Transmutation doping of (In, Ga, Al)N compounds by neutron irradiation is a promising and totally unexplored field to date. It is much more effective than that of Si due to large neutron capture cross section and abundance of In, Al and Ga isotopes participating in reaction. This should make the irradiation possible in low-flux reactors and result in smaller radiation damage. Annealing of the radiation damage seems feasible.

Copper Schottky diodes on n-type GaN grown by metal-organic chemical vapor deposition were achieved and investigated. Ti/Al was used as the ohmic contact. The copper metal is deposited by the Sputter system. The barrier height was determined to be as high as (ΦB =1.13eV by current-voltage (I-V) method and corrected to be ΦB =1.35eV as considered the ideality factor, n, with the value of 1.2. By the capacitance-voltage (C-V) method, the barrier height is determined to be ΦB =1.41eV. Both results indicate that the sputtered copper metal is a high barrier height Schottky metal for n-type GaN.

Thin films of ScN and YN were grown on silicon, quartz and sapphire using metal evaporation and an RF atomic nitrogen source. YN decomposes on contact with water vapor, and only AlN capped films could be stabilized. ScN is stable in air and water, and thin films of this material deposited at temperatures between 300 and 900 °C show a substrate-dependent film texture. Typical growth rates were ∼ 0.1 nm/second with a 300W N discharge at about 0.1 mTorr Nitrogen pressure. Structural characterization by x-ray diffraction, infrared transmission spectroscopy and Hall effect measurements on n-type ScN and the fabrication of p-n junctions of n- type ScN with silicon are presented.

Gallium nitride (GaN) has been under intense investigation due to its unique qualities (wide band gap, chemical and temperature stability) for optoelectronic and high temperature/high power applications. To this end, reactive ion etching (RIE) experiments were performed on GaN thin films using BCl3/Cl2/Ar. These resulted in etch rates of 1400 Å/min at −400 V dc bias. However, rough etched surfaces, nitrogen surface depletion and high chlorine content were observed. In order to remedy these shortcomings, a photo-assisted RIE process using a filtered Xe lamp beam was developed, resulting in higher etch rates but again in nitrogen depleted surfaces. Preliminary results on using nitrogen instead of argon in the process chemistry show a big improvement in photo-asssisted etch rates (50%) and Ga/N ratio (0.78 versus 1.25). In this paper, the effects of epilayer doping, dc bias, nitrogen flow rate and photo-irradiation flux on GaN etch rates, surface morphology and composition are presented. Finally, preliminary results on the use of a KrF excimer laser beam in the GaN photo-assisted RIE process are presented.

The drain-current response to short (<1μs) gate pulses has been measured for a series of GaN HEMT wafers that have similar dc and small-signal characteristics. This response has been found to correlate well with the measured microwave power output. For example, for devices where the pulsed drain current is greater than 70% of the dc value, output power densities of up to 2.3 W/mm are attained. This is in contrast with 0.5 W/mm measured for devices with low pulse response (less than 20% of the dc value). These results, which can be explained by the presence of traps in the device structure, provide a convenient test which is predictive of power performance.

Current collapse in GaN MESFETS is believed to result from the trapping of carriers in the high resistivity GaN layer, and can be reversed by the application of light. Light photoionizes (or photoneutralizes) the carriers, releasing them from the traps and restoring all or part of the original I-V characteristics of the device. In these investigations we have taken advantage of this effect to characterize the traps responsible for current collapse in an n-channel GaN MESFET. At fixed source-drain voltage, the incremental light-induced drain current, above that measured in the dark, and normalized per incident photon, is measured as a function of wavelength. The resulting photoionization spectrum reflects two absorption thresholds corresponding to two distinct electron traps. Because of the nature of the measurement, these traps can be identified as those responsible for current collapse in the device.