I. Introduction
Today’s state-of-the-art technology for efficient high-power radio frequency (RF) applications is based on gallium nitride (GaN) high electron mobility transistors (HEMTs) due to the superior properties of GaN: high electron mobility, high dielectric strength, high current density, and ability to work at high temperatures. However, the current GaN HEMT technology is thermally limited by device self-heating constraining the achievable saturated drain current densities, output powers, and degrading device reliability. For instance, for an increase in the channel temperature from room temperature to 187 °C, the transconductance (), cutoff frequencies (, and the saturation current () fall by around 35% due to the reduced saturated electron velocity and mobility in the channel at the higher temperature [1]. For improved thermal management of AlGaN/GaN HEMTs, high thermal conductivity silicon carbide (SiC) substrates are widely used in devices for high power density RF applications [2], while diamond substrates, with the highest thermal conductivity of , have been actively researched over a number of years now [3], [4]. Additional methods include packaging [5], [6], liquid cooling [7]–[11], and high thermal conductivity material incorporation in the device [12]. To date, however, thermal management through heat removal through substrate alone has proven to be inadequate for GaN technology, and so to advance the technology we need to think about ways to generate less heat in the first place.