Prof. R. J. Trew (Case Western Reserve U) started the Symposium with a review of wide bandgap semiconductor amplifier devices. He noted that GaN FETs have performance similar to the more established 4H SiC for frequencies less than 10 GHz, the range pred ominantly studied to date, probably because of the similarity in mobility for a given electron concentration in the two materials. From analysis of device models, the problems with p-type doping in GaN (difficulty in achieving low resistivity) may limit HBT applications at 300 K but not at elevated temperatures. Mr. S. Binari (Naval Research Laboratory) presented preliminary barrier height vs metal work function data for Schottky metallizations (Au, Cr, Ni, Pd, Pt, and Ti: data from NRL and other groups) which suggested that the GaN-metal interface was not following a simple Schottky-Mott model but that some degree of Fermi level pinning was occurring. Also presented were GaN-based FET performance data of fT vs gate length, shown in the Figure. The MESFET and MODFET data fall into two groups, with fT approximately following the expected inverse gate length functional dependence. The GaAs MESFET functional dependence is also shown for comparison. Recent data by Khan and collaborators (APA Optics, diamond symbols in the Figure), suggests that GaAs-like performance may be achievable.
A demonstration of an OMVPE grown single QW GaN/InGaN/GaN LED structure emitting in the blue region by Prof. S. Denbaars and co-workers (UCSB) was the direct result of extensive nucleation and growth studies (portions of which were first reported at the Fall 1995 MRS Meeting by Prof. J. Speck et al.). Superior electronic properties were obtained for a short sapphire nitridation time, 60 s at 600 C, before growing the 20 nm GaN nucleation layer. The performance of a MESFET device fell on the Binari curve and had 7W mm-1 of power gain.
Dr. J. Redwing (ATMI) presented data on the transport properties of unintentionally doped Al0.15Ga0.85 /GaN 2DEG structures grown on 6H SiC substrates by OMVPE. Shubnikov de Haas oscillations were convincingly demonstrated for the first time and a record Hall mobility of 7500 cm2V-1s-1 at 20 K was recorded, suggesting further improvements in MODFET device performance may be expected in the future. Similar intentionally Si-doped 2DEG structures did not exhibit increased sheet charge density as compared to the unintentionally doped case and hence the source of electrons in 2DEG structures remains unresolved.
Mr. J. C. Roberts (NCSU) of Prof. Bedair's group presented work on InxGa1-xN alloys, heterostructures and QWs grown by the OMVPE ALE technique. In particular the QWs, cladded with AlGaN, exhibited 300 K PL emission ranging from 390 (uv, x=10%) through 456 (blue), 505 (green), and 550 (yellow) to 558 nm (orange, x=64%). As the In mole fraction was increased to the largest values, PL intensity got weaker and the FWHM increased. The key to growing these structures involves the formation of a suitable InGaN "buffer layer" that is grown through a combination of step and temperature gradings; details to be forthcoming at future conferences.
With respect to substrates, Dr. T. Suski (Lawrence Berkeley Laboratory) presented MBE results on GaN platelets grown by UNIPRESS (Research Center, PAS). The PL was dominated by a bound exciton peak, FWHM of 0.6 meV, and little or no yellow band was observed. HRTEM investigation of the homoepitaxial interface showed a concentration of dislocations within 100 nm of the interface and 106 - 107 cm-2 near the top of the epilayer.
Efforts on HVPE growth of GaN were reported by Prof. T. Kuech and co-workers (U. Wisconsin), Dr. R. Molnar and co-workers (MIT Lincoln Laboratory), and Prof. J. Harris and co-workers (Stanford). State-of-the-art transport properties and 5 x 10 7 cm-2 dislocations were reported by Molnar for 40 µm layers grown on "seeded" ZnO/sapphire heterostructures. Since no ZnO was detected at the interface by TEM nor was any Zn signature recorded in PL, it was conjectured that the ZnO evaporated during the growth process. The effects of gas phase interactions and gas flow behavior was investigated and modeled by Safvi, Kuech and company to optimize their HVPE reactor for improved film morphology on 2 inch sapphire substrates. Growth rates up to 180 µm hr-1 were obtained with excellent crystallinity, < 300 arc sec FWHM double crystal x-ray data.
Dr. J. C. Zolper (Sandia National Laboratory) achieved p-type doping in GaN by co-implanting Ca with P and subsequently annealing the sample to 1100 C for p-type conversion (similar to Mg). The activation energy of holes in the layer was 169 meV for Ca and 171 meV for Mg, similar to the 160-180 meV range reported by Dr. W. Goetz (Xerox) for Mg dopants incorporated during OMVPE growth. Dr. Zolper then made a JFET structure via implantation of Ca with an fT of 2.7 GHz for a 1.7 µm gate, gm =7.5 mS mm-1, which falls on the MESFET portion of the Binari curve.
Mr. A. Y. Polyakov and co-workers (Carnegie Mellon U) reported that native donors are effectively passivated by H from a plasma in unintentionally doped OMVPE AlGaN for Al mole fractions up to 100%. They noted that the passivation process decreased the electron concentration while the mobility remained unchanged or increased. These native donors are thought to be responsible for the "softness" of the absorption edge since after exposure to the H plasma the absorption edge became more abrupt.
NSOM investigations by Mr. Liu and co-workers (U. Wisconsin) showed that the yellow luminescence in HVPE GaN is enhanced by step edges on the surface. It was proposed the yellow band PL is related to chemical impurities and that the NSOM observations implies that impurity incorporation at step edges is enhanced.
Yung-Chung Kao and collaborators (Texas Instruments) reported the first growth of InAsN lattice matched to GaAs. This ternary system has considerable bowing in the bandgap with far infrared or semi-metallic absorption behavior predicted. The films were grown by MBE at 0.6 to 1.0 µm hr-1 at a growth temperature of 450 to 600 C. Best results were achieved for temperatures below 500 C but films contained significant polycrystalline InN and InAs components for temperatures near 450 C. For an In mole fraction of 38.6%, the lattice mismatch was less than 1%. Growth on GaAs buffer layers resulted in the best films with x-ray rocking curve FWHM of 270 arc sec reported. For the best film, a mobility of 100 cm2V-1s-1 for an electron concentration of 10 20 cm-3 was found. First attempts to grow the lattice matched composition to Si (60 % In mole fraction) resulted in polycrystalline material with poor morphology.
During the well-attended panel session, the effects of nanopipes on GaN devices was discussed. No data is available at this time to ascertain whether nanopipes effect device performance, although it was suggested (based on 6H SiC experience) that breakdown voltages may drop up to two orders of magnitude. OMVPE reactor designs were discussed and an informal "vote" by the audience favored a vertical, high speed, low pressure design. It was pointed out that sparse information exists about the kinetics/dynamics of the GaN growth process. For example, Denbaars relates that Dr. S. Nakamura (Niichia) claims that his reactor design results in a 1 mm boundary layer which he believes to be critical to the process, yet no detailed information is available in the literature. Until this essential information is available, improvements in current reactor designs may be difficult.
Contributed by Kurt Gaskill