外文翻譯--線鋸過(guò)程中凹凸不平面的損傷情況 英文版.pdf

INTERNATIONALJOURNALOFPRECISIONENGINEERINGANDMANUFACTURINGVol.12,No.6,pp.941-947DECEMBER2011/941DOI:10.1007/s12541-011-0126-4NOMENCLATUREFz=VerticalforceonagritFzg=ForceonasinglegritFzs=Totalverticalforceactiononsampley=YieldstressR=Gritradiush=CutdepthforasingleparticleVx=WirespeedVz=FeedspeedT=WiretensionAp=ProjectedareaofthecuttrenchLo=CutlengthofsampleLg=DistancebetweencuttingparticlesD=WidthofcuttrenchS=Slidingdistancec=Mediancracklength=HalfoftheincludedangleofthegritsE=ModulusofelasticityofingotH=HardnessoftheingotP=IndentationforceKc=Fracturetoughnessoftheingotw=DistributedwireloadonthesampleN=Numberofcuttingparticlesinthecutlength=Wirebowangle1.IntroductionSiliconwafersusedinthesolarcellandmicroelectronicsindustriescanbecutfromsiliconcrystalsusinginnerdiameter(ID)saworwiresaw.WiresawhasadvantagesoverIDsaw.Theseadvantagesarehigherproductivity,lesswafer-surfacedamage,andlowerkerfloss.1Moreover,thediameterofwaferthatcanbeslicedbyawiresawishigherthanthatobtainablebyanIDsaw.Wiresawsareusedtocutsapphire,siliconcarbide,lithiumniobate,wood,rock,andalmostallkindsofceramics,includingfoamceramics.1-3Moller4statedthatthewiresawprocessisresponsiblefor30%RoughnessDamageEvolutionDuetoWireSawProcessEgemenTeomete1,#1Dept.CivilEngineering,DokuzEylulUniversity,KaynaklarCampusBuca,Izmir,Turkey,35160#CorrespondingAuthor/E-mail:,TEL:+90-232-4127060,FAX:+90-232-4531192KEYWORDS:Ceramic,Damagemodel,Ductileregimemachining,Roughness,WiresawThewiresawprocessiswidelyusedforsiliconwaferproductionwithhighyieldandlowsurfacedamageinsolarcellandmicroelectronicsindustries.Thewiresawprocessisusedtomachinebrittlematerialsintheductileregimewherehighyieldandlowsurfacedamagearedesired.Thewiresawprocessisalsousedtocutconcreteandrocksincivilengineering.Inthisstudy,anexperimentalparametricstudywasconductedbyvaryingprocessparameterstodeterminesurfaceroughnessdamage.Ductileregimematerialremovalbytrans-granularfailureandbrittlefracturebyinter-granularfailureareobservedinelectronmicrographsofthecutsurfaces.Adamagemodelthatrelatestheroughnessdamagetoprocessparameterswasderived.Thedamagemodelpredictstheroughnessdamagesatisfactorily.Themodelshowsthattheroughnessdamageisproportionaltotheratiooffeedspeedtowirespeed.Improvementintheefficiencyoftheprocesswithoutincreasingtheroughnessdamagecanbeattainedbyincreasingthefeedspeedproportionallytowirespeed.Wiretensiondoesnotaffectroughnessdamage.Roughnessdamage,however,isaffectedbypropertiesofthewire.Wireshavingsmallergritradiusandsmallgritspacingcauselessroughnessdamage.Manuscriptreceived:May4,2010/Accepted:May15,2011KSPEandSpringer2011942/DECEMBER2011INTERNATIONALJOURNALOFPRECISIONENGINEERINGANDMANUFACTURINGVol.12,No.6ofthetotalsiliconwafer-productioncost,whichdirectlyaffectsindustry.Thereisaneedtooptimizetheprocessbydevelopingmodelsrelatingprocessparameterstoproductqualityandprocessefficiencymeasures.4Earlywiresawprocessesforwaferproductiondevelopedinthe1990sconsistedofabaresteelwireandabrasive-carryingslurry,resultinginfree-abrasivemachiningusingelasto-hydrodynamicforces.5,6TheabrasiveparticlescanbeSiCordiamond.Themeangritsizeofabrasiveparticlescanbe5to30mwitha30%to60%volumefractionintheslurry.Averagewirediameteris180m,leadingtoakerflossof200to250m.Theslurrycanbewaterbasedoroilbased.Oil-basedslurrycausesthewaferstosticktoeachother,anditishardtoseparatethem,whileremovaloftheoilfromthewafersurfaceisanotherproblem.Disposaloftheoil-basedslurryafteruseisalsoaproblem.Hydrogengasproducedfromtheinteractionofwater-basedslurryandsiliconmaycauseexplosions.However,fromanenvironmentalpointofview,consideringthehighamountofslurrydisposedofduringtheprocess,water-basedslurriesaregenerallypreferred.4Clarketal.5statedthatinordertoincreasetheproductivityandtobeabletocutharderceramics,diamond-impregnatedwire,whichleadstofixed-abrasivemachining,wasdeveloped.Inwiresawingwithfreeabrasives,wirespeedisbetween5to15m/sandwiretensionis20to30N.Thefeedintotheingotresultsinawirebowsothatthewiremakes2oto6owiththehorizontal.6Inthefixed-abrasivemachiningwire-sawprocess,thewirespeedislowerasmaterialremovalisnotoccurringbyhydrodynamicaction.Inmulti-wiretechnology,asinglewireiswindedtoatensioncontrolunitandseveralguidepulleys,whicharegroovedwithconstantpitch.Fivetosevenhundredparallelwiresruntogetherandarecollectedatatake-upspool.Theingotisslicedintohundredsofwafersasitisfedintothewireweb.Thewafersinsolar-cellindustryarecutbyrunningthewireinonlyonedirectionatahighspeedbetween5to20m/s,whilethewafersinthemicroelectronicsindustryarecutbyrunningthewireinbothdirectionswithalowerspeed(oscillatingthewirefromonespooltoanother).4Researchonthewiresawprocesshasbeenongoinginthreemainareas:materialremovalmechanisms,kinematicsofwires,andparametricstudiesbetweentheprocessinputsandoutputs.Lietal.7presentedthestressesunderanabrasiveparticle,whichisrollingandindentinginawiresawprocess.Materialremovalmechanismsforfree-abrasivemachiningweredevelopedusingfracturemechanicsandhydrodynamicbehaviorofslurrybyMoller.4ThematerialremovalrateisdefinedasafunctionofpowersuppliedtotheabrasivebyhydrodynamiceffectandthehydrodynamicfilmpropertiesarecalculatedusingthefiniteelementmethodwhichcouplesReynoldsequationofhydrodynamicswiththeelasticityequationofwire.6Liuetal.8statedthatthematerialremovalmechanismofbead-impregnatedwire-sawcuttingofrockisaHertziantypefractureinwhichthefractureoccursduetothetensilefieldbehindtheslidingbead.WeiandKao9workedonstiffnessanalysesofstraightandbowedwiresundertension.Vibrationcharacteristicsofwirewithrespecttowirespeed,tension,andslurryviscositywasinvestigated.Theincreaseofwiretensionandslurryviscositydecreasesvibrationamplitudeandkerfloss,whilethewirespeedhasalmostnoaffectwhenitisbelow25m/s.1,10Processmonitoringofthewiresawforforces,wirespeed,feedrate,wirebow,andwiretensionwasdevelopedbyClarketal.5Parametricstudiesrelatingprocessparameterstoforces,andsurfaceroughnessandwirewearforcuttingfoamceramicsandwoodwereconductedbyClarketal.2Hardinetal.11conductedaparametricstudyforslicingsinglecrystalSiCwithafixed-abrasivediamondwire,relatingwirespeed,rockingfrequency,anddown-feedratewithsurfaceandsubsurfacedamage.Closed-loopdiamond-impregnatedwiresawcuttingofAl2O3andTiCceramicswasstudiedbyMengetal.12HardnessanisotropyofLithiumNiobatewafershasbeeninvestigatedusingnano-indentation.13BhagavatandKao14determinedthedirectionofapproachforthreemostcommonlyslicedorientationsofsiliconconsideringcrystalanisotropy.Damageevolutionduetowiresawingofsiliconwafersisofsignificantinterestasthephotovoltaicandsemiconductorindustrieshavestricttolerancesforsurfacequality.Theprocess-induceddamageonbrittlematerialscanbemodeledstartingwithexistingdamagemodelsofindentationofbrittlematerials.Thereexistseveralmodelsforthefailuremechanismsinbrittlematerialsduetoindentation.15-20Ryuetal.studiedindentationonsiliconwafer,glassandsiliconcarbide.21Zhaoetal.observedtheindentationdamagemodesongroundsurfaceofopticalglass.22Ductileregimegrindingofbrittlematerialshasbeeninvestigatedexperimentallybydifferentresearchers.23-28Bifanoetal.24statedthatwhenthefeedisdecreasedbelowacertainamountingrinding,atransitionofwearmechanismfrombrittletoductilemodecanbeachieved.Inthisstudy,adamagemodelforwiresawprocessinducedroughnessdamageisdeveloped.Thedamagemodelisbasedonductilemodematerialremovalandbrittlemodedamage,asobservedinSEMimagesofcutsurfaces.Thedamagemodelpredictstheexperimentallymeasureddamagesuccessfully.Theexperimentalworkispresentedinsection2.Themodelispresentedinsection3.Theresultsanddiscussionofthestudyarepresentedinsection4.Theconclusionsarepresentedinsection5.2.ExperimentalProcessWiresawexperimentswereconductedonaluminaceramic.Thewirebowangle,wireaxialspeed,Vxandfeedrate,Vzweremeasuredduringthewiresawcuttingtests.Thesurfaceroughnessofcutsurfaceswasalsomeasured.TheSEMimagingofcutsurfaceswasobtained.Theequipmentusedinthesemeasurementsandtheprocessparametersarepresentedinthissection.2.1WireSawCuttingandWireBowAngleMeasurementAwiresawmachine(MillenniummodelproducedbyDiamondWireTechnologyinColorado,Springs)wasusedintheexperiments.Thisspool-to-spoolwiresawmachinewithrockingmotionofthewirecanbecontrolledbythewirespeed,Vx,down-INTERNATIONALJOURNALOFPRECISIONENGINEERINGANDMANUFACTURINGVol.12,No.6DECEMBER2011/943feedspeed,Vz,andwiretension,T.Thetensionwascontrolledbywiretensionpulleyspoweredbyairpressure,whiletherockingmotionwascontrolledbywireguidepulleysascanbeseeninFig.1.Thecutlengthofthewirewas300ft(91.4m).Thus,ateverydirectionreversal,300ftofwirewastransferredfromonespooltotheother.Acoolantconsistsofwater-to-lubricantSawzit(ProductofSyntheticLubricants,Inc.)ratioof50/1wasusedduringcuttingtests.Fourdifferentdiamondgritcoatedsteelwireswereusedinthewiresawexperiments.Theaveragehalf-includedangleofthegritsonDWS2was=71o.Thediamond-grit-coatedsteelwireDWS3wasaproductofWellDiamondWireSawsInc.Diamond-grit-coatedsteelwiresDWS4andDWS5wereproductsofSaint-GobainAbrasivesInc.TheDWS4andDWS5weremanufacturedbynickelelectroplatingonsteel.Thegritswereaffixedintotheelectroplatednickellayer,whilethecoreremainsintact.Aluminaceramicsampleshavingtensilestrengthoffr=300MPa,fracturetoughnessKIC=4MPam1/2,YoungsmodulusofE=370GPa,29andhardnessofH=22GPa20wereusedinthecuttingtests.ThecutlengthofthesampleswasbetweenLo=1520mmandtheheightwasHs=7.1mm.AgroupoftestsweredonewithDWS2withthewirespeedvariedoverVx=1.3,1.8,2.95,3.5m/s,thewiretensionvariedoverT=13.3,17.8,22.4,26.7N,andthedownfeedvariedoverVz=5,6.35,10.16m/sec.Inordertoexploretheeffectofdifferentwirescharacteristicsonsurfacequality,twelvetestsweredonewithprocessparametersVx=1.35,2,3,4m/s,Vz=6.35m/sec,andT=13.3NusingthewiresDWS3,DWS4,andDWS5；fourtestswereconductedwitheachwire.Amegapixeldigitalcamera(KodakEasyShareDX7630)of28562142pixelswasusedtomeasurethewirebowangleseeninFig.2.TheimagesofthewireandsamplewerecollectedduringthetestandanalyzedusingMatlab(Mathworks)toobtaintheanglebetweenthewireandthehorizontal.Theaverageofthesteadystatewirebowangles,wasattainedtothetestasthesteadystatewirebowangleofthattest.2.2SurfaceRoughnessMeasurementsandSEMImagingThesurfaceroughnessofthecutsurfacesweremeasuredbyusinganopticalnon-contactprofilometer,ZygoNewView6000,manufacturedbyZygoCorporation.A10xlenswasusedforthemeasurements.Theprofilometerhadaverticalresolutionontheorderof3nanometer；theresolutioninthehorizontalplanewas1.1m,whilethefieldofviewusedwas0.70.53mm.Inastitchmeasurement,theprofilometertakescontinuousmeasurementseach0.70.53mmandstitchesthemtogetherintoonedataset.Threestitchmeasurements,eachof0.73mmdimensions,wereappliedinthedirectionofcuttingforeachsampleontheleft-middle-rightofthecutsurface.Afterthemeasurementsweretaken,thedatawasprocessedusingthesoftwareMetroProVersion8.1.5developedbyZygoCo.Ahighpassfilteringwasappliedtoremovethesurfacewaviness.Arithmeticaveragedeviationfromthecenterline(bestfitplane)wasobtained.Theaverageofthreemeasurementswastakenassurfaceroughness(Ra)ofthetest.AScanningElectronMicroscope(SEM),JEOLJSM-606LV,Fig.1Singlewire,spool-to-spoolwiresawmachine.Thewiretrackismarkedbythedashedline.(DWTInc.,MillenniumModel,Colorado,Springs,USA)Fig.2WirebowangleinwiresawtestsFig.3SEMimageofawiresawcutsurfaceofaluminaceramic(Vx=1.3m/sec,Vz=5m/sec,T=13N)944/DECEMBER2011INTERNATIONALJOURNALOFPRECISIONENGINEERINGANDMANUFACTURINGVol.12,No.6wasusedtoimagethecut-surfacetopology.TheSEMimagesweretakenfromthelowerhalfofthesample,onthecenterlineofthecutsurface.Itisseenfromtheimagesthatthematerialremovalmechanismisthetrans-granularfailure.Inter-granularfailure,inwhichgrainboundaryfailureresultsingraindislodgementinabrittlemode,isalsoobserved.BothmechanismscanbeseeninFig.3.3.RoughnessModelDerivationDuctilematerialremovalandbrittlefractureisobservedinSEMimages.TheproposedmodelisshowninFig.4.ThematerialremovaloccursinaductilemodeasseeninSEMimages,whilethedamageoccursduetomediancrackingasinFig.4.AsdiscussedbyEvansandMarshall,15removalofplasticallydeformedmaterialinthecuttingzonereducesresidualstress.Thisreducesthetendencyoflateralcrackformationinbrittlematerials.Fuetal.30derivedtheforceonasinglegritinductilemodematerialremovalaspresentedinEq.(1),whereyisyieldstress,Riscuttingparticleradius,andhiscutdepthforasingleparticle.zzgyFFRh=(1)ThemasscontinuityofthecuttingprocessgivesusEq.(2).ogzxpoggLDhsLdVolddShhVVdtAdtLDdtLL=(2)Volumeisthetotalamountofmaterialremoved,Apistheprojectedareaofthecuttrench,Loisthecutlengthofsample,Lgisthedistancebetweencuttingparticles,Diswidthofcuttrenchthatcanbetakenasdiameterofwire,Sisslidingdistance,Vxistheaxialspeedofwire,andVzisthefeedofwire.Theforceonasinglegrit,Fzg,canbeobtainedintermsofprocessparametersbyusingEq.(1)andEq.(2).zzgygxVFRLV=(3)Thedamageresultingfromwiresawcuttingiscorrelatedwithmediancrackdepth.Lawnetal.16derivedthemediancracklengthusingfracturemechanicsprinciples.ThemediancracklengthispresentedinEq.(4).Lawnetal.16calibratedtheindentationcoefficients0.032and0.017inEq.(4)usingindentationdataofsoda-limeglassandnotedthattheyareapplicabletoallbrittlematerials.2132230.0320.017(cot)cEPcHK=+(4)InsertingEq.(3)inplaceofP=FzginEq.(4)givesusEq.(5).221323230.0320.017(cot)ygzcxRLEVcHKV=+(5)Fig.4Wiresawroughnessdamagemodel:ductilematerialremovalandbrittlefractureThedamageduetothewiresawprocessispresentedintermsoftheprocessparametersinEq.(5).Thedamageisafunctionofthehalfoftheincludedangleofthegrits,；themodulusofelasticityofingot,E；thehardnessoftheingot,H；thefracturetoughnessoftheingot,Kc；andwireproperties,feedspeed,andwirespeed.4.ResultsandDiscu