论文部分内容阅读
随着微电子技术的需求和发展,倒装芯片技术在高密度微型化封装领域得到了快速发展和广泛应用,而现有的一些倒装芯片检测方法存在一定的不足之处。为此,研究了主动红外的倒装芯片缺陷检测方法。实验中使用激光加热对倒装样片施加非接触热激励,通过红外热像仪获取样片温度分布。采用小波分析方法提取包括小波熵在内的信号特征,采用自组织神经网络对不同类型焊球进行聚类识别。研究表明,通过自组织神经网络可以有效地将不同缺陷焊球与参考焊球通过距离映射法映射到不同区域从而区分开,并且可以将未知焊球信号映射到相应的区域实现聚类识别。因此该方法可以有效实现倒装芯片的缺陷检测。
With the demand and development of microelectronic technology, flip-chip technology has been rapidly developed and widely used in the field of high-density miniaturization. Some existing flip-chip detection methods have some shortcomings. To this end, the active infrared detection method of flip chip defects. In the experiment, non-contact thermal excitation was applied to the flip-chip by using laser heating, and the temperature distribution of the sample was obtained by infrared thermal imager. Wavelet analysis was used to extract the signal features including wavelet entropy, and the self-organizing neural network was used to cluster different types of solder balls. The research shows that by using self-organizing neural network, different defect solder balls and reference solder balls can be effectively mapped to different regions by distance mapping method to distinguish them, and the unknown solder ball signals can be mapped to the corresponding regions for clustering recognition. Therefore, this method can effectively detect the defects of flip chip.