1、工艺可靠性设计方法与实践 华为公司工艺可靠性研究业务介绍,工艺基础研究部工艺可靠性研究组 20061124,Page 2,目录,工艺可靠性业务的需求分析 工艺可靠性业务行业分析 工艺可靠性业务组织流程 工艺可靠性业务技术内容 工艺可靠性技术在产品中的应用案例,Page 3,产品中工艺可靠性需求,High Performance Systems 1025年的使用寿命对工艺可靠性设计的要求； 高频、高速信号； 热可靠性问题 高复杂度单板； 封装技术的极限化发展带来的失效。Mid Range Performance Electronics 5年以上使用寿命； 混合组装； 无铅可靠性问题。 Hand-
2、held/Wireless Electronics 4年以上使用寿命；振动、冲击等特殊应用环境； 高密； 无铅可靠性问题； 低成本问题。,Page 4,工艺可靠性行业分析业界趋势分析： ECTC、SMTA 近年来在工艺可靠性领域的研究,1、不同表面处理条件下的脆性断裂问题； 2、快速可靠性评估方法研究； 3、无铅、密间距焊点可靠性研究； 4、腐蚀与迁移失效机理研究； 5、无损分析和故障诊断技术。,业界采用仿真、可靠性试验以及失效机理的研究方法对典型的板级互连可靠性问题进行研究，包括：,Page 5,Reliability Engineering,FEM,Page 6,业界常见可靠性业务流程,D
3、esign Verification,Failure analysis Root cause analysis,results,Predictions / failures,Manufacture & Market product,Field failures,Component count reliability analysis,Long term life test & group testing,Design “rules”,Product Qualification,Product design,results,Parametric distributionyield analysi
4、s,Demonstration of target reliability,SHIP,BETA,ALPHA,FEM,Page 7,工艺可靠性业务内容,工艺对象,环境,可靠性研究方法,结论,焊点 PCB/PTH 辅料 器件 ,基于POF仿真分析工具 可靠性试验标准 失效分析方法,工作热循环 工作机械应力 功率循环 组装热过程 组装机械应力 存储 运输 气候 环境气氛,设计方法可靠性规格DFR设计流程技术积累,Page 8,工艺可靠性研究内容,挑战:高密高速、小型化、低成本的市场需求新技术、新工艺、新材料的不断应用研发周期不断缩短新型法律法规出台/环保政策的推行工艺可靠性研究就是从工艺的角度，控制
5、工艺风险，保证产品的质量波动在允许的范围之内,工艺可靠性研究的主要课题：PCBA的环境适应性电气互连可靠性。PCB可靠性；连接器互连的可靠性。,Page 9,工艺可靠性分析测试和评估能力,无损分析技术微观组织结构分析表面形貌分析技术材料力学性能测试有限元模拟和仿真技术,Page 10,工艺可靠性设计与产品开发 有限元仿真分析技术,采用Ansys仿真技术对板级热、机械应力进行分析，评估各种工艺对象包括焊点、封装、PCB的热机械可靠性问题，并进行优化设计。,Page 11,工艺可靠性技术与产品开发 工艺失效分析,失效现象,失效模式,失效机理,根本原因,改进预防,芯片级 器件级 单板级 系统级 ,开
6、路 断路 时断时续 ,物理 化学 电学 热学 力学 材料学 ,可靠性设计 物料 制造过程 应用环境 人为损坏 ,DFM DFR 来料监控 ,Page 12,工艺可靠性设计与产品开发 新材料/新工艺/新封装工艺可靠性认证与评估,Influence of Ni-Sn-Cu ternary intermetallic compound on solder joint reliability,Page 14,PRESENTATION OUTLINE,Overview Experiments results Discussion Conclusion,Page 15,Overview,As we all
7、 know, Cu-Sn or Ni-Sn binary compound forms at the interface of pad and solder material after soldering.,Scallop Cu6Sn5 1,thin Ni3Sn4 2,Cu6Sn5,Page 16,Overview,But sometimes, things dont happened exactly like this.,Ni-Sn-Cu,Ni-Sn,SAC+NiAu BGA after soldering 3,Page 17,Overview,Somebody found solder
8、joint brittle fracture between two layer IMC and then attributed to ternary compound.,Ni,Cu,Ni-Sn,Ni-Sn-Cu,Solder joint open case in Qualcomm 4,Page 18,Experiments,Assembly: PCB（HASL / OSP / IAg / NiAu / ENIG）solder（SnPb eutectic / SAC ）BGA（NiAu）Etch: 5%HNO3 + alcohol to remove solder without cross-
9、section SEM & EDSSolder joint fracture case analysis,Page 19,Results,NiSn binary IMC formed in NiAu/SnPb/NiAu Solder joint,granular sand-like,6000x,Solder joint after solder removed,Ni-Sn only,NiSn binary IMC,Page 20,Disconnected rock formed upon sand in HASL/SnPb/NiAu,Rock like column lies upon san
10、d like granules,Just change PCB finishing from NiAu to HASL, things become different at BGA side?,Results,2000x,Page 21,5000x,Sand-like mainly still Ni-Sn（contains a little copper）; rock-like column mainly Contains Ni-Sn-Cu,Results,EDS indicates Ni-Sn-Cu ternary compound,They do exist!,Page 22,If co
11、pper exist in ether side of solder joint or inside of solder joint, ternary compoud will form upon nickel layer!,Results,They even exist in almost all kinds of solder joint!,Page 23,Results,It seems like this:,SnPb,NiAu,HASL/OSP,SAC,NiAu,Any, even NiAu,reflow,Ni-Sn(Cu),Ni-Sn-Cu,Page 24,Discussion,Wh
12、y copper can migrate so long distance? Copper is so active with much solubility in tin. Copper can form Cu-Sn IMC more easily and quickly than Ni-Sn.,At the touch second, great many copper diffuse into melt solder even exceed stable solubility,Due to unstable, some copper atom come back to decrease
13、energy,Copper migration,Touch melt solder,reflow,Cooling begin,IMC nucleate at interface and ternary IMC forms,Page 25,1000x,3000x,Hollow hexagon,Hollow hexagon Cu6Sn5,3000x,2000x,solid trigon,solid hexagon,Discussion,Variable morphology of ternary compound, sometimes like Cu6Sn5,Looks similar!,Page
14、 26,So the column can be identified as (CuNi)Sn5,(a),(b),Diffraction pattern of (a) Cu6Sn5 binary compound (b) column-like ternary compound found in solder joint 5,Discussion,The crystal structure of rock-like column is close to Cu6Sn5,Page 27,Field failure of BGA solder joint crack , found trigon a
15、nd hexagon rock at fracture surface without any etching,Discussion,Some solder joint failure cases show abnormal brittle fracture,1000x,Can ternary compound be so terrible?,Page 28,Top side BGA solder joint crack after wave solder, found more strong trigon and hexagon rock at fracture surface.,Some
16、solder joint failure cases show abnormal brittle fracture,Discussion,1000x,Is ternary compound a trouble maker?,Page 29,Discussion,It seems that crack develop between two IMC layers, Ni-Sn(Cu) and Ni-Sn-Cu！,Yes, there do exist micro cracks in IMC layer!,Ni-Sn(Cu),Ni-Sn-Cu,5000x,Page 30,Discussion,Fa
17、ilure joint ternary compound didnt show any difference to others,Butthere also exist evidence to prove his innocence.,Fracture didnt occur in IMC but at interface between solder and IMC,Page 31,Discussion,How about other companies ?,Page 32,Discussion,How about other companies ?,Interfacial failure
18、by S company,dewet failure by S company,Interfacial failure by UIC,Page 33,Conclusion,when copper exist inside or at ether side of solder joint, Ni-Sn-Cu ternary compound will form at interface between nickel and solder. The Ni-Sn-Cu ternary compound have two layers, Ni-Sn(Cu) and（CuNi）6Sn5, the for
19、mer is continuous and dense granular sand-like, and later is discontinuous column like. the bonding strength of the two layer compound might not good enough, and sometimes can lead to micro crack, but almost never cause solder joint open in this way. Some company found solder joint brittle fracture
20、and attributed it to ternary compound, but there do exist contrary opinions and many experiment result dont show any weakness of ternary compound. UIC plan to do a full research and we can participate.,Assessment on Reliability of 1.0mm pitch BGA Package double-side assembly,HUAWEI TECHNOLOGIES Co.,
21、 Ltd.,Process Reliability Research Group of AATC Board Design Engineering Dept. 20061124,Page 35,OUTLINE,Overview Reliability Test Sample Design Reliability Test results Simulation Analysis Conclusion,Page 36,Overview,Advantage: layout simple, Optimum routing, noise isolation, shorten line length Pr
22、oblem: rework, long reliability,Page 37,Reliability Test,BGA Package and PCB Test board,1.0mm pitch 2525mm 256 solder balls,1.6mm PCB OSP surface finish Common Tg material,Page 38,Reliability Test,BGA Assembly,Double-side mirror assembly,Double-side offset assembly,Page 39,Reliability Test,Test samp
23、le,240C,Page 40,Reliability Test,Thermal Cycle condition,1hour per cycle, 15min dwell time, 200 failure criterion,Page 41,Reliability Test,Reliability results,Single side SnAgCu solder joints have the strongest reliabilityDouble-side mirror assembly decrease the life by halfDouble-side offset assemb
24、ly will improve the reliability, but depend on the distance of offset,Page 42,Reliability Test,Failure analysis,Crack initial from the solder joints of cornerRed means solder joint open entirelyGreen means solder joint fail in electric function, but not separate from PCB,Page 43,Simulation Analysis,
25、Material Property,Implicit Creep Model,Page 44,Simulation Analysis,Crack growth model,For Sn63Pb37, Darveaux model:,For SnAgCu, Syed model:,Page 45,Simulation Analysis,Finite Element Model,diagonal,Geometry sketch,Single side assembly,Page 46,Simulation Analysis,Finite Element Model,Double side mirr
26、or assembly,Double side offset assembly,Page 47,Simulation Analysis,FEM Results-dangerous points,Single side,Double side,Initial crack location have different for single side and double side,Page 48,Simulation Analysis,FEM Results-Estimation life,failure location by FEM prediction is same to experim
27、ental resultsprediction life have a good correlation with experimental results,Page 49,Conclusion,For PCBA structure using in ATC experiments, we found: for all assembly shape, the first crack will initialize from solder joint of corner; for SnAgCu alloy, double side mirror will reduce life by over
28、3 times compared with single-side whether by using experiment or FEM; while for SnPb alloy, FEM analysis show the reduction degree will increase more; Double side offset assembly shape will increase the solder joints reliability to some extent which depend on the distance of offset Reasons of solder
29、 joints reliability reduction for BGA double-side assembly maybe include package structure, solder alloy, PCB thickness, offset distance, and so on，these factors will interact.,Page 50,DSP C6203器件工艺失效分析,整机测试功能信号时断时续，用手轻轻按压或者开机一段时间故障现象会暂时消失 之前曾经出现过此类失效，二者的故障现象和失效定位几乎完全一致，当时的给出的结论是在拿板的时候该处DSP器件是承受外力最大
30、的位置，从而容易造成失效。如此相似的失效现象重复出现，有必要进行更加深入的分析，以确定起失效的根本原因。,Page 51,无损检测分析XRay分析,最初的故障定位在U21位置芯片的A1角附近，对该芯片A1附近区域进行XRay分析，结果如图1所示。发现除了存在较多的气孔，没有其他的异常现象。,Page 52,无损检测分析IV曲线,重点对BGA器件的四边引脚进行分析 AA19和W22开路， AB21和W21时断时续。,Page 53,无损检测分析ERSASCOP分析,芯片AB22位置角附近存在多处焊点的开裂现象，裂纹产生在器件侧焊盘同焊料球之间，当在该芯片处施加外力或者热应力的情况下，裂纹可能会暂
31、时的愈合或者断开，因此，功能测试表现出时断时续的现象。,Page 54,有损检测分析起拔试验,几乎所有焊点的断裂面都在器件侧焊盘与焊料球之间 断面光滑平整，属于脆性断裂 AB22角部分焊点端面颜色发暗，应该为断裂面在空气中曝露,Page 55,SEM显示断口为脆性断口,Page 56,韧性断口图象（推力试验）,Page 57,一个正常焊点起拔后的各种断面位置,一般断面在焊料球中间或者是由于器件侧和PCB侧的焊盘从基板脱离,Page 58,有损检测分析切片,器件侧焊盘为ENIG镀层 个别焊点器件侧出现的裂纹 裂纹产生在Ni层/IMC之间，并几乎贯穿整个界面。 裂纹的宽度为24微米左右。,裂纹在N
32、i层/IMC之间产生并扩展,Page 59,有损检测分析切片,时断时续焊点器件侧出现明显裂纹,所有焊点PCB侧连接正常,Page 60,SEM,时断时续焊点器件侧裂纹背散射图象,所有焊点PCBIMC背散射图象,Page 61,有损检测分析SEM和EDX,能谱分析的结果显示了断面Ni层的P含量过高。一般来说业界通用的标准认为ENIG焊盘Ni层中P的含量在9wt10wt左右为宜。而从图中可以看出该批次器件样品的P含量已经达到了20wt25wt。断面P含量超高,Page 62,有损检测分析线扫描,Page 63,有损检测分析线扫描,从线扫描的结果看出，在Ni/SnPb的界面位置出现了大约一个微米厚度
33、的P的峰值，说明在该层面存在着P富集的现象 解释：在焊点服役过程中界面处IMC不断生成和长大，会不断的消耗界面附近两侧的Ni和Sn，这就使原本P含量偏高的Ni（P）层在靠近IMC的地方形成含量更高的富P层；同理，IMC的另一侧会形成一个富Pb的区域。富P层的形成会造成焊点界面的严重弱化，在工艺操作或运输过程中即使是受到一些微小的应力，也容易导致器件焊盘从焊点脱离。,Page 64,改进措施,短期解决措施： 尽量减少单板加工过程中的热过程（比如减少回流焊液态以上温度时间，降低峰值温度等） 减小扣板在装配环节的受力，如规范操作、使用工装 后续新单板设计使用到该器件，需要确认不能布局在高应力区域，例
34、如螺钉附近、扣板连接器附近、板边等。 根本解决措施 敦促加强DSP ENIG镀层工艺过程的控制 敦促尽快进行DSP 基板表面处理方式的改进（由ENIG镀层工艺转变为OSP镀层工艺）,Page 65,附：黑盘判断依据4X方法,Page 66,外观检查,市场返回单板异物腐蚀分析,来源：市场运行近一年返修 失效现象：复位频繁，可能为某处短路导致 外观检查： 从外观上看比较明显的现象是有异物流过的痕迹,Page 67,失效分析,外观观察,如果流体导电的话完全可以造成器件引脚之间的短路,Page 68,失效分析,IC分析（离子色谱）从返修样品上取少量异物，进行IC检测，结果显示为异物中含有较多的硫根离子
35、。 EDX分析 对图中的1，2，3位置进行了EDX分析，结果见后页,Page 69,失效分析,EDX分析,Processing option : All elements analyzed (Normalised)Spectrum C O S Ca Total1 17.77 59.31 12.43 10.50 100.00 2 17.44 60.44 11.55 10.57 100.00 3 12.76 62.86 12.90 11.48 100.00 Max. 17.77 62.86 12.90 11.48 Min. 12.76 59.31 11.55 10.50 All results i
36、n Atomic Percent,EDX结果显示异物中含有较多的C，O，S等物质,Page 70,失效分析,FTIR分析,用红外光谱分析表明，该痕迹的含N-H,C=O,C-N,O-S等基团。而且其中NH的含量很多。没有检验出C-H，表明里面含有其他有机成分的可能性不大。 尿素的结构分子式为 O| H2N- C NH 由此可以判断该痕迹中有尿素。 O-S的存在应该是空气中的SO2溶解造成的。,Page 71,结论和改进措施,结论： 该痕迹为鼠尿的遗留痕迹。 鼠尿流过使器件内部和器件之间部分短路，造成失效。 该失效属于个别现象。 改进措施： 敦促局方加强机柜的防鼠措施,Thank You,,联系人：刘桑 075589651218 28780808,
