The 10th Asian Symposium on Computational Heat Transfer and Fluid Flow ( ASCHT2025)

Gallium nitride high electron mobility transistors (GaN HEMTs) have attracted widespread attention in high-frequency and high-power applications due to their superior electrical properties. However, their performance is significantly limited by severe self-heating effects, making accurate junction temperature prediction critical. In most existing studies, the internal heat source of GaN HEMTs under different thermal resistance conditions is simplified as a constant heat flux. Although this assumption facilitates thermal modeling, it neglects the intrinsic thermal–electrical coupling, wherein the power dissipation of the device decreases with rising temperature, thus reducing the accuracy of temperature prediction. To address this issue, we develop a bidirectional electro-thermal coupling model based on semiconductor device physics. This model allows the heat generation to respond dynamically to the device temperature, enabling more precise junction temperature estimation. Three cases are set up to examine the difference in temperature prediction between the conventional constant heat flux method and the proposed coupled model. Under the tested conditions, the constant heat flux assumption yields a temperature prediction error of up to 17.4%, highlighting the necessity of employing a bidirectional electro-thermal approach in GaN HEMT thermal analysis.
 

GaN HEMT,Electro-thermal coupling,Junction temperature,Microchannel cooling