Accelerate Your 3D X-ray Failure Analysis by Deep Learning High Resolution Reconstruction

Over the past decade, 3D X-ray technique has played a critical role in semiconductor package failure analysis (FA), primarily owing to its non-destructive nature and high resolution capability [1,2]. As novel complex IC packages soar in recent years [3,4], X-ray failure analysis faces increasing challenges in imaging new advanced packages because IC interconnects are more densely packed in larger platforms. It takes several hours to overnight to image fault regions at high resolution or the crucial details of a defect remain undetected. A high-productivity X-ray solution is required to substantially speed up data acquisition while maintaining image quality. In this paper, we propose a new deep learning high-resolution reconstruction (DLHRR) method, capable of speeding up data acquisition by at least a factor of four through the implementation of pretrained neural networks. We will demonstrate that DLHRR extracts signals from low-dose data more efficiently than the conventional Feldkamp-Davis-Kress (FDK) method, which is sensitive to noise and prone to the aliasing image artifacts. Several semiconductor packages and a commercial smartwatch battery module will be analyzed using the proposed technique. Up to 10x scan throughput improvement was demonstrated on a commercial IC package. Without the need of any additional X-ray beam-line hardware, the proposed method can provide a viable and affordable solution to turbocharge X-ray failure analysis.

Using Element Birth and Death Technique in Modeling Cumulative Molded Substrate Expansion before Singulation

Semiconductor packages are commonly assembled and molded in array format on a substrate strip before they are singulated into individual units. However, cumulative substrate expansion causes problems such as machine vacuum error or misaligned cut during singulation if the substrate expansion is not factored in. This study uses element birth and death technique in modeling the overall expansion of the molded substrate strip so that the predicted expansion could be considered in the singulation tooling design offsets. The expansion of the substrate was modeled with the different package assembly processes and thermal conditions. Modeling results showed that there is a cumulative increase in the length of the substrate as it passes through the different processes. The results are in agreement with actual substrate expansion prior to package singulation. This would not be captured when simulation is done only for the molded substrate without considering the cumulative contribution of the preceding processes. With the element birth and death technique in process-based thermomechanical modeling, substrate expansion could already be forecasted, and package assembly problems avoided.

Viscoelastic Material Characterization: A Realistic Approach to Stress Analysis

This paper presents an advanced method in materials characterization for the mold compound material in semiconductor packages to build models that can technically explain the actual warpage or stress observations under different thermal conditions and time history. In the study, the mold compound material characterization was conducted using Dynamic Mechanical Analyzer (DMA) followed by curve fitting to obtain parameters for the computer modeling input requirement. Thermo-mechanical modeling using viscoelastic material properties was conducted on a bi-material test sample model. Results showed that the new characterized viscoelastic material properties exhibited dependence on time and temperature. Slow cool down from post mold cure (PMC) to room temperature resulted in lower warpage or stress. This observed rate dependent response was explained using viscoelastic material properties in contrast to the usual linear elastic material simplification. Thus, a realistic result from stress or warpage analysis could be achieved using viscoelastic material characterization.