论文部分内容阅读
Al丝超声引线键合工艺被广泛地应用在大功率器件封装中,以实现大功率芯片与引线框架之间的电互连。Al丝引线键合的质量严重影响功率器件的整体封装水平,对其工艺参数的优化具有重要工业应用意义。利用正交实验设计方法,对Al丝引线键合工艺中的三个最重要影响因数(超声功率P/DAC、键合时间t/ms、键合压力F/g)进行了正交实验设计,实验表明拉力优化后的工艺参数为:键合时间为40 ms,超声功率为25 DAC,键合压力为120 g;剪切推力优化的工艺参数为:键合时间为50 ms,超声功率为40 DAC,键合压力为120 g。基于BP神经网络系统,建立了铝丝超声引线键合工艺的预测模型,揭示了Al丝超声键合工艺参数与键合质量之间的内在联系。网络训练结果表明训练预测值与实验值之间符合很好,检验样本的结果也符合较好,其误差基本控制在10%以内。
Al wire ultrasonic wire bonding process is widely used in high-power device package, in order to achieve high-power chip and lead frame electrical interconnection. The quality of Al wire lead bonding seriously affects the overall package level of power devices and has important industrial application significance for the optimization of its process parameters. Orthogonal experimental design was used to design the most important influence factors (ultrasonic power P / DAC, bonding time t / ms, bonding pressure F / g) of Al wire wire bonding process. The experimental results show that the optimized process parameters are as follows: the bonding time is 40 ms, the ultrasonic power is 25 DAC and the bonding pressure is 120 g; the optimized process parameters of shear thrust are: the bonding time is 50 ms and the ultrasonic power is 40 DAC, bonding pressure of 120 g. Based on the BP neural network system, a prediction model of aluminum wire ultrasonic bonding process was established, and the inherent relationship between the bonding process parameters and bonding quality of Al wire was revealed. The results of network training show that the predicted value of training is in good agreement with the experimental value, and the result of test sample is also in good agreement with the error of 10%.