• Keine Ergebnisse gefunden

How the interface type manipulates the thermochemanical response of nanostructured metals: A case study on nickel: A case study on nickel

N/A
N/A
Protected

Academic year: 2022

Aktie "How the interface type manipulates the thermochemanical response of nanostructured metals: A case study on nickel: A case study on nickel"

Copied!
13
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

ContentslistsavailableatScienceDirect

Materialia

journalhomepage:www.elsevier.com/locate/mtla

How the interface type manipulates the thermomechanical response of nanostructured metals: A case study on nickel

O. Renk

a,

, V. Maier-Kiener

b

, C. Motz

c

, J. Eckert

a,d

, D. Kiener

d

, R. Pippan

a

aErich Schmid Institute of Materials Science, Austrian Academy of Sciences, Jahnstraße 12, 8700 Leoben, Austria

bDepartment of Materials Science, Chair of Physical Metallurgy and Metallic Materials, Montanuniversität Leoben, Roseggerstrasse 12, 8700 Leoben, Austria

cSaarland University, Institute of Material Science and Methods, Saarbrücken, Germany

dDepartment of Materials Science, Chair of Materials Physics, Montanuniversität Leoben, Jahnstraße 12, 8700 Leoben, Austria

a r t i c le i n f o

Keywords:

Interface structure Boundary character Nanotwinned

Intergranular stress relaxation Nanoindentation

Nickel

Rate controlling process

a b s t r a ct

Thepresenceofinterfaceswithnanoscalespacingsignificantlyenhancesthestrengthofmaterials,butalsothe ratecontrollingprocessesofplasticflowaresubjecttochange.Duetotheconfinedgrainvolumes,intragranular dislocation-dislocationinteractions,thepredominantprocessesatthemicrometerscale,arereplacedbyemission ofdislocationsfromandtheirsubsequentaccommodationattheinterfaces.Bothprocessesnotonlydependon theinterfacialspacing,butalsoontheatomisticstructureoftheinterface.Hence,athoroughunderstanding how theseprocessesareaffectedbytheinterfacestructureisrequiredtopredictandimprovethebehavior ofnanomaterials.Thepresentstudyattemptstorationalizethiseffectbyinvestigatingthethermomechanical behaviorofsamplesconsistingofthreedifferentinterfaces.Purenickelsampleswithpredominantfractionsof low-andhigh-angleaswellastwinboundarieswithasimilaraveragespacingaround150nmareinvestigated usinghightemperaturenanoindentationstrainratejumptests.Dependingontheinterfacestructure,hardness, strainratesensitivityandapparentactivationvolumesevolvedistinctivelydifferentwithtestingtemperature.

Whileincaseofhigh-angleboundariesforallquantitiesapronouncedthermaldependenceisfound,theother twointerfacetypesbehavealmostathermalinthesametemperaturerange.Thesedifferencescanberationalized basedonthedifferentinterfacialdiffusivity,affectingthepredominantprocessofinterfacialstressrelaxation.

1. Introduction

Theenhanced fractions of grain boundaries (GBs) and interfaces makingup nanocrystalline(NC) materialsornanocompositessignifi- cantlyaltertheirpropertyspectrum.Thetremendousdecreaseofthe meanobstaclespacingfordislocationsenhancesstrengthtounprece- dented levels [1–4]. Besides that, the rate-controlling processes of plasticflowarealteredcomparedtothecoarsegrainedreferencestate, astheprobabilityforstorageorinteractionofdefectswithinthegrains orphasesdiminisheswhentheirvolumeisreducedtothenanoscale.

ForFCCmetals,thisimpliesashiftfromforestcutting(i.e.dislocation- dislocationinteractions)todislocationemissionfromandinteraction with GBs or interfaces [5]. Accordingly, the measured strain rate sensitivity(SRS),m,ofplasticflowissignificantlyenhancedforNCFCC metals,accompaniedbyareductionofthe(apparent)activationvolume, V,beingonlyontheorderofseveral10b3[6–9].Itisevidentthatthe interactionofdefectswithinterfacesstronglydependsontheiratomistic structure,consequentlyaffectingtheresultingproperties.Nanotwinned

Correspondingauthor.

E-mailaddress:oliver.renk@oeaw.ac.at(O.Renk).

(NT)metalsareoneofthemostprominentexamplesdemonstratingthis effect.Theintroductionofnanoscaledtwinsintomicrometer-sizedcop- pergrainsallowstounitehighstrengthwithexcellenttensileductility [10,11].Thissuperiorpropertycombinationisinclearcontrasttocon- ventionalNCorultra-finegrained(UFG)copper,offeringcomparable strengthbutextremelylimitedtensileductility[12,13].Thesynergyof strengthandductilityincaseofNTcopperisassociatedwiththesimul- taneouspossibilityofeasysliptransferandgenerationofsessiledisloca- tionsatthetwinboundaries(TBs).Thisensuresworkhardeningcapa- bility,notaccessibleforrandomhigh-anglegrainboundaries(HAGBs), wheredislocationseasilyannihilateattheoppositeGBs[14,15].Sim- ilarly,moleculardynamics(MD)simulationscomparingsampleswith HAGBsandlow-anglegrainboundaries(LAGBs)suggestastrongerwork hardeningabilityinthelattercase[16],inlinewithexperimentalobser- vationsonaluminum[17,18].Apartfrommechanicalproperties,also thethermalstabilityissignificantlyaffectedbyachangeoftheinterface structure.NanomaterialswithlargefractionsofLAGBsorTBsshowed significantlyimprovedthermalstabilitycomparedtostructuresconsist- ingmainlyofHAGBs,rationalizedbytheirreducedinterfacialenergies anddiffusivities[19–22].This emphasizesthatnot onlythemechan-

https://doi.org/10.1016/j.mtla.2021.101020 Received20January2021;Accepted22January2021 Availableonline24January2021

2589-1529/© 2021ActaMaterialiaInc.PublishedbyElsevierB.V.Allrightsreserved.

(2)

icalbehavior,butalsootherproperties suchas thethermalstability arestronglydeterminedbytheinterfacetype.Hence,alsothethermo- mechanicalresponsewillbereadilyaffected,buthasremainedwidely unexploredsofar,especiallythecasefornanostructuresbuiltofLAGBs.

Focusingonthemechanicalbehavior,theinfluenceoftheinterface charactershouldbereadilyreflectedintherate-controllingprocesses ofplasticflow.However,studiesunravelingtheeffectoftheinterface characterandtakingapotentialsizedependence(i.e.effectsresulting fromtheinterfacialspacing)intoaccount,remain,exceptforthoseon NTcopper[14],scarceandarecurrentlymostlylimitedtoambienttem- perature.Unexpectedly,acomparisonofNTcopperwithconventional NCcoppersuggestedthattheSRSandactivationvolumeVofplastic flowdonotdifferforagiveninterfacialspacingandfollowthesame scalinglaws[14,23].Basedontheirfundamentallydifferentproperties discussedaboveandrecentresultsemphasizingtheimportanceofinter- granularstressrelaxation[9],thisagreementisastonishing.Infact,the measurementsobtainedonvariousUFGandNCFCCmaterialsshowed thattheSRSremainedconstantatlowlevelsuntilaboveacertaintem- peratureapronouncedincreaseoccurred[9].Abovethistransitiontem- peraturealsotheapparentactivationvolumesstartedtodecrease,ap- proachingsingledigit values,indicative ofdiffusion-basedprocesses.

Thisissupportedbyactivationenergiesatelevatedtestingtemperatures beingcomparabletothosereportedforGBdiffusion[9].Moreover,the transitiontemperatureswerefoundtobeonly50– 70Kbelowtheav- erageonesmeasuredforthermally-induceddislocationannihilationat randomHAGBsintheseparticularmaterials,cf.Refs.[9,24].Abovethe mentionedtransitionpoint,thetemperatureseemssufficienttoallow forthermallyfacilitatedstressrelaxation,reflectedinapronouncedtem- peratureandtimedependenceofplasticflow.Asawhole,theseresults pointtointergranularstressrelaxationasthecontrollingprocess.How- ever,modelsdescribingthekineticsofpurelythermally-inducedstress relaxation(Eq.(1))[25,26],indicatethatapartfrommaterialconstants suchastheshearmodulus,G,theatomicvolumeΩ,theboundarywidth 𝛿,andfundamentalconstantssuchastheBoltzmanconstant,kB,anda pre-factorA(reportedtobeabout1/200forseverelydeformednanoma- terials[25,27]),therelaxationtime𝜏relaxonlydependsontheboundary diffusivityDGBandthespreadingdistances,whichisnativelyrestricted tothegrainsizeasanupperlimit.

𝜏𝑟𝑒𝑙𝑎𝑥=𝐴 𝑘𝐵𝑇𝑠3

𝐺Ω𝐷𝐺𝐵𝛿 (1)

BasedonEq.(1),itisevidentthatforagivenmaterialandinter- facialspacing,therelaxation timemainlydependson theinterfacial diffusivity.Accordingly,differentrelaxationtimesandhencedeforma- tionbehaviorareexpectedfordifferentinterfacetypes.Inotherwords, differentvaluesoftheSRSorworkhardeningrateswouldbeexpected, contrastingthementionedagreementinpreviousstudies[14].Although thereportedsimilaritiesbetweenHAGBsandTBscouldresultfromthe restrictiontoambienttestingtemperatures,theobviouscontroversies motivatedadetailedstudyunravelingtheeffectoftheinterfacetypeon thedeformationbehaviorofnanostructuredmetals.Duetoitshigher meltingpointweusednickelasamodelmaterialtoallowformeasure- mentsinawidertemperaturerangewithouttheoccurrenceofsignif- icantmicrostructuralmodifications.Inaddition,NCnickelandnickel alloyshavebeenwidelystudiedatroomtemperature(e.g.Refs.[8,28–

31])whatallowsforacomparisonwiththegathereddata.Hightem- peraturenanoindentationstrainratejumptestsperformedonsamples withLAGBs,HAGBsandTBsofsimilarspacingallowedtoassesstheef- fectoftheinterfacetypeonthedeformationresponseandtoprovethe raisedhypothesis thatinterfacialstressrelaxationisthepredominant rate-controllingprocess.

2. Experimental

Tounderstand howtheinterface typeaffectsthedeformationbe- havior, nanostructurednickel sampleswithdifferent boundarytypes

butsimilarinterfacialspacingweresynthesized.Thesamplesconsidered inthisworkconsistpredominatelyofHAGBs,LAGBsandTBs,labelled HAGB,LAGBandNTsamplehereafter.TheHAGBandLAGBsamples wereprocessedbyhighpressuretorsion(HPT)andcyclichighpressure torsion(CHPT)atambienttemperature.Nickel(99.99%,fromGoodfel- low)sampleswith10mmdiameterand1 mmheightweredeformed byquasi-constrainedmonotonicHPT[32]atambienttemperaturefor 10 rotationsatarotationalspeedof 0.2rotmin1underanapplied nominalpressureof5.1GPa.Suchseveremonotonicstrainsresultinan UFGstructureconsistingofrandomHAGBswhichmakeupto75–80%

ofthetotalboundarylength[33,34].Tocreateananoscaledstructure ofsimilardimensionsbutconsistingmainlyofLAGBS,recrystallizedNi samples(773K/1h)havingthesamedimensions(10mmdiameterand 1mmheight)werecyclicallyHPTdeformed(5° twistangle)forfive cyclesusingasetupintroducedearlier[35,36].Thelargeplasticstrain amplitudesappliedallowedtocreateadistinctfractionofdislocation cellshavingapproximatelythesamespacingasobtainedincaseofthe monotonicallydeformedsamples.TheNTnickelsamplesweresynthe- sizedusingpulsedelectrodeposition.Sheetsofseveralmillimeterthick- nessconsistingofNTnickelstructuresweregrownoncoppersubstrates.

Butindiol(0.02gl1)andsodiumsaccharin(3gl1)wereaddedtothe electrolyteasgrainrefiners.Adetailedchemicalcompositionoftheelec- trolytecanbefoundelsewhere[37,38].Depositionwasperformedus- ingsquarepulseswithacathodiccurrentdensityof45mAcm2and ananodiccurrentdensityof65mAcm2withapulsedurationof5ms, respectively.

BoundaryspacingsandfractionsofboundarycharactersoftheHAGB andLAGBsampleswereanalyzedusinganelectronbackscatterdiffrac- tion(EBSD)detectorattachedtoaLEO1525fieldemissiongunscanning electronmicroscope(SEM,CarlZeissMicroscopy,Germany),whilefor theNTnickelsamplestransmissionKikuchidiffraction(TKD)inanon- axisconfiguration[39]withasetupfromBruker,Germanywasused.

ThestepsizeincaseoftheTKDscanswassetto10nm,while25 nm wereusedfortheconventionalEBSDscans.ForconventionalEBSD,a finalelectrolyticalpolishingstep(StruerselectrolyteA2)wasusedto remove anydeformationlayerpresentfrommechanicalgrindingand polishing.ForTKD,anelectrontransparentNTsamplewasprepared followingstandardTEMpreparationroutinesofmechanicalgrinding, polishinganddimpling,followedbygentleionpolishing.Theobtained datawasanalyzedusingastandardsoftwarepackage(OIManalysis5.3, EDAX).FortheHAGBandLAGBsamplesthemicrostructureswerean- alyzedinradial(RAD)directionoftheHPTdiskataradiusof4mm, whiletheNTstructurewasobservedperpendiculartothegrowthdirec- tion.Toexaminethethermalstabilityofthesamplesandtodetermine themeaningfultemperaturerangeforthehightemperaturenanoinden- tation protocolusedtoassessthedeformationbehavior,thesamples wereisochronallyannealed(30min)atvarioustemperatures.Vickers microhardnessmeasurements(0.5gfload,15s)atambienttemperature accompaniedbymicrostructuralanalysisasdescribedbeforeenabledto determinethethermalstabilityrange.PleasenotethatincaseofVick- ershardness,nottheprojectedarea,butthecontactareaisconsidered forevaluatingthehardness.Hence,forthesametesttemperatureand materialcondition,thesevaluesareslightlyreducedcomparedtothe nanoindentationdata.Minoreffectscouldfurtherarisefromdifferences inappliedstrainrate.

Toobtaininsightsintothedeformationbehavior,nanoindentation strainratejumptestsatroomtemperatureandelevateddeformation temperatureswereperformed[40].IncaseoftheHPTdisks(HAGBand LAGBsamples)theindentationdirectionwaschosenalongtheaxialdi- rection,whilethenanotwinnedsampleswereindentedalongthegrowth direction,i.e.perpendiculartotheTBs.Priortotheteststhesamplesur- facesweremechanicallygroundandpolishedfollowedbyafinalelec- trolyticalpolishingstep(StruerselectrolyteA2)toremoveanydeforma- tionlayers.IncaseoftheHPTdisks(LAGBandHAGBstructures),the indentswereperformedataradiusofabout4mm,i.e.whereallmi- crostructuralinvestigationswereperformed.Fromthenanoindentation

(3)

strainratejumptests hardness,H,strainratesensitivity, m, andap- parentactivationvolumes,V,asafunctionofthetestingtemperature canbeextracted.Hence,insightsintotherate-controllingdeformation processesofthedifferentnanostructuresandinterfacescanbegained.

AlltestswereperformedonaplatformNanoindenterG200(KLA,CA, USA)equippedwitha continuousstiffness measurement(CSM)unit.

TheCSMunitsuperimposesasinusoidalloadsignal(2nmamplitude and45Hzfrequencyusedhere)thatallowstodeducecontactstiffness andthusmodulusandhardnesscontinuouslythroughoutthetest.For elevatedtestingtemperatures,alaserheatingstage(SurfaceTec,Hück- elhoven,Germany)wasused.Both,thetipandthesampleareheated independently,allowingtoadjustandstabilizethecontacttemperature carefully,minimizingthermaldrift[41].Duetothelaserheating,the selectedtemperaturesandsubsequentstabilizationarereachedwithina fewminutes,reducingtheexposurepriortotheindent.Thisalsoguaran- teedawell-definedandhomogenoustemperaturedistributionthrough- outtheindentationsequence.Thewholesystemincludingsampletray andthecoppercoolingshieldsurroundingtheindentationtipwaswa- tercooledandthechambertemperaturewascloseto291K.Aninert gasatmosphere(forminggas– N2containing5%H2)wasobtainedby twoindividualvalvesthatcontrolledthegasflowaroundthesample trayandtheheatedtip.Thiscreatedanoven-likeatmosphereanden- suredthatoxidationofthesamplescouldbeprevented.Afterthehigh temperaturetests,allsampleswereinspectedvisuallyand,althoughnot quantified,noobviousoxidationwasrecognized.Forallmeasurements Berkovichtipswereused.Incaseoftheelevatedtestingtemperatures sapphiretips(Synton-MDP,Nidau,SwitzerlandtogetherwithSurface Tec,Hückelhoven,Germany)wereusedtominimizechemicalinterac- tionsbetweentip andsample[42],whileatambienttemperaturedi- amondtips(Synton-MDP,Nidau,Switzerland)wereused.Framestiff- nessaswellastipshapecalibrationsweremadeonfusedsilicaregularly betweentheindividualsamplesaccordingtotheOliver-Pharrmethod [43].Athermaldriftoflessthan0.1nm/swasallowedpriortoanytest andwasdeterminedateachtestingtemperaturebeforeaswellasafter eacharrayofindents.

Ateverychosentesttemperatureatleastfourindentations,separated by50μm,wereperformedwithintheheatingcycle,rangingbetween roomtemperature(RT)and573K,depending onthethermalstabil- ityoftheindividualsamples,respectively(NiHAGBs:298,373,398, 425,448,473and498K;NiLAGBs:298,373,423,473,498,523,548 and573K;NiNT:298,373,398,423,448,473,498,523,548and 573K). Thenanoindentationtests wereperformedinconstantstrain rate-controlledmode[44].Theappliedstrainratewasalteredabruptly afterevery500nmindentationdepth,starting with0.05s1,reduc- ingitfurtherto0.01s1,increasingitbackto0.05s1,droppingto 0.005s1beforejumpingbacktotheinitialstrainrateof0.05s1.All hardnessandmodulusdatareportedinthefollowingareaveragedval- uesobtainedatastrainrateof0.05s1between 1050and1450nm indentationdepth.Basedonthestrainratejumptests,thestrainrate sensitivity,m,andtheapparentactivationvolumes,V,canbecalcu- latedaccordingtoEqs.(2)and(3)[30],respectively,whereHdenotes thehardness, ̇𝜀thestrainrate,Caconstraintfactorbeing2.8,kBthe Boltzmanconstant,andTtheabsolutetemperature.

𝑚= 𝜕(ln(𝐻))

𝜕(ln(̇𝜀)) (2)

𝑉 =𝐶∗√

3∗𝑘𝐵𝑇

𝑚𝐻 (3)

ItshouldbenotedthatthevaluescalculatedaccordingtoEqs.(2) and(3)coulddependontheappliedstrain(i.e.indentergeometry)and strainrate,withthelatterchosenratherhighwithinthisstudy.Notably, forhighstrainratetestsconstanthardness(flowstresses)priortoan- otherstrainratejumpmaynotalwaysevolve,complicatingthisanalysis.

Thiswas,however,notthecasehere,asforthemajorityoftheapplied ratejumpsrather constanthardnesslevelsevolved,compareFig.6b.

Furthermore,whilethedislocationdensitycanaffectthecalculatedSRS orVvaluesformetalssubjectedtocomparablylowstrains[45],this issueisnotpronouncedfornanostructuredmaterialstestedhere.Previ- ousresultsindicatethatBerkovichtipsimposingastrainof~7%strain alreadynecessitatealargenumberofdislocations.Thus,nosignificant differenceoftheSRSorVatagiventemperaturewasevidenced,inde- pendentifmeasuredduringtheheatingorcoolingsequence(i.e.testing alreadyarelaxedstate)[9].

Asthetestprotocolincludestwostrainratejumpsbacktoapre- viousstrainrate(̇𝜀=0.05s1),thehardness-contactdepthcurvescan beusedasafirstindicatorforstructuralmodificationsduetothether- momechanicalexposureofthenanostructures.Aswillbedetailedlater, forHAGBsamplestestedatspecifictemperatures,thesecurvesalready suggestedgraingrowthbeneaththeindentertip.Therefore,thecross- sectionsofparticularindentswereextractedandinspectedaswell.After depositingaprotectiveplatinumlayerontopoftheindentusingagas injectionsysteminafocusedionbeam(FIB)workstation(ZeissAuriga, Zeiss,Germany),aslicethroughtheresidualimpressionwithathickness ofabout2μmwasextractedfromthesampleusinganOmniprobe200 micromanipulator(OxfordInstruments,UK).Subsequently,thesamples wereattachedtoacoppergridbyplatinumdepositionandanalyzed usingconventionalEBSDasspecifiedbefore.

AllresidualimprintswerefurtherimagedusingaKeyencelaserscan- ningconfocalmicroscopetodeterminetheheightofthepile-upsand theirevolutionwithtemperature.Therecordeddatawasanalyzedus- ingthesoftwarepackageGwyddion2.56.Linearprofileswereapplied totheimagedindentstomeasurethepile-upheights.Thereportedval- uesareaverageswiththestandarddeviationastheerrorbar.Itshould benotedthatforsomeconditions(testtemperature,interfacetype)the errorbarseemsratherlarge,whichisnotaresultofpronouncedscatter betweentheindentstakenatagiventemperature,butratherduetodis- tinctdeformationdifferencesbetweenthethreefacesoftheBerkovich tip.

3. Results

3.1. Microstructuralcharacteristicsandthermalstabilityofthethree differenttypesofnickelsamples

Fig.1showsrepresentativemicrographsofthethreedifferenttypes ofnickelsamplesunderinvestigation.TheHAGB samples(monotoni- callyHPTdeformed,Fig.1a)exhibitatypicalUFGmicrostructurewith slightlyelongatedgrainsalongthetangential(TAN)direction,consist- ingpredominantlyofHAGBswithfractionsofupto75%ofthetotal boundarylength.Itshouldbementionedthatafterseverestrainingno preferredmisorientationangleoraxispaircanbedetermined,asre- portedearlier[33,34,46,47].Thus,thegeneratedHAGBscanbecon- sidered asrandom. This is alsoevident from themisorientationdis- tribution,Fig.2a.Averageboundaryspacingsweredeterminedusing thelineinterceptmethodandsettingacriticalmisorientationangleof morethan15°,yieldingaverageminimumdimensions(alongtheaxial direction)ofabout190nm.TheLAGBsamples(Fig.1b)appearsimilar toconventionalcold-workedstructures.Insidethestilldistinguishable coarsegrains,well-defineddislocationboundarieswithlowmisorienta- tionanglesdeveloped.Thefractionofboundarieswithamisorientation anglebelow15° inthesesamplesmakesup75%ofthetotalboundary length.BecausetheaveragespacingoftheLAGBswasdeterminedby conventionalEBSD,thethresholdangletoidentifyagrainwassetto 2°,althoughdislocationboundarieswithanevenlowermisorientation anglemayexist.Accordingly,theaveragespacingalongtheaxialdi- rection(minimumdimensions)determinedforthiscondition(265nm) canonlybeconsideredasanupperboundfortheactualspacing.For both,theHAGBsamplesandtheLAGBsamples,dislocation-basedplas- ticityisevidentfromtheODFsections,whichshowagoodagreement withtheidealtexturecomponents[48]inFig.2c.TheNTnickelsam- plesconsistofcolumnargrainsingrowthdirection,havinganaverage

(4)

Fig.1. Representativemicrostructuresofthetestednickelsampleshavingthreedistinctinterfacetypes.(a)HPTdeformednickelconsistingpredominantlyofHAGBs;

(b)Recrystallizedandcyclicallycold-workednickelwithamajorityofLAGBs(boundariesintheinsetimageinblackdenoteHAGBs),and(c)electrodepositedNT nickel.Forallspecimentypestheaverageinterfacialspacingiswith150–250nmrathersimilar.

Fig.2. (a)UncorrelatedmisorientationdistributionoftheHAGBandLAGBsamplessynthesizedbyHPTandcyclicHPT,respectively,withtherandomdistribution (opencircles)givenforcomparison;(b)distributionofthelineinterceptsforthedifferentinterfacetypes;(c)RepresentativeODFsectionsoftheHAGBandLAGB sampleswiththeblackdotsindicatingthepositionofidealtexturecomponentsexpectedfordislocation-basedplasticity[48].

thicknessofabout 2.5μm,withthelargest onesbeingup to10μm thick.Insidethecolumnargrainshighdensitiesoftwinboundariescan befound,seeFig.1c.Thisresultsina{111}fibretextureingrowthdi- rection.Themajorityofthetwinboundariesarecoherenttwinbound- aries(fraction~ 85%)andtheiraveragespacingbasedonlineinter- ceptsisabout120nm.Thus,sampleswithrelativelysimilarboundary spacingscouldbe synthesizedandassessedwithrespecttotheirme-

chanicalproperties andthermalstability.Itshouldbe notedthatthe purityoftheNTnickelsamplesislowercomparedtotheothertwoin- terfacetypes,withsulphurbeingknownasmainimpurityinEDnickel samples.However,asthepreferredsegregationsitesarethecolumnar HAGBs,ratherthanthecloselyspacedTBswhichdeterminetheme- chanical properties,this isnot expectedtocause aneffect, compare Ref.[49].

(5)

Fig.3. ThermalstabilityofthethreedifferentnickelmicrostructuresasrevealedfromroomtemperatureVickersmicrohardness(0.5gfload)testsperformedafter isochronal(30min)annealingandcomplementarymicrostructuralinvestigations.Circlescorrespondtotheimagesinb).DataoftheHAGBspecimensistakenfrom Ref.[50].PleasenotethatthedifferenceoftheRThardnesscomparedtoFig.6originatesfromtheuseofthecontactareainsteadoftheprojectedareatodetermine theVickershardness.

Fig.4. RepresentativeSEMimagesoftheindents;

samples containing HAGBs (upperrow), LAGBs (middlerow)andTBs(bottomrow)atindicated testingtemperatures.

Asimilarboundaryspacingforallthreetypesofspecimensisalsoev- identfromthehardnessmeasurementstakenatroomtemperature.For alldifferenttypesofspecimensacomparablehardnessof2.5–3.2 GPa wasmeasured,seeFig.3.Totesttheirthermalstability,thesamples weresubjectedtoisochronal(30min)annealingtreatmentsatdifferent temperatures.Vickersmicrohardnessmeasurementsatroomtempera- tureaccompaniedwithstructuralcharacterizationallowedtodetermine anystructuralchanges.Thesethermalstabilitylimitswereusedtoset thetestingtemperaturerangeforthenanoindentationexperiments.It canbenoticedthattheHAGBsampleshaveasignificantlylowerthermal stabilitycomparedtotheLAGBandTBsamples.Alreadyabovetemper- aturesof423KmassivegraingrowthoccurredfortheHAGBspecimens, whileatthistemperaturepronouncedchangesstillremainabsentforthe LAGBandTBspecimens.Atannealingtemperaturesof523Kalsointhe LAGBsamplesfirstrecrystallizationnucleicanbe observed(Fig.3b), explainingtheratherlargestandarddeviationofhardnessforthispar- ticularannealingcondition.Atthisannealingtemperaturethehardness slightlydropsalsoin caseof theNTsamples, butchangesof theTB

spacingremainsmall.Evenafterannealingat773K,theTBspacingre- mainswellbelowamicron,indicatingasuperiorthermalstability.The improvedthermalstabilityincaseoftheLAGBandTBspecimenscom- paredtotheHAGBsamplesisinlinewithinvestigationsonNTcopper [20,21]andnanolaminatednickelconsistingofnanoscaledLAGBs[19]. Itcanberationalizedbytheirlowerinterfacialenergyanddiffusivity comparedtogeneralHAGBs,consequentlyreducingthedrivingforces forgrowthandtheatomicmobilitywithintheinterfaces[19,22].Apart fromthermalstability,pronounceddifferences betweenthethreedif- ferenttypesofnickelsamplesarealsoevidentwhenconsideringtheir mechanicalbehavior,aswillbeoutlinedinthefollowingsection.

3.2. Deformationbehavioratambientandelevatedtemperatures

Alreadyfromtheshapesoftheremanentimpressionsafterroomand elevatedtestingtemperatures,adistinctivelydifferentdeformationbe- haviorforthethreeinterfacetypesisevidenced.RepresentativeSEMim- agesoftheindentsfromthethreedifferenttypesofsamplesatdifferent

(6)

Fig.5. Analysisofthepile-upheightofthedifferentnickelsamplesasafunction oftemperature.ThelargeerrorbarsfortheLAGBandTBsamplesarejusta consequenceofthedifferenceofthepile-upheightalongthethreesidesofthe Berkovichtip.

testingtemperaturesarepresentedinFig.4.Itshouldbenotedthatthe micron-sizedgrainstructurevisibleinsomeimagesisjustaconsequence of recrystallizationoccurring in thesample, whichwas continuously heatedduringtheindentationsequencesandcertainlydoesnotaffect anyinterpretation.Forallthreeinterfacetypes,distinctpile-upforma- tion(i.e.aconvexdeviationfromtheidealindentshape)occursduring indentationatroomtemperature.Thisisexpectedformaterialshaving alreadyhighstrengthand,accordingly,limitedwork-hardening capa- bility[51,52].Quantitativeanalysisoftheindents(Fig.5)revealedthat foralltestingtemperaturespile-upformationismorepronouncedwhen LAGBsandTBsareinvolved.Thelargeerrorbarforthesetwosamples isnotcausedbyscatterbetweenmultipleindentstakenatagiventem- perature,butbythepronounceddifferenceofthepile-upheightalong thethreesidesoftheBerkovichtip.Uptotemperaturesof475K,the pile-upheightoftheLAGBandTBspecimensismoreorlessconstant, beforereducingtolowerlevels.Incontrast,thepile-upheightmeasured ontheHAGBspecimensalreadydecreasedatmuchlowertemperatures, Fig.5.Attestingtemperaturesof423K,stillbelowthethermalstability limit,theindentsresemblealreadyanalmostperfectandstraightshape, withthepile-upheightbeingsimilartotheonestakenat473Kwhere graingrowthalreadyoccurred.

Apartfromthecleardifferencesoftheindentshapesforthethree sampletypesandtheirevolutionwithtemperature,alsothetempera- turedependenceofthemeasuredhardnessvaluesexhibitsquitediffer- enttrends.WhileforthesamplescontainingLAGBsandTBsthehardness doesnotchangesignificantlywithanincreaseofthetestingtempera- ture,theHAGBsampleshowsadistincttemperaturedependence,see Fig.6a.DatatakenfromRef.[9]obtainedonaUFGNi4.5Alalloycon- sistingpredominantlyofHAGBsisaddedforcomparison.Asforthepure nickelsamplewithHAGBsalsoforthealloyedvariantthehardnessis extremelytemperaturesensitive.Thealloyingcontentonlyshiftsthe testingtemperaturesthatcauseasubstantialdropinhardnesstowards highervalues.Althoughmeasuringwellbelowthethermalstabilitylimit (~ 0.25Tm),thehardnessoftheHAGBsamplesdecreasesalreadyre- markably.Becauseboth,theLAGBandNTsamplesdonotshowadis- tincttemperaturedependence,above~ 0.22Tmthehardnessofthese twospecimensexceedsthoseoftheHAGBsamples.However,itiswell- establishedthatnanostructuresconsistingpredominantlyofHAGBsare pronetograingrowthduringdeformation,evenundercryogeniccon- ditions,asevidencedforavarietyoftestingtemperaturesanddeforma-

tionmodes,seeforinstanceRefs.[36,53–58].Therefore,graingrowth incaseoftheHAGBsamplesmayalreadyoccurbelowthethermalsta- bilitylimitduetothedeformation(7%equivalentstrain)imposedby theBerkovichindenter,explainingthereducedhardnessvalueswithin- creasingtestingtemperature.Astheindentationprotocolconsistedofa seriesofstrainratejumps,includingtwojumpsbacktoapreviousstrain rate,therecordedhardness-contactdepthcurves(Fig.6b)alreadyallow todeducemicrostructuralinstabilities.Nevertheless,fromthehardness- contactdepthdata,thisisnotexpecteduptotestingtemperaturesof 398K,sincefortestingtemperaturesbelow398K,astrainratejump backtoapreviousstrainrate(i.e.0.05s1)alwaysleadstoidentical hardnesslevels,indicatingmicrostructuralstability.Thisisdifferentfor highertestingtemperatures.Alreadyat423K,thehardnessforthelast strainratesegmentremainsbelowtheonesobtainedintheprevioustwo segments,presumablyalreadyaconsequenceofmeasuringclosetothe thermalstabilitylimit.

However, already for testing temperatures of 398 K, where the hardness-contactdepthsignalsuggestsastablemicrostructure,thehard- nessdiffersalreadymorethan30%comparedtoroomtemperature.To elucidatethereasonsforthishardnessreduction,anotherHAGBsam- plewas preparedandindentedat398K.Alift-out ofathin lamella throughtheindentperformedat398Kwastakenandthevolumebe- neath theindentsubsequently analyzed. Theobtained backscattered electronimagesanddetailedcolorIPFmapsofthemicrostructuresare showninFig.7.Obviously,thegrainsizebeneaththeindentissignifi- cantlylargerthaninunaffectedregionsfurtheraway.Onacloserlook theregionwherenoticeablegraingrowthoccurredislimitedtoacircu- larshapearoundtheindent.Estimatingtheplasticzonesizeaccording toRefs.[59,60]indicatesthatgraingrowthisconfinedtothisvolume, highlightedbythedashedline.Despitetheindentation-inducedgrowth issignificant, itisnotreflectedinthehardness-contactdepthcurves, showingidenticalhardnesslevelsforajumpbacktothesamestrain rate(i.e.0.05s1).Thiscouldberelatedtothefactthatfortheself- similarBerkovichindenttherealizedequivalentstrainand,hence,the amountofgrowthisindependentofthepenetrationdepth,butjustthe sizeoftheplasticzoneincreases.

3.2.1. Strainratejumptests

Fromthehightemperaturenanoindentationtests,strainratesensi- tivity, mandapparentactivationvolumes, V,weredeterminedasa functionoftemperatureaccordingtoEqs.(2)and(3).Thevaluesre- portedin thefollowingareaverages,with thestandarddeviationas theerrorbar.Fig.8summarizesthesequantitiesforthethreedifferent specimentypesasafunctionofthetesttemperature.Inaddition,data obtainedonaNCNi4.5AlalloyconsistingofamajorityofHAGBs(Ref.

[9])isplottedforcomparison.Asforhardnessandindentmorphology, alsothetemperaturedependenttrendsofstrainratesensitivityandap- parentactivationvolumedifferremarkablyforthethreeinterfacetypes investigated.Again,thesetrendsarecomparablefortheLAGBandNT specimens butdiffersubstantially fortheHAGB samples,Fig.8a.At ambienttemperature,SRSislow(0.01– 0.02)andrathersimilarforall threeinterfacetypes,althoughslightlyhighervaluesweremeasuredfor theHAGBnickelandNi4.5Alsamples.Alreadyforslightlyelevatedtest- ingtemperatures,thedifferencesbetweenthesamplesbecomeapparent.

WhileforthepurenickelsampleswithHAGBstheSRSquicklyincreases withtemperature,theLAGBandNTspecimensbehaveratherathermal.

Thisalsoholdstrueforthealloyednickelsampleconsistingpredomi- nantlyofHAGBs.However,similartothepurenickelHAGBsamples, above~ 0.25TmtheSRSoftheNi4.5Alalloyquicklyincreases.Incase ofthepurenickelHAGBsamples,theSRSreachesvaluesofabout0.05 at423K,beforedecreasingagain.ThisdropoftheSRScanbeattributed totheoccurrenceofsignificantgraingrowth,asthetestingtemperature exceededthethermalstabilityofthesamples,Fig.3.Evenhighervalues of0.1canbemeasuredincaseoftheNi4.5AlHAGBsamples,although foragiventemperaturetheSRSofthepurenickelsamplesisalways higher.Incontrast,fortheLAGBandNTsamples,nochangeoftheSRS

(7)

Fig.6. (a)Hardnessasafunctionoftestingtemperaturefornickelsampleswithdifferenttypesofinterfaces.DataobtainedonaNCNi4.5Alalloy(Ref.[9])witha predominantfractionofHAGBsisaddedforcomparison.Circleddatapointsin(a)refertothehardness-contactdepthcurvesoftheHAGBsamplesdisplayedin(b).

Fig. 7. Backscattered electron image showingmicrostructuralchangesbeneath anindent ofaHAGBsampleindentedat 398 K. The dashed line corresponds to the maximumdepth of theplastic zone, estimated according to Refs. [59,60].

Pleasenotethedifferentscalebarsofthe images.

couldbemeasureduptotemperaturesof473K(~ 0.25Tm).Evenfor highertestingtemperatures,theincreaseremainsrathersmallandfor bothtypesofsampleswithinthewholetemperaturerangetested,the SRSexceedsonlyslightlythatmeasuredincaseofHAGBsalreadyatam- bienttemperature(m~ 0.02).Nevertheless,alsoforthesetwointerface typesasimilartrendasmeasuredfortheHAGBsamplescanbeexpected, giventhatthestructuresremainstableuptowaylargertemperatures thantestedhere.Thisissupportedbydataobtainedduringlargestrain deformationofpurecoarse-grainednickel(i.e.incompressionandtor- sion),suggestingthatatleasttemperaturesof873Kwouldberequired tomeasureadistinctratedependenceofflowstress[61,62].Whilethis wouldbepossiblefortheNTsamples(compareFig.3),incaseofthe LAGBsamplesrecrystallizationoccurredalreadyatabout0.32Tmand causedadropoftheSRStoevenlowervalues.

Ascanbeexpectedfornanostructures,theapparentactivationvol- umesVareontheorderofseveraltensofb3forallspecimenstested

atroomtemperature,seeFig.8b.With~ 40b3theapparentactivation volumesincaseoftheLAGBandNTspecimensareslightlylargerthan theonescalculatedfortheHAGB samples.Forthepureandalloyed HAGBsamples,valuesofabout20b3weremeasured.AstheSRSofthe LAGBandNTsamplesdoesnotchangesignificantlywithtemperature accordingtoEq.(3),theapparentactivationvolumesofthesesamples increasemoderatelyuptoabout60 b3 withtestingtemperature.Sig- nificantly largervaluesof almost400b3 aremeasuredfor theLAGB specimenafterrecrystallization,asexpectedforacoarse-grainedFCC metalatmoderatetestingtemperatures,withforestcuttingbeingthe ratecontrollingprocess.Incontrast,Vofboth,thepureandalloyed HAGB specimens,remainsratherconstantbeforestartingtodecrease oncereachingthetemperatureswheretheSRSquicklyincreases.This trendofdecreasingVisrestrictedtoanarrowtemperaturerangein caseofthepurenickelHAGBsamplesasgraingrowthoccurs,reflected inadistinctincreaseoftheactivationvolume.However,incaseofthe

(8)

Fig.8. (a)Strainratesensitivity,m,and(b)apparentactivationvolume,V,ofthenickelsampleshavingdifferentinterfacesasafunctionoftestingtemperature.

DataforaNCNi4.5Alalloy(Ref.[9])consistingpredominantlyofHAGBsandEDNCNi(Refs.[28,31])areaddedforcomparison.Trendlinesareaddedasguide fortheeye.

morestableNi4.5AlHAGBsamples,thedecreaseisclearlyvisibleand singledigitactivationvolumes,indicativefordiffusiveprocesses,are approached,Fig.8b.

Basedonthemeasuredhardnessandactivationvolumes,theappar- entactivationenergiesoftheratelimitingprocessescanbeestimated accordingtoEq.(4)[63].Thisapproachdescribestheactivationbarrier, Q,asstress-dependentGibbsfreeenergy,ΔG,toactivateplasticflow.

Sinceitisgenerallyassumedthatcontributionsfromtheentropyterm arenegligible[29,64],ΔGiscloselyrelatedtotheactivationenthalpy ΔHact,expressedas:

𝑄 ∼ Δ𝐻𝑎𝑐𝑡= 𝐻𝑉

√3𝐶𝑇

𝜕(ln𝐻)

𝜕(1

𝑇

) (4)

withHandVbeingthehardnessandapparentactivationvolume,re- spectively,T,theabsolutetemperatureandCaconstraintfactorof2.8.

AscanbeseenfromFig.9a,forallthreeinterfacetypesthelogarithm ofthehardnessasafunctionoftheinversetemperaturecanbereason- ablydescribedbyalinearfitwithintwodistincttemperatureranges.In caseoftheHAGBsamples,onlytheoneatlowertemperatureswascon- sideredfurther,asthatathighertemperaturesalreadycorrespondsto a(partially)coarsenedmicrostructure,compareFigs.6and8.Thecal- culatedactivationenthalpiesandtheirchangewithtestingtemperature aredisplayedinFig.9b.FortheHAGBspecimensatRTanactivation enthalpyofabout2eVcanbedetermined,whichimmediatelydecreases toabout1eVwithanincreaseofthetesttemperature.FortheLAGBand NTsamples,activationenthalpiesofabout3eVarederivedatelevated testingtemperatures,wherepropertiesbecomeslightlytemperaturede- pendent.ThesevaluesarecomparabletothoseexpectedforGBdiffusion (~ 1eV)andbulkdiffusioninnickel(~ 3eV),respectively,compare Refs.[65,66].

4. Discussion

4.1. Effectoftheinterfacetypeandthedeformationtemperatureonthe indentshape

Distinctdifferencesarenotonlyobservedforthemechanicalprop- erties,butalreadyfortheindentshapes,Fig.4.Whilepile-upforma- tionforhighstrengthmaterialswithlowwork-hardeningcapacityis

expected[51,52], thedifferences betweenthethreetypesof samples (Fig.5)needtobebrieflydiscussedintheremainder.AlthoughLAGBs andTBsareknowntohavehigherwork-hardeningratescomparedto HAGBs[10,13],largerpile-upsareformedforthesetwosamples.More- over,pile-upformationincaseoftheHAGBsamplescontinuouslydimin- isheswithincreasingtestingtemperature, whichisunexpectedgiven thereductionofhardnessandwork-hardeningratewithtemperature [51].

Independentofthesampletype,duetofixedgeometryandindenta- tiondepthaconsistentvolumeofmaterialneedstobedisplacedbythe indentertip.However,thisoccursmorelocalizedincaseoftheLAGB andNTsamples,presumablybecausetheseinterfacesaremorerigid comparedtotheHAGBs.Whilemoredetailedstudiesarerequiredto assessthisissueindetail,investigationsonNTcoppersupportthisno- tion[67].Asdetailedinthenextchapter,detwinningandlocalcollapse of thetwinstructurewasreported, favoringlocalplasticflow,hence largerpile-upheights.AsthemechanicalpropertiesincaseoftheLAGBs andTBsarelargelyunaffectedbythetesttemperature,apronounced changeofthepile-upheightisthusnotexpected.Incontrast,thermo- mechanicallyinducedmigrationoftheHAGBscaneasilyoccur(Fig.7).

Thisisinlinewithdetailedobservationsshowingthatatlowhomolo- goustemperaturesunderanappliedstressfield,foragivenGBcrystal- lography,HAGBsmigratemuchfasterthanLAGBsdo[68].Asanymov- inginterfacerealizesplasticstrain,theeasiermigrationofHAGBscould facilitateamorehomogenousplasticflowoverlargervolumes,leading tosmallerpile-upsheights.Thelargeaffectedzonebeneaththeindent atmoderatetestingtemperaturestogetherwiththeenhancedratesen- sitivityincaseoftheHAGBspecimens(Fig.7andFig.8)supportthis picture andalsoexplainsthefurtherdiminishingpile-upheightwith increasingtesttemperature.

Apartfromalargeraveragepile-upheightincaseoftheLAGBand NTspecimens,alsotheheightdifferencebetweenthethreefacesofthe Berkovichtip,givingrisetothelargeerrorbarsinFig.5,ismorepro- nounced.Thiscanbe rationalizedbytwofactors.First,forthesetwo specimens,onlyasinglecrystallographicorientationisprobed(i.e.the sizeofthegraincontainingtherigidinterfacesismuchlargerthanthe volumeaffectedbytheindent).Indentationstudiesonsinglecrystalsin- dicatethattheorientationofthefacetsoftheindentertipwithrespect totheavailableslipsystemsdeterminestheanisotropyofthepile-up

(9)

Fig.9. (a)PlotoflnH(hardness)asafunctionoftheinversetemperature,withslopeofthelinearfitsintherespectivetemperatureintervalsdisplayed.(b)Calculated activationenthalpies,ΔHactasafunctionoftesttemperatureforthethreeinterfacetypes.

height[69–71].Thisanisotropymightbeenhancedifadditionallyin- terfacesrestrictingplasticflowareintroduced[71],especiallyifthey areperfectly alignedandhardlymobileastheLAGBsandTBsinves- tigatedhere(i.e.paralleltothesheardirectionorperpendiculartothe growthdirection,respectively).However,asthepronouncedanisotropy diminishessomewhatatelevatedtestingtemperatures,theeffectofthe interfaceseemsmoreimportant.

4.2. Pronouncedtemperaturedependenceofmechanicalpropertiesincase ofHAGBs

Besidethedifferentappearanceoftheindentshape,thepresented resultsalsoreveal,dependingontheinterfacetype,adistinctivelydif- ferentthermomechanicalresponse,cf.Figs.6and8.Alreadythehard- nessvaluesmeasuredatRTindicatethatnotsolelytheinterfacialspac- ing,butalsotheinterfacetypeplaysapredominantroleindetermining themechanicalresponse.AlthoughtheaveragespacingoftheTBswas smallerthanthatoftheHAGBs,theRThardnessoftheHAGB speci- menswasthehighest,seeFig.6a.Basedonpreviousresearch,thiscan beexplainedby(partial)collapseoftheNTstructure.Straininduced detwinningwasclearlyobservedbeneathnanoindentsperformedonNT copper,withdetwinningbecomingmorepronouncedforhigherequiva- lentstrainsapplied[67].Further,forNTstructuresahugeanisotropyof theflowstresscanbemeasured,differingbyafactorofthreebetween thehardandsofttestingmodesforNTcopper[72].Althoughlesspro- nounced,suchanisotropicflowbehaviorwasalsoreportedformetallic nanolaminates[73,74].Beneaththeindentsbendingorrealignmentof thelayersand/orshearbandformationwasfrequentlyobserved[75–

77].Alignmentofthelayerswithrespecttothefacesoftheindentercor- respondstotheweakesttestingdirection,thusfacilitatesplasticflow.

Convertedhardnessvalues(usingaconversionfactorofthree)yielding resultsclosetotheuniaxialflowstressoftheweakesttestingorientation, furthersupportthispicture.Hence,theslightlylowerhardnessvalues measuredincaseoftheNTnickelaremostlikelyaconsequenceofsuch instabilitiesoccurringduringindentation.

Withincreasingtestingtemperature,thedifferencesbetweenthein- dividualsampletypesbecomeevenmorepronounced.WhiletheLAGB and NTsamples behave quite similar andtheir properties are only weaklytemperature-dependent,thesamepropertiesvarystronglywith temperatureincaseoftheHAGBsamples.Ascanbenoticedfromthe

Ni4.5AlHAGBsample,thispronouncedtemperaturedependencecan- notbeavoidedbyalloying,butjustthetemperatureswherepronounced changesoccurareshiftedtowardshighervalueswithalloyingcontent, cf.Figs.6aand8.Thisemphasizesthatalreadyatcomparablylowtem- peraturesof~ 0.2Tm,mechanicalpropertiesarenotsolelydetermined bytheaverageinterfacialspacing,buttheinterfacetypeplaysadeci- siverole.Thisstrongtemperaturedependenceofhardnessincaseofthe HAGBsamplesisunexpectedforFCCmetalsthatconventionallyhave atemperature-independent yieldstress[78–80].Althoughduetothe temperaturedependenceoftheworkhardeningratetheflowstresses determinedatlargerstrainscandiffersubstantiallyalsoforFCCmetals [80],thisisnotexpectedforseverelydeformedmaterialshavingalready anegligiblework-hardeningcapacity.Moreover,consideringthelimited equivalentstrainsrealizedwiththeBerkovichtip(i.e.~ 7%),adecrease ofhardnessbymorethan30%whilemarginallyincreasingthetesting temperaturefrom0.17Tmto0.23Tmwouldnotevenbeexpectedfora FCCmaterialwithapronouncedwork-hardeningrate[80].

Thepronouncedtemperaturedependenceofhardnessisalsoevident forthealloyedNi4.5AlHAGBsamples(Fig.6),inperfectagreement withliteraturefindingson variousnanostructuredmaterials,butalso fornanocompositesornanolaminates[28,31,81–85]. Thesameholds trueformetallic thinfilms,usuallycomposedofNCgrainstructures.

Earlynanoindentationstudiesalreadyreportedareductionofhardness by morethan50%whenthetesting temperaturefor goldorcopper thin filmswas increasedfromRTto400K,althoughthemicrostruc- tureswereclaimedtobethermallystableuptothispoint[86],similar towhatwasfoundinthepresentstudy.Contrary,thehardnessofthe LAGBorTBsamplesremainsalmostunaffected(i.e.reductionby8%) inthesametestingtemperatureinterval.Thissuggeststhatthedistinct temperaturedependenceoftheflowstresshasitsorigininthepresence of large fractionsofrandomHAGBs orinterfaces.Astrongtempera- ture dependencein caseof theHAGB samplesis,however,alsoevi- dentfortheotherpropertiesdetermined.SRSoftheHAGBspecimens showsindependentofpurityapronouncedincreasewithtemperature, accompaniedbyadropintheapparentactivationvolume,V,aslongas themicrostructureremainsstable,Fig.8.Incontrast,theSRSvaluesof theLAGBandNTspecimensarealmostathermal(Fig.8a),causingto- getherwiththealmostconstanthardnessslightlyincreasingactivation volumeswithincreasingtemperature.FocusingontheHAGBspecimens, itappearsthatpronouncedpropertychangesofH,mandVwithtem-

(10)

peratureoccuratsimilar testingtemperatures,suggesting acommon origin,cf.Figs.6and8.Fromthedeformationtextures(Fig.2)ofthe LAGBandHAGBsamplesanddetailedresultsonNTmetals(e.g.Refs.

[72,87,88]),itisevidentthatforallsamplesdislocation-basedplasticity prevails.Hence,asthedeformationmechanismremainsthesameforall threetypesofsamples,thedifferencesinthemechanicalbehaviorneed tooriginatefromachangeoftheinteractionofdislocationswiththe respectiveinterfaces.Thestrongtemperaturedependenceofproperties incaseofHAGBsseemstobeapeculiarityoftheinteractionofdislo- cationswiththem,occurringonlyatmuchhighertemperaturesincase ofinterfacesconsistingofdislocationarrangements(LAGBs)oralmost atomicallysharpinterfacessuchasTBs.

Interactionsofdislocationswithdifferentinterfaceshavebeenstud- ied previously, andgiven thenecessary temperatureand/or applied stress,latticedislocationscanenter aGB,forminganextrinsicgrain boundarydislocation(EGBD),i.e.adislocationbeingnotpartoftheGB structurebutcomingfromoutsidetheGB[26].Variousprocessescan becoupledtotheabsorptionofanEGBD,suchasdislocationemission, GBmigrationorgrainrotation,justtonameafew[26].Althoughthe actualprocessescandifferdependingonboundarycrystallographyand theBurgersvectoroftheincomingdislocation[26,87,89,90],thisaspect isnotconsideredintheremainderbecauseindividualinteractionevents arenotaccessiblewiththeexperimentsusedinthiswork.Nevertheless, probingmultipleboundariesofagiventypeatthesametimeallowsto drawconclusionsabouttheiraveragebehaviorandrespectivedistinc- tions.Inanycase,aftertheinjectionorabsorptionofalatticedisloca- tion,theGBwillbeinastateofhigherenergy.Eveninthecaseofdirect transmission,aresidualBurgersvectorcouldbeleftattheinterface,if thoseofthein-andoutbounddislocationsdiffer.Consequently,thein- terfaceaimsatreducingitsenergyagainbyintergranularstressrelease, involvingthermallyactivatedprocesses.Theseprocessesdeterminethe behaviorofpolycrystalstoagreatextent,asdislocationemission,GB migrationorgrainrotationarecloselyrelatedtothem[26].Indeed,our recentstudyfocusingonUFG/NCFCCmetalsreportedgoodagreement betweentheaveragetemperaturesforthermally-induceddislocationan- nihilationinrandomHAGBswiththosewhereSRSincreasedtoremark- ablyhighvalues[9].Thesameholdstrueforthenickelsamplestested here.AverageannihilationtemperaturesfordislocationsinHAGBsof purenickelof about490 Kwerereported [24], agreeing reasonably wellwiththetemperatureswheremaximumratesensitivityandinitia- tionofgraingrowthoccurred,compareFig.8.Itshouldbenotedthat theincreasingtemperaturedependenceofpropertiesintheHAGBsam- plesalreadybelow490Kisnotcontradictingthisidea.Astheaccom- modationofdislocationsatGBsandtheassociatedrelaxationoftheir stressfieldsinvolvethermallyactivatedprocesses,thethermalenergy requiredcanbereducedbythepresenceofmechanicalstresses,com- pareEq.(5).There,GdenotestheGibbsfreeenergy,FtheHelmholtz energy,𝜏theresolvedshearstressandVthe(physicalorreal)activation volumeoftheprocess[91].

𝐺=𝐹𝜏𝑉 (5)

Accordingly,partoftheenergycanbeprovidedbythemechanical workdonebytheappliedshearstress,whilemodelsdescribingtheki- neticsofintergranularstressrelaxation(i.e.Eq.(1))orthementioned temperaturesfordislocationannihilation [24]justconsiderthepure thermallydrivencase.Hence,theappliedstressesduringindentation cansignificantlyreducetherequiredthermalcontributionstoovercome theactivationbarrier.Thiswasindeedobservedinin-situTEMstraining experimentsonUFGaluminum,wherethedislocationcontrastatthe GBsdecreasedmorerapidlywhensmalldislocationpile-ups,i.e.locally higherresolvedshearstresses,werepresent[92].Thisisinlinewiththe developmentofapronouncedtemperaturedependencealreadybelow theaverageannihilationtemperaturesreported.

However,despiteconsideringthethermalpartonly,thementioned modelsdescribingthekineticsofinterfacialstressrelaxationcanstill beusedtoqualitativelyrationalizetheobviousdifferencesbetweenthe

HAGBandtheLAGB/NTsamples,astheappliedstressisroughlythe sameforallthreeinterfacetypes.Asstatedpreviously,therequiredre- laxationtimedependsmainlyonmaterialproperties(i.e.shearmodulus, atomicvolume,interfacialspacingandwidth)andinterfacialdiffusivity [25,26],cf.Eq.(1).Consideringthatthenickelsamplestestedherehave aquitesimilarboundaryspacingandassumingaratherconstantbound- arywidth,theonlyparameterthatfundamentallydiffersistheinterfa- cialdiffusivity.FortheLAGBandNTsamples,theactivationenergies forboundarydiffusionareexpectedtobesimilartobulkdiffusion,i.e.

adifferencebyafactorofabouttwocomparedtotheHAGBspecimens [93].Availableinterfacialdiffusiondatafornickelalongdeformation- inducedLAGBs[22]andHAGBs[27]attemperaturesclosetothehigh- estonestestedhereyieldvaluesofDLAGB(458K)=5.14×1019m2 s1 andDHAGB(440K)=5.03×1017m2s1,respectively.Forcomparison, thebulkdiffusivityatthistemperaturewouldbeonlyontheorderof DV(450K)=5×1037m2s1[65].Withallotherparametersbeingap- proximatelythesameforthenickelsamplesstudiedhere(Eq.(1)),the expectedrelaxationtimesdifferbyatleasttwoordersofmagnitudebe- tweenthesetwosampletypes.Theaccordinglyreducedrelaxationtime- scalesincaseoftheHAGBspecimensarereflectedintheirpronounced temperaturedependenceofmechanicalproperties,whilethelowdif- fusive,‘dense’TBsorLAGBsrequirehighertemperaturestothermally activatesuchprocesses.Thisisreflectedintheiralmosttemperature- independentpropertiesintheinvestigatedtemperaturerange,butalso in activationenthalpiesbeingcomparabletobulkdiffusion,compare Fig.9bandRef.[65].Thesimilaractivationenthalpiescalculatedfor theTBsandLAGBsissupportedbysimulationwork,emphasizingthat both,TBsandLAGBs,canprovideeasysliptransferacrossthebound- ary,butcanalsoresistdislocationslip.Exceptforthecaseofascrew dislocationimpingingaTB,oradislocationwithproperBurgersvector impingingattherightpositionoftheLAGB,locksorjunctionsareeas- ilycreated.Thesesessileresidualsoftheinteractionprocessmayonly beremovedorovercomefromsubsequentdislocationsbyclimb,neces- sitatinginbothcasesascenarioclosetobulkself-diffusion[89,94,95].

Similarly,althoughnotaspronounced,thereduceddiffusivityofHAGBs causedbyalloyingorimpuritiesinducesashifttowardshighertesting temperaturesuntilrelaxationprocessescanbethermallyactivated.The reductionofVtowardssingledigitvalues,foundfortheHAGBspeci- mensonceapronouncedtemperaturedependenceofpropertiesismea- sured(Fig.8b),andtheactivationenthalpiesbeingclosetoexpectations forGBdiffusion(compareFig.9bandRef.[66]),furtherhighlightsthe importanceofdiffusiveprocessesattheGBsatthesetestingtempera- tures.

Thermally-inducedstressrelaxationattheboundariesthusnotonly explainstheobservedpropertydifferencesbetweenthethreeinterface types,butalsoallowstorationalizethepronouncedtemperaturedepen- dentpropertiesincaseoftheHAGBsample.IncaseofHAGBs,relaxation canoccuratmuchsmallertimescales,thusenhancedtestingtempera- turesorprolongedtestingtimes(i.e.reductionofthestrainrate)already allowfor diminishingstressfields.Their interactionwithsubsequent latticedislocationsishencereducedandlowerflowstressesaremea- sured.ContrarytheretardedrelaxationatLAGBsorTBs,whichoccurs bydislocationclimb[96],explainsnotonlythereportedhigherwork- hardeningratesinthiscase[14–18],butalsotherathertemperature- invariantpropertiesmeasuredinFigs.6and8.Thisinterpretationisin linewithrecentinvestigationsonNTcoppersuggestingthattherate limitingprocessisdislocationclimbalongtheTBs,causingalsoanin- creaseoftheratesensitivityatsufficientlyhightemperatures[88].Since stressrelaxationatGBscanalsoresultintheirmigration,theobserved graingrowthbeneaththeindentoftheHAGBsampleswell-belowthe thermalstabilitylimitmaynotbetoosurprising(Fig.7).Althoughfre- quentlyobservedfornanostructuressubjectedtovariousloadingsitu- ations(e.g.Refs.[36,53–55,97]),initiationofdistinctgraingrowthby applyingjust7%additionalstrainmayappearintriguingatfirstview foraseverely deformedsample. However,theequilibriumgrain size wasadjustedatRT,whileHPTdeformationat418Kalreadyyieldsan

(11)

increaseoftheminimumgraindimensionsby60%[98].Inadditionto theincreaseofthetestingtemperature,thechangeoftheloadingsit- uationitselfcouldacceleratethemigrationrates[99].Disconnections, GBlinedefectsresponsiblefortheirmigration[100–102],canalsobe generatedbyimpinginglatticedislocations[100,103].Thestepheight andBurgersvectorofthedisconnectionwill,however,notonlydepend onboundarycrystallography,butalsoontheinteracting dislocation.

Hence,astrainpathchange,activatingdifferentslipsystems,couldal- tertheactivedisconnectionmodesandaccordinglytheGBmigration rates.Thethermalandmechanicalstabilityofananomaterialcanthus largelydiffer,compareFigs.3and6.

5. Summaryandconclusions

Thepresentresultsrevealapronouncedinfluenceoftheinterface type on the deformation behaviorof nanostructured nickel. Despite havingasimilarinterfacialspacing,thepresenceoftwinorlow-angle boundariesresultsinanalmostathermalbehaviorofhardness,ratesen- sitivityandactivationvolume.Incontrast,thepropertiesofHAGBsbe- comestronglytemperature-sensitivealreadyatlow homologoustem- peratures(≥0.2Tm),whilealloyingonlyinducesashifttowardshigher onsettemperatures.Thesedifferencescanberationalizedbasedonthe importantroleofinterfacialstressrelaxationforthedeformationbe- haviorofnanomaterialsanditsdependenceonboundarydiffusivity.As observedfortheHAGBspecimens,interfacialstressreleasecandistinc- tivelymodifythenanostructurestested.Noticeablegraingrowthbelow thethermalstabilitylimitisreflectedinapronouncedtemperaturede- pendenceofhardness.Theseobviousdifferencesinducedbytheinter- facetypecannotbecapturedinpredictionsusingmodelssolelybased ontheinterfacialspacing(i.e.modified Hall-Petchconcepts).Hence, conceptsincludingkineticsofaccommodationandstressrelaxationpro- cessesofdislocationsatinterfacesneedtobetargetedanddevelopedto ensuremorereliablepropertydescriptions.Whiletheirkineticsarerea- sonablydescribedforthepurethermallyactivatedcase,itsstressdepen- denceyetneedstobeassessed.However,withfutureworkinthisfield, amajorsteptowardsageneralizedmodelcapturingeffectsofchemistry, interfacetypeorloadingsituationand,hence,ageneralpredictionof mechanicalpropertiesofnanomaterials,seemsaccessible.

DeclarationofCompetingInterest

Theauthorsdeclarethattheyhavenoknowncompetingfinancial interestsorpersonalrelationshipsthatcouldhaveappearedtoinfluence theworkreportedinthispaper.

Acknowledgments

Financial support by the European Research Council under ERC GrantAgreementsNo.340185USMSandNo.771146TOUGHIT,the AustrianAcademyofSciencesviatheInnovationFundIF2019-37and underthescope of theCOMETprogram withinthe K2Center‘Inte- gratedComputational Material,ProcessandProductEngineering(IC- MPPE)’(ProjectNo859480)supportedbytheAustrianFederalMin- istriesforClimateAction,Environment,Mobility,InnovationandTech- nology(BMK) andfor DigitalandEconomicAffairs(BMDW),repre- sentedbytheAustrianresearchfundingassociation(FFG),andthefed- eralstatesofStyria,UpperAustriaandTyrolisgratefullyacknowledged.

References

[1] M.A. Meyers, A. Mishra, D.J. Benson, Mechanical properties of nanocrystalline ma- terials, Prog. Mater. Sci. 51 (2006) 427–556, doi: 10.1016/j.pmatsci.2005.08.003 . [2] M. Dao, L. Lu, R.J. Asaro, J.T.M. De Hosson, E. Ma, Toward a quantitative under- standing of mechanical behavior of nanocrystalline metals, Acta Mater. 55 (2007) 4041–4065, doi: 10.1016/j.actamat.2007.01.038 .

[3] Y. Li, D. Raabe, M. Herbig, P.P. Choi, S. Goto, A. Kostka, H. Yarita, C. Borchers, R. Kirchheim, Segregation stabilizes nanocrystalline bulk steel with near the- oretical strength, Phys. Rev. Lett. 113 (2014) 106104, doi: 10.1103/Phys- RevLett.113.106104 .

[4] J. Wang, Q. Zhou, S. Shao, A. Misra, Strength and plasticity of nanolaminated ma- terials, Mater. Res. Lett. 5 (2017) 1–19, doi: 10.1080/21663831.2016.1225321 . [5] H. Van Swygenhoven, P.M. Derlet, A.G. Frøseth, Nucleation and propagation of

dislocations in nanocrystalline FCC metals, Acta Mater. 54 (2006) 1975–1983, doi: 10.1016/j.actamat.2005.12.026 .

[6] Q. Wei, S. Cheng, K.T. Ramesh, E. Ma, Effect of nanocrystalline and ultrafine grain sizes on the strain rate sensitivity and activation volume: FCC versus bcc metals, Mater. Sci. Eng. A. 381 (2004) 71–79, doi: 10.1016/j.msea.2004.03.064 . [7] H.W. Höppel, J. May, P. Eisenlohr, M. Göken, Strain-rate sensitivity of ultrafine-

grained materials, Zeitschrift Für Mater. Res. Adv. Technol. 96 (2005) 566–571, doi: 10.1016/j.scriptamat.2005.03.043 .

[8] H. Vehoff, D. Lemaire, K. Schüler, T. Waschkies, B. Yang, The effect of grain size on strain rate sensitivity and activation volume - from nano to ufg nickel, Int. J.

Mater. Res. 98 (2007) 259–268, doi: 10.3139/146.101464 .

[9] O. Renk, V. Maier-Kiener, I. Issa, J.H. Li, D. Kiener, R. Pippan, Anneal hardening and elevated temperature strain rate sensitivity of nanostructured metals: their relation to intergranular dislocation accommodation, Acta Mater. 165 (2019) 409–

419, doi: 10.1016/j.actamat.2018.12.002 .

[10] L. Lu, Y. Shen, X. Chen, L. Qian, K. Lu, Ultrahigh strength and high electrical con- ductivity in copper, Science 304 (2004) 422–426, doi: 10.1126/science.1092905 . [11] K. Lu, L. Lu, S. Suresh, Strengthening materials by engineering coher-

ent internal boundaries at the nanoscale, Science. 324 (2009) 349–352.

doi:10.1126/science.1159610.

[12] Y.M. Wang, K. Wang, D. Pan, K. Lu, K.J. Hemker, E. Ma, Microsample tensile testing of nanocrystalline copper, Scr. Mater. 48 (2003) 1581–1586, doi: 10.1016/S1359-6462(03)00159-3 .

[13] S. Cheng, E. Ma, Y.M. Wang, L.J. Kecskes, K.M. Youssef, C.C. Koch, U.P. Trociewitz, K. Han, Tensile properties of in situ consolidated nanocrystalline Cu, Acta Mater.

53 (2005) 1521–1533, doi: 10.1016/j.actamat.2004.12.005 .

[14] L. Lu, M. Dao, T. Zhu, J. Li, Size dependence of rate-controlling deforma- tion mechanisms in nanotwinned copper, Scr. Mater. 60 (2009) 1062–1066, doi: 10.1016/j.scriptamat.2008.12.039 .

[15] Z.X. Wu, Y.W. Zhang, D.J. Srolovitz, Dislocation-twin interaction mechanisms for ultrahigh strength and ductility in nanotwinned metals, Acta Mater. 57 (2009) 4508–4518, doi: 10.1016/j.actamat.2009.06.015 .

[16] M.J. Caturla, T.G. Nieh, J.S. Stolken, Differences in deformation processes in nanocrystalline nickel with low- and high-angle boundaries from atomistic sim- ulations, Appl. Phys. Lett. 84 (2004) 598–600, doi: 10.1063/1.1640464 . [17] P.L. Sun, C.Y. Yu, P.W. Kao, C.P. Chang, Influence of boundary characters on the

tensile behavior of sub-micron-grained aluminum, Scr. Mater. 52 (2005) 265–269, doi: 10.1016/j.scriptamat.2004.10.022 .

[18] P.L. Sun, E.K. Cerreta, J.F. Bingert, G.T. Gray, M.F. Hundley, Enhanced tensile ductility through boundary structure engineering in ultrafine-grained aluminum, Mater. Sci. Eng. A. 464 (2007) 343–350, doi: 10.1016/j.msea.2007.02.007 . [19] X.C. Liu, H.W. Zhang, K. Lu, Strain-induced ultrahard and ultrastable nanolam-

inated structure in nickel, Science 342 (2013) 337–340, doi: 10.1126/sci- ence.1242578 .

[20] Y. Zhao, T.A. Furnish, M.E. Kassner, A.M. Hodge, Thermal stability of highly nan- otwinned copper: the role of grain boundaries and texture, J. Mater. Res. 27 (2012) 3049–3057, doi: 10.1557/jmr.2012.376 .

[21] X. Zhang, A. Misra, Superior thermal stability of coherent twin boundaries in nanotwinned metals, Scr. Mater. 66 (2012) 860–865, doi: 10.1016/j.scriptamat.2012.01.026 .

[22] Z.B. Wang, S.V. Divinski, Z.P. Luo, Y. Buranova, G. Wilde, K. Lu, Revealing interfa- cial diffusion kinetics in ultra-fine-laminated Ni with low-angle grain boundaries, Mater. Res. Lett. 5 (2017) 577–583, doi: 10.1080/21663831.2017.1368036 . [23] M. Dao, L. Lu, Y.F. Shen, S. Suresh, Strength, strain-rate sensitivity and duc-

tility of copper with nanoscale twins, Acta Mater. 54 (2006) 5421–5432, doi: 10.1016/j.actamat.2006.06.062 .

[24] P.H. Pumphrey, H. Gleiter, The annealing of dislocations in high-angle grain bound- aries, Philos. Mag. 30 (1974) 593–602, doi: 10.1080/14786439808206584 . [25] A.A. Nazarov, Kinetics of grain boundary recovery in deformed polycrystals, Inter-

face Sci. 8 (2000) 315–322, doi: 10.1023/A:1008720710330 .

[26] L. Priester, Grain Boundaries From Theory To Engineering, 1st ed, Springer, Nether- lands, 2013, doi: 10.1007/978-94-007-4969-6 .

[27] S.V. Divinski, G. Reglitz, H. Rösner, Y. Estrin, G. Wilde, Ultra-fast diffusion channels in pure Ni severely deformed by equal-channel angular pressing, Acta Mater. 59 (2011) 1974–1985, doi: 10.1016/j.actamat.2010.11.063 .

[28] Y.M. Wang, A.V. Hamza, E. Ma, Temperature-dependent strain rate sensitivity and activation volume of nanocrystalline Ni, Acta Mater. 54 (2006) 2715–2726, doi: 10.1016/j.actamat.2006.02.013 .

[29] J.R. Trelewicz, C.A. Schuh, Hot nanoindentation of nanocrystalline Ni-W alloys, Scr. Mater. 61 (2009) 1056–1059, doi: 10.1016/j.scriptamat.2009.08.026 . [30] V. Maier, K. Durst, J. Mueller, B. Backes, H.W. Höppel, M. Göken, Nanoinden-

tation strain-rate jump tests for determining the local strain-rate sensitivity in nanocrystalline Ni and ultrafine-grained Al, J. Mater. Res. 26 (2011) 1421–1430, doi: 10.1557/jmr.2011.156 .

[31] G. Mohanty, J.M. Wheeler, R. Raghavan, J. Wehrs, M. Hasegawa, S. Mis- chler, L. Philippe, J. Michler, Elevated temperature, strain rate jump micro- compression of nanocrystalline nickel, Philos. Mag. 95 (2014) 1878–1895, doi: 10.1080/14786435.2014.951709 .

[32] A. Vorhauer, R. Pippan, On the onset of a steady state in body-centered cubic iron during severe plastic deformation at low homologous tempera- tures, Metall. Mater. Trans. A Phys. Metall. Mater. Sci. 39 (2008) 417–429, doi: 10.1007/s11661-007-9413-1 .

[33] Z.C. Wang, P.B. Prangnell, Microstructure refinement and mechanical properties

Referenzen

ÄHNLICHE DOKUMENTE

However, few studies actually examine how the new rules of the international system impact developing countries’ choice of industrial policies, and those that do fail to consider

Gordon (1986) tests the Keynesians and monetarists argument and reaches the interesting results. He states that members of these groups will be disappointed with his

(To put the point in another way, the central bank should maintain a discount rate that was no higher, and normally equal to, the market rate of interest.) “Yet while the ultimate

Wavelet Coherence (WTC) is applied as to provide an overall multi-scale analysis on the relationship of the variables, and further analyzed using Maximal Overlap

In this study, it was concluded that reforms to the economic and legal framework of the country and concluding treaties with other countries cannot, on their

In any case, the overall concentration dependence of excimer formation is by far too high to be explained by a static distribution model and indicates that the presented

available elevation data sets are considered: ASTER GDEM2, SRTM at 3 arc-second and 1 arc-second resolution as well as a DEM derived from digitised contour lines of the

To simu- late the ship–bank interaction, the computational domain requires a 33 × 2.3 L pp (length × width) in the present study to obtain the quasi-steady result in both deep