Thermal Conductance of Buried AlN Interfaces Measured by Dual-Frequency Time-Domain Thermoreflectance

Thermal management is a critical concern in microelectronic devices, especially for third-generation semiconductor devices that are based on GaN and AlN due to high power density during operations. An in-depth understanding of their thermal properties is essential for thermal management, such as interfacial thermal conductance between different layers. However, accurately measuring the thermal conductance in buried interfaces, particularly in epitaxial growth interfaces, is substantially challenging due to its low measurement sensitivity. Conventional time-domain thermoreflectance (TDTR) technique can only achieve around 40% of uncertainty, or not able to separate the thin film thermal conductivity and the buried interfacial thermal conductance. In this study, we utilize dual-frequency TDTR technique with Monte Carlo simulations to precisely quantify the thermal conductance of buried AlN interfaces with various substrates.<br/>Firstly, we deposit AlN on common substrates, specifically, sapphire, silicon, and 4H-SiC, using molecular beam epitaxy (MBE). We carefully control the thickness of AlN to be 100 nm for the best experiment sensitivity. The quality of the AlN films is assessed by atomic force microscopy (AFM) and the picosecond acoustics technique of TDTR. Subsequently, TDTR measurements are performed at modulation frequencies of 18.9 MHz and 5.4 MHz, which leads to two different thermal penetration depths. To enhance the sensitivity of thermal conductance of AlN/substrate interfaces, we analyse the ratio of TDTR signals between 18.9 MHz and 5.4 MHz measurements. In addition, we employ Monte Carlo simulations to statistically estimate the uncertainty by considering results from both measurements and their ratio. The measurement uncertainty is reduced to approximately 1/3 of the conventional single-frequency TDTR measurements.<br/>Our method can accurately measure the interfacial thermal conductance between AlN and different substrates with each result exhibiting an uncertainty of less than 16%. Among these substrates, the high interfacial thermal conductance of AlN/4H-SiC (~450 MW/m²K) provides a promising prospect for effective heat dissipation across the interface.<br/>In summary, this study demonstrates a dual-frequency TDTR technique with Monte Carlo simulations that significantly enhances the precision by 3 times in interfacial thermal conductance measurements of buried interfaces with just 100 nm of AlN. This technique contributes to a guidance of interfacial thermal conductance measurements and exploring thermal management further in electronic devices.