Identifying Hot Spots in Electronics Packages with a Sensitivity-Coefficient Based Inverse Heat Conduction Method

Abstract

Inverse heat conduction methods can be used to estimate the location and intensity of heat sources in electronics, but often there is a tradeoff in computational cost and accuracy of the retrieved heat fluxes and temperatures. This is exacerbated when the exact size and locations of heat sources within the device are unknown. This paper demonstrates the applicability of an inverse sensitivity coefficient method to locate hotspots and estimate associate hot spot temperatures for a commercial electronics package. The sensitivity coefficients are computed with a steady-state, 3-D finite volume model in FloTHERM. In this work, the locations and sizes of the heat sources are not initially known, thus, we impose a grid of potential heater source locations on one surface, and activate the heaters located under the hot spots visible in the temperature profile. The inverse model is validated with the results of a “numerical experiment” (i.e., solving the direct problem in FloTHERM and using the temperature maps as input to the inverse solver) and used to identify hot spots in a commercial device using temperature maps acquired experimentally with infrared microscopy. While demonstrated here for a microelectronic device, this method is broadly applicable to other systems where the distribution of heat generation is unknown and limited only temperature measurements are experimentally possible including batteries and manufacturing processes.