Abstract

Silicon technology, which is the most mainstream semiconductor technology, poses serious limitations on fulfilling the market demands in high-frequency and high-power applications. In response to these limitations, wide bandgap III-nitride devices, including AlxGa1-xN/GaN heterojunction field effect transistors (HFETs), were introduced at about two decades ago to satisfy these rapidly growing market demands for high-power/high-frequency amplifiers and high-voltage/high-temperature switches. The most appealing features of III-nitride technologies, and particularly AlxGa1-xN/GaN HFETs, in these applications, are the polarization-induced high sheet-carrier-concentration, high breakdown-voltage, high electron saturation-velocity, and high maximum operating temperature. Therefore, the development of enhancement-mode AlGaN/GaN HFETs is one of the most important endeavours in the past two decades.
Low-frequency noise (LFN) spectroscopy, empowered by a proper physics-based model, is received as a capable tool for reliability studies. As a result, devising a physics-based LFN model for AlGaN/GaN HFETs can be capable of not only evaluating the alternative techniques proposed for realization of enhancement-mode AlGaN/GaN HFETs, but also more importantly forecasting the reliability, and noise performance of these devices.
In this dissertation, for the first time, a physics-based model for the low-frequency drain noise-current of AlGaN/GaN HFETs is proposed. The proposed model, through including the thermally-activated and quantum tunneling processes of trapping/de-trapping of electrons of channel into and out of the trap-sites located both in the barrier- and buffer-layer of these HFETs, provides a descriptive picture for the LFN behavior of these devices. This work also aims to experimentally investigate the low-frequency noise-current characteristics of both conventional and newly-proposed devices (i.e., fin-, and island-isolated AlGaN/GaN HFETs) at various temperatures (i.e., 150, 300, and 450 K) and bias points in order to address the possible difficulties in performance of these devices. Matching of the trends proposed by the physics-based model to the experimentally recorded LFN spectra of AlGaN/GaN HFETs designed according to a newly-proposed technological variant for positive-shifting the threshold-voltage, confirms the accuracy and predicting power of the proposed model. The insights gained from this model on the latter group of devices provide evidence for the challenges of the aforementioned technological variants, and as a result offer assistance in proposing remedies for those challenges.
In formulating the LFN model, a massive discrepancy between the predictions of the existing analytical relationships used by others in evaluating the subband energy levels of AlGaN/GaN HFETs and the realities of the polarization-induced electron concentration of these HFETs was spotted. Careful evaluation of the polarization properties of these heterostructures unmasked the inaccuracy of the assumption of zero penetration of the electron wave into both the AlGaN barrier-layer and the GaN buffer-layer as the culprit in this discrepancy. In response to this observation, a model based on the variational-method for calculating the first and second subband energy levels of AlGaN/GaN HFETs is developed. On the basis of this model, more accurate analytical frameworks for calculating these subband energy levels in AlGaN/GaN HFETs for a variety of barrier thicknesses and Al mole-fractions in the barrier-layer are proposed.