ESD培训资料(Hitachi).pdf

《ESD培训资料(Hitachi).pdf》由会员分享，可在线阅读，更多相关《ESD培训资料(Hitachi).pdf（23页珍藏版）》请在得力文库 - 分享文档赚钱的网站上搜索。

1、Hitachi Confidential1February 12,2009CDK 2004 Hitachi Global Storage TechnologiesHSPC Training SeriesHSPC 培训系列培训系列Part 3:The ContextClass 13:Understanding ESD 第第13课：理解课：理解ESDSam LuoChris KeenerTranslated by Yuan XiaoyuHitachi Confidential2February 12,2009CDK 2004 Hitachi Global Storage TechnologiesU

2、nderstanding ESD:Outline 概述概述ZAP!?Thermal Stability 热稳定性?Facts about thermal stability&ESD 热稳定性与ESD事例?Human Body Model HB模型?Symptoms in the head as a function of ESD magnitude ESD脉冲强度对磁头的效应?Effect of varying ESD pulse duration 变化ESD脉冲持续时间的效应?Pinned Layer Flipping 钉扎层翻转?Quiz 问题Hitachi Confidential3Fe

3、bruary 12,2009CDK 2004 Hitachi Global Storage TechnologiesThermal Stability 热稳定性热稳定性?Thermal instability is amplitude decay that occurs when heads are stressed at high temperatures 热不稳定性指磁头在高温热应力下幅值衰减的现象热不稳定性指磁头在高温热应力下幅值衰减的现象?The predominant mechanism for thermal instability is diffusion of atoms wi

4、thin the sensor 热不稳定性主要来自读头内部原子扩散效应热不稳定性主要来自读头内部原子扩散效应?Diffusion is temperature driven:热扩散热扩散?T=T Tambient I2R?where T is the maximum temperature at the stripe center 此处此处T指读头中心处的最高温度指读头中心处的最高温度?Note that Lifetime is extremely sensitive to bias voltage 注意读头寿命对偏压极为敏感注意读头寿命对偏压极为敏感?GMR 5.0 has better t

5、hermal stability than GMR 4.2 相对于相对于GMR4.2,GMR5.0具有更好的热稳定性具有更好的热稳定性d d It can run hotter 能工作在更高的温度能工作在更高的温度d d Bias Voltage can be increased 可提高偏压可提高偏压d d Performance is enhanced even with similar R/R 相同相同R/R条件下，综合性能增强条件下，综合性能增强?ESD is thermal decay on a short time scale/ESD是瞬时热衰减是瞬时热衰减?Human body m

6、odel 150 nsec 人体人体150纳秒纳秒?Machine model 10 nsec:equilibrium T not reached 机械机械10纳秒：未达到平衡温度纳秒：未达到平衡温度()273+TkEaeLifetimeHitachi Confidential4February 12,2009CDK 2004 Hitachi Global Storage TechnologiesThermal Stability:General Observations 热稳定性的一般现象热稳定性的一般现象?Thermal stress causes heads to change in 2

7、 ways 热应力对磁头的影响有两种方式：Amplitude decay(irreversible)幅值衰减（不可逆）Caused by diffusion of atoms in the sensor 来自读头内的原子扩散 Follows Arrhenius Law with activation energies consistent with atomic diffusion 依从Arrhenius 法则：激活能源于原子扩散 Bias voltage in HDD is designed to maintain a maximum sensor temperature 110 C 设计时

8、，硬盘中的偏压维持最大读头温度110 C We want heads to maintain high amplitude for 100,000 hours of continuous use 期望读头连续使用100,000 中保持高幅值Amplitude spreading(reversible)幅值分散（可逆）Most likely caused by settling of the free layer orientation into equilibrium 极可能来自于自由层磁化方向 This happens relatively quickly even at low stres

9、s temperatures(HDD operating temperatures and typical usage times,maybe on the order of 100-1000 hours)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）Amplitude can increase or decrease,depending on the heads bias point and details of its stress and mechanical defect(e.g.,grain boundary)distribution幅值可能增大也可能减小，依赖于

10、磁头的偏压点和其应力机械缺陷（如：晶粒边界）的分布Hitachi Confidential5February 12,2009CDK 2004 Hitachi Global Storage TechnologiesESD:General Observations ESD的一般现象的一般现象?ESDTypical heads are damaged in 100 nsec by currents of 13 mA(20 VHBM)一般磁头在电流13 mA(20 VHBM)下100 nsec便被损坏HBM threshold varies inversely with resistance(as s

11、tripe height is changed)HBM门限值与电阻成反比（当SH 改变时）This means that the ESD threshold is approximately a constant Voltage for all heads of a given design(GMR type,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ESD threshold across sensor itself varies linearly with MRW 穿过读头本身的ESD门限值与读宽变化成线性关系 However,le

12、ad resistance is a significant component of total resistance 然而，引导层的电阻占总电阻的大部分 So ESD Voltage threshold varies roughly as MRW0.65(empirical function)因此，ESD电压门限值粗略随MRW0.65变化（经验公式）Hitachi Confidential6February 12,2009CDK 2004 Hitachi Global Storage TechnologiesHuman Body Model HB模型模型?Human Body Model

13、is a standardized ESD experiment HB模型是标准的ESD实验The head is placed in series with 1.5 k 磁头与一系列1.5 k电阻相连HBM Voltage is discharged from a 100 pF capacitor through the resistor and the head HB模型电压来自100pf电容器释电流过电阻和磁头HBM Voltage is actually proportional to the current through the head 实际HBM电压正比于流过磁头的电流Head

14、 resistance is much less than 1.5 kOhm 磁头电阻远小于1.5 kOhmConsider head resistance constant 50?Current=VHBM/1.55k 假设磁头电阻为常数50?电流=VHBM/1.55k 时间常数Time constant =1.55 k x 100 pF=155 nsec?This is a low frequency pulse that allows the head to nearly reach thermal equilibrium 此为低频脉冲允许磁头几乎接近热平衡Hitachi Confiden

15、tial7February 12,2009CDK 2004 Hitachi Global Storage TechnologiesWhat happens to the head as ESD becomes more severe 严重严重ESD对磁头的影响对磁头的影响?Typical amplitude and resistance profile as a function of ESD 典型幅值与电阻随ESD电压增强的变化图HBM Voltage0%20%40%60%80%100%120%N.Amp-50510152025Delta RScope of the workthis wor

16、kother worksSmall pulsesNo damageNo problem小脉冲没问题小脉冲没问题SevereDamageDead headAmp=0Large Resistance m严重m严重ESD磁头损坏幅值为磁头损坏幅值为0电阻增大电阻增大Mild ESDInstabilityAmp m中等m中等ESD不稳定性幅值增大不稳定性幅值增大PermanentESD damageAmp o oResist.m mSlightly永久永久ESD磁头受损幅值降低电阻增大磁头受损幅值降低电阻增大Sam Luo paperSEM FA:Invisible 扫描电镜失效分析：未见扫描电镜失效

17、分析：未见ESD|See spot at sensor 读头存在读头存在ESDHitachi Confidential8February 12,2009CDK 2004 Hitachi Global Storage TechnologiesWhat a Dead Head Looks Like 受损读头的外观受损读头的外观?Only dead heads(zero amp&high resistance)show visible evidence of physical damage 仅仅读头（0幅值和高电阻）表现出可见的物理缺陷Dead Head死头死头(RH open)Sensor mel

18、ted读头融化读头融化Almost dead head(RH open)Good head好头好头Hitachi Confidential9February 12,2009CDK 2004 Hitachi Global Storage TechnologiesDistribution of ESD pulses produces heads with different symptoms 不同征兆的不同征兆的ESD磁头的分布磁头的分布?Assume a hypothetical distribution of ESD pulses 假设ESD脉冲的近似分布?The long tail of h

19、igh-current pulses produces a range of symptoms 高电流脉冲部分?Thus we see some heads with:High amplitude(many of which are unstable)高幅值(许多为不稳定磁头）Negative amplitude(flipped,discussed a few pages later)负幅值（钉扎层翻转）Low amplitude 低的幅值Resistance(RH)very high(open),dead heads)极高的电阻（磁头损坏）Pulse amplitude脉冲幅值Number

20、of pulses Produced in processHitachi Confidential10February 12,2009CDK 2004 Hitachi Global Storage TechnologiesESD evident in Rc change(mild ESD?neg amp w/o Rc)Note amplitude increase correlated with asym failures:ESD-induced instability电阻变化（中等电阻变化（中等ESD?负幅值，无?负幅值，无Rc）注意：幅值增加的同时不对称性变差；）注意：幅值增加的同时不对称

21、性变差；ESD引起不稳定引起不稳定Hitachi Confidential11February 12,2009CDK 2004 Hitachi Global Storage TechnologiesEffect of frequency 频率效应频率效应?Thermal equilibrium reached after about 100 nsec/100纳秒之后达到热平衡?Human Body Model pulse is 150 nsec/HB模型脉冲约150纳秒Thus the sensor reaches equilibrium 因此读头达到热平衡HBM is basically t

22、he same as long term thermal decay/HB模型基本等同于永久热衰减The physics is diffusion and it follows the Arhennius Law(thermally activated)物理机制来源于热激活（Arhennius Law)原子扩散?As a result,very short pulses(10 nsec)can be very large without causing the sensor to melt.因此，很大的瞬时脉冲(读头无原子位移High current pulses?large magnetic

23、 fields 高电流脉冲-大磁场Large magnetic fields?disrupt hard bias?magnetic instability in the free layer 大磁场-破坏硬偏置层-自由层磁不稳定性timeIncrease bias voltage nowTime constant 100 nsecNot actual dataVerbal descriptionfrom Sam LuoMR ResistanceHitachi Confidential12February 12,2009CDK 2004 Hitachi Global Storage Techno

24、logiesWhat happens to the head as ESD becomes more severe 严重严重ESD对磁头的影响对磁头的影响?Short pulses瞬时脉冲(10 nsec)?expands region of magnetic instability 磁不稳定性区域?Permanent ESD damage occurs at higher voltage(current)高电压（电流）导致永久ESD损坏Expanded range Instability more likely不稳定区域不稳定区域Conceptual DiagramSmall pulsesN

25、o damageNo problem小脉冲没问题小脉冲没问题SevereDamageDead headAmp=0Large Resistance m严重m严重ESD磁头损坏幅值为磁头损坏幅值为0电阻极大电阻极大Mild ESDInstabilityAmp m中等m中等ESD不稳定性幅值增大不稳定性幅值增大PermanentESD damageAmp o oResist.m mSlightly永久永久ESD磁头受损幅值降低电阻增大磁头受损幅值降低电阻增大Hitachi Confidential13February 12,2009CDK 2004 Hitachi Global Storage Te

26、chnologiesFailure Thresholds vs Pulse Lengths 失效极限失效极限 vs.脉冲长度脉冲长度?Davide Guarisco and Monica Lin Li,“ESD Sensitivity of GMR Heads at Variable Pulse Lengths,”EOS/ESD Symposium 00-322(3.B.2.1)(Fig.9).?Note that the thresholds for magnetic&physical damage separate as the time scale becomes shorter 值得注

27、意的是：当脉冲时间变短时，磁损失门限值和结构损失门限值分离Hitachi Confidential14February 12,2009CDK 2004 Hitachi Global Storage TechnologiesESD Pulse Shapes ESD脉冲波形脉冲波形?ESD pulse shapes for long and short pulses 长/短脉冲的ESD波形Experimental:Three different waveforms are used for the investigations:1.HBM/1zap2.1nS/10K zaps3.1nS/1zap-

28、50510152025303540Time(nS)-50510152025Current(mA)Data B1nS/1zap(+1000mV into Hd 31 with UQ Rcold 34.55)050100150200Time(nS)-2024681012Current(mA)Data BHBM Waveform-50510152025303540Time(nS)-6-5-4-3-2-1012Current(mA)Data B1nS/10K zaps(around 120mV)4 of the following 5 pp from Sam LuoHitachi Technical

29、Exchange Poster,2003Hitachi Confidential15February 12,2009CDK 2004 Hitachi Global Storage TechnologiesPhenomenon of Pinned Layer Reversal 钉扎层翻转现象钉扎层翻转现象?Transfer curves after ESD pulses of increasing magnitudes 磁头传递曲线随ESD脉冲信号增强的变化图?Some are reversed as the pinned layer flips 由于钉扎层翻转，部分信号反向Hitachi Co

30、nfidential16February 12,2009CDK 2004 Hitachi Global Storage TechnologiesExplanation of Flipping 解释钉扎层磁化方向翻转解释钉扎层磁化方向翻转?ESD current can oppose or support pinned layer orientation ESD电流产生与钉扎层磁化方向同向或反向的磁场?When the sensor is hot,Pinned layer coercivity decreases?easy to flip 读头受热，钉扎层矫顽力减小?易于翻转?Self-pinn

31、ed sensor is easier to flip also by mechanical stress(pinning by stress)自钉扎型读头在机械应力下易于翻转（钉扎效应来自应力）H opposes PL orientation外磁场与钉扎层磁场反向外磁场与钉扎层磁场反向H supports PL orientation外磁场与钉扎层磁场同向外磁场与钉扎层磁场同向Hitachi Confidential17February 12,2009CDK 2004 Hitachi Global Storage TechnologiesPolarity flipping 极性翻转极性翻转H

32、itachi Confidential18February 12,2009CDK 2004 Hitachi Global Storage TechnologiesFlipping and Damage for Many Heads 磁头极性翻转和损坏磁头极性翻转和损坏?Amp vs VHBM?Many hds flip before amp vanishes 许多磁头在幅值变成0之前发生极性翻转?Neg amp problem 负幅值问题?In HDD,neg amp results in pulse polarity reversal 在HDD中，负幅值来源于极性翻转Not expected

33、 by channel 不期望Bias point also can change so that channel is no longer optimized 偏置点也有可能改变，不利于信道优化Hitachi Confidential19February 12,2009CDK 2004 Hitachi Global Storage TechnologiesFlipping Depends on Head Design 极性翻转依赖于磁头设计极性翻转依赖于磁头设计?Flipping dependent on pulse polarity(supporting or against pinned

34、 layer orientation)极性翻转依赖于脉冲极性（与钉扎层方向相反或相同）Jump to:Sam Luo paperHitachi Confidential20February 12,2009CDK 2004 Hitachi Global Storage TechnologiesESD Quiz?Which direction current is more likely to cause flipping?哪个方向电流易于造成极性翻转？?What is the typical damage threshold for ESD current?Voltage across MR?E

35、SD电流损坏门限是多少？通过读头电压多大？10-13 mA,500 mV across sensor?Identify the regions of ESD damage on the diagram 标示出此图表中ESD损坏的区域If the pulse duration is reduced to 1 nsec,what happens to the range of ESD where instability is a problem?如果脉冲持续1纳秒，哪个范围会存在不稳定性问题？Link to answerCome to class to find outCome to classt

36、o find outH B M V o lta g e0%2 0%4 0%6 0%8 0%1 0 0%1 2 0%N.Amp-5051 01 52 02 5Delta RS c o p e o f th e w o r kth is w o rko th e r w o rk sHitachi Confidential21February 12,2009CDK 2004 Hitachi Global Storage TechnologiesABS View of TMR Sensor(TEM)?Insulator separates Hard bias from sensor?Thin ins

37、ulating barrier?Cr below and above HB layer Top of sensorTop of Cr in HB carShoulder/hornTop of Cr cap near junctionHard BiasCrBarrierInsulatorHitachi Confidential22February 12,2009CDK 2004 Hitachi Global Storage Technologies TMR ESD MechanismVdial down resistance:Less TMRTMRVdial up resistance:More

38、 TMRTMRPinhole typeHitachi Confidential23February 12,2009CDK 2004 Hitachi Global Storage TechnologiesESD Homework?Plot data to show the relationship between ESD and instabilitySelect parameters carefully to logically demonstrate your conclusionHints are given below.?ESD for TMR is characterized by:L

39、ow MR resistance 200 Ohms(dead heads)The mechanism is pinhole formation and expansion Pinholes are invasion of conducting shield material into the oxide barrier,making it conductiveLow DET LFTAA compared with slider Quasi amp?Instability is characterized by:High or low asymmetryUL_noiseSlider Quasi parameters:HF range,kink mD52