ContentslistsavailableatScienceDirect
International Journal of Mechanical Sciences
journalhomepage:www.elsevier.com/locate/ijmecsci
Effect of nodal mass on macroscopic mechanical properties of nanoporous metals
J. Jiao
a,∗, N. Huber
a,baInstitute of Materials Research, Materials Mechanics, Helmholtz-Zentrum Geesthacht, Germany
bInstitute of Materials Physics and Technology, Hamburg University of Technology, Germany
a r t i c le i n f o
Keywords:
Nanoporous metal Finite element method Nodal mass Macroscopic response Scaling laws Size dependent strength
a b s t r a ct
Thecurrentworkinvestigatestheeffectofthenodalmassonthemacroscopicmechanicalbehaviorofnanoporous metalsusingtheFiniteElementMethod.Anodalcorrectedbeammodelingconceptisintroducedthatallows localincorporationoftheeffectiveelastoplasticmechanicalbehaviorofthenodalmassinthenodalareaofa representativevolumeelement(RVE).ThecalibrationtothecorrespondingFiniteElementsolidmodelisachieved byintegratingadditionalgeometryandmaterialparameterstotheso-callednodalareasinthebeammodel.With thistechniqueanexcellentpredictioncanbeachievedoveralargerangeofdeformationfordifferenttypes ofRVEs.Fromtheresultsofthenodalcorrectedbeammodel,modifiedleadingconstantsaredeterminedin thescalinglawsforYoung’smodulusandyieldstrength.Theeffectofthenodalcorrectionisalsostudiedwith respecttovariousrandomizationlevels.Finally,theligamentsizedependentstrengthisanalyzedbyapplyingthe proposedmodeltoexperimentaldata.Itcouldbeshownthatthenodalcorrectionimprovestheoverallagreement withliteraturedata,particularlyforsuchdatapointsthatarerelatedtosampleswithahighsolidfraction.
© 2017TheAuthors.PublishedbyElsevierLtd.
ThisisanopenaccessarticleundertheCCBY-NC-NDlicense.
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
1. Introduction
Nanoporousgold(NPG)madeby de-alloyingcan beproducedas macroscopicobjectsthatexhibitabi-continuousnetworkofnanoscale poresandsolid‘ligaments’ whichareconnectedinnodes.Thesolidfrac- tion𝜑oftheporousbodyisapprox.30%[1–4].𝜑isusedasthemajor parameterinseveraltheoreticalmodelsforpredictingthemacroscopic mechanicalbehavioroftheporousmaterials[5–9].TheGibson-Ashby model[10], asthemostcommonlyusedone amongthese modelsis reportedtosignificantlyoverestimatethemacroscopicmechanicalre- sponseofnanoporousmetals[11–14].Theoverestimationindicatesthat themassutilizationfordeformationinsuchamaterialisnotasefficient asassumedbytheGibson-Ashbystructuralmodelforopenporefoams.
Inawider spectrumof attemptsforunderstanding theextraordi- narymechanicalresponsesofnanoporousmetals,extensivemodeling approacheshavebeenconducted. Atomisticandmoleculardynamics simulationshavebeenimplementedtoinvestigatethedeformationbe- haviorundertensionandcompression[6,15–17].Itwasfoundthatthe surfacestresshasasubstantialimpactonthetension/compressionasym- metry,anomalouscomplianceandearlyyield.Thesimulationsreported in[16]revealedsignificantstackingfaultformationanddislocationac- cumulationwithinthenanosizedligaments,confirmingtheexistence
∗Corresponding author.
E-mail address: jingsi.jiao@hzg.de (J. Jiao).
ofsubstantialworkhardeningunderplasticdeformationassuggested in[8].Further modelingworkonthemicrostructuralandcontinuum level[18,19]wasconductedtoexploretheorigin oftheunusuallow Poisson’sratioobservedduringmacroscopiccompressionofnanoporous goldsamples.ItwasfoundthatontheonehandtheelasticPossion’sra- tio isindependentoftheligamentsizebutdecreaseswithincreasing degreeofrandomizationthroughanincreasingpercentageoftorsionof theligaments[18].Ontheotherhand,theplasticPoisson’sratioshowed astrongdependencyontheligamentsize,whichcouldbesuccessfully reproducedwiththeDeshpande–Fleckmodel[19].
Otherworks[20–23]thatstudiedsurfaceelasticityorsurfacebound- aryconditionshavebeenconductedaimingtoexplainthemacroscopic mechanicalbehaviorsofthismaterialfromamicroscopicpointofview, withaparticularemphasisonsizeeffects.Veryrecently,thesignature ofthesurfaceenergywasstudiedbycombiningmacroscopiccompres- sionexperimentswithaFiniteElementbeammodelofarandomized diamondstructure,enriched bycoaxialthin-walledtubularelements for modelinga switchablesurfacestress[24].Theresults showedin conjunctionwiththeexperimentalfindingsthat,contrarytotheelastic Poisson’sratio,theplasticPoisson’sratiorespondsstronglytoelectri- calsurfacemodulation.Thisbehaviorwasidentifiedasthesignature ofasurface-inducedtension-compressionasymmetryoftheflowstress
https://doi.org/10.1016/j.ijmecsci.2017.10.011
Received 23 June 2017; Received in revised form 20 September 2017; Accepted 7 October 2017 Available online 9 October 2017
0020-7403/© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )
sizeofthenodesconnectingtheligamentsisintroducedintheserefined models.Theextramassofthenodesisinvestigatedregardingthecal- culationofthesolidfraction𝜑andthemechanicalresponseoftheunit cell.Itisconcludedthatthenodalmassshouldbecountedasanother importantfactorthatisassociatedwiththerelationbetweenaspectratio oftheligaments,solidfractionandmacroscopicmechanicalproperties.
Morespecifically,itwasindicatedthatthediscrepancybetweentheex- perimentalmeasurementsandtheGibson-Ashbymodelpredictionisdue tothatthereisamassivemassaccumulatedatthenodes,whichdoes notcontributetotheelasticdeformationoftheunitcell[9,12,25].
Theeffectfromthenodesontheelastic-plasticdeformationwasin- vestigatedbycomparingresultsfromfiniteelementsimulationsfora finiteelementsolidmodelwithsphericalnodesandafiniteelement beammodel,thatreducesthenodestoacouplingconstraintofthecon- nectedbeamelements,basedonthediamondunitcell[8].Itwasfound thattheelasticstiffeningcausedbythenodesispartiallycompensated bytheeffectofincreasingsheardeformationforthickligaments.Con- cerningtheplasticdeformationbehavior,theshorteningoftheligament lengthduetothenodalmasswasidentifiedasanimportanteffectthat wasincorporatedasacorrectionfactorintheproposedscalinglaws.
Thus,forthescalinglawofthemacroscopicmechanicalbehavior,the structuralarrangementoftheligamentsisnottheonlyinfluentialpa- rameter;thedifferentapproachesformodelingthenodesintermsof theircontributiontothesolidfractionaswellaselasticandplasticde- formationbehaviorwillalsoleadtodifferentscalinglawscomparedto theGibson–Ashbymodel.
Thestudiesonnanoporousmetalscannotberestrictedtoasimpli- fied,geometricallyperfectunitcellstructure,duetothattherealstruc- tureofnanoporousmetalsisaspatialnetworkstructurewithcomplex topologicalandmorphologicalcharacteristics[7,13,26,27].Inorderto investigatethestructuralparametersandobtainamorerealisticdefor- mationoftheligaments,highlygeneralizedbeamelementsareused, whichofferexcellentcomputationalefficiencyevenwiththousandsof ligamentsmodeledinarepresentativevolumeelement(RVE).Basedon thediamondlatticestructurethatwasestablishedintheworkof[8]and furtherrefinedby[29],theeffectsofligamentshapevariationandran- domizationonthevariousdeformationmodesofbending,torsion,and tension/compressionwasinvestigated[18].Itcouldbeshown,thatthe beamelementB31inABAQUS[28]iscapableofcapturingallfunda- mentaldeformationsforeventhick beamswithaligamentradiusto lengthratio,r/l,upto0.5correspondingtoasolidfractionof69%.So faroneremainingdrawbackoftheRVEbuiltfrombeamelementsisthe unsolvedquestionofhowtomodelthemassinthenodalareas,because theligamentsinthemodelareconjugatedatvirtualnodes.Thisleadsto alargercomplianceandlowerstrengthofthebeammodel.Inthestudy of[29],acorrectionofthemacroscopicyieldstrengthwasconsidered byincreasingtheyieldstrengthofthesolidphasebyafactorthatwas derivedfromtheavailableleverforbendingoftheligamentwhichis shortenedbythenodalmass,basedonthecorrectionderivedin[8]. Itishoweverunclear,howaccurateandhowgeneralthisapproachis, particularlywhentheRVEhasincreasingcomplexity throughadding randomizationorvariationsofligamentshape.
lyzedandcomparedtopreviousstudies.Gibson-Ashbyscalinglawsfor elasticityandplasticitywithmodifiedleadingconstantsareproposed thatincorporatetheeffectofthenodalmassonthemacroscopicme- chanicalresponse.Theroleofthenodalmassonthemacroscopicme- chanicalresponseisalsoanalyzedwithrespecttovariousrandomization levels.Finally,theligamentsizedependentstrengthofnanoporousgold is determinedbyapplyingtheproposedmodeltoexperimentaldata.
Theresultsarediscussedinthelightofpreviousstudies[29]. 2. Extensionofthebeammodel
ThegoalofthissectionistodevelopaRVEbuiltwithbeamelements, wherethemechanicsoftheconnectingnodesisphysicallyandlocally includedineachofthenodesofthebeamstructure.Itshallserveforpre- dictingtheelastic-plasticmacroscopicresponseofananoporousmetal withhighcomputationalefficiency.Thisapproachwouldallowelimi- natingthephenomenologicalcorrectionfactorthathaslimitedaccuracy anddoesnotallowconsideringeffectsoflocalstructuralvariationsin thevicinityofthenodes.
Inwhatfollows,abeamnodalcorrectionapproachwillbeproposed thatisbasedonatetrahedronstructure– thebuildingblockofthedi- amondstructure– thatallowsstudyingthemechanicalresponseona moreadvancedlevelcomparedtoasingleligamentRVE.Thetetrahe- dronstructureservesforadjustingtheelastic-plasticdeformationbe- haviorofthebeammodelinrelationtoasolidmodelofsamegeometry againstbending.Theapproachwillbevalidatedwithrespecttotorsion onthelevelofthetetrahedronstructureand,inasecondstep,forlarger RVEsthatarecomparedtosolidRVEs.
2.1. Tetrahedralbuildingblock
Atetrahedronstructureconsistingoffourligamentsissketchedin Fig.1.Itserves asthefundamentalbuildingblockfortheRVE,pro- posedby[8].Fig.1ashowsthestructuremodeledwithsolidelements.
Thegeometryconsistsof thenodalareain formof asphericalmass connectedwithfourligaments,of whichonlyhalfof theligamentis modeled.Fig.1bdemonstratestheequivalentstructurethatismodeled with beamelements. Theligamentsare connectedbytheir common nodevialinking alldisplacementandrotationaldegreesoffreedom.
Asthenatureofthebeammodel,thenodalareacontainsintersecting volumesoftheligaments.Thus,thesolidfractionhastobecalculated independentofthevolumeoftheelementsinthefiniteelementmodel basedontheinitialgeometryshowninFig.1a.Whilethecalculation ofthesolidfractioncanbeeasilycarriedoutfortheperfectlyordered diamondstructure[8]andalsofortherandomizeddiamondstructure [29], thecorrectmechanicalresponserequiresamodification of the beammodel.
Asbendingisthedominantdeformationmechanisminnanoporous metals,thecalibrationofthemodelparameterswillbecarriedoutfor suchadeformation.Thisisachievedbyapplyingatransversedisplace- mentwattheendofthetopligamentwhilefixingtheotherthreelig- amentendsinspace(showninFig.1aandb).Thefollowingapproach
Fig. 1. Geometry of the tetrahedron building block for (a) solid model and (b) beam model.
Fig. 2. Parameters defining the geometry of the ligament-nodal structure (a) solid model and (b) beam model.
Table 1
Geometry parameters and resulting solid fractions for simulations with variation of r / l and c R.
r / l = 0.19 r / l = 0.25 r / l = 0.31
c R= 1 c R= 1.05 c R= 1.1 c R= 1 c R= 1.05 c R= 1.1 c R= 1 c R= 1.05 c R= 1.1 𝜑 = 0.126 𝜑 = 0.127 𝜑 = 0.129 𝜑 = 0.207 𝜑 = 0.210 𝜑 = 0.214 𝜑 = 0.301 𝜑 = 0.307 𝜑 = 0.313
aimsatadjustinggeometricalandmaterialparametersofthebeamele- mentsinthenodalarea,suchthatthemechanicalresponseofthebeam modelisequivalenttothatofthesolidmodel.
Forintroducing thegeometricalparametersofthemodifiedbeam model,oneligamentwithitsconnectingnodeisschematicallyshownin Fig.2aandb.ForthesolidmodelinFig.2a,theparametersRandln arethenodalradiusandthelengthbetweennodalcenterandtheendof theligamentattachedtothenode,respectively.Inthenodalcorrected beammodel,shown inFig. 2b,𝑟∗𝑛 and𝑙∗𝑛 areadjustablegeometrical parametersdescribingthelargercross-sectionofthebeamelementsin thenodalareaandthelengthbetweenthenodecenterandtheendofthe ligament,respectively.Theligamentradiusrandligamentlengthlthat isdefinedasthedistancebetweenthetwonodecenters,arecommon parametersforthesolidmodelandthebeammodel.Anotherparameter cR isintroduced,whichis aconstantgoverningtheadjustablenodal massbylinkingtheligamentradiusrandnodalradiusRintheformof 𝑅=√
3∕2𝑐𝑅⋅𝑟[8].
AllthecalculationsinthestudyareconductedwiththeFEAcode ABAQUS/Implicit[28].Thegeometryofthesolidandthebeammodel ismeshedwithR3D3andB31elements.InTable1threer/lratiosand cRvaluesarelistedthatresultinninemodelrealizations.Therangeof r/lisselectedbasedontherangefromthemorphologicalstudyforthe ligamentsofNPGintheworkof[9],whichincludesthetypicalrangeof thesolidfractionofNPGfrom0.25to0.3[8,29].AvalueofcR=1leads totheminimumpossible sizeof theconnectingnodewherethefour ligamentstouch,whilecR=1.1isreportedtorepresentamorerealistic estimationofthenodesizeinnanoporousmetals[8].
Toapplythesameloadingconditiontothesolidmodelastothe beammodel,thesolidligamentcrosssectionalsurface(Fig.1a)iscou- pledtoarigidplatewitha‘Coupling’ constraintthathasa‘Continuum distribution’.Theloadingcanbethereforeassignedtothecenterpoint of therigidplateanddistributedtothewholesurface[28],ensuring themaximumlevelofthesimilaritybetweenthesolidandbeammod- elswithrespecttotheloadapplication[18].
Thematerialbehaviorofthesolidfractionisassumedtobeelastic- perfectlyplasticwithaYoung’smodulusESof81GPaandaPoisson’s ratio𝜈of0.42.Theyieldstrength𝜎y,Sis500MPaandplasticdeforma- tionevolveswithoutworkhardening[8,29].
2.2. Calibrationofthenodalbeamelements
Asmentionedintheprevioussection,thecalibrationofthebeam modelisconductedwiththereferencetothemechanicalresponseofthe solidmodelunderbendingdeformation.Fig.3illustratesthevonMises stressdistributionforasolidtetrahedronstructure(r/l=0.25,cR=1.1) underthetransverseloadingdisplacementwaccordingtoFig.1athat leadstoplasticbendingandshearingofthetopligament,butalsoofthe upperpartoftheconnectingnodeanditssurroundingligaments.
Thecorrectionforthemechanicalresponseofthebeamtetrahedron structure iscarried outbyfirstlyadjustingitsadditionalgeometrical parameters𝑟∗𝑛 and𝑙∗𝑛.Tothatend,thestiffnessofthetetrahedron(for bothofthebeamandsolidmodels),k=F/w,iscomputedfromthefirst loadingincrementintheelasticregime,suchthatthestructureisun- dergoingonlysmallelasticdeformation.ThestrengthFisreadasthe reactionforceofthetetrahedronstructureintheplasticregimewhen
Fig. 3. Mises stress distribution of solid model with r / l = 0.25, c R= 1.1; top ligament and upper part of the nodal mass and lower ligaments are plastically deformed.
w/l=0.017.Thisdeformationvalueleadstoamacroscopicdeformation withintheplasticregimeoftheforcedisplacementcurve,whichissuf- ficientlyfarfromtheelastic-plastictransition.ThedataksolidandFsolid, determinedfromthesolidmodel,areusedasreferencesforthecali- brationofthebeammodel’selasticandplasticresponse,represented bykbeamandFbeam,respectively.Fig.4ashowstheratiosofkbeam/ksolid (blackcrosses)andFbeam/Fsolid(redcircles)forthecaseofr/l=0.19and cR=1.1.Theanalyzisisconductedwithsystematicallyvaryingthepa- rameters𝑟∗𝑛and𝑙∗𝑛;anditisdoneforallgeometrieslistedinTable1.The bestfitisfoundwherestiffnessandstrengthresultsaresimultaneously closesttothevalueof1,representedbytheyellowplaneinFig.4a.
ItcanbeseenfromFig.4athatthestiffnesskbeamsmoothlyincreases with𝑟∗𝑛.Incontrasttothat,thestrengthratiofirstincreasesinthesame waywith𝑟∗𝑛butthenarrivesataplateau.Thisisbecauseplasticyielding wouldalwaysinitiateattheendoftheligamentoncethenodalregion issufficientlystrong.Untilthen,thenodalregionalsocontributestothe plasticdeformation,asshowninFig.3.Themagnitudeoftheplateau isrelatedtotheleverlengthoftheligamentthatcanbeobtainedinthe formof𝑙−𝑙∗𝑛.Alargervaluefor𝑙∗𝑛resultsinsmallerleverlengthwith theconsequencefortheyieldingtorequirealargerexternalforce,which againleadstoanincreaseoftheplateauvalueasshowninFig.4a.
valueFbeam/Fsolid ≥1representedbytheyellowplaneinFig.4a.
Thisisdonebytuningthenumberofbeamelementsincludedinthe nodalarea,assignedwiththenodalradius,𝑟∗𝑛,whiletheremaining elementskeeptheradiusoftheligament,r.
b) 𝑟∗𝑛issecondlyadjustedforthebeammodeltoreproducethecorrect strengthofthesolidmodelsuchthatFbeam/Fsolid =1.
c) Thefinalcalibrationof𝐸∗𝑛willresolvetheremainingcalibrationof thestiffnesskbeam/ksolid =1withoutaffectingthestrengththatwas calibratedbefore.
Accordingtothestepslistedabove,forthecaseshowninFig.4a, firstly𝑙∗𝑛∕𝑙𝑛=2.23isdetermined,whichisfollowedbythedetermination of𝑟∗𝑛∕𝑟=1.4.Thefittingiscompletedbyadjusting𝐸𝑛∗to𝐸𝑛∗∕𝐸𝑠=0.75 and, as illustrated in Fig. 4b, a 99% agreement is simultaneously reachedforthestiffness andstrengthratio, markedbytherectangle inFig.4b.Itshouldnotedthatanexactmatchinstepa)isnorrequired neitherdesired,becausewewanttousetheinitialdiscretizationofthe ligamenttoformtheelementsinthenodalareasintherandomizedRVEs structures.Otherapproacheswithvariableelementlengtharealsopos- sible,butwouldneedaremeshingofnodalareaaswellastheremaining ligament.
AllninegeometrieslistedinTable1wereanalyzedfollowingthe stepsintroducedabove.Anoverallfittingaccuracyof98%isachieved simultaneouslyforstiffnessandstrength.TheresultsshowninFig.5a–d arepresentedinformoftheratiosoftheadjustableparametersofthe nodalcorrected beammodeltothecorresponding parametersofthe solidmodel,𝑙∗𝑛∕𝑙𝑛,𝑙∗𝑛∕𝑙,𝑟∗𝑛∕𝑟,and𝐸𝑛∗∕𝐸𝑠respectively.Theseratioscan beinterpretedasindicatorsforthegeometricaldifferenceofthenodal
Fig. 4. Comparison of the mechanical response obtained for the beam and solid model for the case r / l = 0.19 and c R= 1.1. (a) Scan with variation of geometrical parameters 𝑟 ∗𝑛and 𝑙 ∗𝑛; best agreement is achieved along the curve 𝑙 ∗𝑛∕ 𝑙 𝑛= 2.23 at 𝑟 ∗𝑛∕ 𝑟 = 1.4; (b) Final fitting with 99% accuracy of stiffness and strength by additionally adjusting the material parameter 𝐸 ∗𝑛to 𝐸 𝑛∗∕ 𝐸= 0.75.
Fig. 5. Identified values for the adjustable parameters of the nodal beam elements providing an 98% agreement of the tetrahedron beam model with the solid model with respect to stiffness and strength for the cases listed Table 1 (a) 𝑙 𝑛∗∕ 𝑙 𝑛; (b) 𝑙 ∗𝑛∕ 𝑙; (c) 𝑟 ∗𝑛∕ 𝑟 ; and (d) 𝐸 𝑛∗∕ 𝐸 𝑠ratio.
areamodeledbybeamelementssuchthatitpredictsthemacroscopicre- sponseofasolidmodel.Fig.5ashowsthatthevalueof𝑙∗𝑛∕𝑙𝑛isstrongly dependentonther/lratioandcR.AsitisfurthershowninFig.5b,the magnitudeof𝑙∗𝑛∕𝑙(liskeptconstant)foragivenr/ldoesnotdependon cR.Thisshowsthatthedependencyof𝑙𝑛∗∕𝑙𝑛 oncRshowninFig.5ais causedbythecalculationofthelnvaluesasfunctionofcR.Thedecrease of𝑙∗𝑛∕𝑙𝑛 from2±0.2to1±0.1inFig.5asuggeststhatthesizeofthe nodalareaisconvergingwithincreasingligamentsize.Inotherwords, thebeammodel’splasticbehaviorbasedonthesamegeometryiscloser tothatofthesolidmodelforlargerligamentsizes(thisisfurtherdis- cussedinsect.4.1).Thesamelineofargumentsappliestotheratio𝑟∗𝑛∕𝑟, seeFig.5c.
Thethirdadjustablematerialparameter,𝐸𝑛∗,controlstheelasticbe- haviorof the nodalarea.Fig. 5ddemonstrates a generaltrend that theratioof𝐸∗𝑛∕𝐸𝑠increasesfrom0.75forr/l=0.19to2.5±0.04for r/l=0.31.Therequiredcorrectionof theeffectivenodalstiffness in- creasesprogressivelywithincreasingr/l ratioand,atthesametime, itshowsanincreasingsensitivitywithregardtothenodalextension, representedbycR.Thegeneraltrendfromavaluebelow1toincreas- ingvaluessimplyresultsfromthediscrepancybetweenthestiffnessand strengthratiothathasbeendiscussedincontextofFig.4a.Thatratio followstheoppositeandhastobecompensatedbyanincreasinglocal Young’smodulusassignedtothenodalbeamelements.
Eqs(1)–(3)arethefittingfunctionsthataregeneratedbasedonthe datesetsfromFig.5(a),(c)and(d),respectively.Thefittingconstants arelistedinTable2.
𝑙∗𝑛∕𝑙𝑛=𝑎0+𝑎1⋅𝑟∕𝑙 (1)
𝑟∗𝑛∕𝑟𝑛=𝑎0+𝑎1⋅𝑟∕𝑙 (2)
𝐸𝑛∗∕𝐸=𝑎0+𝑎1⋅exp( 𝑎2⋅𝑟∕𝑙)
(3)
2.3. Modelvalidation
Itwasreportedintheworkof[18]thatbesidesbendingasthemajor deformationmechanism,torsionrepresentsanotherfundamentaldefor- mationthatiscausedbythespatiallycurvedligamentsandshouldnot beneglected.Thetorqueintheligamentsoccupiesapprox.20%ofliga- mentloadingundermacroscopiccompressingofnanoporousgold.The validationoftheproposedmodelisthereforefirstlycomparingthenodal correctedbeamtetrahedronmodeltothesolidmodel,seeFig.1,under externaltorsionloading.
Fig.6showsthecomparisonbetweensolidmodel(SM)andbeam tetrahedronundertorsion.ThetorqueMthasbeennormalizedwithEr3.
Fig. 6. The mechanical responses of solid model (SM), beam model (BM) and nodal cor- rected beam model (NCBM) for the tetrahedron structure under external torsion loading for the case of r / l = 0.25, c R= 1.1.
Theoverallfittingofthenodalcorrectedbeammodel(NCBM)iscom- parabletothatofthebeammodel(BM),reachingabeamtosolidstiff- nessratioof130%andanapprox.80%overallaccuracyfortheplastic deformation.Theperformancefortorsionwasfoundtobeconsistent acrossthewholerangeofr/lstudiedinthecurrentwork.Asinrandom- izedRVEsapprox.80%oftheligamentdeformationisoriginatingfrom bendingasthegoverningdeformationmechanismandapprox.20%is relatedtotorsion[18],itisreasonabletoassumethattheoverallerror inthemacroscopicresponseoftheRVEiskeptwithin10%.
Asdiscussedintheintroduction,theRVEbeammodelprovidesan idealaccessforstudyingthemechanicsofnanoporousmaterials.Incon- trasttotheanalyzesbasedonabuildingblock,anRVEhowevercon- tainsthousandsofligamentsinterconnectedinanetworkstructure.Such anetworkstructureis muchmorerepresentativeoftherealmaterial intermsofstructuralfeatures,localloadingoftheligamentsaswell asobservabledeformationmechanisms.Thecurrentnodalcorrection methodisthereforeintegratedintotheRVEbeammodelusingthedi- amondstructure,developedintheworkof[8].ThreeRVEshavebeen generatedassolidandbeammodelsforcR=1.1andligamentaspect ratiosofr/l=0.19,0.25,and0.31.InFig.7aandbtheRVEsareshown forr/l=0.25forthesolidandthebeammodel(inrenderingmode), respectively.TheRVEsareloadedwiththesamecompressiondeforma- tionof25%engineeringstrainonthetopsurface.Althoughthestructure isperfectlyordered,itrepresentsafurtherstepofvalidation,because themacroscopiccompressionoftheRVEtranslatesintoadifferentlocal loaddistributionanddeformationinthetetrahedronsasappliedduring thecalibrationofthenodalcorrectedelements.
Theresulting agreementfor themacroscopic elasticmodulusbe- tweenthebeamandsolidRVEis100±5%forallthree cases.Thus, thenodalcorrectedbeammodeliscapableofpredictingtheelasticme- chanicalresponseoftheRVEequivalentlytothesolidmodel.
Becauseanelastic-perfectlyplastic materialmodelis usedso far, earlyconvergenceproblemshavebeenfacedinallbeamRVEsimula- tions,possiblyowingtobuckling.Therefore,amore realisticelastic- linearplasticconstitutivelawisemployedtothesimulationsofRVEs forbothofthebeamandsolidmodels,usingthesameelasticproper-
examiningthegeneralityofthenodalcorrectedbeammodelingconcept inpresenceofworkhardening.
Fig.8ademonstratestheresultingmacroscopicstress-strainresponse fromthenodalcorrectedbeamRVEandthesolidmodelRVEfordif- ferentr/lratiosincomparisonwiththeresultsfromtheoriginalbeam model.Thenodalcorrectedbeammodel(NCBM)significantlyimproves thepredictionsofthemacroscopicmechanicalresponsebyeliminating thesystematicunderestimationofstiffnessandstrengthoftheoriginal beammodel(BM).Furthermore,averygoodagreementwiththesolid model(SM)isachievedoverthewholestrainrange.Fig.8bshowsthe ratioof theresultingreaction forceobtained forthenodalcorrected beamRVEandthesolidmodelRVEagainstthemacroscopiccompres- sionstrain,𝜀eng.Theoverallerroroftheelastic-plasticresponseofthe nodalcorrectedbeamRVEiswithinabout10%forallsolidfractions(i.e.
allconsideredr/lratios).Morespecifically,theaccuracyis101%±5%
forr/l=0.19and0.31,and103%±7%forr/l=0.25.Itcanbeconcluded thatthenodalcorrected beammodelprovidesasufficientlyaccurate elastic-plasticmechanicalresponsethathasthesamepredictivequality ofacorrespondingsolidmodel.Atthesametimeitcostsmuchlesscom- putationtimeandprovidesmuchmoredegreesoffreedomconcerning theintroductionofarandomstructure[8,29]aswell asvariationof ligamentshapesintheRVE[18].
3. Discussionofscalinglaws
InthissectionwewillapplythenodalcorrectedbeamRVEproposed andvalidatedintheprevioussectionsforanalyzingtheimpactofthe nodalcorrectiononthescalinglawsforYoung’smodulusandstrength.
Theresultingcurvesarecomparedtoscalinglawsfromliterature.For thenodalcorrectedbeamRVE,thesolidfraction𝜑isderivedfromthe originalgeometrythatcorrespondstothesolidRVE.
Fig. 9a andbrepresent the scalingbehaviorof the macroscopic Young’smodulusandstrengththat arecomputed fromtheRVEnor- malizedtothecorrespondingmaterialpropertiesofthesolidfractionin theformE/ES,and𝜎y/𝜎y,S,respectively.Thecurvesareshownincom- parisonwithscalinglawssuggestedinpreviousworks[8,29],aswellas theoriginalGibson-Ashbyscalinglawsthatserveasthecommonrefer- enceforallmodels.Foropenporefoamsthatimplybendingasmajor deformationmechanism,theleadingconstantsinEqs.(4)and(5)are CE=1andC𝜎=0.3respectively[10].
𝐸∕𝐸𝑆=𝐶𝐸𝜑2 (4)
𝜎𝑦∕𝜎𝑦,𝑆=𝐶𝜎𝜑3∕2 (5)
InFig.9a,thescalingbehavioroftheYoung’smodulus,E/ES,calcu- latedfromthedifferentRVEsisplottedagainstthesolidfraction𝜑.The different𝜑valuesaretheresultsofvaryingr/lratioswith0.19,0.25and 0.31,giveninTable1.TheblacksolidlinecorrespondstoEq.(11)in [8]whichdescribesthescalinglawoftheRVEbeammodelundercon- siderationoftheeffectofsheardeformationforthickbeams,i.e.itisthe approximationofthelowerboundforthediamondRVE.Thediamond
Fig. 7. RVEs for r / l = 0.25 and c R= 1.1 (a) solid model; (b) nodal corrected beam model with the same effective elastoplastic response.
Fig. 8. (a) Macroscopic stress-strain response for RVEs built as solid model, beam model, and nodal corrected beam model for various r / l ratios; (b) RVE reaction force ratios F beam/ F solid
for different r / l values covering the whole range of elastic and elastic-plastic deformation.
Fig. 9. Comparison between the nodal corrected beam model and previous studies for (a) Young’s modulus; (b) yield strength. RVE-SM, RVE-NCBM, and RVE-BM denote the results from the RVE solid model (this work, Fig. 7 a), the RVE nodal corrected beam model (this work, Fig. 7 b) and RVE beam model [29] , respectively. All results are for perfectly ordered diamond structure ( A = 0).
studiesshowedthatrandomizationofthestructureleadstoafurtherin- creaseincompliance[8,29].Inrecentstudies[14,30],experimentally constructedscalingrelationsaredevelopedforthemacroscopicYoung’s modulusofnanoporousgold,whichsuggestthepowerinEq.(4)being 2.5or2.8withCE=1or0.86.However,theincorporationofthenodal massinthisworkdoesnotleadtoanexponentdifferentto2thatac- cordingto[10]representsthehighestpowerofalltypesofdeformation (tension,compression,shear,bending).Atthispointthequestionre- mainsunsolvedhowhighervaluesintheexponent,asobservedfrom experiments,canbeexplainedbyastructuralmodelsuchastheRVE presentedinthiswork.
Fig.9bshowstheresultsforthescalingbehavioroftheyieldstrength 𝜎y,normalizedbytheyieldstrengthofthesolidphase𝜎y,S.Again,the analyticalsolutionforthebeammodelproposedasEq.(9)in[8]pro- videsalowerlimitthatcorrespondstothenumericalresultsfromthe beammodel(RVE-BM).ItcanbealsoobservedthattheGibson-Ashby modelwithC𝜎=0.3resultsinagoodagreementwiththatscalingrela- tion.
Theanalyticalnodalcorrection(blacksolid line)proposedin the workof[8]asEq.(10)formstheupperlimitwhichagreeswellwith theresultsfromtheworkof[29]representedbyrhombussymbolsin Fig.9b.In[8]themacroscopicstrengthfortheRVEbeammodelwas correctedbyreducingtheavailableleverforbendingoftheligament bytheradiusofthenodalmass.In[29]thiseffectwastranslatedinto thebeammodelbyincreasingtheyieldstrengthofthesolidfraction accordingtothisstructuralstrengtheningeffect. Thelatter approach hastheadvantage,thatitdoesnotaffectthestiffnessofthenetwork structure.Forbothoftheworks,thenodalmassistakenaccountasa shorteningoftheeffectivebendingleveroftheligamentsandassuming thatyieldoccursattheedgeformedbytheligamentandthenode.The resultsshowninFig.9bleadtotheconclusionthatthistypeofcorrec- tionprovidesreasonableresultsforsolidfractions𝜑≤0.15.Ithowever significantlyoverestimatestheeffectofthenodalmassonthemacro- scopicstrengthfor𝜑>0.15,whichiswherethenanoporousmetalsare located.Inviewoftheseinsights,wewillrevisittheanalysisof[8]and [29]withsupportofthenodalcorrectedbeammodelinSect.4.2.
TheresultsofthenodalcorrectedbeamRVE-NCBM(blackcircles) initiallyfollowthesolutionfortheupperlimitbutthendeviatebycon- tinuinginamuchlessprogressivewayfor𝜑>0.15.Thisisbecause thattheupperlimitisconstructedundertheassumptionthattheplas- ticdeformationwillalwayshappenattheendoftheligamentthatis analyticallyshortenedforthepresenceofanode.Thismeansthatthe nodalareaforthisupperlimitissettobeinfinitelystrong,e.g.noplastic deformationcanhappenwithinthenodalarea.However,thevisualin- vestigationonthelocalplasticdeformationofthenodalcorrectedbeam modelandsolidmodelindicatesthat,thenodalareaisalsoplastically deformed.Thistrendbecomesmoresignificantwithincreasing𝜑,be- causelargersolidfractionsleadtolessgeometricaldistinction𝑟∗𝑛∕𝑟be- tweenthenodalareaandtheligament asshownin Fig.5c.Sothat thenodalcorrectedmodeltendstopredictamuchlowerstrengthata higher𝜑.Afitofthedatausingthescalinglaw,Eq.(5)(greendashed line)leadstoaleadingconstantC𝜎=0.72.Theincreaseoftheleading
curvatureoftheconnectedligaments.TheparameterAdefinesthefrac- tionoftheamplitudeoftheequallydistributedrandomdisplacementin relationtotheunitcellsizeoftheundistorteddiamondunitcell.The effectoftherandomizationlevelonthesolidfraction𝜑,causedbythe spatiallengtheningoftheligaments,hasbeentakenintoaccountinthe followinganalyzisassuggestedby[8,29].Concerningthemacroscopic responseoftheRVE,increasingrandomizationdecreasesthevaluesof Eand𝜎y,aswellasthedegreeoflateralextension.Formoredetailswe referto[8,29].
Forthenodalcorrectedbeammodel,therandomizationcharacter- isticsarealsointroducedtothenodalareasbecausethelengtheningof theelementsappliestoallelementsofthemodel.Thisgivesrisetothe questionifthelengtheningoftheelementscausedbytherandomiza- tionshouldbekeptasisorifitshouldbeavoidedinthenodalareasof thenodalcorrectedbeammodel.Forwhatfollows,weassumethatthe nodalelementpropertiesfollowtherelationshipsthatshowninFig.5. Thismeansthatforagivenratioof𝑙∗𝑛∕𝑙𝑛anincreaseintheligament lengthln,causedbytherandomization,leadstoaproportionalincrease inthenodalelementlength𝑙∗𝑛.Itisthusconsistenttoletthenodalele- mentselongateinthesamewayastheelementsthatformtheligament.
Asthepurposeoftheintroductionofnodalcorrectedbeamelementsex- clusivelyaimsatamechanicallyequivalentbehaviorwithreferenceto thesolidmodel,thecalculationofthesolidfractionremainsunaffected.
AllsimulationsforrandomizedstructuresarecarriedoutforcR=1.1.
EachdatapointplottedinFig.10representstheaverageoffivereal- izations; thesizeof theerrorbar correspondstothestandarddevia- tion.Fig.10ashowstheratioofthemacroscopicYoung’smodulusof thenodalcorrectedbeammodel,ENCBM,tobeammodel,EBM,withre- specttovariousrandomizationlevelsfromA=0thatisperfectordered toA=0.3thatisstronglydisordered.Itisshownthatforallrandom- izationlevelsthestiffnessratioincreasesnearlylinearlywiththesolid fraction,indicatingthecontributionofthenodalmasstothestiffness.
Moreover,withincreasingrandomizationlevelthecurvesshiftrightand down.Therightshiftisduetotheincreaseofsolidfractionthatiscaused bytheincreaseoftherandomizationlevel.Thetrendofslightlyshift- ingdownwardsrevealstheeffectofthedegreeofrandomizationonthe complianceinthenodalareas,whichisslightlyreducingthenodalmass contributiononthemacroscopicelasticity.Thiseffectisabout10%if weconsiderthefullrangefromzerotomaximumrandomization.
The verticalaxisof Fig. 10bdemonstrates theratio of theyield strengthofthenodalcorrectedbeammodel𝜎y,NCBMtothebeammodel 𝜎y,BM withrespecttovariousrandomizationlevels.Thegeneraltrend ofFig.10btoaisopposite,namely,increasingsolidfractionleadstoa relativedecreaseofthecontributionofnodalmasstothemacroscopic strength.Forbetterunderstandingofthis effect,thelinkagebetween thecurrentdiscussiontothediscussionsofFig.5a,bandcisrequired.
InthediscussionofFig.5itwasstatedthatwithincreasingr/lthegeo- metricalgapbetweenacylindricalshapednodeandanactualspherical noderesponseisclosingwithincreasingsolidfraction,i.e.𝑙𝑛∗∕𝑙𝑛→1and 𝑟∗𝑛∕𝑟𝑛→1.Itcanbealsobeobservedthatthestrengthratiogradually shiftsdownwardsfrom2.5to2forA=0andfrom1.8to1.5forA=0.3.
Becausetheplasticdeformationinitiatesinthetransitionfromthenodal
Fig. 10. Results for the macroscopic properties of the randomized RVE as ratio of the nodal corrected beam model (NCBM) to the beam model (BM) (a) Young’s modulus; (b) yield strength.
Fig. 11. Determined yield strength of the solid phase vs. ligament size, L . Experimental data are taken from [32,33] ; model data from [29] .
areatotheligament,theeffectonthestrengthreduceswithreducing sizeofthenodalelements.Withincreasingdegreeofrandomization,the correlationbetweenlocalgeometryandmacroscopicstrengthismore andmorereducedandbecomesweakforA=0.3andsolidfractionsof 𝜑≤0.3.
4.2. Analysisofexperimentaldata
Toevaluatetheperformanceofthenodalcorrectedbeammodelthe sizedependentyieldstrengthoftheligamentsisdeterminedfromlit- eraturedataonmacroscopiccompressionofnanoporousgoldsamples withdifferentligamentsize.
Intheworkof[29],thescalingrelationforthemacroscopicyield strengthwasusedasthestartingpointforfittingthetruestress-strain curvesobtainedfrommacroscopiccompressingnanoporousgoldsam- pleswithvariousligamentsize[31,32]usingtheRVEmodel.Theout- comesthatwerefromthefittingwereadatasetofyieldstrengthsforthe solidphase𝜎y,Sindependenceofligamentsize,representedbyhollow circlesinFig.11.Thisdatasetisplottedagainstresultsfromtheworks of[32,33]assolidblacksymbolsinFig.11,thatweredeterminedus- ingindependentapproaches.Roschningetal.’sresultsareofthesame magnitudeasthosedatareportedinthereferencedliteratures,butone resultforthe50nmligamentsizeappearedtobeanoutlierwhichcould notbeexplainedby[29].Theonlydifferencecomparedtotheother datawasthatithadthehighestsolidfractionofallsamplesof𝜑=0.3.
Were-analyzedthedatawiththenodalcorrectedRVEfollowingthe sameapproachproposedby[29]witharandomizationlevelofA=0.23 andaligamentaspectratioofr/l=0.25.Theresults,shownassolidred symbolsinFig.11,leadtoanelevationofRoschningetal.’sresults.The nodalcorrectedbeamRVEseemstoprovideanoverallcloseralignment withtheresultsfrom[32,33].Inparticular,thepreviousoutlierisnow closertotheoveralltrend.
Thedifferentelevationintheyieldstrengthforthedifferentdata pointscan beunderstoodwiththehelpofFig.9b.Thenodalcorrec- tionappliedin[29]wasbasedonacorrectiontermwhich– asaresult oftheassumedrigidityofthenodalmass– increasinglyoverestimates thegeometricalstrengtheningforincreasingsolidfraction.Becausethe materialsyield strengthandthegeometrical strengtheningeffect are multiplicativewithrespecttothemacroscopicstrength,areductionof thegeometricalstrengtheningeffecttotheaccuratevaluemustbecom- pensatedbyanincreasedyieldstrengthofthesolidfraction,𝜎y,S.The higherthesolidfraction,themoretheidentifiedyieldstrengthisele- vated. Consequently,theoriginaloutlier,whichhasthehighestsolid fraction,getsmostlyelevated.
5. Conclusions
Inthecurrentwork,anewbeammodelingconceptfornanoporous metalshasbeenproposedforpredictingthemacroscopicelastic-plastic responsebyincorporatingtheconnectingnodalmassinthejunctions ofthenetworkstructure.Thenodalcorrectedbeammodeliscalibrated basedonthemechanicalresponseforbendingofabeamtetrahedron structureinrelationtoitscorrespondingsolidmodelbyadjustingthree additionalparametersspecificallyassignedtothenodalarea.Anexcel- lent agreementbetween thenodalcorrectedbeamandsolidRVEfor elasticmodulus andyieldstrength hasbeenachieved, whichis con- ducted withoutlosing theadvantage of thehighcomputationaleffi- ciency.
Thenodalcorrectionleadstoanincreaseinthemacroscopicstiff- nessandstrengthofthebeammodel.Theleadingconstantintheresult- ingscalinglawforYoung’smodulusisCE=0.57and,comparedtothe beammodelwithoutnodalcorrection,aboutafactoroftwolarger.It ishoweverstilllessstiff thantheoriginalGibson-Ashbymodelthathas aleadingconstantofCE=1.Theincorporationofthenodalmassdoes notchangetheexponentof2inthescalinglaw.
Withrespecttomacroscopicplasticdeformation,thecorrectionsug- gestedin[8,29]significantlyoverestimatestheeffectofthenodalmass for solid fractionsabove 15%. Itwas found, thatthescaling lawof Gibson-Ashbyverywellfitsthedatafromthesimulationswiththenodal correctedRVE,whentheleadingconstantissettobeC𝜎=0.72.Thisdif- ferenceto[8]isnowunderstoodasaconsequenceoftheoverestimation
SupportwasprovidedbyDeutscheForschungsgemeinschaftwithin SFB986‘‘Tailor-MadeMulti-ScaleMaterialsSystems:M3”,projectB4.
AppendixA.Nomenclatures
Parameters Descriptions 𝜑 Solid fraction
w Transverse loading displacement
R Nodal radius
l n Length between nodal center and the end of the ligament attached to the node
𝑟 ∗𝑛 Larger cross-section radius of beam element in the nodal area (adjustable)
𝑙 ∗𝑛 The length between the node center and the end of the ligament (adjustable)
r Ligament radius
l Ligament length
c R Constant governing nodal mass
R Nodal radius
E S Young’s modulus 𝜈 Poisson’s ratio 𝜎y,S Yield strength
k solid Stiffness of the solid tetrahedron F solid Solid tetrahedron strength k beam Stiffness of the beam tetrahedron F beam Beam tetrahedron strength
𝐸 𝑛∗ Young’s modulus for the beam elements of nodal area (adjustable) a 0, a 1, and a 2 Constants for fitting functions Eqs (1) –(3)
𝜀 eng RVE macroscopic compression engineering strain C Eand C 𝜎 Constants of Gibson-Ashby model
A Randomization parameter
E NCBM Macroscopic Young’s modulus of nodal corrected beam model E BM Macroscopic Young’s modulus of beam model
𝜎y,NCBM Macroscopic yield strength of nodal corrected beam model 𝜎y,BM Macroscopic yield strength of beam model
References
[1] Weissmüller J , Newman RC , Jin H-J , Hodge AM , Kysar JW . Nanoporous Met- als by Alloy Corrosion: Formation and Mechanical Properties. MRS Bulletin 2009;34(08):577–86 .
[2] Balk TJ , Eberl C , Sun Y , Hemker K , Gianola D . Tensile and compressive microspeci- men testing of bulk nanoporous gold. JOM 2009;61(12):26–31 .
[3] Biener J , Hodge AM , Hayes JR , Volkert CA , Zepeda-Ruiz LA , Hamza AV , et al. Size ef- fects on the mechanical behavior of nanoporous Au. Nano Lett 2006;6(10):2379–82 .
[12] Liu R , Antoniou A . A relationship between the geometrical structure of a nanoporous metal foam and its modulus. Acta Materialia 2013;61(7):2390–402 .
[13] Hu K , Ziehmer M , Wang K , Lilleodden ET . Nanoporous gold: 3D structural analyses of representative volumes and their implications on scaling relations of mechanical behaviour. Philosophical Magazine 2016:1–14 .
[14] Badwe N , Chen X , Sieradzki K . Mechanical properties of nanoporous gold in tension.
Acta Materialia 2017;129:251–8 .
[15] Farkas D , Caro A , Bringa E , Crowson D . Mechanical response of nanoporous gold.
Acta Materialia 2013;61(9):3249–56 .
[16] Ngô B-ND , Stukowski A , Mameka N , Markmann J , Albe K , Weissmüller J . Anomalous compliance and early yielding of nanoporous gold. Acta Materialia 2015;93:144–55 . [17] Ngô BND , Roschning B , Albe K , Weissmüller J , Markmann J . On the origin of the anomalous compliance of dealloying-derived nanoporous gold. Scripta Materialia 2017;130:74–7 .
[18] Jiao J , Huber N . Deformation mechanisms in nanoporous metals: Effect of ligament shape and disorder. Comput Mater Sci 2017;127:194–203 .
[19] Lührs L , Soyarslan C , Markmann J , Bargmann S , Weissmüller J . Elastic and plastic Poisson’s ratios of nanoporous gold. Scripta Materialia 2016;110:65–9 .
[20] Saane SSR , Mangipudi KR , Loos KU , De Hosson JTM , Onck PR . Multiscale modeling of charge-induced deformation of nanoporous gold structures. J Mech Phys Solids 2014;66:1–15 .
[21] Soyarslan C , Husser E , Bargmann S . Effect of Surface Elasticity on the Elastic Re- sponse of Nanoporous Gold. Journal of Nanomech Micromech 2017;7(4):04017013 . [22] Husser E , Soyarslan C , Bargmann S . Size affected dislocation activity in crystals: Ad- vanced surface and grain boundary conditions. Extreme Mech Lett 2017;13:36–41 . [23] Bargmann S , Soyarslan C , Husser E , Konchakova N . Materials based design of structures: Computational modeling of the mechanical behavior of gold-polymer nanocomposites. Mech Mater 2016;94:53–65 .
[24] Lührs L , Müller B , Huber N , Weissmüller J . Plastic Poisson’s Ratio of Nanoporous Metals: A Macroscopic Signature of Tension-Compression Asymmetry at the Nanoscale. Nano Lett 2017;17:6258–66 .
[25] Pia G , Delogu F . Mechanical behavior of nanoporous Au with fine ligaments. Chem Phys Lett 2015;635:35–9 .
[26] Ziehmer M , Hu K , Wang K , Lilleodden ET . A principle curvatures analysis of the isothermal evolution of nanoporous gold: Quantifying the characteristic length-s- cales. Acta Materialia 2016;120:24–31 .
[27] Mangipudi KR , Epler E , Volkert CA . Topology-dependent scaling laws for the stiff- ness and strength of nanoporous gold. Acta Materialia 2016;119:115–22 . [28] DassaultSystèmes. Abaqus Documentation 2014 .
[29] Roschning B , Huber N . Scaling laws of nanoporous gold under uniaxial compression:
Effects of structural disorder on the solid fraction, elastic Poisson’s ratio, Young’s modulus and yield strength. J Mech Phys Solids 2016(92):55–71 .
[30] Zabihzadeh S , Van Petegem S , Holler M , Diaz A , Duarte LI , Van Swygenhoven H . Deformation behavior of nanoporous polycrystalline silver. Part I: Microstructure and mechanical properties. Acta Materialia 2017;131:467–74 .
[31] Wang K , Weissmüller J . Composites of Nanoporous Gold and Polymer. Adv Mater 2013;25(9):1280–4 .
[32] Wang K , Kobler A , Kübel C , Jelitto H , Schneider G , Weissmüller J . Nanoporous–
gold-based composites: toward tensile ductility. NPG Asia Mater 2015;7(6):e187 . [33] Jin H-J , Kurmanaeva L , Schmauch J , Rösner H , Ivanisenko Y , Weissmüller J .
Deforming nanoporous metal: Role of lattice coherency. Acta Materialia 2009;57(9):2665–72 .