No abstract

No abstract

The authors investigate 2μm gate-length InAlN∕GaN metal-oxide-semiconductor high-electron-mobility transistors (MOS HEMTs) with 12nm thick Al2O3 gate insulation. Compared to the Schottky barrier (SB) HEMT with similar design, the MOS HEMT exhibits a gate leakage reduction by six to ten orders of magnitude. A maximal drain current density (IDS=0.9A∕mm) and an extrinsic transconductance (gme=115mS∕mm) of the MOS HEMT also show improvements despite the threshold voltage shift. An analytical modeling shows that a higher mobility of electrons in the channel of the MOS HEMT and consequently a higher number of electrons attaining the velocity saturation may explain the observed increase in gme after the gate insulation.

Pulse behavior of insulated-gate double-field-plate power AlGaN/GaN HEMTs with C-doped buffers showing small current-collapse effects and dynamic RDS,on increase can accurately be reproduced by numerical device simulations that assume the CN-CGa autocompensation model as carbon doping mechanism. Current-collapse effects much larger than experimentally observed are instead predicted by simulations if C doping is accounted by dominant acceptor states. This suggests that buffer growth conditions favoring CN-CGa autocompensation can allow for the fabrication of power AlGaN/GaN HEMTs with reduced current-collapse effects. The drain-source capacitance of these devices is found to be a sensitive function of the C doping model, suggesting that its monitoring can be adopted as a fast technique to assess buffer compensation properties

We address degradation aspects of lattice-matched unpassivated InAlN/GaN high-electron-mobility transistors (HEMTs). Stress conditions include an off-state stress, a semi-on stress (with a partially opened channel), and a negative gate bias stress (with source and drain contacts grounded). Degradation is analyzed by measuring the drain current, a threshold voltage, a Schottky contact barrier height, a gate leakage and an ideality factor, an access, and an intrinsic channel resistance, respectively. For the drain-gate bias <38 V parameters are only reversibly degraded due to charging of the pre-existing surface states. This is in a clear contrast to reported AlGaN/GaN HEMTs where an irreversible damage and a lattice relaxation have been found for similar conditions. For drain-gate biases over 38 V InAlN/GaN HEMTs show again only temporal changes for the negative gate bias stresses; however, irreversible damage was found for the off-state and for the semi-on stresses. Most severe changes, an increase in the intrinsic channel resistance by one order of magnitude and a decrease in the drain current by ∼70%, are found after the off-state ∼50 V drain-gate bias stresses. We conclude that in the off-state condition hot electrons may create defects or ionize deep states in the GaN buffer or at the InAlN/GaN interface. If an InAlN/GaN HEMT channel is opened during the stress, lack of the strain in the barrier layer is beneficial for enhancing the device stability.