A Novel Temperature Estimation Technique Exploiting Carrier Emission from Buffer Traps

We propose a novel technique for temperature estimation in electron devices based on the mutual correlation between emission time constant from traps ( τ ) and temperature ( T ). Arrhenius equation is employed as the physical model relating τ and T The reference system used to present the technique is AlGaN/GaN high electron mobility transistors (HEMTs) with Fe-doping in the buffer. Drain Current Transients (DCTs) are used for extracting the emission time constant ( τ ) from Fe traps and non-linear regression through Trust Region Reflective (TRR) optimization algorithm is used to learn the model parameters from data and infer device temperature. Electro-thermal device simulations are employed for validating the proposed technique, showing that this method is able to provide an improved accuracy with respect to conventional electrical techniques (e.g., McAlister method) promoting it as a valid alternative to state of-the-art optical techniques in GaN HEMTs.