A Numerical Methodology for Thermo-Mechanical Characterization of High Power Electronic Packages

Abstract

In this study, an effective numerical methodology that combines computational fluid dynamics (CFD) and finite element method (FEM) modeling is presented for thermo-mechanical characterization of electronic packages in transient state under a forced convection environment. Experiments were performed on a thermal test chip under a power cycling condition to measure transient temperature distribution in the package. A power profile was designed and applied to the test die to generate a typical temperature profile of a high power electronic package under field use conditions which includes ramp up, mini-cycle, dwell, and ramp down periods. CFD model was then built with the Icepak software to simulate the performed experiment. The CFD simulation reproduced the experimentally-obtained temperature profile accurately. The nodal heat flux and heat transfer coefficients (HTC) were extracted from the CFD simulation results. Then, a separate thermal FEM model which contains all the components of the package except the heat sink was built in ANSYS. The heat flux from the package to the heat sink obtained from the CDF simulation results was applied to the thermal FEM model as the boundary conditions (BC) at the top surface of the package. The HTC from the CFD were used for BC’s at all other surfaces. This equivalent BC’s eliminated the necessity of complex modeling for forced convection. Transient temperature distribution obtained from this simplified model thermal FEM simulation matched very well with that of CFD simulations. Consequently, a structural simulation was performed with temperature distribution obtained from this thermal FEM simulation as the thermal loading. Strains, stresses, strain energy density, and other mechanical index were obtained and used for reliability assessments of the package. The proposed methodology enables to assess reliability of high power packages under real power cycling conditions with any desired power and temperature patterns efficiently.