ESD培训资料(Hitachi)

HitachiConfidential1February12,2009CDK©2004HitachiGlobalStorageTechnologiesHSPCTrainingSeriesHSPC培训系列Part3:TheContextClass13:UnderstandingESD第13课：理解ESDSamLuoChrisKeenerTranslatedbyYuanXiaoyuHitachiConfidential2February12,2009CDK©2004HitachiGlobalStorageTechnologiesUnderstandingESD:Outline概述ZAP!!!„ThermalStability热稳定性„Factsaboutthermalstability&ESD热稳定性与ESD事例„HumanBodyModelHB模型„SymptomsintheheadasafunctionofESDmagnitudeESD脉冲强度对磁头的效应„EffectofvaryingESDpulseduration变化ESD脉冲持续时间的效应„PinnedLayerFlipping钉扎层翻转„Quiz问题HitachiConfidential3February12,2009CDK©2004HitachiGlobalStorageTechnologiesThermalStability热稳定性¾Thermalinstabilityisamplitudedecaythatoccurswhenheadsarestressedathightemperatures热不稳定性指磁头在高温热应力下幅值衰减的现象¾Thepredominantmechanismforthermalinstabilityisdiffusionofatomswithinthesensor热不稳定性主要来自读头内部原子扩散效应¾Diffusionistemperaturedriven:热扩散¾∆T=T–Tambient~I2R¾whereTisthemaximumtemperatureatthestripecenter此处T指读头中心处的最高温度¾NotethatLifetimeisextremelysensitivetobiasvoltage注意读头寿命对偏压极为敏感¾GMR5.0hasbetterthermalstabilitythanGMR4.2相对于GMR4.2,GMR5.0具有更好的热稳定性dItcanrunhotter能工作在更高的温度dBiasVoltagecanbeincreased可提高偏压dPerformanceisenhancedevenwithsimilar∆R/R相同∆R/R条件下，综合性能增强¾ESDisthermaldecayonashorttimescale/ESD是瞬时热衰减¾Humanbodymodel~150nsec人体~150纳秒¾Machinemodel~10nsec:equilibriumTnotreached机械~10纳秒：未达到平衡温度()273+∝TkEaeLifetimeHitachiConfidential4February12,2009CDK©2004HitachiGlobalStorageTechnologiesThermalStability:GeneralObservations热稳定性的一般现象„Thermalstresscausesheadstochangein2ways热应力对磁头的影响有两种方式：•Amplitudedecay(irreversible)幅值衰减（不可逆）–Causedbydiffusionofatomsinthesensor来自读头内的原子扩散–FollowsArrheniusLawwithactivationenergiesconsistentwithatomicdiffusion依从Arrhenius法则：激活能源于原子扩散–BiasvoltageinHDDisdesignedtomaintainamaximumsensortemperatureé110˚C设计时，硬盘中的偏压维持最大读头温度é110˚C–Wewantheadstomaintainhighamplitudefor100,000hoursofcontinuoususe期望读头连续使用100,000中保持高幅值•Amplitudespreading(reversible)幅值分散（可逆）–Mostlikelycausedbysettlingofthefreelayerorientationintoequilibrium极可能来自于自由层磁化方向–Thishappensrelativelyquicklyevenatlowstresstemperatures(HDDoperatingtemperaturesandtypicalusagetimes,maybeontheorderof100-1000hours)即便在低的热应力下也可能发生此现象(HDD长时间运行会产生热量）–Amplitudecanincreaseordecrease,dependingonthehead’sbiaspointanddetailsofitsstressandmechanicaldefect(e.g.,grainboundary)distribution幅值可能增大也可能减小，依赖于磁头的偏压点和其应力机械缺陷（如：晶粒边界）的分布HitachiConfidential5February12,2009CDK©2004HitachiGlobalStorageTechnologiesESD:GeneralObservationsESD的一般现象„ESD•Typicalheadsaredamagedin100nsecbycurrentsof13mA(20VHBM)一般磁头在电流13mA(20VHBM)下100nsec便被损坏•HBMthresholdvariesinverselywithresistance(asstripeheightischanged)HBM门限值与电阻成反比（当SH改变时）–ThismeansthattheESDthresholdisapproximatelyaconstantVoltageforallheadsofagivendesign(GMRtype,MRW,gap)此意味着对于给定设计（GMR类型，读宽，间隙），ESD电压门限值对所有磁头是一个常数•ESDthresholdacrosssensoritselfvarieslinearlywithMRW穿过读头本身的ESD门限值与读宽变化成线性关系–However,leadresistanceisasignificantcomponentoftotalresistance然而，引导层的电阻占总电阻的大部分–SoESDVoltagethresholdvariesroughlyasMRW0.65(empiricalfunction)因此，ESD电压门限值粗略随MRW0.65变化（经验公式）HitachiConfidential6February12,2009CDK©2004HitachiGlobalStorageTechnologiesHumanBodyModelHB模型„HumanBodyModelisastandardizedESDexperimentHB模型是标准的ESD实验•Theheadisplacedinserieswith1.5kΩ磁头与一系列1.5kΩ电阻相连•HBMVoltageisdischargedfroma100pFcapacitorthroughtheresistorandtheheadHB模型电压来自100pf电容器释电流过电阻和磁头•HBMVoltageisactuallyproportionaltothecurrentthroughthehead实际HBM电压正比于流过磁头的电流–Headresistanceismuchlessthan1.5kOhm磁头电阻远小于1.5kOhm–Considerheadresistanceconstant50ΩÆCurrent=VHBM/1.55kΩ假设磁头电阻为常数50ΩÆ电流=VHBM/1.55kΩ–时间常数Timeconstantτ=1.55kΩx100pF=155nsec„Thisisalowfrequencypulsethatallowstheheadtonearlyreachthermalequilibrium此为低频脉冲允许磁头几乎接近热平衡HitachiConfidential7February12,2009CDK©2004HitachiGlobalStorageTechnologiesWhathappenstotheheadasESDbecomesmoresevere严重ESD对磁头的影响„TypicalamplitudeandresistanceprofileasafunctionofESD典型幅值与电阻随ESD电压增强的变化图HBMVoltage0%20%40%60%80%100%120%N.Amp-50510152025DeltaRScopeoftheworkthisworkotherworksSmallpulsesNodamageNoproblem小脉冲没问题SevereDamageDeadheadAmp=0LargeResistancem严重ESD磁头损坏幅值为0电阻增大MildESDInstabilityAmpm中等ESD不稳定性幅值增大PermanentESDdamageAmpoResist.mSlightly永久ESD磁头受损幅值降低电阻增大SamLuopaperSEMFA:Invisible扫描电镜失效分析：未见ESD|Seespotatsensor读头存在ESDHitachiConfidential8February12,2009CDK©2004HitachiGlobalStorageTechnologiesWhataDeadHeadLooksLike受损读头的外观„Onlydeadheads(zeroamp&highresistance)showvisibleevidenceofphysicaldamage仅仅读头（0幅值和高电阻）表现出可见的物理缺陷DeadHead死头(RHopen)—Sensormelted读头融化Almostdeadhead(RHopen)Goodhead好头HitachiConfidential9February12,2009CDK©2004HitachiGlobalStorageTechnologiesDistributionofESDpulsesproducesheadswithdifferentsymptoms不同征兆的ESD磁头的分布„AssumeahypotheticaldistributionofESDpulses假设ESD脉冲的近似分布„Thelongtailofhigh-currentpulsesproducesarangeofsymptoms高电流脉冲部分„Thusweseesomeheadswith:•Highamplitude(manyofwhichareunstable)高幅值(许多为不稳定磁头）•Negativeamplitude(flipped,discussedafewpageslater)负幅值（钉扎层翻转）•Lowamplitude低的幅值•Resistance(RH)veryhigh(open),deadheads)极高的电阻（磁头损坏）Pulseamplitude脉冲幅值NumberofpulsesProducedinprocessHitachiConfidential10February12,2009CDK©2004HitachiGlobalStorageTechnol