Free and bound excitons have been studied by photoluminescence in thin (1-10 µm) wurtzite-undoped GaN, n-type GaN:Si as well as p-type GaN:Mg and GaN:Zn layers grown by metal-organic chemical vapour phase deposition (MOCVD). An accurate value for the free A exciton binding energy and an estimate for the isotropically averaged hole mass of the uppermost 9 valence band are deduced from the data on undoped samples. The acceptor-doped samples reveal recombination lines which are attributed to excitons bound to Mg 0 and Zn 0 respectively. These lines are spectrally clearly separated and the exciton localization energies are in line with Haynes' rule. Whenever a comparison is possible, it is found that the exciton lines in these thin MOCVD layers are ultraviolet-shifted by 20 to 25 meV as compared to quasi-bulk (≥ 100 µm) samples. This effect is interpreted in terms of the compressive hydrostatic stress component which thin GaN layers experience when grown on sapphire with an AlN buffer layer.

The temperature and excitation power dependence of a bound exciton photoluminescence line S with a localization energy Q=11.5 meV has been studied in undoped and moderately Mg-doped wurtzite GaN of high resistivity. The data provide strong evidence that line S is due to recombination of excitons bound to ionized shallow donors. The consistency of this assignment with theoretical predictions is demonstrated

Oxygen doped GaN has been grown by metalorganic chemical vapor deposition using N2O as oxygen dopant source. The layers were deposited on 2" sapphire substrates from trimethylgallium and especially dried ammonia using nitrogen (N2) as carrier gas. Prior to the growth of the films, an AlN nucleation layer with a thickness of about 300 AA was grown using trimethylaluminum. The films were deposited at 1085 degrees C at a growth rate of 1.0 mu m/h and showed a specular, mirrorlike surface. Not intentionally doped layers have high resistivity (>20 kW/square). The gas phase concentration of the N2O was varied between 25 and 400 ppm with respect to the total gas volume. The doped layers were n-type with carrier concentrations in the range of 4*1016 cm-3 to 4*1018 cm-3 as measured by Hall effect. The observed carrier concentration increased with increasing N2O concentration. Low temperature photoluminescence experiments performed on the doped layers revealed besides free A and B exciton emissi on an exciton bound to a shallow donor. With increasing N2O concentration in the gas phase, the intensity of the donor bound exciton increased relative to that of the free excitons. These observations indicate that oxygen behaves as a shallow donor in GaN. This interpretation is supported by covalent radius and electronegativity arguments

The effects of annealing pressure, temperature, and time (p,T,t) on residual levels of n-channel insulated gate field effect transistor (IGFET) process induced insulator radiation damage has been examined, and the behavior over the range p =1-5 arm H2, T = 300~-450~ and times up to 14 hr at 300~ have been defined for polysilicon gated devices. It has been found that higher temperatures, higher pressures, and longer times all lead to decreased residual neutral trap and positive charge densities levels, but' that neutral traps are more difficult to remove. Removal of positive charge does not appear to depend on initial damage level while residual neutral trap level is sensitive to the initial damage level. It has been found also that with the specific damage technique employed, which involves electron beam (E-beam) metal evaporation, positive trap levels saturate at about 5 x 10 ~ rads Si, while apparent negative charge and neutral traps appear to increase slowly with exposure of the gate insulator to this ionizing radiation. Adequate removal of all charged and neutral centers can be achieved in 50 atm of forming gas at 400~ for 30 min for doses as great as 5 x 107 rads SIO2. If damage levels are not too high, 25 atm treatments are useful for the same temperature and time.When subjected to ionizing radiation during fabrication, the insulators in insulated gate field effect transistor (IGFET's) exhibit damage in the form of fixed positive charge and neutral traps (1). This damage requires thermal annealing treatments in the 550~176 range for its complete removal. Metallized shallow junction devices cannot, however, be subjected to annealing temperatures much in excess of 400~ for perhaps 30 min because of the danger of junction penetration by the interconnection metal. Such an inadequate temperature-time cycle results in residual damage levels of 100-200 mV of equivalent net positive charge generally, and even larger levels of neutral traps which pose both yield and reliability problems in small dimension IGFET's (in the 1 sm ground-rule range).Recently (2), first results were presented on the use of elevated pressure annealing cycles to remove such damage in polysilicon gated capacitors and n-channel IGFET's. The cited results were for a single damage level, ca. 107 rads Si, introduced in a specific way, and a single set of annealing conditions, 50 atm 10% H2-90% Ar, 400~ 30 min. Most of the preliminary data were in the form of capacitor flatband shifts, only a small number dealing directly with trapping of charge in insulator defects as a measure of damage. Commercially available elevated pressure systems have an upper operating pressure range of 20-25 arm, and it was of interest to learn how effective this pressure range might be in damage removal. In addition, FET threshold measurements are of greater practical significance than capacitor flatbands. Equally important is the observation that the annealing rate of FET gate insulators is different than that of the insulator of the much larger capacitor structure us...

No abstract