Warpage Optimization of Package Substrates Using Metamodels - A Review

This review analyzes the state-of-the-art in metamodel-based simulation and optimization of microelectronics package substrate warpage. The work gives an overview on the general procedure, including metamodel selection, Design of Experiment approaches, or sampling. It attempts to not only focus on material property optimization, but also highlight approaches in structural and process optimization, which are largely underexplored. Additionally, the review reveals a critical gap, as no research explicitly incorporates geometry, boundary conditions and constraints as parameters within the metamodel itself. It also summarizes literature on neural network approaches and shows that although there is a significant influence of feature position on substrate warpage, e.g. convolutional neural networks are not well represented in current research.