In this paper, we report an analysis of reverse current mechanisms observed in GaN Schottky barrier diodes leading to hysteretic behavior of the I–V curves at low temperature. By means of DC measurements from 33 to 475 K, we demonstrate the presence of two leakage mechanisms when comparing the experiments with the results obtained using a unified model to predict the ideal reverse current of the diode. Poole–Frenkel emission is the dominant mechanism for temperatures above 200 K, while trap-assisted tunneling prevails for lower temperatures, where also, hysteresis cycles are revealed by means of DC dual-sweep voltage measurements. The energy of the corresponding traps has also been determined, being around 0.2 and 0.45 eV, respectively. The hysteresis phenomenon is attributed to the bias-induced occupancy of the energy states originating the leakage-current processes, which leads to the reduction of the reverse current after a high negative voltage is applied to the diode.

The hysteresis cycles observed in the reverse leakage current measured at low temperatures in GaN-on-sapphire Schottky barrier diodes (SBDs) have been deeply studied and interpreted in terms of trap-assisted tunneling (TAT) compatible with the existence of a trap energy band (TEB) near the metal-semiconductor interface. The analysis of the energies for which the maximum tunneling current occurs, both direct and through this trap band, allows us to explain the behaviors found at different temperatures. Starting from empty trap states by previous illumination, transient current measurements performed under different preconditioning voltages evidence a progressive partial filling/release of the trap energy levels, confirmed as well by pulsed measurements. Captured/released electrons modify the number of trap states available for tunneling and thus the current level. As a consequence, tunneling processes in the go and return paths take place through levels with different occupations within the TEB, originating the hysteresis cycle.

Transfer technology is now becoming very attractive not only for new technologies such as flexible technology but also for solid state technologies when performances are limited by technological barriers that have to be overcome. In this last context, the transfer of high electron mobility transistors (HEMTs) on diamond substrates represents an opportunity to improve the thermal dissipation when the device operates at high radio frequency power levels. Up to now, the technological process for the transfer of these transistors is not detailed in the literature. In this article, the first demonstration of AlGaN/GaN HEMTs on diamond substrates by transfer technology obtained through sputtered aluminum nitride (AlN) layers bonding at low temperature is reported. Devices are first fabricated on AlGaN/GaN epilayers grown on silicon (Si) substrates. Afterward, AlGaN/GaN thin films with devices are released from the Si growth substrate and transferred at 160 °C onto a diamond substrate thanks to an AlN bonding layer. A full description of the transfer technology and all the technological limits and risks are presented. The transferred device provides a maximum DC drain current density IDS Max of 690 mA mm−1 at VGS = 0 V. Furthermore, a cutoff frequency fT of 85 GHz and a maximum oscillation frequency fMAX of 106 GHz are extracted from S-parameter measurements.

Planar Gunn diodes based on doped GaN active layers with different geometries have been fabricated and characterized. Gunn oscillations have not been observed due to the catastrophic breakdown of the diodes for applied voltages around 20–25 V, much below the bias theoretically needed for the onset of Gunn oscillations. The breakdown of the diodes has been analyzed by pulsed ${I}$ – ${V}$ measurements at low temperatures, and it has been observed to be almost independent of the geometry of the channels, thus allowing to discard self-heating effects as the origin of the device burning. The other possible mechanism for the device failure is an impact-ionization avalanche due to the high electric fields present at the anode corner of the isolating trenches. However, Monte Carlo simulations using the typical value of the intervalley energy separation of GaN, $\varepsilon _{{1}-{2}}$ = 2.2 eV, show that impact ionization mechanisms are not significant for the voltages for which the experimental failure is observed. But recent experiments showed that $\varepsilon _{{1}-{2}}$ is lower, around 0.9 eV. This lower intervalley separation leads to a much lower threshold voltage for the Gunn oscillations, not far from the experimental breakdown. Therefore, we attribute the device’s failure to an avalanche process just when Gunn domains start to form, since they increase the population of electrons at the high electric field region, thus strongly enhancing impact ionization mechanisms that lead to the diode failure.

A Differential Hall Effect method has been developed that allows to determine of the dopant activation level of SiGe alloys in the surface region. The technique is first optimised in terms of etch rate (∼ 1Å/min), surface roughness, strain and stoichiometry stability during etch. We then show that this DHE method allows to experimentally access the dopant activation of ultra-shallow surface layers of less than 0.5 nm.