainscomparedto theothers.ResultsobtainedfortheAA2050-T34alloysuggested thatmetallurgicalparametersatthepolycrystalscaleshould con-tributetoexplainthecorrosionsusceptibilityofAA2050-T8alloy also.First,attentionwaspaidtotheinfluenceofthegrainsizeon thecorrosionsusceptibility.AsfortheAA2050-T34alloy,astrong reactivityoflargegrainswasalsoobservedfortheAA2050-T8 sam-ple.Thedistributionofgrainsaccordingtotheirareaispresented inFig.10aforthis metallurgicalstate.Thedistributionobtained fortheAA 2050-T8samplewassimilartothatoftheAA 2050-T34samplebecausetheageingtreatmentdidnotmodifythesize ofthegrains.Forthe-T8state,whichissusceptibleto intragran-ularcorrosion, theanalysis performedafter thecorrosion tests showedthepreferentialdissolutionofgrainswithanareagreater than10,000mm2₍_Fig.₁₀_a)._This_result_was_in_agreement_with_the

dataobtainedfortheAA2050-T34sampleforcorrosionat sub-grainboundariesandwasbetterexplainedbythepresenceofmany subgrainboundariesanddislocationsinsidethelargepolygonized grains,leadingtoasignificantamountofstoredenergyand con-sequently,astrongreactivity.Theliteraturereportsanevolution ofthecorrosionpotentialwiththedislocationdensitybutwithout anyidentifiedtrends[25].Itwasassumedinthepresentworkthat thegrainswiththehighestdislocationdensityhavethemost nega-tivecorrosionpotential.Incontrast,thesmallrecrystallisedgrains didnotexhibitsubgrainboundariesandtheircorrosionresistance wasconsequentlyhigher.

Then,theinfluenceoftheinternalmisorientationofagrainon itscorrosionsusceptibilitywasconsidered.Concerningthe rela-tionshipbetweenthegrainsizeanddislocationdensity,thesame resultwasobtainedforanAA2050-T8sampleasforanon-corroded AA 2050-T34 sample(Fig.8a), i.e. thesmallestgrains havethe smallestinternalgrainmisorientation.Moreover,asforthe-T34, thenon-corroded-T8metallurgicalstatesexhibitedahigh propor-tionofgrainswithalowinternalmisorientation(<2◦_,_Fig.₁₀_b).

Afterthecorrosiontests,resultsshowedthatintragranular corro-sionoccurredingrainscontainingahighinternalmisorientation (Fig.10b)showingthat themostrelevantparametercapableof explaining the intragranular corrosion susceptibility of the AA 2050-T8alloywastheinternalmisorientationofthegrains,asfor theintergranularcorrosionsusceptibilityoftheAA2050-T34alloy. AnotherparameterwasinvestigatedintheAA2050-T8sample: thegrain orientationrelated to theplaneexposed tothe elec-trolyte.Theanalysisperformedafterthecorrosiontestsonthe-T8 statesamplesshowedthatthemajorityofthecorrodedgrainsare orientedaccordingto(111)planesrelativetotheexposedplane (Fig.11).Itisworthnotingthatthe(111)planesarethedenser planesandtheyarethehabitplanesoftheT1phaseintheface

cen-trecubicstructureofaluminium.Theliteraturereportedthatthe dissolutionbehaviourofthegrainsdependedontheorientation ofthecrystalrelativetotheplaneexposedtotheelectrolyte[28]. Consequently,thepreferentialcorrosionofgrainsorientedwith the(111)planesexposedtotheelectrolytemaybeexplainedby amorenegativecorrosionpotentialthanforgrainsorientedtothe othercrystalplanes.Itissuggestedthatgalvaniccouplingbetween grainswithdifferentorientationsledtothepreferentialcorrosion ofsomegrains.Thishypothesisshouldbeadvancedtoexplainthe susceptibilityofthe-T34statetointergranularcorrosion.Inthis case,galvaniccouplingbetweengrainswithdifferentorientations

(10)

Fig.10.AA2050-T8alloy—(a)Distributionofthegrainsurfaceareasand(b)distributionoftheinternalgrainmisorientationbeforeandafterthecorrosiontestsfor intragranularcorrosion.

(11)

mayincreasethecorrosionsusceptibilityoftheirsharedinterface. FurthermeasurementsoftheVoltapotentialofthegrains accord-ingtotheircrystallineorientationusingKelvinForceMicroscopy shouldconfirmthishypothesis.

4. Conclusions

ThecorrosionbehaviourofAA2050wasstudiedforboth nat-urallyaged(-T34)andartificiallyaged(-T8)metallurgicalstates. Theimpactofthemicrostructuralparametersonbothintergranular (-T34)andintragranular(-T8)corrosionmechanismswas investi-gated.Theconclusionsareasfollows:

1Thestructureofprecipitationcontributedtoexplainthe suscepti-bilitytocorrosionoftheAA2050alloy.However,thepresenceof T1phaseprecipitatesatthegrainboundariesofa-T34alloywas

notnecessarytosensitisethealloytointergranularcorrosion. Theintragranularcorrosionofa-T8alloycanbeexplained par-tiallybyT1precipitateshomogeneouslydistributedinthegrains

andatthegrainboundariesbutsuchahomogeneousdistribution didnotexplainthestrongreactivityofsomegrainscomparedto others.

2Othermetallurgicalparametersatthepolycrystalscalewereto betakenintoaccounttoexplainthecorrosionmechanismsinthe AA2050alloy.

3Thenatureoftheinterfaceswasfoundtoplayadominantroleon theintergranularcorrosionsusceptibilityofthe-T34alloywitha highercorrosionresistanceforinterfacescharacterisedbyalow levelofmisorientation.

4Forboththe-T34and-T8samples,theinternalmisorientationof thegrainsisadrivingfactorintheintergranularand intragran-ularcorrosion mechanisms.Galvanic couplingbetweengrains withstronglydifferentinternalmisorientationshouldleadtothe corrosionoftheirsharedinterface.Grainswithhighinternal mis-orientationwerefoundtobethemostsusceptibletocorrosion. 5For-T8samples,grainsorientedinthe(111)planes,inwhichthe

T1phaseprecipitated,werethemostsusceptibletointragranular

corrosion. Acknowledgments

Thisworkwasfinanciallysupportedbythe“PRES/Région Midi-Pyrénées”.TheauthorswouldliketothanktheAirbusGroupfor manyfruitfuldiscussions,Constelliumforsupplyingthematerial. TheTEManalyseswereperformedwiththehelpofMarie-Christine Lafont.

References

[1]J.G.Rinker,M.Marek,Microstructure,toughnessandstresscorrosioncracking behaviorofaluminumalloy2020,Mater.Sci.Eng.64(1984)203–221. [2]R.G.Buchheit,J.P.Moran,G.E.Stoner,Localizedcorrosionbehaviorofalloy

2090-theroleofmicrostructureheterogeneity,Corrosion46(1990)610–617. [3]P.Niskanen,T.H.Sanders,J.G.Rinker,M.Marek,Corrosionofaluminumalloys

containinglithium,Corros.Sci.22(1982)283–304.

[4]V.Proton,J.Alexis,E.Andrieu,J.Delfosse,A.Deschamps,F.DeGeuser,M.C. Lafont,C.Blanc,Theinfluenceofartificialageingonthecorrosionbehaviour ofa2050aluminium–copper–lithiumalloy,Corros.Sci.80(2014)494–502.

[5]C.Kumai,J.Kusinski,G.Thomas,T.M.Devine,Influenceofagingat200◦_C_on

thecorrosionresistanceofAl–LiandAl–Li–Cualloys,Corros.Sci.45(1989) 294–302.

[6]R.G.Buchheit,J.P.Moran,G.E.Stoner,ElectrochemicalbehaviorofT1(Al2CuLi)

intermetalliccompoundanditsroleinlocalizedcorrosionofAl–2%Li–3%Cu alloys,Corrosion50(1994)120–130.

[7]J.E.Kertz,P.I.Gouma,R.G.Buchheit,Localizedcorrosionsusceptibilityof Al–Li–Cu–Mg–ZnalloyAF/C458duetointerruptedquenchingfrom solutionizingtemperatures,Metall.Mater.Trans.A33(2001)2561–2573. [8]H.Y.Li,Y.Tang,Z.D.Zeng,F.Zheng,ExfoliationcorrosionofT6-andT3-aged

AlxCuyLizalloy,Trans.NonFerrousMater.Soc.China18(2008)778–783. [9]M.Guérin,E.Andrieu,G.Odemer,J.Alexis,C.Blanc,Effectofvarying

conditionsofexposuretoanaggressivemediumonthecorrosionbehaviorof the2050Al–Cu–Lialloy,Corros.Sci.85(2014)455–470.

[10]C.Luo,X.Zhou,G.E.Thompson,A.E.Hughes,Observationsofintergranular corrosioninAA2024-T351:theinfluenceofgrainstoredenergy,Corros.Sci. 61(2012)35–44.

[11]S.H.Kim,U.Erb,K.T.Aust,G.Palumbo,Grainboundarycharacterdistribution andintergranularcorrosionbehaviourinhighpurityaluminium,Scr.Mater. 44(2001)835–839.

[12]J.G.Brunner,N.Birbilis,K.D.Ralston,S.Virtanen,Impactofultrafine-grained microstructureonthecorrosionofaluminiumalloyAA2024,Corros.Sci.57 (2012)209–214.

[13]M.Winning,A.D.Rollett,Transitionbetweenlowandhighanglegrain boundaries,ActaMater.53(2005)2901–2907.

[14]M.Tanaka,K.Higashida,K.Kaneko,S.Hata,M.Mitsuhara,Cracktip dislocationsrevealedbyelectrontomographyinsiliconsinglecrystal,Scr. Mater.59(2008)901–904.

[15]P.Donnadieu,Y.Shao,F.DeGeuser,G.A.Botton,S.Lazar,M.Cheynet,M.De Boissieu,A.Deschamps,AtomicstructureofT1precipitatesinAl–Li–Cualloys

revisitedwithHAADF-STEMimagingandsmall-angleX-rayscattering,Acta Mater.59(2011)462–472.

[16]C.Dwyer,M.Weyland,L.Y.Chang,B.C.Muddle,Combinedelectronbeam imagingandab-initiomodelingofT1precipitatesinAl–Li–Cualloys,Appl.

Phys.Lett.98(2011)201909.

[17]Z.Gao,J.Z.Liu,J.H.Chen,S.Y.Duan,Z.R.Liu,W.Q.Ming,C.L.Wu,Formation mechanismofprecipitateT1inAlCuLialloys,J.AlloyCompd.624(2015)

22–26.

[18]V.Araullo-Peters,B.Gault,F.deGeuser,A.Deschamps,J.M.Cairney, MicrostructuralevolutionduringageingofAl–Cu–Li–xalloys,ActaMater.66 (2014)199–208.

[19]M.Froment,Surlemécanismedelacorrosionintergranulairedesmatériaux métalliques,J.Phys.Colloq.36(1975)371–385.

[20]S.H.Kim,U.Erb,K.Aust,G.Palumbo,Grainboundarycharacterdistribution andintergranularcorrosionbehaviorinhighpurityaluminum,Scr.Mater.44 (2001)835–839.

[21]V.Keast,D.Williams,Grainboundarychemistry,Curr.Opin.SolidStateMater. Sci.5(2001)23–30.

[22]B.W.Bennett,H.W.Pickering,Effectofgrainboundarystructureon sensitizationandcorrosionofstainlesssteel,Metall.Trans.A18(1987) 1117–1124.

[23]S.R.Ortner,V.Randle,Astudyoftherelationbetweengrainboundarytype andsensitisationinpartially-sensitisedAISI304stainlessstellusingelectron back-scatteringpatterns,Scr.Metall.23(1989)1903–1908.

[24]C.Luo,X.Zhou,G.E.Thompson,A.E.Hughes,Observationsofintergranular corrosioninAA2024-T351:theinfluenceofgrainstoredenergy,Corros.Sci. 61(2012)35–44.

[25]S.R.Salimon,A.I.Salimon,A.M.Korsunsky,Theevolutionofelectrochemical, microstructural,andmechanicalpropertiesofaluminiumalloy2024-T4 (D16AT)duringfatiguecycling,Proc.Inst.Mech.Eng.PartG(2010)339–353. [26]K.D.Ralston,D.Fabijanic,N.Birbilis,Effectofgrainsizeoncorrosionofhigh

purityaluminium,Electrochim.Acta56(2011)1729–1736. [27]V.Proton,J.Alexis,E.Andrieu,J.Delfosse,M.-C.Lafont,C.Blanc,

Characterisationandunderstandingofthecorrosionbehaviourofthenugget ina2050aluminiumalloyfrictionstirweldingjoint,Corros.Sci.73(2013) 130–142.

[28]L.Lapeire,E.MartinezLombardia,K.Verbeken,I.DeGraeve,L.A.I.Kestens,H. Terryn,Effectofneighboringgrainsonthemicroscopiccorrosionbehaviorof agraininpolycrystallinecopper,Corros.Sci.67(2013)179–183.