High thermal boundary conductance across a GaN/SiC interface characterized via signal-ratio-based dual-frequency time-domain thermoreflectance

Abstract

To realize the full potential of GaN high electron mobility transistors (HEMTs), device-level thermal management is necessary by reducing the thermal resistance of the constituent layers, substrate, and interfaces. Accurate time-domain thermoreflectance (TDTR) measurement of a high thermal boundary conductance (TBC) between a high thermal conductivity buffer layer and substrate remains a challenge due to the low measurement sensitivity and the interdependence between thermal resistances associated with the adjacent materials and interfaces during the data fitting process. In this work, a dual-frequency TDTR approach is demonstrated that overcomes this limitation by analyzing the ratio of the TDTR signals acquired from high and low modulation frequency measurements. This TDTR signal-ratio-based approach enables accurate determination of the TBC with high precision by suppressing the measurement sensitivity to parameters other than the TBC that exhibit weak frequency dependence. Using this approach, the measurement uncertainty of the TBC across the GaN/SiC interface improves by more than a factor of two to three compared to that for a conventional TDTR method. The measured high TBC of 420 − 75/ + 105 MW m−2 K−1 across the GaN/SiC interface agrees with previously calculated values in the range of 480–545 MW m−2 K−1.