Skip to Main Content
In this paper, principal component regression models (PCR) have been investigated for reliability prediction and part selection of area package architectures under thermo- mechanical loads in conjunction with stepwise regression methods. Package architectures studied include, BGA packages mounted on CU-CORE and NO-CORE printed circuit assemblies in harsh environments. The models have been developed based on thermo-mechanical reliability data acquired on copper-core and no-core assemblies in four different thermal cycling conditions. Solder alloys examined include SnPb and SAC Alloys. The models presented in this paper provide decision guidance for smart selection and substitution to address component obsolescence by perturbing product designs for minimal risk insertion of new packaging technologies. It is conceivable for commercial off the shelf parts to become unavailable during the production-life of a product. Typical Commercial-of-the-Shelf parts are manufactured for a period of two to four years, and IC manufacturing processes are available for five to six years. It is envisioned that the reliability assessment models will enable turn-key evaluation of geometric architecture, material properties, and operating conditions effects on thermo-mechanical reliability. The presented approach enables the evaluation of qualitative parameter interaction effects, which are often ignored in closed-form modeling, have been incorporated in this work. Previously, the feasibility of using multiple linear regression models for reliability prediction has been demonstrated for flex-substrate BGA packages [1, 2], flip-chip packages [3, 4] and ceramic BGA packages  Convergence of statistical models with experimental data and finite element models has been demonstrated using a single factor design of experiment study. In addition, the power-law dependencies of individual variables have been correlated with established failure mechanics models. PCR approach uses the potentially im- portant variables from stepwise regression. The statistics models are based on accelerated test data acquired as part of this paper, in harsh environments, while finite-element models are based on damage mechanics and material constitutive behavior. Sensitivity relations for geometry, materials, and architectures based on statistical models, and FEA models have been developed.