Combining Process Modeling and Machine Learning Technology to Predict the Warpage of the Panel-Level Packaging After Debonding

Fan-Out Panel-Level Packaging (FO-PLP) has emerged as a high-usage efficiency and cost-effective solution for advanced packaging. However, the challenge of excessive warpage significantly impacts its yield and reliability. To address this issue, this study focuses on analyzing the warping behavior using the Finite Element Method (FEM). Additionally, Process Modeling technology is employed in the simulations to enhance their realism and alignment with actual processes. The primary objective of this study is to investigate the factors contributing to the emergence of asymmetrical patterns in PLP. Various influential factors are taken into consideration, including geometrical uncertainty, material uncertainty, working platform inclination, and the influence of vacuum release paths. While simulations offer advantages in terms of time and cost, inconsistent results may arise due to variations in modelling approaches used by different researchers in finite element analysis. To overcome this challenge, the study proposes the utilization of artificial neural network (ANN) algorithms for warpage prediction in FO-PLP. The methodology involves constructing finite element models of different scales and establishing a training database based on these models. Subsequently, the ANN model is optimized by fine-tuning hyperparameters to obtain the most accurate predictive model. The proposed ANN model is expected to provide precise predictions of FO-PLP warpage in a shorter time compared to conventional simulations and experiments while ensuring reasonable accuracy. By harnessing the capabilities of machine learning, this study aims to overcome the limitations of traditional simulation methods and enhance the efficiency of FO-PLP warpage prediction. The findings of this research are anticipated to contribute to optimizing and improving the FO-PLP process, thereby enhancing the yield and reliability of advanced packaging applications.