Publication: Machine learning study of void detection using machine learning
Loading...
Date
2023-07-03
Authors
Goh, Min Shing
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Underfilling process is a common technique used in current Surface Mount Technology (SMT) for Ball Grid Array (BGA) package. However, void will form during this process causing the integrity of the solder joint to be questioned. As there are air vapour in the void, ‘Popcorn effect’ might occur when the solder is experiencing thermal shock causing the chip to be defective. Therefore, the purpose of this study is to develop a machine learning model that can detect the void in the carrier chip and to predict the propagation of crack. Convolutional Neural Network (CNN) has been used as an approach for the machine learning model to detect the void in chip. Moreover, the prediction of crack propagation from void has been done under various conditions such as elastic plastic condition and creep fatigue condition. From the results of the machine learning conducted, the voids in the carrier chip have been successfully detected. Additionally, the mAP and the AR of the machine learning is 0.4885 and 0.4730 respectively which is considered magnificent based on typical machine learning evaluation rating. From the results obtained, the J-integral for the void in the carrier chip that exceed 25 000 𝑁/𝑚𝑚 will propagate into crack under elastic plastic condition whereas the strain rate of the void exceed 4 𝑥 10−3𝑚𝑚 will cause crack propagation under creep fatigue condition. Furthermore, the area of void that will propagate into crack is 15mm2. In short, the research done can help the end user of Surface Mount Technology (SMT) identifying the void in the chip and also predicting the propagation of crack from void so that such problem can be resolved