With the upcoming semiconductor devices, becoming more and more complex, also the failure analysis becomes more challenging. Advanced signal processing methods can improve the abilities of scanning acoustic microscopy significantly. On the one hand, they can reduce the measurement effort, on the other hand, they can improve detection capabilities.

In contrast to conventional imaging, which requires each layer needs to be scanned separately, this paper proposes the application of synthetic aperture focusing technique to the scan data. By superposing the signals of several scan positions a virtual, much larger aperture is generated. This allows a reconstruction of the full volume and provides information about structures in any depths. Reconstruction is done in time domain and frequency domain. The method is applied on a reference sample and real microelectronic samples are examined.

Especially for structures with a huge number of interfaces and a periodic structure the signal to noise ratio is very low for the deeper interfaces. Wavelet transform using fast Mexican hat and its coefficient selection filtering techniques are used for improving the SNR and by this enhancing defect detection. The method is demonstrated on a reference sample and on high bandwidth memory modules consist of a stack of small memory dies.

随着即将面世的半导体器件变得越来越复杂，故障分析也变得更具挑战性。先进的信号处理方法可以显着提高扫描声学显微镜的能力。一方面，它们可以减少测量工作量，另一方面，它们可以提高检测能力。

与传统成像需要单独扫描每一层相比，本文提出将合成孔径聚焦技术应用于扫描数据。通过叠加几个扫描位置的信号，可以生成一个更大的虚拟孔径。这允许重建整个体积并提供有关任何深度的结构的信息。重构是在时域和频域中完成的。该方法应用于参考样品并检查真实的微电子样品。

特别是对于具有大量界面和周期性结构的结构，对于较深的界面，信噪比非常低。使用快速墨西哥帽的小波变换及其系数选择滤波技术用于提高 SNR 并由此增强缺陷检测。该方法在参考样本上进行了演示，并且在由一堆小内存裸片组成的高带宽内存模块上进行了演示。