Implanted ohmic contacts were made on molecular beam epitaxy grown GaN materials. Si was implanted at a doping density of about 4×1020  cm-3 to decrease the contact resistance of the contact, followed by an activation anneal at 1150 °C for 30 s. The overlay metal Ti/Au was evaporated. Four-probe measurements were performed on transmission line model patterns. The measured maximum contact resistance was 0.097 Ω mm and the apparent specific contact resistance was 3.6×10−8 Ω cm2.

High-quality AlGaN/GaN heterostructures have been grown on sapphire substrates by plasma-assisted molecular-beam epitaxy. Polarization effects are exploited to achieve a two-dimensional electron-gas sheet density of 8.8×1012 cm−2 and greater on intentionally undoped material with a measured room-temperature mobility as high as 1478 cm2/V s. Transistors were then fabricated from this material, yielding a unity current gain frequency of 50 GHz and a unity power gain frequency of 97 GHz. By increasing the buffer layer thickness, output powers of 1.88 W/mm at 4 GHz with an efficiency of 34% were achieved. These results prove that the polarization effects in the nitrides are as enormous as theory predicts. The key to the improved mobility and operation of the devices of the all-molecular-beam-epitaxy-grown material, the AlN nucleation layer, will be discussed.

Two-dimensional electron gases in AlGaN/GaN heterostructures grown by plasma induced molecular beam epitaxy have been formed by polarization induced interface charge effects. Growth of the “normal” structure (AlGaN on GaN) has formed a two-dimensional electron gas confined in the GaN when grown on sapphire with an AlN nucleation layer, SiC (Si face), or on GaN nucleated by organometallic vapor phase epitaxy on sapphire. Hall mobilities in GaN for normal interfaces are as high as 1238 cm2/V s at 300 K and 3182 cm2/V s at 77 K. Direct current results from field effect transistors fabricated from this material yield a maximum transconductance of 210 mS/mm and a current density of 710 mA/mm. Microwave measurements gave a ft of 27 GHz and a fmax of 37 GHz. When nucleating GaN directly on sapphire, an “inverted” structure (GaN on AlGaN) was used to create a two-dimensional electron gas. The difference in the normal and inverted structures is the polarity of the (Al)GaN layers. The flipping of the (Al)GaN polarity on sapphire is achieved through the use of the AlN nucleation layer.

Short-gate MODFET's of AlGaN/GaN on Sapphire have been fabricated and characterized with gate lengths in the .12 - .25 μm range. Values of ft = 50 GHz and fmax = 100 GHz have been obtained. Analyzing the performance, the average electron transit velocity is shown to be 1.25 × 107 cm/s and in some cases well under that value. This compares with theoretical predictions of ~ 2.0 × 107 cm/s. The electron scattering effects of dislocations, which are charged, are modeled to explain the lower mobility. Ion bombardment or dry etching is used for mesa isolation. Ti/Al/Ti/Au sintered for 100 seconds at 800 °C is used to yield ohmic contacts of .5 - 1.0 Ω-mm. Pt/Au Schottky gates are used. A high breakdown voltage, exceeding 100 V even for short gate MODFET's, shows that ten times higher load resistance values are possible, compared with GaAs MODFET's. Normalized output power levels well over 10 W/ mm are thus projected for GaN MODFET's on SiC substrates, where the thermal conductivity is about 5W/cm-°C. with future integrated traveling-wave, power-combining circuits, output power > 100 W at 10 GHz is predicted.

In this letter, an alternative approach to determine the polarity of GaN thin films based on the atomic location by channeling-enhanced microanalysis technique is described. Theoretical calculations provide a straightforward criterion for polarity determination that is a major advantage of this method. At the Bragg position, the thickness-averaged incident electron intensity, and hence, electron induced characteristic x-ray yield, is higher on the N plane than on the Ga if the g vector of the diffraction beam is parallel to the Ga–N bond direction, and vice versa. Experimental results support the theoretical predictions. The possible errors in the experiments are also discussed.