外文翻譯--對(duì)由微注塑模型的聚合物微結(jié)構(gòu)復(fù)制的實(shí)施和分析 英文版.pdf

INSTITUTEOFPHYSICSPUBLISHINGJOURNALOFMICROMECHANICSANDMICROENGINEERINGJ.Micromech.Microeng.14(2004)415422PII:S0960-1317(04)69783-XImplementationandanalysisofpolymericmicrostructurereplicationbymicroinjectionmoldingYu-ChuanSu1,2,JatanShah3andLiweiLin1,21BerkeleySensor&ActuatorCenter,UniversityofCalifornia,Berkeley,CA94720,USA2DepartmentofMechanicalEngineering,UniversityofCalifornia,Berkeley,CA94720,USA3DepartmentofMechanicalEngineering,UniversityofMichigan,AnnArbor,MI48105,USAE-mail:Received30September2003Published17December2003O/JMM/14/415(DOI:10.1088/0960-1317/14/3/015)AbstractThispaperpresentstheadaptationofaconventionalinjectionmoldingprocesstothemassreplicationofpolymericmicrostructureswithappropriatemolddesignandprocesscontrol.Usingwet-etchedsiliconwaferswithmicrostructuresonthesurfacesasmoldinserts,wehavesuccessfullypredicted,improvedandoptimizedthereplicationresults.Theflowbehaviorsofpolymermeltsinmicromold-cavitiesarecharacterizedbybothsimulationandexperiments.Amongvariousprocessparameters,temperatureisidentifiedasthekeyfactorthatdecisivelydeterminesthequalityofinjection-moldedmicrostructures.Basedonthecollectedexperimentalandsimulationresults,processoptimizationisperformedtoimprovereplicationqualityandtoestablishguidelinesforpotentialapplications.Becauseofitshighspeedandlowcost,theadaptationoftheinjectionmoldingprocesstomicrofabricationwillleadtoapromisingtechnologyforMEMSapplications.1.IntroductionBecauseoftheiruniqueproperties,polymershavebeenincreasinglyusedinawiderangeofapplicationsincludingbothmacro-andmicro-devices.InordertoexpandthefieldofMEMStopolymer-baseddevices,itisimportanttointroduceeffectivetechniquesforthefabricationofpolymericmicrostructuresatalowcostandwithhighprecision.Inrecentyears,anumberoftechnologiesforpolymericmicrostructurereplicationhavebeenproposed,includingtheLIGAprocess1,2thatuseseitherhotembossing3orinjectionmolding4toduplicatepolymericmicrostructures.Usingmoldinsertsfabricatedbyx-raylithography,theLIGAprocessprovidesthepossibilitytomanufacturemicrostructureswitharbitrarylateralgeometryandhighdepthforhighaspectratiodevicesfromavarietyofmaterialssuchasmetals,polymersandceramicsbyvariousmoldingprocesses.Amongdifferentmoldingtechniques,injectionmoldingisthemostprominentonewithadvantagesoflowcostandhighprecisionformassproduction.Successfulresultsforthereplicationofpolymericmicrostructureshavebeenachievedbyusingspecialinjectionmoldingprocesses512andconventionalCD-injectionmoldingtechniques13,14.However,theflowbehaviorsofpolymermeltsinmicromold-cavitiesarenotfullyunderstood.Itisbelievedthatduetothelargesurface-to-volumeratio,surfaceeffectswilldominatetheflowbehavioratthemicroscale15.Thispaperaimstoinvestigatetheflowbehaviorofpolymermeltinthemicromold-cavityanddeterminethenecessarystrategiestoadaptthetraditionalinjectionmoldingprocessforthereplicationofpolymericmicrostructures.First,thedirectapplicationoftheconventionalinjectionmoldingprocessinthereplicationofpolymericmicrostructuresisanalyzedusingasimulationsoftwareC-MOLD16.Differentcombinationsofprocessparametersarethensimulatedtoinvestigatetheflowbehaviorofpolymermelt,therelationshipbetweenprocessparametersandthequalityofmoldedmicrostructures.Usingtheseresults,themostsignificantparameterscanbeidentifiedandpossibleprocessingstrategiescanbeproposedandsimulatedtotestthefeasibility.Finally,0960-1317/04/030415+08$30.002004IOPPublishingLtdPrintedintheUK415Y-CSuetal2bxyzyzVelocityprofilePolymermeltPressureandmaterialsupplyFigure1.Schematicofpolymermeltflowinginathincavity.thesestrategiesareappliedinmoldtrialstoevaluatetheirvalidity.2.TheoreticalmodelsBecausemostinjectionmoldedpolymericpartshavecomplicatedthree-dimensional(3D)configurationsandtherheologicalresponseofpolymermeltisgenerallynon-Newtonianandnon-isothermal,itisextremelydifficulttoanalyzethefillingprocesswithoutsimplifications.ThegeneralizedHele-Shaw(GHS)flowmodelintroducedbyHieberandShen17isthemostcommonapproximationthatprovidessimplifiedgoverningequationsfornon-isothermal,non-Newtonianandinelasticflowsinathincavity,asshowninfigure1.TheassumptionsoftheGHSflowmodelare(1)Thethicknessofthecavityismuchsmallerthantheotherdimensions.(2)Thevelocitycomponentinthedirectionofthicknessisneglected,andpressureisafunctionofxandyonly.(3)TheflowregionsareconsideredtobefullydevelopedHele-Shawflowsinwhichinertiaandgravitationalforcesaremuchsmallerthanviscousforces.(4)Theflowkinematicsisshear-dominatedandtheshearviscosityistakentobebothtemperatureandshearratedependent.ThedetailedderivationshavebeendevelopedbyHieberandShen,andtheseassumptionsapplywellforthemicroinjectionmoldingprocess.Inviewoftheseassumptionsandneglectingcompressibilityduringthefillingstages,themomentumequationintheCartesiancoordinatesystemreducesto17zbracketleftbiggvxzbracketrightbigg=Pxzbracketleftbiggvyzbracketrightbigg=Py(1)wherevxandvyarevelocitycomponentsinthexandydirections,respectively；P(x,y)isthepressure,(prime,T)istheshearviscosity,primeistheshearrateandTistemperature.Underthepresentassumptions,primeisgivenbyprime=braceleftBiggbracketleftbiggvxzbracketrightbigg2+bracketleftbiggvyzbracketrightbigg2bracerightBigg1/2.(2)Applyingthelubricationapproximation,thethickness-averagedcontinuityequationresultsin(bvx)x+(bvy)y=0(3)wherevxandvyareaveragedvelocitiesoverz,andbishalfofthethickness.Afterseveralderivativesteps,thegoverningequationfortheflowofthepolymermeltcanbereducedtothecelebratedReynoldsequation:xbracketleftbiggSPxbracketrightbigg+ybracketleftbiggSPybracketrightbigg=0(4)whereSistheflowconductancewhichisdefinedasS=integraldisplayb0z2dz.(5)Thevelocitiesandshearratecanbeobtainedasvx=Lambda1xintegraldisplaybzz1dz1vy=Lambda1yintegraldisplaybzz1dz1prime=zLambda1(6)whereLambda1x=Px,Lambda1y=PyandLambda1=bracketleftbigLambda12x+Lambda12ybracketrightbig1/2.Becauseofthetemperaturedifferencebetweenmoldandpolymermeltandtheviscousheatinginsidetheflow,thefillingprocessshouldbetreatedasanon-isothermalcase.Heatconductioninthedirectionofflowisneglectedbasedontheassumptionthatthethickness2bismuchsmallerthantheothertwodimensions.TheenergyequationinthemeltregionbecomescpbracketleftbiggTt+vxTx+vyTybracketrightbigg=k2Tz2+prime2(7)wheretheprime2istheviscousheatingterm,and,cpandkaredensity,specificheatandthermalconductivity,respectively.Forsimplicity,itisassumedthatthevelocityandtemperaturearesymmetricinthezdirection,thevelocitiesofpolymermeltonthemoldsurfacesarezeroandthetemperatureofmoldremainsatTwduringfilling.Theboundaryconditionsaregivenbyvx=vy=0atz=bvxz=vyz=0atz=0T=Twatz=bTz=0atz=0.(8)Ascanbeseen,theequationsofthismodelarenonlinearandcoupled.Itisdifficulttosolvetheseequationsanalytically.Inthispaper,simulationsoftwareC-MOLDthatemploysnumericalsolversbasedonahybridfiniteelement/finitedifferencemethodisusedtosolvethepressure,velocityandtemperaturefieldsoftheGHSmodel.Becauseoftheseapproximations,aGHSmodelcannotpredicttheexactflowfieldneartheadvancingflowfrontorattheedgesofthemold.Thismightcauseerrorsinpredictingtheflowbehaviornearmicroscalemoldcavities.3.DesignandfabricationofmoldingapparatusAnaluminummoldismanufacturedforthereplicationprocess.Theschematicdiagramandaphotographofthealuminummold,whichconsistsofcavityandcorehalves,areshowninfigure2.Thecavityhalfincorporatesthecavityinwhichamoldinsertiskept.A4-inchsilicon416ImplementationandanalysisofpolymericmicrostructurereplicationbymicroinjectionmoldingMountingplateStripperplateMoldinsert(Siliconwafer)MountingplateCorehousingplateSprueCavityhousingplateInsulationlayerHeaterBaseplateFigure2.Injectionmoldset-up.Figure3.Microstructuresonasiliconmoldinsert.waferwithbulkmicromachinedmicrostructuresisusedasthemoldinsert.Figure3showsthesiliconmicromold-insertthatisetchedtohaveacavitydepthof110m.Thesquarecavitieshaveopeningsof320m,160m,80mand40mandareetchedbymeansofanisotropicsiliconetchinginTMAH(tetramethyl-ammoniumhydroxide).Aheaterisinstalledintheinjectionmoldtocontrolthetemperatureduringthemoldingprocess.Tohavebetterthermalconductivityandshortercoolingtime,weemployedanaluminummoldthatisalsoeasiertomanufactureandmodify.Inaddition,withappropriatethermalinsulationandacoolingsystem,theproblemofdimensionalvariationcausedbythermalexpansioncanbecontrolledandanaluminummoldcanbeusedasamoreeconomicaltoolforthereplicationprocess.Themoldedcomponentcanberemovedfromthemoldmanuallyorbyusingtheejectionsystem.Unliketheprocessesdescribedinthepreviousliterature,asiliconwaferthatservesasthemoldinsertisplacedinthemoldcavity.Usingsiliconwaferasmoldinserthastheadvantageofshortturnaroundtime.Inaddition,thewearofasiliconmoldinsertismuchsmallerascomparedtoatraditionalnickeltool18.However,asiliconmoldinsertismorebrittlethananickelone.Toavoidthebreakageofthewaferduringthemoldingprocess,theedgeofsiliconwafershouldmatchthecavityboundary.Agapbetweenthemoldinsertandcavitycanallowpolymermelttosolidifywithin,whichwouldeventuallyliftthewaferFigure4.ArburgAllrounder221M350-75injectionmoldingmachine.fromthecavityduringmoldopeningandresultinthebreakageofthewafer.4.ExperimentsAnArburgAllrounder221M350-75conventionalinjectionmoldingmachine,asshowninfigure4,withasinglecavity,coldrunnermoldisemployed.ThematerialusedformoldtrialsisBayerMakrolon2205polycarbonate(PC)thermoplasticresin.Becauseofitsexcellentoptical,chemicalandmechanicalproperties,PCcanbeusedinapplicationssuchasmedicalinstruments,biochemicalsensorsanddatastoragesystems.Thepolymerisinjectedintothemoldcavityatapressurerangingfrom40to50MPa.Themelttemperatureinthefeedingzoneismaintainedatabout300C.Themoldtemperatureiscontrolledbyaheaterandmaintainedatatemperaturelowerthan200C.Thecycletimeofthemoldingprocessis65s,andpolymermeltandmoldareallowedtocooldownfor30safterthefillingstage.Figure5showsthetypicalpressureversustimeandcorrespondingflowrateversustimerelationshipofthemoldingprocess.Forthemicro-moldingprocess,injectionpressure,moldtemperatureand417Y-CSuetalHolding0t1t2t30t1t2t3InjectionPackingHoldingInjectionPackingTimeTimePressureFlowrateFigure5.Typicalpressureversustimeandcorrespondingflowrateversustimerelationshipsduringtheinjectionmoldingprocess.Figure6.SEMmicrographofmoldingresults(injectionpressure45MPa,moldtemperature25C).injectionvelocityarerecognizedasthedrivingparameters.Thedepth-to-openingratiosofmoldedmicrostructuresareusedtomeasurethequalityofmoldingresults.Ahigherdepth-to-openingratiomeansbetterfillingstatusandmoldingquality.Thepresenceofvoidsplaysamajorroleinthemoldingprocess.Preheatingofthepolymerpriortothemoldingprocessreducesthechancesofentrapmentofvoids.Conventionalventingmethodsaredifficulttouseformicroinjectionmoldingduetothehighpossibilityofundesiredstructuralchangesinthemoldedcomponent.Hence,anevacuatedmoldisrecommendedtoobtainagoodreplicationprocess.Inthefirstmoldtrial,ordinaryinjectionmoldingparameterswereusedandnoadditionalcontrolunitwasactivated.Themoldingresultisshowninfigure6.Ascanbeseen,themoldingresultshaveasmalldepth-to-openingratiowhichmeanspolymermeltcannotfillthemicromold-cavity.Inthissituation,polymericmicrostructurescannotbesuccessfullyreplicated.Beforedoingmoremoldtrialstoimprovethemoldingresults,asimulationtoolwasusedtounderstandtheflowbehaviorofpolymermeltinthemicromold-cavityforfeasiblemodificationstoimprovethemoldingresults.5.SimulationItiswellknownthatcomputer-aidedengineering(CAE)canimprovethetrial-and-errortechniques,andcomputermodelscanbereliedupontopredictflowbehaviorandmoldresults.Ideally,CAEanalysisprovidesinsightthatisusefulindesigningparts,moldsandmoldingprocesses.ByusingCAEanalysistoiterateandevaluatealternativedesignsandmaterials,engineeringknow-howintheformofdesignguidelinescanbeestablishedrelativelyfastandcost-effectively.TheCAEsoftwareC-MOLDdevelopedbyACTechnologyisemployedasthenumericalcomputationtool.ThemoldfillingprocessismodeledbytheGHSmodeldescribedintheprevioussection.Thenumericalsolutionsarebasedonahybridfiniteelement/finitedifferencemethodtosolveforthepressure,flowandtemperaturefieldsandacontrolvolumemethodtotrackmovingmeltfronts.Afiniteelementmeshisusedtoapproximatethecircular-shapebaseplatewithconvexmicrostructuresononesurface,asshowninfigure7.Thisfiniteelementmode