effect of irradiation damage on gan-based metal...
TRANSCRIPT
-
EFFECT OF IRRADIATION DAMAGE ON GAN-BASED METAL-OXIDE
SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS AND β-GA2O3
By
SHIHYUN AHN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2017
-
© 2017 Shihyun Ahn
-
To my family
-
4
ACKNOWLEDGMENTS
First of all, I would like to thank my advisor and supervisory committee chair, Dr. Fan
Ren. His profound knowledge in the research field and passion guided me through to the right
direction in this PhD journey. Without his dedication and commitment, I would not have made
this far.
I also would like to thank the rest of the committee members, Dr. Stephen J. Pearton, Dr.
Brent P. Gila, and Dr. Peng Jiang for their kind advises and supports, widening my research
perspectives.
I could not make this happened without the support and assistance from former and
current group members, Dr. Byung-Jae Kim, Dr. Ya-hsi Hwang, Dr. Tsung Sheng Kang, Dr. Liu
Lu, Jiangcheng Yang, Patrick Carey, Chien Hsu, Lin Yuan, Yi-Hsuan Lin, Weidi Zhu, Chen
Dong, Shun Li, Yueh-Ling Hsieh, Lei Lei, Guan-Che Ting. We all had good days and bad days
but time we spent together was very meaningful and motivating. Thanks all for your helps and I
wish you all the best lucks in the future career. The support staffs from the chemical engineering
department were exceptional; Craig Smith, Preston Town, and Jim Hinnant in machine shop,
Micah Herron and Carolyn Miller in purchasing orders.
Special thanks to our collaborators; Dr. Jihyun Kim and Gwangseok Yang at Korea
University, Ivan I. Kravchenko at Oak Ridge National Laboratory, Dr. Leonid Charnyak at
University of Central Florida, Dr. Aaron G. Lind, Dr. Kevin S. Jones, Dr. David Cheney, and Dr.
Nancy J. Ruzycki in material science department, Dr. Erin Patrick and Dr. Mark E. Law in
electrical & computer engineering department, Akito Kurama in Tamura Corporation.
Finally, I thank my family for their supports and love that made a person who I am now.
-
5
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ...............................................................................................................4
LIST OF TABLES ...........................................................................................................................8
LIST OF FIGURES .........................................................................................................................9
ABSTRACT ...................................................................................................................................13
CHAPTER
1 INTRODUCTION ..................................................................................................................15
Background .............................................................................................................................15
Dissertation Outline ................................................................................................................19
2 METHODOLOGY .................................................................................................................21
Material Characterization .......................................................................................................21 Scanning Electron Microscopy and Energy Dispersive X-ray Spectroscopy .................21 Transmission Electron Microscopy .................................................................................21
Secondary Ion Mass Spectrometry ..................................................................................22 X-ray Photoelectron Spectroscopy ..................................................................................22
Ellipsometer .....................................................................................................................22
Device Fabrication ..................................................................................................................23
Photolithography .............................................................................................................23 Isolation ...........................................................................................................................23
Reactive Ion Etching .......................................................................................................24 Electron Beam Evaporation .............................................................................................24 Lift-Off ............................................................................................................................24
Rapid Thermal Annealing ...............................................................................................24 Plasma-Enhanced Chemical Vapor Deposition ...............................................................25 Atomic Layer Deposition ................................................................................................25
Device Characterization ..........................................................................................................25
Transmission Line Measurement ....................................................................................25 DC Performance ..............................................................................................................26 Off-State Drain Breakdown Voltage ...............................................................................26
Gate Lag Measurement ....................................................................................................26
3 STUDY OF THE EFFECTS OF GAN BUFFER LAYER QUALITY ON THE DC
CHARACERISTICS OF ALGAN/GAN HIGH ELECTRON MOBILITY
TRANSISTORS .....................................................................................................................28
Introduction to GaN buffer layer quality impact on dc performance .....................................28 Experimental ...........................................................................................................................29
-
6
Result and Discussion .............................................................................................................29
Summary .................................................................................................................................32
4 EFFECT OF PROTON IRRADIATION ENERGY ON ALGAN/GAN METAL-
OXIDE SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS ...............37
Introduction to Effect of Proton Irradiation on AlGaN/GaN MOSHEMT.............................37 Experimental ...........................................................................................................................38 Results and Discussion ...........................................................................................................38 Summary .................................................................................................................................43
5 EFFECT OF PROTON IRRADIATION DOES ON INALN/GAN METAL-OXIDE
SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS WITH AL2O3
GATE OXIDE ........................................................................................................................52
Introduction to Effect of Proton Irradiation on InAlN/GaN MOSHEMT ..............................52
Experimental ...........................................................................................................................52 Results and Discussion ...........................................................................................................53 Summary .................................................................................................................................58
6 DEUTERIUM INCORPORATION AND DIFFUSIVITY IN PLASMA-EXPOSED
BULK GA2O3 .........................................................................................................................64
Introduction to Plasma Exposed Deuterium in bulk Ga2O3....................................................64 Experimental ...........................................................................................................................65 Result and Discussion .............................................................................................................65
Summary .................................................................................................................................67
7 THERMAL STABILITY OF IMPLANTED OR PLASMA EXPOSED DEUTERIUM
IN SINGLE CRYSTAL GA2O3 .............................................................................................71
Introduction to Thermal Stability of Deuterium in bulk Ga2O3 .............................................71
Experimental ...........................................................................................................................74 Result and Discussion .............................................................................................................75 Summary .................................................................................................................................77
8 ELEVATED TEMPERATURE PERFORMANCE OF SI-IMPLANTED SOLAR
BLIND β-GA2O3 PHOTODETECTOR .................................................................................80
Introduction to Solar Blind β-Ga2O3 Photodetector ...............................................................80 Experimental ...........................................................................................................................81 Results and Discussion ...........................................................................................................82 Summary .................................................................................................................................85
9 EFFECT OF 5 MEV PROTON IRRADIATION DAMAGE ON PERFORMANCE OF
β-GA2O3 PHOTODETECTOR ..............................................................................................89
Introduction to Irradiation Damage in β-Ga2O3 Photodetector ..............................................89
-
7
Experimental ...........................................................................................................................90
Result and Discussion .............................................................................................................91 Summary .................................................................................................................................94
10 TERMPERATRUE-DEPENDENT CHARACTERISTICS OF NI/AU AND PT/AU
SCHOTTKY DIODES ON β-GA2O3 ...................................................................................101
Introduction to Schottky Diodes on β-Ga2O3 .......................................................................101
Experimental .........................................................................................................................102 Results and Discussion .........................................................................................................103 Summary ...............................................................................................................................107
11 EFFECT OF FRONT AND BACK GATES ON β-GA2O3 NANO-BELT FIELD-EFFECT TRANSISTORS ....................................................................................................114
Introduction to β-Ga2O3 Nano-Belt Field-Effect Transistors ...............................................114 Experimental .........................................................................................................................116 Result and Discussion ...........................................................................................................117
Summary ...............................................................................................................................119
12 CONCLUSIONS ..................................................................................................................124
LIST OF REFERENCES .............................................................................................................127
BIOGRAPHICAL SKETCH .......................................................................................................138
-
8
LIST OF TABLES
Table page
1-1 Comparison between common semiconductor material properties ...................................19
4-1 Effect of proton irradiation on sheet, transfer and specific contact resistance ..................50
4-2 Effect of proton irradiation on the changes ........................................................................50
4-3 Simulated potential difference ...........................................................................................51
5-1 Effect of proton irradiation ................................................................................................63
5-2 Carrier removal rate of InAlN/GaN MOSHEMTs. ...........................................................63
-
9
LIST OF FIGURES
Figure page
3-1 BF-XTEM showing the AlGaN/GaN interface .................................................................33
3-2 Schematics of wafer bowing for HEMT wafer ..................................................................34
3-3 Drain I-Vs of HEMTs ........................................................................................................34
3-4 Transfer characteristics of HEMTs ....................................................................................35
3-5 Gate-lag pulsed measurements ..........................................................................................36
4-1 SRIM simulation results ....................................................................................................45
4-2 Drain I-V characteristics ....................................................................................................46
4-3 Mobility and sheet carrier concentration reduction ...........................................................47
4-4 Transfer characteristics ......................................................................................................48
4-5 Gate I-V .............................................................................................................................49
5-1 Drain I-V ............................................................................................................................59
5-2 Gate I-V of MOSHEMTs before and after ........................................................................60
5-3 Sub-threshold curve. ..........................................................................................................61
5-4 Gate lag measurement ........................................................................................................62
6-1 SIMS profiles of 2H in Ga2O3 exposed to deuterium plasma ............................................69
6-2 SIMS profiles of 2H in Ga2O3 exposed to deuterium plasma ............................................69
6-3 Fraction of 2H remaining ...................................................................................................70
7-1 SIMS profile of 2H implanted into Ga2O3 (100 keV, 1015 cm-2) before and after ............78
7-2 Percentage of retained 2H implanted into Ga2O3 (100 keV, 1015 cm-2) ............................78
7-4 SIMS profile of 2H in Ga2O3 exposed to deuterium plasma at 270°C. .............................79
8-1 Schematic ...........................................................................................................................86
8-2 Real-time current of the Si implanted β-Ga2O3 photoconductor biased at 5 V .................87
8-3 Temperature dependent net photocurrent increase ............................................................87
-
10
8-4 Temperature dependent photo-to-dark current ratio (PDCR) ............................................88
9-1 Schematics of the Si implanted β-Ga2O3 photo-diode under UV lamp. ............................95
9-2 SRIM simulation ................................................................................................................96
9-3 Real-time current of the Si implanted β-Ga2O3 photo-diode biased at 5V ........................97
9-4 Difference in photo current and dark current. ....................................................................98
9-5 Photo to dark current ratio .................................................................................................99
9-6 Responsivity of β-Ga2O3 photo-diode under 254 nm UV light .......................................100
9-7 Quantum efficiency of β-Ga2O3 photo-diode under 254 nm UV light ............................100
10-1 Schematics .......................................................................................................................108
10-2 I-V characteristics ............................................................................................................109
10-3 Schottky barrier height and ideality factor .......................................................................110
10-4 Reverse biased I-V characteristics ...................................................................................111
10-5 Reverse biased breakdown voltage and figure of merit (FOM), VBR2/Ron ......................112
10-6 Reverse recovery time measurement ...............................................................................113
11-1 Schematic .........................................................................................................................120
11-2 TEM images .....................................................................................................................121
11-3 Drain I-V of β-Ga2O3 flake based FET with front, back, and both gates at 25°C. ..........122
11-4 Typical transfer characteristics of β-Ga2O3 flake FET with front or both gates. .............122
11-5 Gate voltage dependent drain current and gate leakage current ......................................123
-
11
LIST OF ABBREVIATIONS
2DEG 2 Dimensional Electron Gas
AES Auger Electron Spectroscopy
ALD Atomic Layer Deposition
AlGaN Aluminum Gallium Nitride
AlN Aluminum Nitride
BOE Buffer Oxide Etchant
BSD Back Scattering Electron Detector
C-V Capacitance-Voltage
DLTS Deep-Level Transient Spectroscopy
Ec Conduction Band Edge
EDX Energy Dispersive X-Ray Spectroscopy
EV Valence Band Edge
FET Field Effect Transistor
FLOODS FLorida Object Oriented Device Simulator
GaN Gallium Nitride
HEMT High Electron Mobility Transistor
ICP Inductively Coupled Plasma
InAlN Indium Aluminum Nitride
IR Infrared
I-V Current-Voltage
LOR Lift-Off Resist
MBE Molecular Beam Epitaxy
MISHEMT Metal Insulator Semiconductor High Electron Mobility Transistor
MOCVD Metal Organic Chemical Vapor Deposition
-
12
MOSFET Metal Oxide Semiconductor Field Effect Transistor
MOSHEMT Metal Oxide Semiconductor High Electron Mobility Transistor
NSS Surface States Density
PECVD Plasma-Enhanced Chemical Vapor Deposition
PMGI Poly(methyl glutarimide)
PMMA Poly(methyl methacrylate)
PR Photo Resist
Rc Contact Resistance
RI Refractive Index
RIE Reactive Ion Etching
Rs Sheet Resistance
RT Transfer Resistance
RTA Rapid Thermal Annealing
SBH Schottky Barrier Height
SD Secondary Electron Detector
SEM Scanning Electron Microscopy
SiC Silicon Carbide
SRIM Stopping and Range of Ions In Matter
TEM Transmission electron microscopy
TLM Transmission line measurement
UV Ultra Violet
VTH Threshold voltage
XPS X-Ray Photoelectron Spectroscopy
-
13
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EFFECT OF IRRADIATION DAMAGE ON GAN-BASED METAL-OXIDE
SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS AND β-GA2O3
By
Shihyun Ahn
May 2017
Chair: Fan Ren
Major: Chemical Engineering
The effect of buffer layer quality on dc characteristics of AlGaN/GaN high electron
mobility was studied. HEMTs were fabricated on the same aluminum concentration of
AlGaN/GaN substrate with different thickness and defect densities. Their dc performances show
similar results however, one with more defects showed 71% drain current reduction in the gate-
lag pulsed measurements.
The effects of proton irradiation on AlGaN/GaN and InAlN/GaN MOS-HEMTs’ dc
performances were also studied. At lower irradiation energy, devices showed severe current
reduction upto 95.3% as well as with the reduction of transconductance upto 88%. In addition,
devices also showed more degradation in their dc performance at higher irradiation doses. These
degradations are due to the damage done by collisions between the incident protons and the
semiconductor lattice causing the reduction of electron densities and saturated carrier velocities
in the 2DEG.
The further proton irradiation study was conducted on photodetector fabricated on Si-
implanted β-Ga2O3 bulk substrate showing higher responsivity with higher irradiation doses. The
photodetectors were also characterized at the different temperature with no significant
-
14
degradation in its sensitivity upto 350°C, showing its robust material properties to withstand
harsh environmental conditions such as high temperature and radiation abundant atmosphere.
The thermal stability of the differently implanted deuterium into bulk β-Ga2O3 was also
studied. Ion implanted deuterium in β-Ga2O3 showed similar thermal stability in ZnO with ~60%
remaining after 500°C annealing. For plasma treated deuterium, the estimated diffusivity was 6.4
х 10-13 cm2/V∙s.
The electrical properties of the β-Ga2O3 were evaluated through fabrications of Schottky
diodes and field-effect-transistors on the β-Ga2O3 with different metallizations. The barrier
height value of 1.07 eV was obtained for Ni/Au based Schottky diode on Si doped β-Ga2O3 and
1.04 eV was obtained for Pt/Au based Schottky diode on undoped β-Ga2O3. Also, the β-Ga2O3
nano-belt FET was fabricated using Al2O3 and SiO2 as the gate insulator for the front and back
gate. Employing both gates showed its advantages in the current modulation with no electrical
breakdown upto biases of VDS = +100V and VGS = -100V.
-
15
CHAPTER 1
INTRODUCTION
Background
With the increase in needs for higher power and higher frequency as the current
technology advances in various fields including optoelectronics, communications, power devices,
etc., the Si-based (semiconductor) transistors have grown to meet the needs however; there are
still some difficulties that cannot be overcome due to the materials’ limit. Indeed, there have
been a lot of efforts to solve those difficulties through compound semiconductor. In such, group
III-V compound semiconductors show their superior intrinsic properties shown in table 1-1.
GaAs’ higher mobility over Si promises its usage in high speed applications. Also SiC, GaN, and
β-Ga2O3 have superior breakdown field over Si, which are crucial factors for power devices. III-
V compounds semiconductors show their abilities to sustain their material properties at harsh
environment, such as radiation rich surrounding or at elevated temperature.
Transistors are semiconductor devices that are to either amplify or to switch electronic
signals and electrical powers. They are fabricated on the semiconductor material with three
electrodes; source (S), drain (D), and gate (G). Source and drain are to form Ohmic contacts to
the semiconductor forming channels for electrons to flow, and gate is to form either Schottky
contact or metal-insulator-semiconductor contact to modulate the channel. With the gate’s
modulation, the output electrical signals or power can be higher than the input power. These days,
Si is commonly used semiconductor material for its easy accessibility and matured technologies.
However, as demands for high power and frequencies as well as for ability to operate in harsh
environments increases, Si based transistors show their limits due to their intrinsic property limit.
In the other hands, the formation of heterojunctions in III-V composite materials such as
AlGaN/GaN or InAlN/GaN material with excellent physical and chemical stabilities as well as
-
16
outstanding electrical properties including high breakdown electric field, electron motilities, and
radiation hardness.
The AlGaN/GaN and InAlN/GaN based transistors high electron mobility transistors
(HEMTs), which have been intensely studied for their promising features in high power, high
temperature, and high frequency applications as mentioned above. These features promise its
candidacy for next generation of power transistor technology in hybrid electric vehicles inverter,
advanced radar system, and ground or in space communication system[1-3]. For those reasons,
several studies focused on pushing the limit of the HEMTs by improving material quality,
optimizing epi-layer structures, and designing and processing device fabrication[4, 5]. These
days, the-state-of-the-art HEMTs can take up to electron mobility of over 1500 cm2/V-s and
electron saturation velocity of around 2.5 × 107 cm/s. A current-gain cutoff frequency (fT) of 225
GHz with a gate length (LG) of 55 nm, and a power-gain cutoff frequency (fmax) of 300 GHz with
a gate length of 60 nm have been demonstrated[6]. Likewise, 2 dimensional electron gas (2DEG)
forms owning to the energy band gap discontinuity. The presence of piezoelectric effect and
spontaneous polarization in III-nitride induce high carrier concentration within 2DEG, which
could reach above 1013 cm-2. This can result in high current density without conventional doping
like in silicon based semiconductor business, thus potentially reduces the cost of doping
implantation. High breakdown electric field due to large band gap of GaN (3.4eV) and Ga2O3
(~4.9eV) also enabled III-V composite materials to handle high voltage which allows their usage
in high power application.
Furthermore, the radiation hardness of III-nitride based transistors highlights their
prospective usages in space missions and military applications. In case of space application,
usually satellites are orbiting above 300 km from the earth, which are within Van Allen belts.
-
17
The Van Allen belts consist of high energy particles ranging from few MeVs to several hundred
MeV, where its inner altitude ranges from 100s to 6,000 km and outer altitude exits up to 60,000
km from the earth[7]. In the space, high energy particles including protons, neutrons, electrons,
and heavy ions, can change the lattice atoms of the irradiated devices; likely degrade
performance of the device[8]. In order to simulate the space application environment, proton
irradiation of HEMTs devices at different MeV energies have been experimented. According to
Lu, no degradation of the device was observed at 5 MeV proton irradiation of fluence up to 2 ×
1013 cm-2, which is equivalent to over 100 years exposure to the low earth orbit’s dose[9]. Higher
energy irradiation showed less degradation in device’s dc performance was reported due to less
damage in the 2 dimensional electron gas region (2DEG)[10, 11].
Despite of those HEMTs’ superior material qualities mentioned previously, the lack of
reliability limits its potential usage. The high gate leakage and trap-related effects during the
device operation results in reduction of drain current and compromises its radio frequency device
performance. The reliability issue of the HEMTs can be accounted for hot electron induced
degradation and current collapse[12].
During the device operation with hot electron-induced degradation[13], under higher
drain bias, the electron obtain high energy in the channel to be trapped in between device’s
surface and AlGaN, InAlN barrier layer or in GaN buffer layer. Hot-electron also can induce
permanent traps in regions between gate to drain which cause surface depletion, increase sheet
resistance, and reduce gate-drain electric field. These results in threshold voltage shift and
increase in drain resistance thus reducing saturation drain current[13]. The current collapse is
another phenomenon that reduces drain current in high voltage device operation. The current
collapse can be attributed to surface trapping on the surface, in AlGaN barrier layer, in GaN
-
18
buffer layer, resulting in dispersion between dc and pulsed current characteristics[13]. Also,
donor-like traps induced by strong polarization of the material capture electrons tunneling from
gate to drain forming a so-called “virtual gate”, depleting electrons in the channel thus, the
reduction of the drain current is observed[14].
In order to overcome the reliability issues of GaN based HEMTs, different methods have
been employed. Surface passivation of different dielectric on the HEMTs or surface treatments to
terminate dangling bonds on surface to reduce the current dispersion showed enhancement in
device reliability [15-19]. Although, the surface passivation and surface treatments were reported
to help to enhance the reliability problems, under dc and rf stress conditions of SiNx and SiO2[20]
passivated AlGaN/GaN HEMTs, the hot-electron induced device degradation has been observed.
Instead of the surface modification, the metal-oxide-semiconductor HEMTs (MOS-HEMTs) are
studied. Different oxides have widely been studied, using SiO2[21, 22], Si3N4[20], and other
oxides[20, 23] as gate dielectrics to solve the problems mention in above. Al2O3 deposited by
atomic layer deposition (ALD) shows its advantage in perfect conformity, low defect density,
low stress, and excellent adhesion[24, 25]. Moreover, Al2O3 has high dielectric constant (k~10),
high breakdown field (5-10 MV/cm), thermal stability, and chemical stability.
Gallium oxide (Ga2O3) is another promising candidate for next generation compounds
semiconductor material with excellence in its material properties including high electric
breakdown field, and solar-blindness. With the presence of large band gap (4.8-4.9 eV), the
expected Baliga’s figure of merits (FOM) are much larger compare to the other compound
semiconductor materials as well. A capability of mass production adds more values to the
material evaluation beyond its material superiority.
-
19
Table 1-1 Comparison between common semiconductor material properties
Material Si GaAs 4H-SiC GaN β-Ga2O3
Band gap (eV) 1.1 1.4 3.3 3.4 4.8-4.9
Electron Mobility
(cm2/Vs) 1,400 8,000 1,000 1,200 300
Breakdown Field (MV/cm) 0.3 0.4 2.5 3.3 8
Thermal Conductivity
(W/cmK) 1.5 0.5 4.9 2.0 13.6, 22.8
Relative dielectric constant 11.8 12.9 9.7 9.0 10
Baliga’s FOM (to Si) 1 15 340 870 3,444
Dissertation Outline
This dissertation covers two main topics; the effects of proton irradiations on GaN-based
metal-oxide-semiconductor high electron mobility transistors (MOS-HEMTs) and β-Ga2O3 bulk
substrates as well as the reliability study of devices fabricated on β-Ga2O3.
Chapter 1 covers the background knowledge of the III-V compound semiconductor
transistors, especially, the properties and current status of gallium nitride (GaN) based high
electron mobility transistors as well as with their reliability issues, which leads into metal-oxide
semiconductor high electron transistors. Further characteristics of β-Ga2O3 are discussed as well.
Chapter 2 provides the general overviews of the methodologies and technics that are
utilized in this dissertation. Chapter 3 covers the effect of GaN buffer layer on HEMTs dc
performance. Chapter 4 covers the effects of proton irradiation energies on AlGaN/GaN
MOSHEMTs, and chapter 5 covers the effect of proton irradiation doses on InAlN/GaN
MOSHEMT. These two chapters are to serve the effect of proton irradiation on these
MOSHEMTs’ dc characteristics and reliabilities.
Chapter 6 and 7 covers deuterium diffusion and implantation to bulk β-Ga2O3 substrates
using various methods and activation energies of differently implanted deuterium are extracted.
-
20
Chapter 8 covers the effect of elevated temperature on performance of Si-implanted solar blind
β-Ga2O3 photodetector. Chapter 9 covers the effect of proton irradiation damage on the
performance solar blind solar blind β-Ga2O3 photodetector. Chapter 10 covers the effect of
temperature on dc characteristics of Ni/Au and Pt/Au Schottky diodes fabricated on β-Ga2O3 and
chapter 11 covers the effect of front and back gate on β-Ga2O3 nanobelt field effect transistor.
Chapter 12 provides a summary and conclusion of all the topics that are discussed in my
dissertation.
-
21
CHAPTER 2
METHODOLOGY
Material Characterization
Scanning Electron Microscopy and Energy Dispersive X-ray Spectroscopy
Scanning Electron Microscopy is a tool that produces images of the target sample by
scanning the surface with beam of electrons. The electron gun of the SEM emits electrons in the
energy range of 1-40 keV and those emitted electrons collides onto the target surface. The
excited atoms of the sample emit either secondary electrons or backscattered electrons upon the
collisions. By analyzing the intensities of emitted electrons from the excited atoms, the machine
produces an image.
Energy Dispersive X-ray Spectroscopy is a technique that is embedded in the SEM
system, where it uses the X-ray intensities and energy generated upon the electron collision to
quantify the elements present in the target. Each element has its unique atomic structure that
emits its unique electromagnetic emission spectrum. Once the high energy electrons knock
elements’ electrons from their orbits, they create vacancies. At higher energy state electron in the
orbit then falls into fill the vacancy, thus emitting excessive energy. Detector monitors the
emission of kinetic energy and number of electrons during the electron irradiation. By analyzing
this electromagnetic emission spectrum, the system gives semi-quantitative elemental analysis.
Transmission Electron Microscopy
Transmission Electron Microscopy is a technique that uses a beam of electrons to
transmit through an ultra-thin sample to produce a high magnification image. As the electrons
pass through the sample, they are scattered by the electrostatic potentials set up by the
constituent elements in the sample and by analyzing the scatters, the system can provide detailed
-
22
images. These images can be used to study the growth of layers, composition, and defects in the
semiconductors.
Secondary Ion Mass Spectrometry
Secondary Ion Mass Spectrometry is an analytical method to examine the composition of
the solid surface with high detection limit by sputtering off the surface under an ultra-high
vacuum environment. It bombards the surface with focused beam of high energetic ions (10 – 40
keV) such as Cs+, O2+, O-, and Ar, on the target specimen’s surface. Small fraction of the ejected
atoms from the bombardment is ionized either positively or negatively, which are secondary
electrons. Those secondary electrons are detected by a mass spectrometer to determine the
elemental, isotopic, or molecular composition of the surface profile.
X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy is a surface-sensitive qualitative and quantitative
spectroscopic technique that can gives elemental composition, empirical formula, chemical state
and electronic state of the elements within a material. The beam of x-rays is irradiated onto the
target material and the kinetic energy and number of electrons are measured simultaneously.
Each element has its own sets of XPS peaks at the characteristic binding energies and by
monitoring the emission of kinetic energy and number of electrons during the x-ray irradiation,
the surface chemistry of the specimen can be projected.
Ellipsometer
Ellipsometer uses an optical technique to measure the dielectric properties of a thin film.
It measures the change in polarization as light reflects or transmits from a material structure and
compares it to a model. Because the signal depends on the film’s thickness and material
properties (refractive index), sample properties can be extracted by fitting an experimental curve
into a model.
-
23
Device Fabrication
Photolithography
Photolithography is the process of the transferring of geometric shapes on a mask (or a
reticle) to a photoresist (PR) covered surface of a semiconductor wafer. The photolithography is
widely used technique in the semiconductor fabrication process that is very critical.
Photolithography requires photoresist, reticles, exposure system, and developer.
Photoresist (PR) commonly consists with base resin, photo active compound (PAC), and
organic solvent. Depending on the solubility of the PR after the exposure, it determines negative
or positive PR. The simple steps of photolithography consist with coating, exposure, and
developing. After coating of PR, the mask with target pattern is placed in between the exposure
light source to the PR covered wafer and the PR covered wafer is exposed then developed in the
developer solution transferring the opposite tone of patterns on the mask.
Photolithography can be used as a mask for either wet or dry etching or creating patterns
for metal deposition, which are crucial process steps in the device fabrication.
Isolation
Isolation is a process to eliminate the leakage current among the devices that are
fabricated on semiconductor materials. This process can be achieved in either way; a mesa
etching or ion implantation. Mesa etching technique is to physically remove a semiconductor
material in between each device. Both wet and dry etching can be utilized, and patterned PR is
commonly used as a mask for a desired pattern. In doing so, a possible leakage path through the
semiconductor channel is removed, providing a good isolation of each device to device. Another
technique is to intentionally damage the semiconductor surface (a possible leakage path) between
devices by implanting high energy ions such as H, He, N, O, Zn. The implantation causes the
vacancies formation within the semiconductor leading towards a good isolation.
-
24
Reactive Ion Etching
Reactive Ion Etching (RIE) is one of dry etching techniques that utilize the plasma to etch
the target sample’s surface. The oscillating electric field created from RIE chamber electrodes
ionizes the gas molecules, move their outer shell electrons from parent atom thus igniting plasma.
Depending on the volatility of the etched product of gas precursor and target wafer, this plasma
can both chemically and physically removes particles on the wafer surface. RIE can provide very
anisotropic etch profiles but can leave the ion bombardment damages on the specimen.
Electron Beam Evaporation
Electron beam evaporation is a common deposition method to deposit metals on the
sample. In a high vacuum chamber (
-
25
throughout the semiconductor. The system consists of either high intensity lamps or lasers[26]
allowing system to reach the target temperature in a short period of time. For Ohmic contact
formation, the annealing temperature, time, and the flowing gas plays important roles in
achieving low resistance metal contact on semiconductor.
Plasma-Enhanced Chemical Vapor Deposition
Plasma-Enhanced Chemical Vapor Deposition is a widely used process that is used to
deposit conformal dielectrics. These dielectrics can provide passivation as well as improve
device performance by changing the semiconductor surface chemistry. The use of plasma
provides a high deposition rate of dielectric as well as its ability to operate at high enough
pressure, lowering down the equipment cost. The dielectric film formation can be controlled by
the different precursor gases, flow rate of each gas, temperature, pressure, and rf power of
plasma generation.
Atomic Layer Deposition
Atomic Layer Deposition is another dielectric film deposition technique that is based on
the sequential gas phase chemical process on a solid surface. The dielectric film is deposited
atomically layer by layer on a surface that provides high quality of film. As a result, the
thickness and quality of the deposited film is well controlled, however, due to its layer by layer
deposition method, the deposition rate is slow.
Device Characterization
Transmission Line Measurement
Transmission Line Measurement (TLM) is a method to quantify Ohmic contacts. A
typical transmission line consists with series of rectangular Ohmic contacts with different gaps
on a single mesa. By following the Ohmic law (V = IR), each resistance can be measured on
each gap. By plotting the resistance versus the gap, it provides a linear curve. The slope is the
-
26
sheet resistance, Rs reported in Ω/□, of the channel, which is a normalized resistance of
semiconductor material. In the ideal case of the plot, the intercept should be 0 if there is no
additional resistance besides the material. However, the contact has resistance itself and the
intercept represents either transfer resistance (Rs) in Ω/mm or contact resistance (Rc) in Ω-cm2
depending on the orientation of the device (lateral or vertical respectively).
DC Performance
DC performance is a way to characterize the performance of field effect transistor. It
consists of drain and gate current-voltage (I-V) and transfer characteristics. The voltage is
applied to the three terminals, source, gate, and drain of the FET and current is monitored. Drain
I-V shows the current modulation of the device including its pinch off and saturation current.
Gate I-V shows the blocking voltage and Schottky characteristics to evaluate the effectiveness of
gate modulation.
Off-State Drain Breakdown Voltage
Off-State Drain Breakdown Voltage is a key feature for high power switching
applications which can be operated at lower current level to reduce the diameter of cable. To
measure the device’s off-state drain breakdown voltage, the gate voltage (VG) is held constant
voltage where the current is pinched off. Meanwhile, the current is monitored as the drain
voltage (VD) is increased. Off-state breakdown voltage is a drain voltage that the current will
abruptly increases. The typical off-state drain breakdown voltage’s current set point for lateral
HEMT device is 1 mA/mm.
Gate Lag Measurement
Gat lag measurement is a technique that uses pulsed gate voltage to estimate traps on the
HEMT surface. While the drain voltage is held at constant, the gate voltage is pulsed from
beyond pinch-off to a voltage of an on-state at a frequency range. The presence traps in the
-
27
material will cause the device to response differently from the gate pulse to dc measurements.
When the device is pulsed from off-state to on-state at high frequency, the trapped charges does
not have enough time to responds resulting in the reduction of current as compared to the current
in dc mode operation.
-
28
CHAPTER 3
STUDY OF THE EFFECTS OF GAN BUFFER LAYER QUALITY ON THE DC
CHARACERISTICS OF ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS
Introduction to GaN buffer layer quality impact on dc performance
Owing to the large bandgaps in the AlGaN/GaN heterostructure as well as the high
electron mobility and breakdown fields, AlGaN/GaN high electron mobility transistors (HEMTs)
are promising candidates for high power and high current applications in advanced radar
systems, inverter units in hybrid electric vehicles, space and satellite communication
networks[27-29]. AlGaN/GaN HEMTs are commonly grown on silicon carbide (SiC) or
sapphire. It is well established that epitaxial films grown on sapphire or SiC typically have
dislocation densities in the order of 108-1010 cm-2 due to their thermal coefficient and lattice
mismatches[21, 30, 31]. SiC is an excellent substrate candidate for power applications because of
its superior thermal conductivity, however, the cost of SiC substrates is high. Therefore, despite
its low thermal conductivity, sapphire has an advantage in its lower price and often superior
surface quality.
A number of studies have reported that optimizing the parameters of GaN buffer layer
growth such as process pressure, precursor flow rate, additional AlN interlayer and GaN buffer
layer thickness can reduce the defect densities which are caused by lattice mismatch[21, 30, 31].
Also, the thinner GaN buffer layers on SiC have been reported to have higher threading
dislocation density in the GaN, but lead to higher off-state breakdown voltages[32, 33].
However, thick GaN buffer layers on sapphire and their impact on dc performance has not been
widely studied.
In this work, the impact of GaN buffer layer quality on dc and gate-lag pulse performance
of AlGaN/GaN HEMTs was investigated. Transmission electron microscopy (TEM) was used to
determine the dislocation density in the GaN buffer layer of the HEMT samples. Drain I-Vs,
-
29
transfer characteristics and gate pulsed drain I-V characteristics of the HEMTs were measured to
establish a correlation between the AlGaN/GaN HEMTs’ dc and gate-pulsed performance and
GaN buffer layer quality.
Experimental
AlGaN/GaN HEMT structures grown by metal organic chemical vapor deposition
(MOCVD) on sapphire substrates were acquired from two different vendors. Both types of
wafers had 24% Al concentration in the AlGaN barrier layer. One type of wafer had a 5 μm GaN
buffer and the other had a 2 μm GaN buffer layer. HEMT fabrication was started with mesa
patterning by conventional optical lithography. The mesa-etching was achieved using a Unaxis
Shuttle-lock Reactive Ion Etcher with Inductively Coupled Plasma Module (ICP) for device
isolation with a Cl2/Ar plasma. Ohmic metallization of electron-beam evaporated Ti/Al/Ni/Au
(25nm/125nm/45nm/100nm) was patterned by lift-off and subsequently rapid thermally annealed
at 850ºC in a flowing N2 ambient for 45seconds. Schottky gates of e-beam evaporated Ni/Au
(20nm/80nm) with 100 µm width and 1 µm length were obtained by standard lift-off patterning.
DC current-voltage (I-V) characteristics were measured using an HP4156 parameter analyzer,
and gate-lag pulse measurements were measured using an Agilent 8114A pulse generator.
Result and Discussion
Bright-field cross-sectional TEM (BF-XTEM) was used to quantify the defect density in
the HEMT wafers. Figure 3-1 displays the cross-sectional transmission electron microscopy
(XTEM) pictures of AlGaN/GaN surface for the top layers of A) HEMT structure with 2 μm
GaN buffer layer and B) HEMT structure with 5 μm GaN buffer layer as well as the
GaN/sapphire interface for the C) HEMT structure with 2 µm GaN buffer layer and D) HEMT
structure with 5 µm GaN buffer layer. Dislocations are seen shown to originate near the
-
30
GaN/sapphire interface and propagate up through the AlGaN/GaN interface to the surface of
both HEMT wafers. However, it is clear that the HEMT structure with 2 μm GaN buffer layer
exhibited a much higher defect density. The defect density near the surface for the two substrates
was calculated from XTEM taken along the length of the TEM lamella based on the thickness
measured during lamella formation with a focused ion beam (FIB). In the case of the 5 μm GaN
buffer layer substrate, the measured defect density was 5×108 cm-2, while an order of magnitude
higher measured defect density of 7×109 cm-2 was observed for the 2 μm GaN buffer layer
sample.
Although the HEMT wafer with 5 µm GaN buffer had one order less defect density, this
wafer was under a high strain. As shown in Figure 3-2, the HEMT wafer with 5 μm GaN buffer
layer was significantly bowed. A height difference of 70 µm across the 2” wafer was measured
with a much smaller radius of curvature of 5 m. On the contrary, the HEMT wafer with 2 μm
GaN buffer layer was flat and the wafer height difference across 3” wafer was around 21.25 µm
with a radius of curvature of 45 m. These results indicate that the HEMT wafer with 5 μm GaN
buffer layer was highly strained by with lower defect density.
Figure 3-3 shows the drain current-voltage (I-V) characteristics of HEMTs fabricated on
the different thickness of buffer layers. The HEMTs were measured by sweeping VDS from 0V to
5V while VG started from +1V with a step of -1V till reaching pinch-off. There was little
difference in the saturation current observed at VDS = +5V. Transfer characteristics of HEMTs
fabricated on different buffer layers were measured at VDS = +5V and Vg was swept from -5V to
0V, as illustrated in Figure 3-4. The thicker buffer GaN layer HEMTs showed marginally higher
drain current at VG = 0V than the thinner GaN layer HEMTs, with current densities of 305 ± 35
mA/mm and 302 ± 36 mA/mm respectively. The peak transconductance for 5 µm GaN buffer
-
31
layer HEMTs was 111 ± 7 mS/mm and for the thinner GaN layer HEMTs was 107 ± 4 mS/mm,
showing only a slightly higher value for the thicker GaN layer HEMT. Using the linear region of
the drain I-V curve, the mobility and carrier concentration were calculated using a charge control
model[12, 34]. The calculated mobility and carrier concentration for thicker GaN layer HEMTs
were 989 cm2/V·S and 8.45 х1012 cm-2 and for thinner GaN layer HEMTs were 907 cm2/V·S and
8.63 х 1012 cm-2. These similarities in their dc characteristics suggest that the defect density in
GaN buffer layer has a minimal influence on HEMT dc performance over this range of buffer
thicknesses.
Furthermore, gate-lag pulse measurements were applied to evaluate the effect of the
different GaN buffer layers on material electrical quality. In this method, the drain current (IDS)
was recorded while gate voltage (VG) was pulsed. The normalized drain current-gate voltage
(VG) and pulsed measurements are shown in Figure 3-5 A and B. VG was pulsed from -5V to the
voltage indicated on the x-axis in Figure 3-5 at 100 kHz frequency with 10% duty cycle while
VDS was kept +5V. As shown in Figure 3-5 A, a dramatic reduction of drain current was
observed during the gate-pulsed measurement for the HEMT with 2 µm GaN buffer layer. When
the gate was pulsed at a fixed drain voltage, some of the surface traps become charged by
trapping hot electrons[12]. If the gate is pulsed above a certain high frequency, the traps do not
have enough time to detrap. Those charged traps acted as a virtual gate and formed an additional
depletion region between gate and drain electrodes, and reduced the drain current. However,
there was no drain current reduction during the gate pulsed measurement for the HEMT with 5
µm GaN buffer layers. On the other hand, there was 79 ± 10 % of drain current reduction in the
gate pulsed measurement for the HEMT with 2 µm GaN buffer layer at VG = 0V. This indicates
-
32
that the excess number of defects in the thinner GaN buffer layer influenced the generation of
surface trap and induced drain current collapse at high frequency operation.
Summary
The effects of defect density on HEMT dc and rf performance were studied. HEMTs
were fabricated on 2 and 5μm of GaN buffer layers with the same composition of active layers.
TEM showed that the thinner GaN layers had an order of magnitude higher defect densities.
HEMTs on thicker GaN buffer layers showed negligible difference in the drain I-V
characteristics and showed a 6% higher value in peak transconductance. Significant reduction in
pulse measurements were observed on thinner GaN layer HEMTs. These results suggest that the
defects created in GaN buffer layer do not largely influence device dc performance but are
closely related to high frequency performance.
-
33
Figure 3-1. BF-XTEM showing the AlGaN/GaN interface for the A) 2 μm GaN and B) 5 μm
GaN substrates and the GaN/sapphire interface for the C) 2 μm GaN and D) 5 μm
GaN substrates.
-
34
Figure 3-2. Schematics of wafer bowing for HEMT wafer with A) 5 μm and B) 2 μm GaN buffer
layer.
0 1 2 3 4 50
100
200
300
400
500
600 5m GaN
2m GaN
VG = +1V Step = -1V
I DS (
mA
/mm
)
VDS
(V)
Figure 3-3. Drain I-Vs of HEMTs fabricated on different GaN buffer layer structures.
A
B
-
35
-5 -4 -3 -2 -1 00
100
200
300
400
500
5m GaN
2m GaN
VDS
= +5V
I DS (
mA
/mm
)
VG (V)
0
30
60
90
120
150
GM (
mS
/mm
)
Figure 3-4. Transfer characteristics of HEMTs fabricated on different GaN buffer layer
structures.
-
36
-5 -4 -3 -2 -1 00
20
40
60
80
100
120 DC measurement 100 kHz Pulse
VDS
= +5V
Norm
aliz
ed I
DS (
%)
VG (V)
-5 -4 -3 -2 -1 00
20
40
60
80
100
120 DC measurement 100 kHz Pulse
VDS
= +5V
No
rma
lize
d I
DS (
%)
VG (V)
Figure 3-5. Gate-lag pulsed measurements on A) HEMTs fabricated with a 2 μm GaN buffer
layer. B) Gate-lag pulsed measurements on HEMTs fabricated with a 5 μm GaN
buffer layer.
A
B
-
37
CHAPTER 4
EFFECT OF PROTON IRRADIATION ENERGY ON ALGAN/GAN METAL-OXIDE
SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS
Introduction to Effect of Proton Irradiation on AlGaN/GaN MOSHEMT
Due to their wide energy bandgaps, high sheet carrier concentration and high electron
mobility, AlGaN/GaN high-electron mobility transistors (HEMTs) are promising for high power
and high frequency applications, including inverter units in hybrid electric vehicles, advanced
radar systems, satellite-based communication networks and space communication systems[4, 35,
36]. However, the high gate leakage and drain current collapse in these conventional Schottky
gate metal devices limits the stability and performance of the HEMTs[37, 38]. In order to solve
these problems, oxides have been widely employed to reduce the metal-oxide-semiconductor
(MOS) diode leakage current or passivate surface traps, with Sc2O3, MgO, SiO2 and Si3N4[20,
22, 39-41], as the most effective gate dielectrics. Al2O3 deposited by atomic layer deposition
(ALD) has also been used as the gate oxide due to its advantages of excellent conformability,
low defect density, low stress, and excellent adhesion[38, 42]. Moreover, Al2O3 has a high
dielectric constant (k~10), high breakdown field (5-10 MV/cm), excellent thermal stability, and
chemical stability against reaction with AlGaN.
In this work, the effect of proton irradiation energy on dc performance of AlGaN/GaN
MOSHEMT with ALD-deposited Al2O3 as the gate dielectric was investigated. The proton
energy was varied from 5 MeV, 10 MeV to 15 MeV at a fixed dose of 5 × 1015 cm-2. The
MOSHEMTs dc characteristics, including drain and gate IV characteristics, sheet and contact
resistance, threshold voltage were compared before and after proton irradiation. The effect of
trap density in AlGaN barrier layer and GaN channel layer on the drain saturation current was
investigated with the finite-element based Florida object oriented device and process simulator
(FLOODs) technology computer aided design (TCAD) simulator.
-
38
Experimental
The AlGaN/GaN heterostructure consisted of 5μm GaN buffer, 21 nm of un-doped
Al0.24Ga0.76N, and a 5nm undoped GaN cap on sapphire substrate grown by metal-organic
chemical vapor deposition (MOCVD). Device isolation was achieved with mesa etching by an
inductively coupled plasma (ICP) system using Cl2/Ar based discharges. Ti/Al/Ni/Au based
metallization was used as the Ohmic metal with the standard lift-off electron-beam deposition,
with subsequently rapid thermal annealing at 850ºC for 45 sec in a N2 ambient. A 10 nm Al2O3
layer was deposited by atomic layer deposition (ALD) at 300ºC using trimethylaluminum and
water as the precursors. Electron-beam-deposited Ti/Au (20 nm/80 nm) was employed as the
gate metal. The gate width and length were 2 × 100 µm and 1 µm, respectively. Proton
irradiations were conducted at the Korean Institute of Radiological & Medical Sciences
(KIRAMS) using a MC 50 (Scanditronix) cyclotron.
MOSHEMT dc characteristics were measured with a 4156 HP parameter analyzer.
Transmission line measurements (TLM) were used to determine sheet resistance (RS), transfer
resistance (RT), and contact resistance (RC) before and after proton irradiation. Stopping and
range of ions in matter simulations (SRIMs) were used to simulate the proton penetration depth
and proton irradiation generated vacancy concentration distribution.
Results and Discussion
Figure 4-1 shows the SRIM simulation of vacancies created as a function of proton
penetration into the Al2O3/AlGaN/GaN MOSHEMT structure grown on a sapphire substrate. The
energy loss of the proton is a maximum near the end-of-range due to the nuclear stopping in that
region. This process causes atomic displacements and is the main cause of carrier loss due to
trapping into these defects. Nearer the surface, the energy loss is dominated by electronic
stopping which leads to ionization and heating. Thus, the damage induced by proton irradiation
-
39
is focused near the end-of-range at the penetration depths of 120, 392 and 794 μm for 5, 10 and
15 MeV, respectively. Since the wafer thickness is around 500 µm as marked in Figure 4-1A, the
15 MeV protons penetrate through the entire wafer with the least amount of damage created in
the active regions of the MOSHEMTs. The thickness of the AlGaN/GaN MOSHEMT structure
is less than 6 µm, and the 2DEG channel is located 36 nm below the Al2O3 surface of the
MOSHEMT. Therefore, the damage created by the high energy proton in the MOSHEMT
structure is uniform but much lower than in the tail region of the damage profile; the simulated
vacancy concentrations are 6.6 × 1018, 3.2 × 1018, 1.8 × 1018 cm-3 for 5, 10 and 15 MeV
irradiation, respectively, as shown in Figure 4-1B. Energetic protons transfer a part of their
kinetic energy to the Ga and N atoms through non-ionizing energy loss (NIEL), and displaces
atoms from their lattice sites to create vacancies. The minimal energies of defect formation were
found to be 18±1 eV for Ga and 22± 1 eV for nitrogen, but the average displacement energy was
much higher, 45±1 eV (Ga) and 109±2 eV (N). Thus the density of the Ga vacancy (VGa) should
be much higher than that of the VN. The VGa behaves as a compensation acceptor-like defect, and
VN is donor-like defect. As illustrated in Table 4-1, the sheet resistance, Rs, of the 5 MeV
proton-irradiated MOSHEMT was increased more than 10 times and it increased around 2.5
times for 10 and 15 MeV proton irradiated MOSHEMTs. These experimental outcomes are
consistent with the SRIM simulation results. Besides the increases of the sheet resistance, the
MOSHEMT transfer resistance, Rt, and specific contact resistance, Rc, also increased inversely
proportional to the proton energy, as shown in Table 4-1.
Figure 4-2A shows the drain current-voltage (I-V) curves of specific MOSHEMTs before
and after proton irradiation at 10MeV fluence of 5 × 1015 cm-2. The gate voltage was varied form
1 V to -3 V with a step of -0.5 V. All the proton-irradiated Al2O3/AlGaN/GaN MOSHEMTs
-
40
exhibited a sub-threshold drain leakage current less than 1 × 10-4 mA/mm. The amount of the
saturation drain current reduction depended on the irradiation energy. For the 10 MeV irradiated
MOSHEMTs, the reduction of saturation drain current at VG = 1V was 68.3%. A much larger
saturation drain current reduction, 95%, was observed for the MOSHEMTs irradiated with
proton energy of 5 MeV, and there was only 59.8% drain current reduction exhibited for the
MOSHEMTs irradiated with proton energy at 15 MeV, as summarized in Figure 4-2B. The
reduction of saturation drain current could be attributed both to the decrease in electron density
in the channel and in saturated carrier velocity. Defect centers are introduced as a result of
displacement damage during collisions between the incident protons and the semiconductor
lattice. These defect centers capture free carriers, resulting in reduction of carrier density and
conductivity of irradiated MOSHEMTs[43-45]. The mobility is affected both by the AlGaN/GaN
interface roughness[46, 47] and the scattering from defect centers created through proton
irradiation in the vicinity of the MOSHEMT channel[44, 48]. The decrease of initial slope of the
drain I-V curves in the linear region is also an indication that the irradiation decreased of carrier
concentration and mobility [49]. The electron mobility and sheet carrier concentration in the
2DEG channel were extracted using the low field linear region of the drain I-V curves using the
charge control model and verified with the Hall measurements[50]. The electron mobility in the
2DEG channel before and after proton irradiation at 5, 10 and 15 MeV were 1363, 1342.5, 944.5
and 865.5 cm2/V s, respectively. The sheet carrier concentration in 2DEG channel before and
after proton irradiation at 5, 10 and 15 MeV were 9.55 × 1012, 4.75 × 1012, 4.33 × 1012 and 2.87 ×
1012 cm-2, respectively. As shown in Figure 4-3 A, the ranges of the extracted percent electron
mobility and sheet carrier concentration reductions were 63.5 to 99.0% and 30.3 to 55.6%,
respectively, depending on the proton energy. The electron mobility and sheet carrier
-
41
concentrations of proton irradiated MOSHEMTs are summarized in Table 4-2. Figure 4-3B
shows that the carrier removal rate was inversely proportional to the irradiated proton energy,
RNC = -16.0·E (MeV) + 362.7, where RNC is carrier removal rate and E is proton energy. The
carrier removal rates were defined as the ratio of carrier concentration decrease divided by the
fluence of irradiated protons. The carrier concentration was obtained by the Hall measurements.
The lower energy proton irradiation created more non-ionizing energy loss and hence more
damage in the 2DEG, producing a higher carrier removal rate. Note that these removal rates are
similar to metal gate HEMTs irradiation under similar conditions and show that the presence of
the dielectric is not influencing the response of the devices to proton irradiation[11].
Figure 4-4A shows transfer characteristics from the MOSHEMTs before and after
irradiation at 10 MeV. The extrinsic transconductance, gm, was reduced by 54.4 % and there was
a positive shift of 1.96 V for the threshold voltage, Vth. These degradations were mainly due to
the displacement damage induced by the ion bombardment that reduced the carrier density and
electron mobility. As shown in Figure 4-4B, more severe degradation of the gm and a larger
positive Vth shift were observed for MOSHEMTs irradiated with lower irradiation energy of 5
MeV, with the gm decreased around 88% and Vth shifted by 2.89V. The lower the irradiation
energy used, then the shallower is the ion penetration depth. Thus, more damage was created
around the AlGaN/GaN interface. The FLOODS TCAD finite-element solver was employed to
simulate the potential difference between Fermi level (EF) and the ground state of sub-bands (E0)
in 2DEG channel and sheet carrier concentration in the 2DEG channel with different
concentrations of the charged traps. The sheet carrier concentration in the 2DEG channel is
proportional to EF - E0 as follows,
)()(
)()(*
2
0 xnxm
xExE sF
(4-1)
-
42
where ћ is the Planck’s constant divide by 2π, x is the Aluminum fraction of barrier layer, m* is
the effective mass of carriers, and is the sheet concentration in the 2DEG channel. Figure 4-4C
shows simulated conduction band-structures of the AlGaN/GaN HEMT by varying the charged
trap concentration in the GaN layer from 1×1015 to 5×1016 cm-3 and keeping charged trap
concentration constant at 1×1014 cm-3 uniformly distributed across the AlGaN barrier layer. As
more negative charges reside in the GaN layer, the E0 moves closer to EF resulted in lower sheet
concentrations in the 2DEG channel. The effect of charged traps in the AlGaN channel layer on
the sheet carrier concentration in the 2 DEG channel was also studied and a similar trend was
observed; less sheet carrier concentration presented in the 2DEG channel for more charged traps
in the AlGaN layer to induce a positive threshold voltage shift. Table 4-3 summarizes the
simulated EF - E0 and ns as a function of trap concentration in the AlGaN and GaN channel
layers. The variations of EF - E0 and ns were more sensitive to charged trap concentration in the
channel GaN layer. A similar trend was observed for the effect of irradiation energy on extrinsic
transconductance reduction, with 88.0 % and 54.4 % reductions after 5 and 10MeV irradiation,
respectively, and 40.7 % for 15 MeV. There was a positive shift of threshold voltage of 2.89 V
after 5 MeV proton irradiation, while the MOSHEMT irradiated with a 15 MeV proton
irradiation exhibited a much smaller shift of 1.26 V.
The reverse and forward gate I-V characteristics of the MOSHEMTs before and after
proton irradiation at 10 MeV are illustrated in Figure 4-5A. There was no degradation observed
on the oxide breakdown strength; around 5 MV/cm, but unexpectedly both forward and reverse
MOS gate leakage current decreased more than three orders of magnitude. The level of MOS
gate leakage current is inversely proportional to the gate oxide thickness and oxide breakdown
strength. Since there was no degradation of the oxide breakdown strength nor changes of the
-
43
oxide thickness after proton irradiation, the forward and reverse MOS gate leakage current
should be very similar prior to and after proton irradiation. Thus, the dramatic MOS gate leakage
current reduction must be related to other factors. It was reported that proton irradiation with a
dose of 4 × 1014 cm-2 induced voids in the Ni layer of Ni/Au based gate electrode and resulted in
decreasing gate area and gate leakage current[51]. Ferric cyanide (FeCN) and potassium iodide
(KI) based Au etch solution was used to remove the Au layer on the top of the Ti/Au based gate
electrode and no voids on Ti layer were observed, as shown in Figure 4-5B. Thus, the possibility
of gate contact area reduction induced gate leakage decrease was ruled out. The MOSHEMT
structure used in this experiment has 5 µm un-doped GaN layer. Typical keV isolation
implantation and mesa etching will not be able to isolate the thicker GaN buffer except the use of
carbon or iron doping in the GaN buffer layer. However, the penetration depth of MeV
implantation is way over 5 µm GaN layer, the traps created by proton implantation would reduce
the leakage current in the GaN buffer layer as well as MOSHEMT isolation current. As shown in
Figure 4-5C, the MOSHEMT isolation current significantly decreased after proton irradiation.
Therefore, the unexpected reduction of MOS gate leakage current after proton irradiation was
resulted from the improvement of MOSHEMT isolation current.
Summary
Fluence of 5 × 1015cm-2 protons at irradiation energies of 5 MeV, 10 MeV, and 15 MeV
were used to study the effect of proton irradiation energy on Al2O3/AlGaN/GaN MOSHEMTs.
There were significant reduction of carrier mobility, carrier concentration, and degradation of dc
characteristics observed for the lower proton energy irradiation. The severe degradation at lower
irradiation energy is due to deep trap generation near the 2DEG channel which increased the
resistance of the channel and this defect generation also influenced the positive shift in the
threshold voltage. The reverse and forward gate I-V was decreased more than an order of
-
44
magnitude after proton irradiation for all energies, which was due to the improvement of
isolation in the GaN buffer layer.
-
45
5 10 15
2x1018
4x1018
6x1018
8x1018
Va
ca
ncie
s (
cm
-3)
Proton Energy (MeV) Figure 4-1. SRIM simulation results showing A) vacancies created as a function of target depth
in the AlGaN/GaN MOSHEMT structure and B) vacancies created at the 2DEG
region as function of irradiation energy.
0 200 400 600 8000
2x1019
4x1019
6x1019
8x1019
1x1020
Va
can
cie
s (
cm
-3)
Target Depth (m)
5 MeV
10 MeV
15 MeV
(a)
Dose
5 x 1015
cm-2
A
B
-
46
0 2 4 6 80
100
200
300
400 Before
After
10MeV
VG = +1V
Step = -0.5V
I DS (
mA
/mm
)
VDS
(V)
5 10 150
20
40
60
80
100
I DS r
ed
uctio
n (
%)
Proton Energy (MeV) Figure 4-2. Drain I-V characteristics of A) MOSHEMT before and after proton irradiation with
fluence of 5 × 1015 cm-2 at an energy of 10 MeV. B) Reduction of saturation current
as function of proton irradiation energies for a fixed fluence of 5 × 1015 cm-2.
A
B
-
47
5 10 1560
70
80
90
100 Mobility
Sheet carrier
concentration
Mo
bili
ty R
ed
uctio
n (
%)
Proton Energy (MeV)
20
40
60
80
Sh
ee
t ca
rrie
r co
nce
ntr
atio
n
Re
du
ctio
n (
%)
Figure 4-3. Mobility and sheet carrier concentration reduction of A) irradiated MOSHEMTs as a
function of proton irradiation energies at a fixed fluence 5 × 1015 cm-2. B) Carrier
removal rate in MOSHEMT as a function of proton irradiation energy (cm-1).
5 10 150
100
200
300
400
500
N
c/P
roto
n D
ose
, R
Nc (
cm
-1)
Proton Energy (MeV)
RNC
= -16.0E(MeV) + 362.7
(b)
A
B
-
48
-5 -4 -3 -2 -1 0 10
100
200
300
400
IDS
IDS
After
GM G
M After
VDS
= +8V
I DS (
mA
/mm
)
VG (V)
10 MeV
0
30
60
90
120
150
GM (
mS
/mm
)5 10 15
0
20
40
60
80
100
Gm reduction
Vth shift
GM r
ed
uctio
n (
%)
Proton Energy (MeV)
1
2
3
Vth s
hift
(V)
0 1 2 3 4 5 6 7 8-0.5
0.0
0.5
1.0
1.5
2.0
510
16 cm-3
AlGaN GaN
EF
11016 cm
-3
En
erg
y (
eV
)
Distance (10-2 m)
11014
cm-3
Figure 4-4. Transfer characteristics of A) MOSHEMTs before and after proton irradiation with
fluence of 5 × 1015 cm-2 at 10 MeV. B) Reduction of gm and threshold voltage shift
as function of proton irradiation energies at fixed fluence of 5 × 1015 cm-2. C) FLOOD
TCAD simulation results for conduction band-structure of the AlGaN/GaN HEMT at
different charged trap concentration in the GaN layer.
A
B
C
-
49
-10 -8 -6 -4 -2 0 210
-9
10-7
10-5
10-3
10-1
101
Before
After
I G (
mA
/mm
)
VG (V)
10MeV
0 10 20 30 40 500
2x10-7
4x10-7
6x10-7
8x10-7
1x10-6 Reference
5MeV
10MeV
15 MeV
Curr
ent
(A)
VDS
(V)
Figure 4-5. Gate I-V of A) MOSHEMTs pre- and post- proton irradiation with 10MeV protons at
a dose of 5 × 1015 cm-2 B) Scanning electron microscopy image of proton irradiated
gate area after the removal of Au on the gate electrode. C) Change of device isolation
of MOSHEMT pre- and post- proton irradiation as a function of different energies.
A
B
C
-
50
Table 4-1. Effect of proton irradiation on sheet, transfer and specific contact resistance of
Al2O3/AlGaN/GaN MOSHEMTs.
Condition Rs (Ω/) Rt (Ω-mm) Rc (Ω-cm2)
5 MeV Before 493 0.48 4.7 × 10-6
After 5713 6.85 8.2 × 10-5
10 MeV Before 489 0.51 5.3 × 10-6
After 1254 1.00 7.9 × 10-6
15 MeV Before 497 0.37 2.8 × 10-6
After 1260 0.76 4.6 × 10-6
Table 4-2. Effect of proton irradiation on the changes of saturation drain current, extrinsic
transconductance, threshold voltage, electron mobility and 2DEG sheet carrier
concentration of Al2O3/AlGaN/GaN MOSHEMTs.
Proton Energy ΔIDSS
(%)
Δgm
(%)
ΔVth
(V)
ΔMobility
(%)
ΔSheet carrier
concentration (%)
5 MeV 95.3 88.0 2.89 99.0 55.6
10 MeV 68.3 54.4 1.96 69.3 45.2
15 MeV 59.8 40.7 1.26 63.5 30.3
-
51
Table 4-3. Simulated potential difference between Fermi level (EF) and the ground state of sub-
bands (E0) in 2DEG channel as well as the sheet carrier (ns) concentration
degradation in the 2DEG channel as a function of ionized acceptor-type trap
concentration in the AlGaN and GaN layers with FLOODS TCAD finite-element
based simulator.
Ionized accept-type
trap concentration
in AlGaN (cm-3)
Ionized accept-type
trap concentration
in GaN (cm-3)
EF-E0 (meV) ns degradation
1.0×1013
1.0×1014 97.910 0.04%
1.0×1015 97.880 0.04%
1.0×1016 97.580 4.50%
1.0×1014
1.0×1013
97.914 0.0002%
1.0×1015 97.900 0.006%
1.0×1016 97.750 0.080%
-
52
CHAPTER 5
EFFECT OF PROTON IRRADIATION DOES ON INALN/GAN METAL-OXIDE
SEMICONDUCTOR HIGH ELECTRON MOBILITY TRANSISTORS WITH AL2O3 GATE
OXIDE
Introduction to Effect of Proton Irradiation on InAlN/GaN MOSHEMT
The lattice mismatch between AlGaN and GaN channel/buffer layers has led to interest in
InAlN/GaN since the InAlN layer can be grown lattice matched to the GaN channel/buffer layer
at a concentration of 17% indium. These InAlN/GaN based HEMTs have a higher polarization-
induced two-dimensional electron gas (2DEG) density and have better carrier confinement of
2DEG channel with a wider energy bandgap of InAlN as compared to AlGaN[52-57]. Al2O3 or
AlN dielectric layers have been employed to reduce gate leakage current, increase gate
breakdown voltage and improve gate modulation for GaN based HEMTs[38, 58, 59]. Many
studies have been conducted on high energy proton irradiation tolerance of both AlGaN/GaN and
InAlN/GaN HEMTs as well as AlGaN/GaN metal oxide semiconductor HEMTs (MOSHEMTs)
but there has not been a study performed for proton irradiation effects on InAlN/GaN
MOSHEMTs[59-61].
In this chapter, the effect of proton irradiation dose on the characteristics of InAlN/GaN
metal-oxide semiconductor high electron transistors (MOSHEMTs) with atomic layer deposited
(ALD) Al2O3 as the gate oxide has been studied. The proton dose was from 1013-1015 cm-2 at 5
MeV energy. The sheet and contact resistance, drain and gate current-voltage characteristics,
sub-threshold swing, and gate-pulsed drain current were investigated before and after the proton
irradiations.
Experimental
The InAlN based MOSHEMT structure consisted of a 2 nm GaN cap layer, a 5.3 nm
In0.17Al0.83N barrier layer, a 1 nm AlN spacer layer, and a 0.85 μm Al0.04Ga0.96N channel/buffer
-
53
layer. These were grown on a sapphire substrate by metal-organic chemical vapor deposition
(MOCVD). Device isolation was achieved with an inductively coupled plasma (ICP) system
using Cl2/Ar mixture. Ohmic metallization was accomplished by standard lift-off process of
electron-beam deposited Ti/Al/Ni/Au followed with a rapid thermal annealing at 850°C for 45
seconds in flowing nitrogen ambient. 5 nm Al2O3 was deposited on the surface using atomic
layer deposition (300°C) with tri-methyl-aluminum and water as the precursors. Afterwards, the
Al2O3 was annealed at 400°C ambient nitrogen to reduce the defect density in the oxide. Gates
with a dimension of 1 μm × 100 μm were defined by standard lift-off processing of electron-
beam deposited Ni/Au. 200 nm of plasma enhanced chemical vapor deposited (PECVD) silicon
nitride was used for device passivation and electron-beam deposited Ti/Au was employed for the
final metal.
Proton irradiations were performed at the Korean Institute of Radiological & Medical
Sciences (KIRAMS) using a MC 50 Scanditronix cyclotron. The proton energy at the exit of the
cyclotron was 30 MeV. The proton energy at the sample was 5 MeV after passing through two
aluminum degraders. The thickness of each aluminum degrader was 2.7mm. The beam currents
were measured using a Faraday-cup to calculate the flux density. Proton doses of 1 × 1013, 7 ×
1013, 3 × 1014, and 1 × 1015 cm-2 were used in this study.
Results and Discussion
The sheet resistance (Rs), contact resistance (Rc), and transfer resistance (Rt) of the
InAlN/GaN MOSHEMTs were extracted by transmission line measurements before and after the
proton irradiation and are summarized in Table 5-1. There was no change of Rs observed for the
MOSHEMTs irradiated at 1 × 1013 cm-2 of protons, and 2.8, 3.3 and 9.7% increases of Rs for
MOSHEMTs irradiated with 7 × 1013, 3 × 1014 and 1 × 1015 cm-2 protons, respectively. However,
much higher degradations in Rc and Rt were exhibited for the proton irradiated MOSHEMTs,
-
54
and the Rc and Rt also increased linearly proportional to the irradiation doses from 27% to 114%
and from 12% to 54%, respectively, over the dose range investigated. The differences in the
degree of degradation induced by proton irradiations between Rs and Rc/Rt were a result of the
relative changes in the Ohmic metal contacts. Since the 5MeV protons have a long projected
range, the resultant nuclear energy loss is minimal around the irradiated surface area, and the
proton energy loss in this region is dominated by the electronic-stopping mechanism and not by
nuclear-stopping. Only the latter leads to lattice damage in the semiconductor and thus, minimal
damage was introduced in the 2 dimensional-electron gas (2DEG) region located 6.3nm below
the top surface of the MOSHEMT. The result is that there is only a small change in Rs. On the
contrary, in the Ohmic contacts, there is 200 nm of metal in the contact stacks on the
MOSHEMT structure, and the nuclear-stopping mechanism for the proton energy loss is now
much more significant both in that area and in the underlying semiconductor layers. The
observed near- linear increases of Rc and Rt as a function of proton dose were a result of the
higher non-ionizing energy loss creating more damage in the metal contact regions.
Figure 5-1A shows the drain I-V characteristics of InAlN/GaN MOSHEMTs before and
after the proton irradiation dose of 1 × 1015 cm-2 at 5 MeV. The drain voltage was swept from 0 V
to 10 V, while the gate bias voltage started at 0 V with a step of -1 V. Figure 5-1 B shows the
reduction of I-V after the proton irradiation as function of doses. The averaged drain current
reductions were 96, 131, 175, and 242 mA/mm for MOSHEMT irradiated with 1 × 1013, 7 ×
1013, 3 × 1014, and 1 × 1015 cm-2, respectively. The IDS reductions were proportional to the proton
irradiation dose. The displacement damage created during the collisions between the incident
protons and the semiconductor lattice causes the reduction of electron density in the channel and
in the saturated carrier velocity[7, 59-61]. The defect centers introduced in the crystal lattice lead
-
55
to carrier removal through carrier trapping. Figure 5-1 B shows the drain current density
reduction as a function of irradiation dose. The drain current reduction increased due to the
increase of contact resistance between the metal to the semiconductor and the displacement
damage within 2DEG, resulting in reduction of electron density and carrier mobility. As more
charged traps reside in the InAlN and GaN, the potential difference between the Fermi level (EF)
and the ground state of sub-bands (E0) in the 2DEG channel became smaller, resulting in lower
sheet carrier concentrations[50, 62]. The low field drain I-Vs (< 0.5 V) were used to extract the
carrier concentrations with the charge control model[50]. Carrier removal rates of the proton
irradiations were calculated using the change of sheet carrier concentration before and after the
proton irradiation divided by the proton dose, as shown in Table 5-2. The carrier removal rate
was independent of proton dose. Ionized traps introduced in the InAlN and GaN layers changed
conduction band bending and lowered the carrier density in the 2-DEG[62]. The average carrier
removal rate of 1287±64 cm-1 was larger than we reported for AlGaN/GaN MOSHEMTs
irradiated under similar conditions (~300)[7, 63-65] and is consistent with the lower average
bond strength of InAlN relative to AlGaN. The same trend was reported previously for standard
metal-gate InAlN/GaN and AlGaN/GaN HEMTs[7, 63-66].
Figure 5-2A and B illustrate the reverse and forward gate I-V characteristics of the
MOSHEMTs before and after proton irradiation at 1 × 1013 cm-2 and 1 × 1015 cm-2, respectively,
at 5 MeV. There was no sign of changes in oxide breakdown strength, which was around 6
MV/cm and defined as the forward voltage at the forward gate leakage current of 1 mA/mm,
after proton irradiation. However, the forward bias gate current was slightly reduced, and there
was an almost one order magnitude decrease in the reverse bias gate leakage current for the
MOSHEMTs irradiated with 1 × 1015 cm-2 of protons. Since the gate oxide breakdown strength
-
56
did not change, the gate leakage current should be similar to the value prior to proton irradiation.
Koehler et al. reported that Ni voids form between semiconductor and gate interface of high dose
proton irradiated AlGaN/GaN HEMTs when using Ni/Au gate as the gate electrode, which
resulted in less gate contact area and less gate leakage current[51]. However, similar work was
performed in our study on AlGaN/GaN MOSHEMT irradiated with 1 × 1015 cm-2 proton by wet-
chemically removing the Ti/Au based gate electrode and there were no voids formed on the
Al2O3 gate oxide when using this metallization scheme. The reduction of the gate leakage current
was attributed to the improvement of device isolation by reducing the conductivity of GaN buffer