|M.Sc Student||Ber Emanuel|
|Subject||Interface Traps Characterization in GaN Devices using UV|
Assisted Gated Van Der Pauw Measurements
|Department||Department of Nanoscience and Nanotechnology||Supervisor||PROF. Dan Ritter|
|Full Thesis text|
In the past few years there has been a constantly growing interest in material systems for the replacement of silicon as the basis of the semiconductor industry. One prominent group of materials is the III-V semiconductors. These materials are all characterized by high electron mobility, compared to silicon, and therefore are considered as prime candidates for its replacement in RF, and power, applications. III-Nitrides in general, and gallium nitride (GaN) in particular, are one of the most interesting III-V sub-groups, as they exhibit high values of spontaneous polarization. Furthermore, the large bandgap of GaN makes it a great material for high switching speed, and high power, devices. The basis for GaN electronics is the GaN/AlGaN heterojunction. The polarization in both materials results in the formation of two dimensional electron gas (2DEG) at their interface, even for undoped materials. This junction is the core of the GaN transistors, as the 2DEG can act as the channel, and the lack of doping contributes to the high electron mobility.
Although the 2DEG unarguably compensates the net polarization charge at the GaN/AlGaN interface, the nature of the compensating charge at the surfaces is still unclear. The most popular model that tries to explain the surface charge is the surface donor model by Ibbetson et al. It relies on the assumption that a high concentration of positively charged empty surface donors acts as the surface charge. A few problems arise when carefully examining that assumption. Namely, the measured surface donor density is too low to compensate the polarization charge, and the lack of Fermi level pinning which is essential to the model.
In this work, we have evaluated the interface states density (Dit) at the surface of a GaN/AlGaN/GaN heterojunction, using the newly developed UV assisted gated van der Pauw method. We introduce a direct manner of measuring the D?it, as no models are required for our performed experiments. Comparisons of results for Schottky and insulated gate devices confirmed that our new method is selective towards interface states. The obtained Dit values were about two orders of magnitude lower than required for the polarization charge compensation. We suggest an alternative polarization self-compensating (PSC) charge. The nature of the PSC is yet unclear but some comments are presented. Its most noticeable feature being that it is fixed, i.e. not affected by perturbations such as the application of an electric field. Contrasting the surface donors charge which is variable, i.e. dependent on the surface donors’ occupancy. We also ran TCAD simulations to reiterate that the Fermi level pinning is an inherent result of using the surface donor model. In contrast, when we used our alternative PSC fixed surface charge, there was no pinning, which matches the common experimental results.
We hence concluded that variable surface charge, such as surface donors, could not compensate the polarization surface charge. Instead, a yet unexplained PSC surface charge is proposed to account for experiments.