搜索
上海交通大学硕士学位论文CopyExactly模式在先进芯片制造行业中成功案例分析姓名：张青申请学位级别：硕士专业：项目管理指导教师：吴祖育;桂军20071201ICopyExactlyCopyExactly,ABSTRACTIICASESTUDYONSUCCESSFUL‘COPYEXACTLY’METHODOLOGYINCHIPMANUFACTURINGINDUSTRIALABSTRACTWiththedevelopmentofscienceandtechnology,thenewproductandprocessarecreatedcontinuouslyinthesemiconductorindustrial.UsuallythoseproductswillbetransferredfromtheR&DtotheHVM(HighVolumeManufacture)siteorfromoneHVMsitetoanothernewone,andmeanwhileensuredtheycanbeputonthemarketassoonaspossibleandbringtheprofittothecompany.Allkindsofunexpectedissuewillbeencounteredduringthetechnologytransferringandequipmenttesting.Sometimes,itmaytakeaquitelongtimetoreachthetargetedyield,evenifthenewproductionlinewascopiedfromtheexistedone.Thiswillnotonlycausetheunnecessaryprofitlossforthecompanybutalsoimpacttheeffectiveexecutionofthestrategy.Isthereanymethodologythatcanbefollowedtoshortenthistechnologytransferringperiodandmovetothehighvolumemanufacturingstagewiththetargetedyieldandprocessstabilityguaranteed?Theanswerisyes.AsthebiggestCPUmanufacturerintheworld,IntelnotonlyhasthestrongR&Dcapabilitybutalsohastheadvancedmanufacturetechnology.ThenewproductusuallycanbeproducedmassivelywithhighqualityandoccupythemarketinaquickmannerthankstoitsCopyExactlystrategy.ThispapertendstoprovideapracticalreferencethroughsummarizingtheCopyExactlymethodologyonthebasisofrealexperiences.AndalsoitwillhaveadiscussiononhowtomakesuretheCopyExactlymethodcanbecarriedouteffectively.Hopethispaperwillofferavaluablereferencetosemiconductormanufacturingindustryandtheapproachdiscussedherecouldequallybeemployedinotherindustries,wherethetechnologyiscomplexandhasmanyinteractingvariableseffectingtheendresult.ABSTRACTIIIKeywords:CopyExactly,SequentialTest,MatchingTest1200712422007124200712411.11CopyExactly1996221.0um#1#2#31#1#2#30.5um2311umFig.1Intel1umtechnologyyield20.5umFig.20.5umtechnologyyieldwithCopyExactlyIntelTechnologyJournal19981.2SixSigmaTotalQualityManagement45ME/5ME151,1.362.12.1.133Fig.3Samplinglevellot(WTW)(WIW)RUN3Lot25LotLotAvg.=1535WTW=std.dev.(wafer1mean,wafer2mean,wafer3mean)7WIW=avg{(std.dev.5pointsonwafer1),(std.dev.5pointsonwafer2),(std.dev.5pointsonwafer3)}wafer#mean5#avgstd.dev.5pointsonwafer11562.462.863.263.664.005101520250.000.150.300.450.6005101520250.300.450.600.750.9005101520254Fig.4Runstatisticsondifferentsamplinglevel2.1.251Fig.5Thereferencetoolandnewtooloutputdiagram1862Fig.6Thereferencetoolandnewtooloutputdiagram2,Η0:µnew=µrefVsΗΑ:µnew≠µrefΗ0:σnew=σrefVsΗΑ:σnew≠σrefΗ0ΗΑµnewσnewµrefσrefδδ7Fig.7Thereferencetoolandnewtoolmeantestdiagram9•δ=µref-µnewδ1.5σrefλ8Fig.8Thereferencetoolandnewtoolvariabilitytestdiagram•λ=σnew/σrefλ2δλΗ0Η0(β)(α)101αnewxµµNEWREF=α/2α/29αFig.9Theαerrordiagramxnewxnew2β10δ1110βFig.10Theβerrordiagramxnewαβαβ≤0.052.1.(NormalDistribution)4453011abcdδαβ1211Fig.11Thedistributionshapeofdifferentsampleaveragen55=nx1n1xu2n1251030121Fig.12Thecenterlimittheoremdiagram1-3.0-2.0-1.0.01.02.03.0-3.0-2.0-1.0.0.51.02.03.0n=5x-3.0-2.0-1.0.0.51.02.03.0n=30x-3.0-2.0-1.0.0.51.02.03.0n=10x(a)(b)(c)(d)13132Fig.13Thecenterlimittheoremdiagram2132.1.1414Fig.14Thecentrallimittheoremdiagramwhenmatchingµnew=µrefn01234567891-3.-2.-2.-1.-1.-0.0.0.1.1.2.2.3.--µNEW=µREFxNEW14sequentialtestmaximumlikelihood15012345678910-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.0-----β=5%δ15Fig.15Thecentrallimittheoremdiagramwhenun-matchingµnew-µref=δδµrefδnxnewxnewµnew-µrefδxnewµnew-µrefδ2.2µrefσrefδ152.2.116δµnew-µref=δ16Fig.16SequentialmeantestdiagramΗ0:µnew-µref=0ΗΑ:|µnew-µref|=δδ1.5σαβ2.2.2λσnew/σref=λ1617Fig.17SequentialvariabilitytestdiagramΗ0:σnew/σref=1ΗΑ:σnew/σref=λλ22.2.3(ConfidenceInterval)confidencelevel18Fig.18Confidenceleveldiagram100(1-α)%17σσ1-sampleconfidenceintervalxznnew±×−()12ασ12.2.4Honewref:µµ=vs.Hanewref:µµ≠t-2txnewrefSnrunrunnew=−−µ()2nxnewµrefSrunrunnew−()nntttσrun-runsrun-runover-rejectsZ-318Zxnewrefnrunrunref=−−µσ()3n19aba(b)19Fig.19Distributionofmatchandun-matchtool2020Fig.20Sequentialtestdiagramαβδµσ012345678910-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.0-----β=5%012345678910-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.0----------α=5%195δδδconfidenceintervalµδref±2121Fig.21Sequentialtest-continuetestdiagram∆95%(-7.3,106.3)δ(-123,+123)95%(493,606)δ500+/-123λ42220σσλ)()(refruntorunnewruntorun−−−−=4816243240715233139780840900960102010801140120022Fig.22Variabilitytestdiagramλ=2Honewref:σσ≤VsHanewref:σλσChi-squaretest5χσ2221=−−−()()()nsrunrunnewrunrunref5nsrunrunnew−()2σrunrunref−()2nnχ2χ2212323Fig.23Sequentialvariabilitytestdiagramσrun-run(new)αβλσrun-runσrun-run(new)σrun-run(new)53022ProcessMatchingHandbookRevision1.0forATM.Intel.1999.233.12424,,?nMax_Ndelta?,,?(JET)?24Fig.24Equipmentqualificationprocedure253.23.2.11234133•αβ0.05•(Cpk)4(δ,λ)•4Processmonitorcapabilityαriskβriskδ=µref-µnewλ=σσnewrefMax#RunsCapableCpk≥1.3*.05.051.5*σrun-run28NotcapableCpk≤1.3*.05.051.0*σrun-run216Nospeclimits.05.051.0to1.5*σrun-run28to1626•2•2••()JET,3••1.2.3.4.)5.3.2.2PWP1•RunAvgWafer-Waferstd.dev.Within-Waferstd.dev.•27UCLLCLCL25-100102030050100150200250UCLCL