A Process Dependent Warpage and Stress Model for 3D Packages Considering Incoming Die/Substrate Warpage and Assembly Process Impacts

Abstract

Warpage control and die stress are critical for 3D package assembly. Currently most of the warpage studies, either measurements or simulations, are mainly evaluated at the end-of-line (EOL) for the finished package. However, for 3D packaging such as multiple die stacking or through-silicon-via (TSV), the interaction between warpage and assembly process steps becomes far more important because: (1) Warpage at each single process step will impact the yield of the next process step. (2) Process induced warpage will accumulated and affect the final end-of-line package warpage, and (3) Die stress in the end-of-line package is highly dependent on the incoming die alone residual stress and the assembly process flow. It is required to develop capability to evaluate warpage at each process step following the process flow. In this paper, we use a 2-die TSV package as a test vehicle to collect warpage data at each assembly step. A process dependent finite element model (FEM) method is developed which is capable of capturing the process induced warpage and stress at each process step based on the process condition. Shadow moire is used to measure warpage data at following check points: (1) Incoming bare dies and bare substrate warpage; (2) Die 1 warpage after underfilled on substrate; (3) Die 2 warpage after TCNCP on Die 1; (4) Completed package after molding (end-of-line). Element birth and kill approach is developed in the FEM model to simulate the assembly process flow. In each process step, only the presented components and materials in the model are active while the others are killed. When new parts or materials are added through the assembly flow, they are activated in the model and the related process condition is applied for each step. Using this process dependent modeling approach, excellent correlation of the model results with the actual measured warpage data is achieved at each process step. One factor often been ignored is the incoming bare die warpage and bare substrate warpage effect. They are often assumed flat and stress free. However, as die and substrate become thinner and thinner, their incoming initial warpage or residual stress plays an important role. In this paper, the measured incoming substrate and die warpage data are presented, their impacts to the warpage and stress evolution through the process flow are evaluated. A novel method is also developed in the simulation model which is able to study the incoming die/substrate warpage and residual stress effect.