Transient Thermal Resistance Analysis for IC Packages

Abstract

The next-generation 5G mobile network causes handheld device will be linked to various things in life, such as wearable devices, electric vehicles, GPS, AR/VR, smart home system, IOT, etc. In the trend of lighter, thinner and more functions, it also suffers from several issues like high power consumption and overheating, and high performance and low power consumption can meet the requirement. In order to meet these expectations, IC need to design low or ultra-low current (nA~μA), low clock speed and command simplification, etc., however, for these milliseconds or microseconds signals, more complicated operation modes and rapid switching frequency, the traditional temperature measurement cannot accurately measure the instantaneous temperature variations. If the temperature rising is evaluated by transient thermal simulation, the correct model size and material characteristics such as density and specific heat are required, but it’s difficult to find the correct material information for multiple material, for example: plating material trace on silicon die, substrate and PCC, etc. Therefore, temperature variations in transient simulation cannot be correctly analyzed.This paper evaluates the difference of thermal resistance measurement between JEDEC traditional measurement method and TDIM (Transient Dual Interface Measurement) in QFN and HFCBGA, and compares the difference and causes of the results under different measurement environments. In addition, TDIM can evaluate the thermal resistance of each structural layer and also measure the instantaneous variations in temperature and thermal resistance. Finally, the transient thermal model can be provided after correction with simulation software to ensure that the transient simulation result is better and more accurate. The results show that the difference between TDIM and traditional method isn’t significant when the heating power is smaller; the larger heating power cause the difference is increases significantly to more than 10%. Through the process of structure function can find out that defects in the heat path may cause large thermal resistance. The corrected thermal model can be closer to the real measurement result in the transient analysis, and also significant difference in system level thermal simulation.