Fracture Simulation of Redistribution Layer in Fan-Out Wafer-Level Package Based on Fatigue Crack Growth Characteristics of Insulating Polymer

For improving and enhancing the performance and reliability of the package, it is quite valuable to research fracture behaviors of the interlayer insulating polymer in the package, which is influenced by thermal and physical stresses during the operation or the manufacturing process. If the reliability of the package can be known in advance by simulation, it can be expected to greatly help in material selections and package designs. While fracture simulation using energy release rate (G) computed by finite element analysis (FEA) and critical energy release rate (Gc) obtained experimentally has been recently reported by many researchers, it has not precisely indicated its fracture mechanism. We discussed fatigue fracture behaviors related to initial crack growth under a thermal cycle, and suggested fracture simulation by FEA based on fatigue crack growth properties of insulating polymer materials in Fan-Out Wafer-Level Package (FOWLP). The photosensitive polymers (2-types of polyimides, and phenolic resin) were evaluated in this study. The relationship between the fatigue crack propagation rate and the energy release rate range (Delta G(i)) at the interface between an insulating polymer film and copper (Cu) on the Si wafer was obtained by the cyclic peel test under room temperature (298K) and arbitrary stress. The relationship between the fatigue crack propagation rate and the energy release rate range (Delta G(b)) in a self-supported bulk film was obtained by the conventional fracture mechanic method. Threshold energy release rate range (Delta G(th)) was found from each crack propagation rate for the Delta G(i) and the Delta G(b). We calculated the energy release rate range (Delta G(sim)) from FEA with the FOWLP model under a thermal cycle stress, predicted that created cracks of the insulating polymer at the side walls of Cu wiring layers in the FOWLP model would grow to which direction to insulating polymer bulk or polymer/Cu interface. As analyzed by the normalized Delta G(sim)/interface-Delta G(th) (Delta G(i_th)) and Delta G(sim)/bulk-Delta G(th) (Delta G(b_th)), it was realized that the value of Delta G(sim)/Delta G(i_th) was higher than that of Delta G(sim)/Delta G(b_th). By this result, we judged that the created cracks at the side walls of Cu wiring layers in FOWLP model would grow to polymer/Cu interface rather than polymer bulk, when the interface crack runs to a corner of a Cu layer. Moreover, it was realized that a polyimide type resin would be more suitable to suppress growing cracks than a phenol resin with the results of peel judgement by Delta G(sim)/Delta G(i_)(th). Also, as Delta G(sim)/VG(i_th) was higher than G/interface-Gc (G(i_c)), it was realized that the peel judgement using G/G(i_c) was insufficient to estimate fracture. This study would help in the selection of optimal materials for designed package structures and design guides for high reliability package structures.