11 Part-to-part and lot-to-lot

TRAD,Tests&Radiations07/03/2017–ESA/CNESPresentationDaysPart-to-partandlot-to-lotvariabilitystudyofTIDeffectsinbipolarlineardevicesJ.Guillermin,N.Sukhaseum,A.Varotsou,P.GarciaandC.PoiveyRadecs2016Proceedings[PC-2]TRAD,Tests&Radiations2IntroductionDefinitionofRadiationDesignMargin(RDM)requirementsforRadiationVerificationTesting(RVT)onflightlotsunderTotalIonizingDose(TID)CriticalforlinearbipolardevicesLikelytoshowpart-to-partandlot-to-lotvariationinTIDsensitivityInvestigationonthewithin-one-lotandinter-lotvariabilityonthreedifferentreferencesExperimentalcharacterizationsDataanalysisOne-sidedtolerancelimit(KTL)methodisusuallyconsideredforTIDtestingAppliedtoourtestdataComparedtoaMaximumLikelihoodRatio(MLR)methodwhichisinvestigatedandprovidesaninterestingaccuracy07/03/2017–ESA/CNESPresentationDaysTRAD,Tests&Radiations3DevicesUnderTestThreedifferentreferenceswereirradiatedThreelotsweretestedforeachreference30deviceswereirradiatedforeachlotCobalt60TIDtestDoserate210rad/hGAMRAYfacility(TRAD–Labège,France)07/03/2017–ESA/CNESPresentationDays0539A1136A1306ATexasInstruments5962R9950401VZALM124AWGGE245074GE334152GE337030STMicroelectronics-TL1431ACZ1052AAnalogDevices5962-3812801VGAAD584SH0125A0226AAnalogDevices5962R3812801VGAAD584SHLotsManufacturerPartnumberDeviceOp.Amp.Volt.Ref.TRAD,Tests&Radiations4StudyFrameworkRadiationhardnessassurancestandardsIntra-lotvariabilitytakenintoaccountwiththe3-sigmaapproachOveratestsample,theelectronicdeviceparameterdriftconsideredfortheworst-caseanalysisismean±3sigmaInter-lotvariabilitytakenintoaccountwithperiodictestingRVTperiodicitydependsonspaceprojectprime-contractorInmostofthecases,the3-sigmaapproachisrepresentativeenoughandwelladpated(whentheassociatedriskisaccepted)totakeintoaccounttheintra-lotvariabilityforaspaceprojectThisstudyfocusesonthefewothercases…Inthispresentation,3worst-cases(1pertestedreference)areshownInvestigatetheapplicationofanotherstatisticalmethodology(MLR)thatcouldbeappliedwhenahigherconfidencelevelisrequired07/03/2017–ESA/CNESPresentationDaysTRAD,Tests&Radiations5StatisticalAnalysisFirstlevelanalysis3-sigmaapproachKTL=2.742forn=5(C=90,P=90)[1]SecondlevelanalysisMaximumLikelihoodRatioTestdatafittedtoanormallaw2-levelanalysisbasedontheuseofexperimentalparameters‘m’and‘s’asdefinedabove,aswellasstatisticalparameters‘µ’and‘σ’07/03/2017–ESA/CNESPresentationDays∑==niixnm11∑=−−=niimxns12)(11()−−=221exp21,,σµπσσµxxfnmExperimentalmean‘m’Standarddeviation‘s’Normallaw(2parametersµandσ)TRAD,Tests&Radiations6AD584SH(AD)AD584SHVrline1(%/V)lot0226A07/03/2017–ESA/CNESPresentationDays-1.0E-02-5.0E-030.0E+005.0E-031.0E-021.5E-022.0E-022.5E-023.0E-020102030405060708090100110120130140150160170180190200210220Dose(krad)Vrline1(%/V)mm±3sSpecification-1.0E-02-5.0E-030.0E+005.0E-031.0E-021.5E-022.0E-022.5E-023.0E-020102030405060708090100110120130140150160170180190200210220Dose(krad)Vrline1(%/V)mm±3sSpecification020406080100120140160180200220Allthelots0125A0226A1052AOutofspecificationTIDlevel(krad)Allcomponents5best5worst4best4worstrandomselection1randomselection2Mean±3sigmacalculatedoverthe30testeddevicesMean±3sigmacalculatedover5randomlyselecteddevicesEstimatedlotbehaviourfordifferentsubgroupsOut-of-specificationTIDlevelbasedonm+3slimitAlldevices,5bestorworst,4bestorworst,randomselections30testeddevices5randomselection(1lotplottedingrey)(1lotplottedingrey)TRAD,Tests&Radiations7TL1431ACZ(STM)TL1431ACZIref(µA)lotGE33415207/03/2017–ESA/CNESPresentationDaysMean±3sigmacalculatedoverthe30testeddevicesMean±3sigmacalculatedover5randomlyselecteddevicesEstimatedlotbehaviourfordifferentsubgroupsOut-of-specificationTIDlevelbasedonm+3slimitAlldevices,5bestorworst,4bestorworst,randomselections012345678910111213141516170102030405060708090100110120130140150160170180190200210220Dose(krad)Iref(µA)mm±3sSpecification30testeddevices01234567891011121314150102030405060708090100110120130140150160170180190200210220Dose(krad)Iref(µA)mm±3sSpecification5randomselection05101520AllthelotsGE245074GE334152GE337030OutofspecificationTIDlevel(krad)Allcomponents5best5worst4best4worstrandomselection1randomselection2(1lotplottedingrey)(1lotplottedingrey)TRAD,Tests&Radiations8LM124AWG(TI)LM124AWGVio3module3(mV)lot1136A07/03/2017–ESA/CNESPresentationDaysMean±3sigmacalculatedoverthe30testeddevicesMean±3sigmacalculatedover5randomlyselecteddevicesEstimatedlotbehaviourfordifferentsubgroupsOut-of-specificationTIDlevelbasedonm+3slimitAlldevices,5bestorworst,4bestorworst,randomselections020406080100120140160180200220Allthelots0539A1136A1306AOutofspecificationTIDlevel(krad)Allcomponents5best5worst4best4worstrandomselection1randomselection2-4-3.5-3-2.5-2-1.5-1-0.500.50102030405060708090100110120130140150160170180190200210220Dose(krad)Vio3module3(mV)mm±3sSpecification-4-3.5-3-2.5-2-1.5-1-0.500.50102030405060708090100110120130140150160170180190200210220Dose(krad)Vio3module3(mV)mm±3sSpecification30testeddevices5randomselection(1lotplottedingrey)(1lotplottedingrey)TRAD,Tests&Radiations9SecondLevelAnalysisSecondlevelanalysisNormallawparametersoptimizedwithmax.likelihoodThemaximumlikelihoodisalsousedtodefineaconfidenceintervalforthenormallawparameters07/03/2017–ESA/CNESPresentationDays()∏==niixfL1,,σµLikelihoodformula()LLmaxmax=Maximumlikelihood()2,121ln2maxαχ−−≥LLConfidenceintervalformula(maximumlikelihoodlog-ratio)01234567896789101112131415MeasuredvalueFrequencyTestdataNormallaw(10;2)ConfidenceintervalfortheoptimizeddatameanRepresentationofthenormalfitconfidenceintervalcalculation(frequencyplottedasafunctionofthemeasuredvaluefortheelectricalparameter)TRAD,Tests&Radiations10AD584SH(AD)AD584SHVrline1(%/V)lot0226A07/03/2017–ESA/CNESPresentationDaysMax.LikelihoodRatiomethodwiththe30devicesinonelotComparisonofMLRconfidenceintervaland3-sigmacalculationwith5devicesNumberofsigmatocovertheconfidenceintervaldelimitedby«90%+1σ»MLRmethod,10devicesselectedineachoneofthe3testedlotsTotakeintoaccounttheinter-lotvariability-1.0E-02-5.0E-030.0E+005.0E-031.0E-021.5E-022.0E-022.5E-023.0E-023.5E-024.0E-020102030405060708090100110120130140150160170180190200210220Dose(krad)Vrline1(%/V)Specificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3s-1.0E-02-5.0E-030.0E+005.0E-031.0E-021.5E-022.0E-022.5E-023.0E-023.5E-020102030405060708090100110120130140150160170180190200210220Dose(krad)Vrline1(%/V)Specificationµµmax+1σmax/µmin-1σminm±3smm±10sm±15sMLRconfidenceintervalComparison-1.0E-02-5.0E-030.0E+005.0E-031.0E-021.5E-022.0E-022.5E-023.0E-020102030405060708090100110120130140150160170180190200210220Dose(krad)Vrline1(%/V)SelectedDataSpecificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3s(1lotplottedingrey)(3lotsplottedingrey,30selecteddevicesinblue)10devicesineachlot(greencurves)TRAD,Tests&Radiations1107/03/2017–ESA/CNESPresentationDaysMax.LikelihoodRatiomethodwiththe30devicesinonelotComparisonofMLRconfidenceintervaland3-sigmacalculationwith5devicesNumberofsigmatocovertheconfidenceintervaldelimitedby«90%+1σ»MLRmethod,10devicesselectedineachoneofthe3testedlotsTotakeintoaccounttheinter-lotvariabilityMLRconfidenceintervalComparison(1lotplottedingrey)(3lotsplottedingrey,30selecteddevicesinblue)TL1431ACZ(STM)TL1431ACZIref(µA)lotGE334152024681012141618200102030405060708090100110120130140150160170180190200210220Dose(krad)Iref(µA)SelectedDataSpecificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3s024681012141618200102030405060708090100110120130140150160170180190200210220Dose(krad)Iref(µA)Specificationµµmax+1σmax/µmin-1σminm±3smm±4sm±5s024681012141618200102030405060708090100110120130140150160170180190200210220Dose(krad)Iref(µA)Specificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3sMLRconfidenceinterval10devicesineachlotTRAD,Tests&Radiations1207/03/2017–ESA/CNESPresentationDaysMax.LikelihoodRatiomethodwiththe30devicesinonelotComparisonofMLRconfidenceintervaland3-sigmacalculationwith5devicesNumberofsigmatocovertheconfidenceintervaldelimitedby«90%+1σ»MLRmethod,10devicesselectedineachoneofthe3testedlotsTotakeintoaccounttheinter-lotvariabilityMLRconfidenceintervalComparisonLM124AWG(TI)LM124AWGVio3module3(mV)lot1136A-5.5-5-4.5-4-3.5-3-2.5-2-1.5-1-0.500.510102030405060708090100110120130140150160170180190200210220Dose(krad)Vio3module3(mV)Specificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3s-4.5-4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.00.51.00102030405060708090100110120130140150160170180190200210220Dose(krad)Vio3module3(mV)Specificationµµmax+1σmax/µmin-1σminm±3smm±5sm±6s-5.5-5-4.5-4-3.5-3-2.5-2-1.5-1-0.500.510102030405060708090100110120130140150160170180190200210220Dose(krad)Vio3module3(mV)SelectedDataSpecificationµµmax90%µmin90%µmax+3σmaxµmin-3σminµmax+1σmaxµmin-1σminm±3s(3lotsplottedingrey,30selecteddevicesinblue)10devicesineachlot(1lotplottedingrey)TRAD,Tests&Radiations1307/03/2017–ESA/CNESPresentationDaysMRLMethodAccuracyMultiplicationfactorappliedtosigma‘s’ateachTIDsteptocovertheconfidenceintervaldelimitedby«90%+1σ»fromMLRcalculation30devicelot(topgraph)5devicerandomselection(bottomgraph)IndicationoftheaccuracywiththeMLRmethodcorrespondingtoKTL=5(P>0.999forC=90)Theimpactofthelot/selectionhomogeneityintermsofdegradationcanbeobservedAD584Vrline1exhibitsatypicaldevicesLargermarginrequiredtocoverthelotbehaviourLM124Vio3morehomogeneousSmallermarginrequiredtocoverthelotbehaviour1223344556050100150200Dose(krad)CoefficientValueAD584Vrline10226AAD584Vrload10226AAD584Ios0226ATL1431IoffGE334152TL1431ZkaGE245074LM124CMRR1306ALM124Iio11306ALM124Vio31136A051015202530050100150200Dose(krad)CoefficientValueAD584Vrline10226AAD584Vrload10226AAD584Ios0226ATL1431IoffGE334152TL1431ZkaGE245074LM124CMRR1306ALM124Iio11306ALM124Vio31136ATRAD,Tests&Radiations14ConclusionTIDtestingwasperformedon3lotsfor3bipolardevicereferencesAnalysisofthelot-to-lotandwithin-one-lotvariabilityDataanalysisobjectivesCharacterizethelotcoveragebasedon5-devicesamplesizeforTIDtestingProposeandevaluateaMaximumLikelihoodRatio(MLR)methodbasedontheuseofconfidenceintervalsMLRisanaccuratewaytoestimatethelotbehaviourLargesamplesizeneeded(30devices)UseofaconfidenceintervalonthemeanμoftheelectricalparametermeasurementCanbeappliedtodatagroupscomposedwithdifferentlotsIncreasethesamplesize,bettercalculationaccuracyTakesintoaccountthelot-to-lotvariabilityPerspectiveCostproblemrelatedtothelargeneededsamplesizeAdaptationoftheMLRmethodtosmallersamplesizesbutkeepingsimilaraccuracy07/03/2017–ESA/CNESPresentationDaysTRAD,Tests&Radiations15References[1]MILITARYHANDBOOK,IonizingDoseandNeutronHardnessAssuranceGuidelinesforMicrocircuitsandSemiconductorDevices,MIL-HDBK-814,08February1994.[2]R.L.Pease,‘Totalionizingdoseeffectsinbipolardevicesandcircuits’,IEEETransactionsonNuclearScience,vol.50,no.3,pp.539–551,Jun.2003.[3]A.NamensonandI.Arimura,‘ALogicalMethodologyforDeterminingElectricalEndPointsforMulti-LotandMulti-ParameterData’,IEEETransactionsonNuclearScience,vol.34,no.6,pp.1726–1729,Dec.1987.[4]I.ArimuraandA.I.Namenson,‘HardnessAssuranceStatisticalMethodologyforSemiconductorDevices’,IEEETransactionsonNuclearScience,vol.30,no.6,pp.4322–4325,Dec.1983.[5]R.Ladbury,J.L.Gorelick,andS.S.McClure,‘StatisticalModelSelectionforTIDHardnessAssurance’,IEEETransactionsonNuclearScience,vol.56,no.6,pp.3354–3360,Dec.2009.[6]R.LadburyandJ.L.Gorelick,‘Statisticalmethodsforlargeflightlotsandultra-highreliabilityapplications’,IEEETransactionsonNuclearScience,vol.52,no.6,pp.2630–2637,Dec.2005.[7]R.Ladbury,‘StatisticalTechniquesforAnalyzingProcessor“Similarity”DatainTIDHardnessAssurance’,IEEETransactionsonNuclearScience,vol.57,no.6,pp.3432–3437,Dec.2010.[8]R.LadburyandB.Triggs,‘ABayesianApproachforTotalIonizingDoseHardnessAssurance’,IEEETransactionsonNuclearScience,vol.58,no.6,pp.3004–3010,Dec.2011.[9]R.L.LadburyandM.J.Campola,‘BayesianMethodsforBoundingSingle-EventRelatedRiskinLow-CostSatelliteMissions’,IEEETransactionsonNuclearScience,vol.60,no.6,pp.4464–4469,Dec.2013.[10]‘Part-to-partandlot-to-lotvariabilitystudyofTIDeffectsinbipolarlineardevices’ESAstudyreportref.TRAD/ESA/IR/VAR/NS/241115of29/11/2015.[11]J.Guillermin,N.Sukhaseum‘Part-to-partandlot-to-lotvariabilitystudyofTIDeffectsinbipolarlineardevices’,Radecs2016ProceedingsPC-207/03/2017–ESA/CNESPresentationDays