Heating rate dependent delamination of metal–polymer interfaces: experiments and modeling

Abstract

Bimaterial interfaces in microelectronics packages are the most common regions of failure under thermo-mechanical excursions. In this work, we report experimentally observed role of heating rate on the delamination initiation and propagation across a metal-polymer interface in a microelectronic package. We observe that the rate of delamination propagation increases with increasing heating rate. When the heating rate increases, in addition to the higher amount of delamination growth per unit time, experimental results suggests that higher growth will also incur per unit temperature (loading). Correspondingly, the temperature at which complete delamination occur decreases. Using finite element modeling with cohesive interfaces, we provide a plausible explanation to this observed phenomenon. The analyses indicate that the mechanical behavior of the bimaterial interface is sensitive to both temperature and thermal rate.

Appendix: Effect of non-physical parameters

The likely spurious effects that may arise from the numerics of the problem will be addressed in this section. Non-physical parameters such as stabilization factor (for viscous regularization damping) or mesh size can have an impact on the delamination propagation. Since the ultimate aim of numerical analyses is typically to predict actual physical behavior, it is important to isolate and eliminate the influence of non-physical behavior on the results obtained from the simulations.

1.1 Stabilization factor

Viscous stabilization is implemented in Abaqus to circumvent convergence difficulties in crack propagation simulations. When the stabilization parameter is sufficiently small, accuracy of results is ensured. Otherwise, toughening of material response will be observed. This suggests that the stabilization factor can have an effect on the results. It will be undesirable if the stabilization factor introduced artificial effects that affect the trends among the simulations. For more information on the viscous stabilization scheme, the reader can refer to (Abaqus 2009; Davila et al. 2008).

The matrix of parameters presented in Table 3 is analysed using the model presented earlier in Sect. 4.1, $\uprho$ is the stabilization factor while $\hbox {t}_\mathrm{f}$ is the pseudo-time at maximum load. The results (Fig. 9) show that $\uprho /\hbox {t}_\mathrm{f}$ (instead of $\uprho )$ needs to be kept constant to eliminate the impact of viscous stabilization when comparing among delamination at different rates.

Table 3 Parameters of the study on stabilization factor

Fig. 9

Effect of stabilization factor

1.2 Mesh size

Mesh can have an effect on the results. To ensure that the mesh used is sufficiently fine to give converged results, a mesh dependence study is performed using the model fast(b) found in Sect. 4.1. The $\uprho /\hbox {t}_\mathrm{f}$ ratio is kept at 0.208. The results (Fig. 10) show that Mesh 1 (Table 4) is sufficiently fine for the problem. Mesh 1 is used in the analyses presented in Sect. 4.

Fig. 10

Results of mesh sensitivity study for heating rate dependent CZM

Table 4 Attributes of mesh at the interface for heating rate dependent CZM