Time Integration Damage Model for Sn3.5Ag Solder Interconnect in Power Electronic Module

Abstract

In this study, existing damage evolution models in the literature for solder layer in microelectronics have been reviewed. A two dimensional approximate semi-analytic time integration damage indicator model for Sn3.5Ag material solder interconnect in power electronic module has been proposed. The proposed time dependent damage model is dependent on the inelastic strain, the accumulated damage at previous time step and the temperature. The strains were approximated semi-analytically. A numerical modelling methodology combined with the data from public domain for crack initiation and crack propagation of Sn3.5Ag solder layer has been adopted to extract the parameter values of the proposed damage model. The proposed model has advantages over fatigue lifetime models as it instantaneously predicts the damage over time for any loading history. The damage model was compared with Ansys FEA tool based damage prediction using Coffin Manson and Paris law fatigue models. The predicted damage value by the model is slightly higher than those models. Furthermore, this damage model does not need a time consuming numerical simulation evaluating the damage model variables, which is an advantage

Physics-informed Machine Learning for Predicting Fatigue Damage of Wire Bonds in Power Electronic Modules

“…Power components based on current packaging architectures are most susceptible to thermally induced fatigue damage of their wire bonds and the die attachment layer. While many studies contributed to the reliability modelling of PEMs [4][5][6][7][8], the informed deployment of these modules in different applications remains a challenge for the end-users. Driven by considerations for protecting intellectual property (IP), data on internal design and dimensions of power components and their bill of materials are not included in manufacturers' technical datasheets and thus are not readily available.…”

“…Finite element (FE) modelling, coupled with lifetime prediction methodologies and models, offers a robust solution for this problem by predicting damage metrics for the expected failure modes, for example, the studies reported in [5][6][7][8]. Lifetime prediction methodologies for PEM have been also extensively developed, with much of the effort on establishing accurate models predicting the wire bonds lifetime under accelerated active and passive temperature cycling conditions that qualify the device for the respective application load.…”

“…For the failure of wire bonds in PEMs, a damage metric is most often obtained with FE simulation, and commonly chosen as the plastic strain range per cycle. This lifetime modelling approach has been demonstrated extensively, and in many studies [6,9]. However, developing and using such FE-based computational models to predict damage, and then lifetime, is a complex task that requires characterisation data, specialist skills, computing hardware, and advanced simulation software tools, all of which are not available to many end-users.…”

Reliability Meta-modelling of Power Components

“…Temperature cycling loads make the die attachment and interconnection layers (commonly solder) and the wire bonds susceptible to failure under modes such as interfacial cracking and lift-off, receptively. Both solder interconnection layer damage modelling, for example [7,8], and wire bond lift-off failure simulations, for example [9,10], have received extensive attention in power electronics reliability studies.…”