Nowadays GaN-based HEMTs are becoming popular for high frequency, high power as well as high temperature applications. GaN-based HEMTs offer low noise performance due to high electron concentration and low carrier scattering. In AlGaN/GaN or AlInN/GaN HEMTs, due to the polarization-induced electric field in the barrier layer, the surface donor states donate electrons to populate the 2-dimensional quantum well at the heterojunction, when the barrier layer is of more than a certain thickness (critical thickness), even in the absence of any applied gate voltage. So GaN-based HEMT devices are normally ON i.e., they are depletion type devices. But in circuit applications enhancement type devices are desirable for fail-free operations. Different techniques, like ion implantation, recess gate etching, and FinHEMT configurations are used to realize enhancement type device. Ion implantation and recess gate etching increases the device on resistance (RON). GaN-based FinHEMT has several advantages over planar HEMT. It shows normally off operation, less on resistance (RON), high ION/IOFF ratio, less gate leakage current, smaller sub-threshold slope (SS), and less RF-DC dispersion than planar HEMT. In FinHEMT, gate has higher control over channel compared to planar HEMT. Transconductance (gm) is more linear with VG in FinHEMT and hence current gain cut-off frequency (fT) is also more linear in FinHEMT compare to planar HEMT. So during RF applications we can operate FinHEMT at higher VG which gives higher output power density. The characterization results of FinHEMTs fabricated in our lab will be discussed.

Another technology of realizing e-mode HEMT is by using floating gate. Using floating gate technology, e-mode AlGaN/GaN HEMT with large positive threshold voltage (greater than 2 volt) is possible. In this technique, a charge storing layer is placed in-between the tunnel dielectric and blocking dielectric just below the gate metal and the charge storing layer is charged before operation by applying high positive gate voltage. This stored charge depletes electrons from the channel and shifts the threshold voltage towards positive direction. Thus e-mode operations is obtained by charge storing without reducing the channel conductance. But a single large charge storing layer is very likely to fail. Instead of using one large charge storing layer, if the charge storing layer is split into 5, the failure probability reduces to 10-12. Hence split floating gate structure is a viable route to e-mode GaN HEMTs. I would like to explore floating gate FinHEMTs during the next phase of my work.