Gallium Nitride (GaN) is a wonder material which has widely transformed the world by enabling
energy-efficient white light-emitting diodes. Over the past decade, GaN has also emerged as one
of the most promising materials for developing power devices which can operate at significantly
higher power densities, higher temperatures, and higher frequencies, thanks to inherently superior
material properties like higher bandgap, 10x higher critical electric field, and 3x higher electron
saturation velocity, compared to silicon.
Lateral GaN high electron mobility transistors (HEMTs) based on the AlGaN/GaN heterostructures
capable of switching at high frequencies over 10 MHz have been already commercialized
and are the device of choice for implementing modern-day adapters and for wireless charging solutions.
However, for high-voltage and high-current applications, it is envisaged that vertical GaN
power devices will play a crucial role given that these devices dont scale in size for increasing
the BV unlike HEMTs, and are not affected by surface trap related reliability issues. The main
bottleneck towards the commercialization of vertical GaN devices on bulk GaN substrates is the high
cost and small size availability of these substrates. Similar to lateral GaN HEMTs, GaN epitaxial
layers grown on silicon substrate could also become a game-changer for vertical GaN power
devices considering that silicon substrates are significantly cheaper and are available in large
sizes up to 12-inch diameters which can greatly accelerate viable commercialization. However,
there are several roadblocks arising from the growth as well as fabrication perspective that has
limited the demonstration of high-performance power devices on GaN-on-Si.
In this thesis, we discuss the key hindrances and our solutions for improving the feasibility
of GaN-on-Si vertical power devices. All the necessary fabrication steps were first optimized
from scratch to develop state-of-the-art power devices. As a first demonstration, we could develop
a GaN p-i-n diode with an ultra-low Ron,sp of 0.33 mohmcm2 and record BV of 820 V with a
voltage blocking GaN layer of just 4 um.
A quasi-vertical MOSFET was then demonstrated for the first time on GaN-on-Si platform with
excellent ON- and OFF-state performances. We then probed the limits of current crowding,
a main deterrent to the current up-scaling of quasi-vertical devices by exploring large area
quasi-vertical MOSFETs. A novel and robust method for achieving a fully-vertical design for
GaN-on-Si devices was developed which led to an exemplary improvement in the ON-state performance
of quasi-vertical MOSFETs.
Device integration has been identified by many leading power semiconductor companies as
the way forward due to significant advantages to be had, as a result of lower parasitics and
simplified packaging. Taking a cue from these developments, we demonstrated vertical GaN power
MOSFETs with integrated freewheeling diodes and reverse blocking capability as described in
Chapter 4.
In the last chapter, we introduce p-type NiO as a possible substitute for p-GaN for realizing
high-performance p-i-n diodes and as junction termination extensions (JTEs) for Schottky barrier
diodes. Our initial results point to a strong future for p-NiO to be used for realizing a myriad of
reliable GaN power devices.
EPFL_TH7710.pdf
N/a
openaccess
Copyright
26.37 MB
Adobe PDF
32d5079e74df216b960e77ac29058c49