Improved finite element modeling strategies with multipoint constraints for BGA packages subjected to thermal cycling

Finite Element simulations are often used to study the reliability of solder joints subjected to thermal cycling. Packaging configurations are becoming more complex to accommodate better functionality and performance. Increased complexity leads to several challenges for FE models including difficulties modeling thin layers and interfaces, as well as keeping the total numbers of nodes and elements to reasonable levels so that computation times can be practical. To reduce the use of high-density meshes and to relax the restrictions of nodal connections, the technique of Multi-Point Constraints (MPC) is often used in finite element analysis. In the MPC method, constraints are enacted between different degrees of freedom of the model to simply transition between finely and coarsely meshed regions. MPC algorithms require additional DOF constraints on a FE model; and extra contact nodes/elements are deployed between the interfaces of contacting elements. MPC methods can be implemented with materials having linear or nonlinear mechanical behavior. The accuracy and efficiency of MPC-based finite element simulations for electronic packages have not been evaluated completely in the literature. In this work, an improved MPC based FE modeling strategy was developed for BGA packages to reduce the total number of elements (including both conventional and MPC elements), and thus reduce the simulation time. In addition, the new method can improve the simulation accuracy relative to models prepared using conventional meshing strategies. The proposed technique allows for different types of mesh patterns (circular pattern from solder joint and rectangular patterns from other component) to be connected in a package assembly while reducing the overall number of elements in the model. The proposed approach works with both symmetric and non-symmetric solder ball arrays, and achieves a good balance between simulation cost and simulation accuracy.