Risk Management Models for Flip-Chip Electronics in Extreme Environments

Abstract

In this work, risk-management and decision-support models for reliability prediction of flip chip packages in harsh environments have been presented. The models presented in this paper provide decision guidance for smart selection of component packaging technologies and perturbing product designs for minimal risk insertion of new packaging technologies. In addition, qualitative parameter interaction effects, which are often ignored in closed-form modeling, have been incorporated in this work. Previous studies have focused on development of modeling tools at sub-scale or component level. The tools are often available only in an offline manner for decision support and risk assessment of advanced technology programs. There is need for a turn key approach, for making trade-offs between geometry and materials and quantitatively evaluating the impact on reliability. Multivariate linear regression and robust principal components regression methods were used for developing these models. The first approach uses the potentially important variables from stepwise regression, and the second approach uses the principal components obtained from the eigen-values and eigen-vectors, for model building. Principal-component models have been included because if their added ability in addressing multi-collinearity. The statistics models are based on accelerated test data in harsh environments, while failure mechanics models are based on damage mechanics and material constitutive behavior. Statistical models developed in the present work are based on failure data collected from the published literature and extensive accelerated test reliability database in harsh environments, collected by center of advanced vehicle electronics. Sensitivity relations for geometry, materials, and architectures based on statistical models, failure mechanics based closed form models and FEA models have been developed. Convergence of statistical, failure mechanics, and FEA based model sensitivities with experimental data has been demonstrated.