楔形焊接工艺简介ppt课件.ppt

焊接质量分析,莫卓亚 2007.8.20,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,1 焊接过程出现的问题以及解决方法,2 Three factors affecting the bonding result,Three factors that will affect the bonding result should be considered in the design of the wire bonder：FAB forming technologyBonding technologyLoop technology(from JauLiang Chen,Member IEEE),2.1 FAB forming technology,The parameters which affect the gold wire ball formation include:1.Tail length left after second bonder2.Type and shape of capillaries used3.Material characteristics of gold wire4.Supplied voltage,current and time of EFO unit5.Gap between tail and electrode plate6.Relative position between capillary and electrode plate,2.1 FAB Modeling for Gold WB for different WD,Different EFO settings can achieve the same FAB size for the same wire diameter Fig1.EFO time-FAB charts for 1.0 mil Au at 30mA,40mA and 50mA EFO currentsThese results conclude that the different levels of EFO settings deliver the same input energy to produce the same FAB or output energy,2.1 FAB Modeling for Gold WB for different WD,EFO输入的能量理想表示为：对于一个确定的bonding系统，R是确定的，当EFO放电击穿空气间隙时，可以将R忽略因此:理想情况，能量完全传递时，n=2因为存在能量损失，所以n=12,2.1 FAB Modeling for Gold WB for different WD,理想情况下，输入能量=输出能量输出能量可由FAB size 表示输入能量可由EFO current and EFO time 表示在相同的设置和环境下,2.1 FAB Modeling for Gold WB for different WD,The energy transfer efficiency charts in Figure show a direct relationship between the FAB size and the EFO input energy for each reference wire diameter.Fig2:Energy Transfer Efficiency charts for:0.6mil,0.8mil,1.0mil,1.5mil,2mil,2.3mil,2.7mil and 3.2mil Au wire,2.1 FAB Modeling for Gold WB for different WD,The EFO settings required to achieve the desired FAB size can be calculated from the Energy Transfer Efficiency chartsShortcoming in using the Energy Transfer Efficiency charts to carry out FAB size predictions for a particular wire size is that reference data points are required for that same wire size.A,B are constants for each particular wire,2.1 FAB Modeling for Gold WB for different WD,利用上面的公式，输入的能量直接对应于wire size Fig3:Constant Energy Gap Chart for 0.6mil to 3.2mil Au wire,2.1 FAB Modeling for Gold WB for different WD,1.There is a constant input energy change as FAB ratio changes across the range of wire diameters2.This is due to the use of FAB ratios,instead of the absolute FAB size.3.A 10%change in FAB ratio will lead to a 10%change in the absolute FAB size,will lead to a proportional change in,2.1 FAB Modeling for Gold WB for different WD,A single characteristic equation can be formulated to calculate the FAB size for any wire diameter from Fig3The equation was found to be more accurate when it describes the energy transfer relationship for fine and heavy wire separately,as shown in Figures 4 and 5.Fig4:Constant Energy Gap Chart for fine wires and Fig5 Constant Energy Gap Chart for heavy medium wires:0.6mil to less than 2mil wire 2.0mil to 3.2mil wire,2.1 FAB Modeling for Gold WB for different WD,The following equation,which is derived from the constant energy gap concept,is the single characteristic equation that describes the relationship between the delivered input EFO energy and the FAB size,which is the measurable output energy.Where the coefficients and variables are shown in Table 1,2.1 FAB Modeling for Gold WB for different WD,Table 1:Coefficients of the semi-empirical model for fine,medium and heavy Au wires.,2.1 FAB Modeling for Gold WB for different WD,对于公式的总结This semi-empirical equation(半经验主义公式),which can be implemented in the wire bonder software,can serve as a tool to calculate the EFO time needed to form the desired FAB size that was keyed in by the operator.,2.1 FAB Modeling for Gold WB for different WD,Shortcoming in using semi-empirical modelGiven changes in bonding environment,setups or conditions that affect the FAB size formed at the same EFO settings and wire diameter,a new set of reference FAB data points can be collected to numerically modify the semi-empirical model.The methodology to build the model will still be the same.Therefore,the application of this model can also be extended to copper wire bonding,as long as a set of reference copper FAB data points are available.,2.1 FAB Modeling for Gold WB for different WD,Verification of Semi-empirical ModelThe verification FAB data points were collected from 0.6mil,0.7mil,0.8mil,0.9mil,1.0mil,1.2mil,1.5mil,2.0mil,2.3mil,2.5mil,2.7mil and 3.0mil and 3.2mil Au wire,over a range of EFO current and time settings.The model proved to be highly consistent and reliable in predicting the Au FAB size.,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,3个疑问？PART2只考虑了EFO supplying time and EFO current 来预测FAB ball size 有没有道理？如何证明EFO supplying time and EFO current are the two parameters affecting the FAB ball size most?PART3 Taguchi DOE 如何实现 FAB 成型工艺参数优化?,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Taguchi方法原理Taguchi方法是日本学者田口玄一提出的一种试验设计方法，它用到一种特别的正交表。正交表的符号为：其中字母L表示正交表，n为正交试验的次数，q为试验的因素数，t为因素的水平数。这些标准的正交矩阵保证了用最少的实验次数来反映影响性能参数的所有因素的全部信息。,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,正交矩阵的特性为均衡分散性和整齐可比性：(1)每一独立变量垂直对应的一列有着特定的取值设置组合；每一变量的各水平出现相同的次数。(2)独立变量的每一个取值都被用到；(3)正交矩阵每一变量的取值顺序不能任意改变。这是因为正交矩阵中的任意两列是相互正交的，向量对权的内积为零,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table2:实验因素和水平设置表Note:EFO Supplying Time:1 scale=1msec EFO Current:1 scale=5mA EFO Voltage:1 scale=450V 1p=4.3umTable1 shows the selection of each parameters level used in this study.,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table3:experiment result正交表的符号为：,2.2Taguchi DOE实验设计法FAB成型工艺参数优化,在Taguchi分析中，采用信噪比S/N来研究各个参数对目标的影响。本实验选用 1 EFO Supplying Time 2 EFO Current 3 EFO Voltage 4 Tail Length 5 Spark Gap作为主要实验因素,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table 4:S/N calculation,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Figure6：sensitivity diagramFigure7:signal-to-noise ratio diagramEFO current and EFO supplying time are the two parameters affecting the FAB ball size most.,A，B,A，B,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,从figure7中可以得到，对信噪比影响最小的组合是：A1B1C1D1E3，这是粗糙得到一组较为优化的参数。如何得到精确的经过优化的工艺参数组合？Using the Taguchi method with neural network，a small number of experiments can easily find the proper parameters setting.,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table5:EBP neural network training pattern for experiment,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table6:EBP Prediction for experiment 1,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Table 7:FAB ball size comparison between EBP prediction and verification resultsFig 8:comparison between prediction and real ball size,2.2 Taguchi DOE实验设计法FAB成型工艺参数优化,Part 3 conclusion in this part,the Taguchi method EBP together with EBP neural network was used to find the best parameters setting for gold wire ball formation.From the experiment results,the following can be concluded:1)By the Taguchi method,through a limited number of experiment,it is very easy to find the effect of process parameters.2)In this part,EFO current and EFO applying time are the two parameters that affect the FAB ball formation significantly.3)With EBP neural network training,one can predict the ball formation precisely.And with proper adjustment,the best process parameters can be set correctly.,3 Looping Technology实验细节,Wire propertiseTwo gold ball bonding wire types,A and B,of 20m diameter,3 Looping Technology 实验细节,Wires Mechanical PropertiesThe elongation to break(%EL),the break load(BL),and elastic modulus(E)of the wire A and wire B are summarized below.Wire A has higher break load compared to wire B.The elastic modulus of both wires is similar,3 Looping TechnologyFAB,Grain Size and HAZ,Fig.8.Average grain width at the neck of the FAB of the wire A and wire B.Wire A has a finer grain size at the neck compared to wire B for all FAB sizes.Increasing the FAB size increases the neck grain size,due to the larger heat input.,3 Looping Technology FAB,Grain Size and HAZ,Fig.9.Vickers hardness(HV)at the neck of the FABWire A has a higher hardness than wire B and the microhardness decreases as the FAB size increases,3 Looping Technology FAB,Grain Size and HAZ,Fig.10.The HAZ length of the FAB of the wire A and BIt was observed that wire A had longer HAZ compared to wire B and HAZ length increases as the FAB size increases,3 Looping Technologyball pull test,The ball pull test results for wire A and B are shown in Fig.11There is no significant difference in the ball pull results betweenwire A and wire B at the same bonding condition.In addition,there seems to be a slight decrease in the ball pull force as the FAB size increases,and slight increase in the pull force as the loop height increases due to geometry effect.,3 Looping TechnologyLooping Profile Measurements,It was clear that wire A and wire B had different looping behavior at 35 and 45 m FAB sizes,3 Looping TechnologyLooping Profile Measurements,It was clear that wire A and wire B had similar looping behavior at 55 and 65 m FAB sizes.,3 Looping TechnologyLooping Profile Measurements,Comparing wire A and wire B,it was clear the difference in the chemical composition between wire A and wire B played a role which contributed to their looping behavior.The different alloying elements in the wire A and wire B were responsible for the different hardness,grain size,and HAZ length in both wires.However,while the difference in hardness,grain size,and HAZ length in wire A and wire B were observed on all FAB sizes,it was not very clear why the difference in the loop profile was only observed in 35 and 45 m FAB sizes only.,3 Looping Technologystacked die package,Comparison of different wire bond looping profilesA normal ball bonding process places a ball bond on the die first,performs a loop motion to the substrate,then places a stitch bond on the lead.A reverse ball bonding process places a bump on the die pad first.After the bump is formed,a ball bond is placed on the substrate and the stitch bond is placed on the bump,3 Looping Technology stacked die package,Figure 14.Folded forward loop with less than 3 mil loop height.待补充,4 补充材料拉力测试,断线点”晶片表面有污秽 参数调较不恰当断线点”参数调较不恰当 焊线质量 邦线周围有污秽断线点”焊线质量 铁钧有尖物,断线点”参数调较不恰当 焊线质量 邦线周围有污秽断线点”厎板表面有污秽 参数调较不恰当,4 补充材料可接受邦线的规格,ASM 标准宽度(W)=1.3-1.8 长度(L)=2-2.2 线尾长度(T)=0.5=邦线的直径,线尾长度,长度,宽度,邦线的直径,4 补充材料邦线品质检查,没有尾巴,尾巴过长,冷邦,偏位,不可接受邦头的出现,4 补充材料the ball bonding process,视频！,