Effect of Thermal Aging on Board Level Drop Reliab

EffectofThermalAgingonBoardLevelDropReliabilityforPb-freeBGAPackagesTz-ChengChiu,KejunZeng,RogerStiermanandDarvinEdwards1,KazuakiAno21TexasInstruments,Inc.13536NorthCentralExpressway,MS940,Dallas,TX75243Phone:[phone]Email:[e-mail]2TexasInstrumentsJapan,Ltd.4260AzaTakao,OazaKawasaki,Hiji-machi,Hayami-gun,Oita,879-15,JapanAbstractThedriveforPb-freesoldersinthemicroelectronicsindustrypresentsseveralnewreliabilitychallenges.Examplesincludepackagecompatibilitywithhigherprocesstemperatures,newsoldercompoundfailuremechanisms,andtheselectionoftheproperPb-freealloytomaximizeproductlifetime.InadditiontothechallengesposedbythePb-freematerialconversion,themigrationofmarketfocusfromdesktopcomputingtoportableapplicationsischangingthecriticalsystemfailuremodeofinterestfromconventionaltemperaturecycling(T/C)inducedsolderfatigueopenstodropimpactinducedsolderjointfracture.Inthispaperastudywasconductedtoinvestigatetheinfluenceofintermetalliccompound(IMC)growthonthesolderjointreliabilityofPb-freeballgridarray(BGA)packagesunderdroploadingconditions.Thermalagingathomologoustemperaturesbetween0.76and0.91withmicrostructuralanalysiswasconductedtoanalyzethesolidphaseIMCgrowthatthesoldertoBGApadinterface.Componentlevelballshearandpulltestswerealsoconductedtoinvestigatetheagingeffectonsolderjointstrength.AkeyfindingfromthisworkisthatKirkendallvoidsformedatthebulksoldertopackagebareCupadinterfaceunderrelativelow100°Caging.Voidformationandcoalesceisshowntobethedominantmechanismforsolderjointstrengthandboardleveldropreliabilitydegradation.IntroductionTwoofthemainfailuremodesforpackagetosystemsolderjointinterconnectarelowcyclesolderfatigueduetothermalexcursionsandthebrittlefracturecausedbystaticanddynamicloadssuchasimpactshock,vibration,andbending.Traditionally,thecriticalfailuremodeofinteresthasbeensolderfatiguewhichisdrivenbythecoefficientofthermalexpansion(CTE)mismatchbetweenthecomponentandmotherboard.However,duetothecontinuouspushfordeviceminiaturizationandnewapplicationsinportableelectronicswhicharesubjectedtohumanclumsiness,solderjointfailureunderdroporshockconditionhasnowbecomethecriticalreliabilityissue.InsteadofductilefatiguecrackingthroughthebulksolderobservedinT/Cloading,thetypicalfailuremodeforasolderjointunderdropimpactisbrittlefractureattheintermetalliccompound(IMC)betweentheBGApadandbulksolderjoint.Inordertopredictthesolderjointreliabilityunderdropconditions,itisimportanttounderstandthemechanicalconstitutivebehaviorofthesystemunderhighstrainratetoproperlycharacterizethedrivingforceforjointfracture.Furthermore,itisalsocriticaltocharacterizethejointtoughnessagainstshockloading.ThelatertaskiscomplicatedbythenatureofmicrostructureevolutionovertimeattheIMCregion.ThevolumetricchangeresultingfromsolidphasediffusionandIMCformationcauseshighresidualstresseslocally.Inaddition,theIMC’stendtobebrittleinnature.Consequently,theIMCgrowthprocessleadstoprogressivedamageandweakeningoftheinterfacewhichisstronglydependentontheagingtime.SeveralstudieshavebeenconductedtoinvestigatetheIMCbetweenCuandSn-Pbsolders.IthasbeenshownthatCu6Sn5formswhenCureactswitheutectic63Sn-37Pbat283°C[1].Ontheotherhand,whenCureactswith5Sn-95Pbat400°C,Cu3SnformsinsteadofCu6Sn5[2].ThermodynamicmodelingshowedthataftertheformationofafirstIMClayer,eitherCu6Sn5orCu3Snwillalsoformattheinterfacebysolidstatediffusion[3-5].Inthisstudy,adesignofexperimentswasperformedtoinvestigatetheeffectofIMCgrowthonboardleveldropreliability.ThetestsampleswereassembledtodroptestboardsandsubjectedtothermalagingtosimulateIMCgrowthunderapplicationconditions.Droptestingwasthenconductedtoestablishtherelationshipbetweendropreliabilityandagingtime.Mechanicalcross-sectioningandmicrostructuralanalysiswerealsoconductedonthermallyagedpackagestostudytheIMCevolutionatBGACupadtobulksolderinterface.CorrelationbetweenIMCgrowthanddropreliabilityevolutionwasthendeterminedfromthesetwoanalyses.Inadditiontothesetwoexperiments,ballshearandpulltestswereconductedtostudythesolderjointstrengthvariationunderagingconditions.Thepurposeofthepullandsheartestswastoestablishacomponentleveltestmethodthatwouldcorrelatetotheboardleveldropreliability.ExperimentSetupandProceduresThetestvehicleconsideredwasachipscalepackagewithbareCuBGApadsandSnAgCusolderball.TheexperimentmatrixandsampleagingconditionsaredescribedinTable1.Thedroptestpackagesweresingle-sidemountedonadroptestprintedwiringboard.Thetestboardwas1mmthickandhadanorganicsurfaceprotectant(OSP)BGApadfinish.AschematicofthetestboardlayoutisshowninFig.1.Thermalagingofthetestboardassemblywasperformedat125°Cfor3,10,20and40daystoinducesolidstateIMCgrowthforsolderjoints.HighertemperatureagingwasnotconsideredduetoPWBFR-4hightemperaturecompatibilityissues.0-7803-8365-6/04/$20.00©2004IEEE2004ElectronicComponentsandTechnologyConference1256Fortheboardleveldropexperiment,thetestboardwasscrew-mountedtothedroptableat4cornerswiththetestpackagesfacingdownwards.TheshockloadingforthetestwassetaccordingtoJESD22-B104-Abyadjustingthedropheight.Theshockpulsefromthedropimpactwasoptimized