|
|
1. The investigation of the field distribution at
high drain biases (up to 200 Volts). This included the
two dimensional simulations of the electric field
distribution. These simulations showed that the electric
field tends to peak near the get edge and the high field
region does not extend that much towards the drain, in
contrast to our own experimental data that showed that
the breakdown voltage increases nearly proportionally to
the drain-to-gate spacing.
2. Simulations of a possible 2D-hole gas enhancement at
the AlGaN/GaN heterointerface. The results show that it
should be possible to induce the 2D-hole gas in p-type
GaN by piezoelectric effects but it might be much more
difficult (or impossible?) to invert n-type GaN.
3. The theory of the gate leakage current (in progress).
The preliminary results point to a low forward leakage
but confirm the conclusions of Eastman's group of a
possible large reverse leakage at high sheet electron
densities. We plan to set up a two-diode gate current
model suitable for the microwave simulations.
Currently, we are calculating bulk band structures
and overlap factors for bulk wurtzite semiconductors
using the full-band empirical pseudopotential method. By
considering piezoelecteric lattice deformation due to
high applied electric fields and the resulting changes in
the structure factors, we plan to compute modifications
to the bandstructure, overlap factors, and scattering
rates at high electric fields. Although we have
previously obtained the wurtzite GaN band structure based
on a tight-binding approach, its accuracy may be
inadequate in the high energy regime.
Dual ram uniaxial wafer bonding system has been
designed and completed. Gallium nitride samples have been
grown on SiC substrates for initial bonding studies.
1. Investigation of the growth conditions, mask
design and mask material on the electrical properties of
LEO GaN with the goal to develop a semi-insulating base
layer for device applications. 2. Fabrication of
AlGaN/GaN HEMT devices on LEO GaN on sapphire wafers. The
conducting LEO GaN layer was separated from the channel
by a thick semi-insulating GaN layer. The LEO based
devices showed a two orders of magnitude lower gate
leakage than HEMTs grown on regular, dislocated GaN base
layers. Measurements to study the effect of dislocations
on the transport properties of the devices are presently
under way. 3. Investigation of the effects of growth
conditions and strain on the formation of defects in thin
AlGaN layers grown on GaN. Under optimum growth
conditions, the formation of additional defects in the
AlGaN layer close to the GaN/AlGaN interface could be
significantly suppressed. In addition, the influence of
the composition and the AlGaN layer thickness on the
properties of the 2DEG has been studied.
We have our HVPE reactor up and running, growing
films of undoped GaN on two inch sapphire wafers. There
is some variation in the quality of the films but they
are uniformly clear with no coloration. We are growing
films of different thicknesses, ranging from ten microns
to over a hundred microns. However, as has been
experienced by other groups, the films can be extensively
cracked and we are trying to get to the bottom of this.
We have concluded that cracking is most probably
occurring during growth and not during subsequent
cooling. The stresses in the films at room temperature
are compressive and our wafer curvature measurements of
the stresses are consistent with those calculated using
the known thermal expansion coefficients and moduli. We
can grow films that are crack-free over large areas but
it is not yet reproducible. The films in these areas have
mobilities of ~ 80 but we have not made any attempts to
maximize this through careful control of impurities etc.
Preliminary thermal conductivity measurements on the LEO
material grown on sapphire at UCSB show values that range
between 30 and 46 W/mK. These are provisional data and
need to be confirmed.
The growth of GaN and GaN-related structures via
lateral epitaxial overgrowth (LEO) has been conducted.
Growth over larger areas has been achieved. Layered
structures consisting of two or more layers of different
materials have also been achieved using the general LEO
technique.
A recurrent issue in the dry etching of the GaN
materials has been a slight overcut slope that persists
under a variety of pressure, power and bias conditions.
By using a specially designed Radical beam Ion Beam Etch
(RBIBE) station, with a heatable and tiltable substrate
holder, we have managed to achieve straight side walls.
We are currently varying etch gas composition, ion energy
and ion current in order to both understand and optimize
the etch chemistry. The Photoelectrochemical Etch station
has been largely reconfigured and redesigned for better
handling of and electrical access to the substrate. We
are currently carrying out and characterizing selective
etch processes which stop on either AlGaN or p-GaN. This
will facilitate device fabrication (e.g. HBTs) in a
variety of ways. For example, large area lift-off of GaN
materials has been achieved in Professor Clarke's group
through laser disintegration of the GaN material near the
GaN-sapphire interface. The surface roughness of the
resulting lifted-off layer may make some subsequent
device processes (such as wafer fusion) difficult. We are
examining the possibility of a post lift-off PEC etch,
for samples that incorporate etch-stop layers, such as
AlGaN.
|
