close

Вход

Забыли?

вход по аккаунту

?

Wet etching studies of aluminum nitride bulk crystals and their sublimation growth by microwaves

код для вставкиСкачать
NOTE TO USERS
This reproduction is the best copy available.
UMI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WET ETCHING STUDIES OF ALN BULK
CRYSTALS AND THEIR SUBLIMATION GROWTH
BY MICROWAVES
by
DEJIN ZHUANG
B. S., Tianjin University, 1996
A DISSERTATION
submitted in partial fulfillment o f the
requirements for the degree
DOCTOR OF PHILOSOPHY
Department of Chemical Engineering
College of Engineering
KANSAS STATE UNIVERISTY
Manhattan, Kansas
2004
Approved by:
lajor Professo
)r. James H. Edgar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3132194
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignm ent can adversely affect reproduction.
In the unlikely event that the author did not send a com plete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
UMI
UMI Microform 3132194
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
The research described in this dissertation was motivated by the need o f bulk AIN
single crystals to improve the quality o f group III nitride based devices. In this
dissertation, first the evolution o f semiconductors is reviewed. Second, historical reviews
and recent advances o f AIN crystal growth are presented. Third, the experimental setup
and characterization methods are described. Finally, four papers regarding wet etching
and sublimation growth o f AIN are attached: (1) AIN bulk crystal growth using
microwaves as heat source; (2) a review o f wet etching o f GaN and AIN; (3) anisotropic
etching technique for identifying AIN crystal polarities; and (4) defect-selective etching
to reveal dislocations in Al-polar crystals.
Single crystalline AIN platelets up to 2 x 3 mm^ and needles 3 mm long were
successfully grown by directly heating the source materials with microwaves. The grown
crystals were characterized by optical microscopy, photoluminescence (PL), Raman
spectroscopy, synchrotron white beam X-ray topography (SWBXT), and defect-selective
etching. The grown crystals have good structural quality, with etch pit density as low as
10^cm~^. A peak positioned at 5.5 eV in PL spectra was attributed to magnesium
impurities, presumably originating from the source materials.
The wet etchings o f GaN and AIN by electrochemical etching and defect-selective
etching are reviewed. The mechanism o f each etching process and etching conditions
resulting in highly anisotropic, dopant-type/bandgap selective, defect-selective, and
smooth surfaces are discussed. The applications o f wet etching techniques in device
fabrication and crystal characterization are also reviewed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The anisotropic etching technique for AIN crystals was successfully developed.
Aqueous KOH solution did not attack Al-polar surfaces, but produced hexagonal hillocks
on N-polar surfaces. The etching results suggested that freely nucleated AIN crystals
predominately have the A1 polarity facing the source materials. The hillocks on N-polar
after etching were bonded by S1 1 0 1 > planes.
Defect-selective etching in eutectic alloys resulted in hexagonal etch pits and
hillocks on Al-polar and N-polar surfaces, respectively. The reliability o f estimating
dislocation density in Al-polar crystals by defect-selective chemical etching was
confirmed by SWBXT. From the defect density point o f view, self-seeded AIN crystals
have a better quality than those grown on Si-face 6 H-SiC (0001) substrates.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
List of Figures
111
List of Tables
iv
Acknowledgments
V
vi
Dedication
Chapter 1
Chapter 2
Chapter 3
Semieonduetor Evolution
1
1.1. Silicon
1
1.2. Gallium Arsenide
2
1.3. Wide Bandgap Semiconductors
3
Group III Nitrides
6
2.1. Structure and Properties o f Group III Nitrides
6
2.2. Applications o f Bulk AIN Single Crystals
8
2.3. Market Studies o f Group III Nitrides
10
Historical Reviews and Recent Advances o f Bulk AIN Sublimation
13
Growth
Chapter 4
3.1. Growth Methods
13
3.2. Crystal Growth by Sublimation
15
3.2.1. Temperature and Pressure
16
3.2.2. Growth Modeling
17
3.2.3. Self Seeding
IS
3.2.4. Seeded Growth
20
3.2.5. Chemical Stability o f the Crucible
22
Experimental and Characterization
26
4.1. Experimental
26
4.1.1 .Sublimation Systems
26
4.1.2.Etching Experiment Setup
28
4.2. Characterization
29
4.2.1. Wet Etching
29
4.2.2. Scanning Electron Microscopy
30
4.2.3. Synchrotron White Beam X-Ray Topography
30
4.2.4. Photolumineseence
31
4.2.5. Raman Spectroscopy
31
33
Reference
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PAPERS
Paper I
Bulk AIN Crystal Growth by Direct Heating o f the Source Using
43
Microwaves
(Journal o f Crystal Growth, Vol. 262, 168, 2004)
Paper II
Wet Etching o f GaN and AIN: A Review
60
(unpublished)
Paper III
Wet Chemical Etching o f AIN Single Crystals
122
(MRS Internet Journal o f Nitride Semiconductor Research, Vol. 7, 4,
2002)
Paper IV
Defect-Selective Etching o f Bulk AIN Single Crystals in Molten
KOH/NaOH Eutectic Alloy
(Journal o f Crystal Growth, Vol. 26, 89, 2004)
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
LIST OF FIGURES
Figure 2.1
Perspective views o f the wurtzite crystal structure along a) [0001]; b)
1120 ; and c) 1010
directions
Figure 2.2
Electricity savings by sector due to SSL market penetration
Figure 3.1
Polished AIN wafer cut from a polycrystalline boule
Figure 3.2
SAM results showing compositions o f each element at AIN and TaC
interface
Figure 4.1
Photograph o f the microwave reactor
Figure 4.2
Schematic view o f the microwave reactor
Figure 4.3
Photograph o f the tungsten resistively heated reactor
111
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table 2.1 Properties o f group 111 nitrides
Table 3.1 Strengths and weaknesses o f MOCVD, MBE and HVPE
IV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
Though the list o f people who contributed their selfless supports to this work is
endless, I would like to first express my sincere thanks to my major advisor. Dr. James H.
Edgar, for his kindness o f introducing me to the field o f wide bandgap semiconductors,
guidance and valuable suggestions in research, assistance in composing this dissertation,
and patience in educating a former chemical engineer in petrochemicals to a Ph.D. of
semiconductor crystal growth. Without getting help and hints tfom his expertise in
applying chemical engineering principles to crystal growth and characterization, and his
broad knowledge o f semiconductors, I would have not been able to complete this work.
Acknowledgments extend to Dr. Keith L. Hohn in Chemical Engineering, Dr.
Andrew Rys in Electrical and Computer Engineering, Dr. Jingyu Lin in Physics, and Dr.
Mark D. Hollingsworth in Chemistry for their time and efforts on reviewing this work.
Help from Dr. Hongxing Jiang and Dr. Jingyu Lin’s group with photo luminescence
measurements. Dr. Martin Kuball’s group for Raman spectroscopy, and Dr. Jhama
Chaudhuri’s group for X-ray topography studies is greatly appreciated.
Additional appreciation goes to my colleagues and friends. Dr. Lianghong Liu,
Dr. Bei Liu, and Ms. Zheng Gu for their friendship and helpful discussions. The
professional help from Mr. David Threewit was indispensable.
The financial support from the Office o f Naval Research through grant number
NOOO14-02-1-0290 is greatly appreciated.
I am most greatly indebted to my wife, Ms. Fei Zhu, who has sacrificed a lot in
helping me go through the Ph.D. study. I must also acknowledge my parents and sister
for their everlasting support and encouragement.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
To my parents, sister, wife,
and (hveCy daughter <Demi...
VI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1
Semiconductor Evolution
The third technology revolution that started in the middle o f twentieth century is
still influencing our lifestyle by producing more and more electronic products. As the
main ingredient o f such electronic and optoelectronic products, the semiconductor has
been at the center o f this technology revolution. Considering the applications o f devices
made o f different semiconductor materials, e.g. elemental and compound, the evolution
o f semiconductors can be divided into three phases.
1.1. Silicon (Si)
The first phase o f semiconductor industry was initiated by the invention o f the Si
integrated circuits (ICs) in 1950s. The pervasive applications o f ICs - computer CPU and
memory - for example, and the ability to integrate tens o f millions electrical components
on a single silicon chip with an area o f a few square centimeters have impelled our
civilization into the information era. Driven by the huge market demands and with
constant shrinking o f device geometries and improving o f manufacturing practice, the
number o f devices integrated on silicon chips continues to grow. By 2012, a single silicon
chip will contain approximately one and half billion microprocessor transistors,
according to the International Technology Roadmap for Semiconductors (ITRS) [1]. In
spite o f the fast growth o f the silicon industry, as a semiconductor, silicon has several
disadvantages. These include, for example, low dielectric breakdown field and indirect
bandgap, both o f which are determined by the nature o f the material. Thus, new
semiconductor materials are needed to fabricate high frequency electronic devices that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
can be used in high temperature/power conditions, such as wireless and high-speed
digital communications, space applications, and the automotive industry. In addition,
conventional silicon in its bulk form is fundamentally incapable o f emitting sufficient
amounts of light for display and illumination purposes.
1.2. Gallium Arsenide (GaAs)
The second phase in semiconductor history is represented by the emergence of
narrow bandgap (Eg < 2.0 eV) compound semiconductors, such as GaAs and indium
phosphide (InP). The electronic properties o f GaAs are superior to those o f Si. It has a
higher saturated electron velocity and higher electron mobility, so the majority carriers
move faster than in silicon. This results in ICs that are faster than those made with
silicon. Consequently, the improved signal speed o f GaAs devices permits them to react
to high frequency microwave signals and accurately convert them into electrical signals
for commnunication systems. Furthermore, GaAs electronic devices generate less noise
than silicon devices. The direct bandgap o f GaAs makes fabrication o f light emitting
diodes (LEDs) possible as well.
In 1964, the first GaAs-based electronic device, a chip capable o f emitting
microwaves at frequencies as high as 90 GHz, was demonstrated by Gunn [2]. Two years
later, GaAs based infrared emitting diodes (IREDs) were reported [3]. The IREDs found
their applications in radar “guns”, automatic door openers and remote controls, and are
still in production today.
The world’s first red-emitting LED (GaAs? as an active layer) with the electricalto-optical conversion efficiency similar to that o f the ordinary incandescent bulb (4%)
was fabricated in 1968 [4]. N ot long after that, orange-yellow and yellow-green LEDs
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were all demonstrated, providing solid-state illumination eovering the wavelengths o f
940-540 nm. In 1970s, LEDs were first widely applied to large screen displays and
transportation signal lights. However, blue and violet LEDs were not available, so full
color displays were not possible.
A drawback o f GaAs and InP (and their related alloys) are their relatively lower
bandgap energies, which limit the emission wavelengths o f GaAs-based LEDs to between
the green and red region o f the spectrum. In addition, GaAs based electronic devices can
not be used at high temperatures. With its larger than Si bandgap, GaAs could be used at
higher temperature than Si, but only slightly higher. Lastly, the fragility o f GaAs makes
wafer handling a major issue in the wafer production.
1.3. Wide Bandgap Semiconductors
The third phase o f semiconductor evolution started with the wide bandgap
semiconductors (Eg > 2 eV). These include gallium nitride (GaN), aluminum nitride
(AIN), indium nitride (InN), silicon carbide (SiC), and diamond, among which GaN,
AIN, and InN are referred as group III nitrides since they are all composed o f group IIIA
elements and nitrogen. Group III nitrides have attracted much interest due to their
potential applications in short wavelength optoelectronic devices and in high temperature,
high power, and high frequency electronic devices.
Group III nitrides are suitable for high power switching devices and high
temperature devices because o f their excellent thermal and eleetrical properties, such as
high thermal conductivities, high dielectric breakdown voltages, high drift velocities in
large electric fields, high current carrying capacities, and good physical and chemical
stabilities [5]. The group III nitride based electronic devices, including the heterostructure
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
field effect transistor (HFET), heterojunction bipolar transistor (HBT), high electron
mobility transistor (HEMT), metal semiconductor field effect transistor (MESFET), and
metal-oxide semiconductor field effect transistor (MOSFET), have a wide variety of
applications in harsh environments (radiation and heat), such as spacecraft electronics, oil
drilling equipments, electric vehicles, chemical reaction monitoring and power supplies.
Moreover, the group III nitrides, especially AIN, have attractive thin film piezoelectric
properties for the fabrication o f surface acoustic wave (SAW) devices [6 , 7], which are
useful for high frequency communications.
Group III nitrides are excellent candidates for violet, blue and green LEDs as
well, due to their direct and wide bandgaps ranging from 0.85 eV (InN) [8 ] (and
references therein) through 3.4 eV (GaN) to 6.2 eV (AIN). Such solid state LEDs have
advantages o f long life, small volume, high luminous efficiency, and quick response
speed. The ability o f GaN to form solid solutions with AIN and InN, making bandgap
engineering possible, is essential for defining the emission wavelengths o f the LEDs. The
potential spectrum coverage o f LEDs fabricated with group III nitrides continuously
varies from infrared (1460 nm) to ultra violet (200 nm). This flexibility in tailoring of
emission wavelengths can lead to LEDs o f various applications: color LEDs can
potentially replace incandescent light bulbs in traffic signals due to their superior
efficiency and reliability; white LEDs could possibly replace conventional light sources,
saving tremendous amounts o f energy; and UV LEDs can be used in medical,
biotechnology, food industry and military applications.
Another exciting application o f group III nitrides is to fabricate short-wavelength
(UV to green spectral region) laser diodes (LDs). Short wavelength LDs can be used to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increase the storage capacity o f optical discs as well as to improve the resolution o f laser
scanners and laser printers. GaN based LDs will be the predominant laser used in the next
generation o f DVDs (Blu-ray Disc). Compared to GaN and InN, AIN has advantages in
fabricating short wavelength LEDs and LDs because o f its larger bandgap.
Though distinguished progress in nitride research has been made in the last two
decades, the development o f group III nitrides is still hindered by the lack o f bulk single
crystal substrates for homoexpitaxial growth, a technique capable o f improving the
quality o f current group III nitride devices dramatically, in the aspects o f both efficiency
and lifetime.
This dissertation reviews the historical research efforts and recent advances in
bulk AIN crystal sublimation growth and presents four papers regarding AIN sublimation
growth by microwaves and wet etching techniques for AIN single crystals: (1) AIN single
crystal sublimation growth via a novel technique using microwaves as the heat source;
(2) a review o f wet etching techniques for GaN and AIN; (3) anisotropic chemical etching
techniques for identifying AIN crystal polarities; and (4) molten KOH etching to reveal
dislocations in Al-polar crystals.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
Group III Nitrides
2.1. Structure and Properties of Group III Nitrides
Unlike other semiconductors, such as Si and GaAs, which have a diamond or
zincblende structure with cubic symmetry, the group III nitrides have a wurtzite structure
in their stable form with hexagonal symmetry. GaN and AIN with the zincblende or
rocksalt structure have been reported imder extreme conditions, though they are not
thermodynamically stable [9]. Fig. 2.1 shows the perspective views o f wurtzite AIN
matrix along various directions. The crystal structures o f wurtzite GaN and InN are
isomorphic to wurtzite AIN.
Figure 2.1 Perspective views o f the wurtzite crystal structure along a) [0001];
b) 1120
and c) 1010
directions
The large difference o f ionic natures between the group IIIA elements and the
nitrogen results in high bond energy in group III nitride materials, which consequently
results in their wide bandgap and high thermal and chemical stability. Table 2.1 [8 , 1014] lists some important properties o f the group III nitrides.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.1 Properties o f group III nitrides (after [8, 10-14])
GaN
AIN
InN
Basic parameters
Crystal structure
Wurtzite
Wurtzite
Wurtzite
C^dv-P^smc
Group o f symmetry
C‘*6 v--Edjwc
C*6v-P63mc
Bandgap energy (eV)
6 .2
3.44
0.85 + 0.1 [8 ]
Number of atoms in 1 cm^
9 .5 8 x 1 0 "
8 .9 x 1 0 ''
6 .4 x 1 0 "
Melting point (°C)
-2300
3000
1 1 0 0
3.255
6.81
Density (g/cm^)
6.15
Dielectric constant e(0)
8.5
8.9
Dielectric constant e(co)
4.6
5.35
9.3
Lattice constant a (A)
3.112
3.189
3.533
4.982
Lattice constant c (A)
5.693
5.186
Electrical properties
Breakdown field (MV/cm)
1 .2 - 1 .4
1 .2
5
Electron saturation velocity ( 1 0 Am/s)
1.4
2
Mobility electrons (cm^ V'* s'*)
< 1 0 0 0
300
<3200
14
Mobility holes (cm^ V* s'*)
< 2 0 0
<80
Diffusion coefficient electrons (cmAs)
7
25
<80
Diffusion coefficient holes (cm^/s)
5
0.3
Electron thermal velocity (m/s)
2 .6 x 1 0 '
1.85x10'
3.4 x 1 0 '
Hole thermal velocity (m/s)
4.0x10'*
9.4x10^
9.0x10^
Thermal properties
Bulk modulus (GPa)
2 1 0
204
140
Thermal diffusivity (cm"*/s)
1.47
0 .2
0.43
3.2
Thermal conductivity (W/cm K)
1.76
1.3
Thermal expansion coefficient Ug (10'*’K'*)
5.27 (20-800°C)
4.2 (17~477**C)
5.6 (-280°C)
Thermal expansion coefficient Uc (10'*’K'*)
4.0 (17~477**C) 4.15 (20-800°C)
3.8 (-280°C)
Heat capacity (cal mol'*K'*)
9.7
7.6
1 0 .0
Piezoelectric properties
Piezoelectric constant eis (C/m )
-0.48
-0.3
Piezoelectric constant 0 3 1 (C/m^)
-0.58
-0.33
-0.57
Piezoelectric constant 0 3 3 (C/m"*)
1.55
0.65
0.97
Other properties
-6 8 . 2
AG** (kcal/mol)
-33.0
-23.0
N i partial pressure at melting point (atm)
1 0 0
3000
» 10'
Note: All values were measured at room temperature (300 K), unless otherwise mentioned
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lacking o f inversion symmetry, the basal plane (i.e. the (0001) plane) o f group III
nitride crystals can be either N- or group III element polar. The polarity o f (0001) planes
is conventionally defined as follows. As the crystal surface is approached from the bulk
along the c-direction, if the long bond goes from the nitrogen atom to the group III atom,
the crystal is nitrogen polar. If instead the long bond goes from the group III atom toward
the nitrogen atom, the crystal is group III polar. The polarity o f the group III nitrides has
significant effects on its surface and bulk properties [15-22] as well as its electrical and
optical properties [23-27]. In addition, such spontaneous and piezoelectric polarizations
are essential for fabricating devices used in high frequency communication systems.
2.2. Applications of Bulk AIN Single Crystals
Due to the high equilibrium nitrogen pressure over GaN and its high melting point
(see Table 2.1), growing bulk GaN crystals via high nitrogen pressure solution (HNPS)
[28-30] is extremely difficult. The process requires high system pressure and temperature
(up to 1700 °C). Though extensive research has been devoted into other approaches, e.g.
hydride vapor phase epitaxy (HVPE), no bulk GaN substrate with high quality has been
reported so far. Thus, most GaN crystals to date have been grown by heteroepitaxy on
commercially available foreign substrates, such as SiC and sapphire. The large lattice
mismatch between the epilayers and substrates results in a high dislocation density in
heteroepitaxial GaN, typically in the range o f lO’ ~ IO ''cw “^ [31, 32]. Such high
dislocation densities, combined with other defects, such as nanopipes, inversion domain
boundaries, and stacking faults, increase the device threshold voltage; reverse bias
leakage currents; deplete sheet charge carrier concentrations in heterostructure field effect
transistor; degrade p-n junction abruptness; and reduce the charge mobility and thermal
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conductivity, and thus significantly increase the cost o f GaN based devices by shortening
their lifetimes [33]. Furthermore, the thermal expansion mismatch between substrate and
epilayer introduces stress and frequently results in cracks in the substrates or epilayers. A
cost-efficient bulk nitride substrate would substantially improve the quality o f epitaxial
GaN films by switching to homoepitaxy and would potentially enhance the performance
o f nitride based devices. An ideal substrate for GaN epitaxy should have following
characteristics:
>
Closely matched crystal structure and lattice parameters with GaN
> A close thermal expansion match with GaN
> Chemically stable
> High mechanical strength
> High thermal conductivity to dissipate excess heat generated by devices
W ith an isomorphic crystal structure; small lattice constant mismatch (-2.4% in
a-axis [5]); and a small difference in thermal expansion coefficient with GaN, AIN is a
promising substrate material for GaN epitaxial growth. Furthermore, AIN is an even
better substrate for high Al-content AlGaN alloy growth, where lattice match is a critical
factor to avoid dislocations and cracks. Recent literature reports on using bulk AIN
crystals to fabricate AlGaN LEDs [34], high quality GaN epilayers [35, 36], AlGaN/GaN
HFET devices [37], deep UV emitters, and multi quantum well structures [38-41],
demonstrate the promising characteristics o f AIN.
In addition to being substrate for epitaxy o f GaN, AIN finds its own applications
in SAW devices and short wavelength LEDs/LDs, where good piezoelectric properties
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and higher bandgap energies are preferred. Moreover, AIN is a good substrate for certain
electronic devices, such as field effect transistors (FET) because its high resistivity (on
the order o f 10*^~10'^ Q-cm [42-45]) simplifies the device isolation process. Its high
resistivity
also
makes
AIN
attractive
as
insulating
films
for
metal-insulator-
semiconductor structures, or as passivating layers. In contrast, unintentionally doped GaN
usually has large n-type carrier concentration. Due to its high thermal conductivity, AIN
is used in thermally conducting packaging and dielectric optical enhancement layers in
magneto-optic multi-layer structures as well. Finally, AIN can be used as an etch stop
layer in the fabrication processes o f GaN base devices (see attached paper “Wet Etching
of GaN and AIN; A Review” for details).
2.3. Market Studies of Group III Nitrides
The major market segment o f group III nitride semiconductors is GaN based
LEDs, also known as solid state lighting (SSL), for general illumination. The merits of
SSL compared to conventional illumination technologies are much greater energy
lU
2015
2025
Figure 2.2 Electricity savings by sector due to SSL market penetration (after [46])
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
efficiency, longer lasting, and potentially cost competitive. W ith huge government and
private investment in nitride research, the Department o f Energy (DOE) o f the United
States has an ambitious program to develop advanced SSL technologies hy the year o f
2015 [46]. If successful, the electricity savings due to the SSL market penetration is
forecasted to be $23 billion in the year o f 2025, as shown in Fig. 2.2 [46]. Foreseen by
the DOE, the niche market for SSL, including motor vehicles and stationary indoor and
outdoor installations, has remarkable market values as well [46].
This forecast was made assuming that SSL will be able to satisfy the requirements
o f its niche market and general lighting by 2025. In other words, the current key
technology problems facing the industry will soon be completely solved, such as the
availability o f high quality bulk substrates, and other challenges in the fabrication
process, such as low damage dry etching. With at least 180 laboratories all over the world
studying GaN and its related compounds and considerable research funding, technology
breakthroughs in group III nitride semiconductors are expected to emerge in the near
future.
According to the projections by Strategies Unlimited o f Mountain View,
California, the total markets o f GaN based LDs and photodetectors will reach $2 billion
and $18.1 million by the year o f 2009 [47]. The market o f electronic devices including
RF/microwave, power switches, power rectifiers, high voltage rectifiers, and high
temperature devices was estimated to be $436 million by 2009 [47]. All o f the above
markets are expected to climb even more dramatically after 2009 [47].
Being an ideal substrate for GaN epitaxy and considering the giant market o f SSL,
AIN is surely a noteworthy material. Its other application markets, such as ultraviolet
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LEDs and high power radiofrequency devices, contribute to the growing research interest
as well, though the total market o f AIN is smaller than that o f GaN.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTERS
Historical Review and Recent Advances of Bulk AIN Sublimation
Growth
3.1. Growth Methods
Epitaxial AIN growth has been investigated using several growth techniques, such
as metal organic chemical vapor deposition (MOCVD) [48-54], molecular beam epitaxy
(MBE) [55-58], and HVPE [59-62]. O f these, MOCVD is the most cost effective, highest
throughput, and highest quality thin film deposition tool. In addition to a high growth rate
(on the order o f 1~10 pm/hr), MOCVD can provide uniform coverage o f non-planar
shapes, and layer thickness can be monitored in situ. MOCVD, especially in combination
with the epitaxial lateral overgrowth (ELO) technique, produces GaN films o f the highest
reported quality to date, with dislocation densities as low as 1 0 ® [ 6 3 ] . MBE has
advantages o f better control o f growth parameters, relatively lower growth temperature,
no hydrogen carrier gas involved, and in situ characterization. However, MOCVD and
MBE are capable o f producing epitaxial thin films, but not bulk crystals. Furthermore, the
large lattice constant mismatch, coupled with the difference in the thermal expansion
coefficient between AIN and the substrates (sapphire or SiC) often lead to AIN films with
a high density o f dislocations (10^ ~ 1 0 ‘“ c m [64, 65]), cracking and low growth
reproducibility. HVPE is favored for growing free-standing GaN bulk crystals because o f
its high growth rate (10-100 pm/hr). Nevertheless, growing bulk AIN by HVPE is
difficult. The complications from excessive impurity incorporation, originating from the
reactions between AICI3 and quartz reactor wall, and severe homogeneous gas phase
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reactions have hindered the development o f HVPE AIN growth. A comparison of
strengths and weaknesses o f each aforementioned growth method is presented in Table
3.1.
Table 3.1 Strengths and weaknesses o f MOCVD, MBE and HVPE
Growth
technologies
MOCVD
MBE
HVPE
Disadvantages
Advantages
Atomically sharp interfaces
High growth rate and throughput
Low dislocation density (using ELO)
Uniform coverage and large area
growth
In situ thickness monitoring
Scalable for mass production_______
Atomically sharp interfaces
Precise control of growth parameters
Relatively low growth temperature
No hydrogen involved
In situ characterization
High purity growth environment
Very high growth rate
Bulk crystal growth
Simple growth technique
o Lack o f precise in situ characterization
o Need large quantities o f NH3
o Thin film growth only
o
o
o
o
Require ultra high vacuum
Low throughput and growth rate
Thin film growth only
High dislocation density
o No sharp interface
o Complicated reactions in vapor phase
o Relatively high process cost (need to
replace quartz reactor wall frequently)
Bulk AIN crystal growth by several processes has been studied including
vaporization [6 6 ], solution routes [67,
6 8
], and physical vapor transport (PVT) [69, 70].
Recently, Schlesser et al. [71, 72] achieved hulk AIN crystal growth via a vaporization
method using high purity aluminum metal and ultra high purity (UHP) nitrogen as source
materials. Growth rate as high as 5 mm/hr in the c-plane and transparent c-plates o f up to
10x5 mm^ surface area were demonstrated within 2 hour o f growth at 2100 °C [71]. The
higher growth rate compared to sublimation growth was attributed to the higher
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
alviminum vapor pressure over pure aluminum than AIN. However, longer-term growth
was counteracted by the formation o f a nitride layer over the metallic aluminum source,
which gradually decreased the aluminum vapor pressure. Growing AIN by solution route
requires extremely high system pressure to enhance the solubility o f nitrogen in the
solution [6 8 ].
Sublimation-recondensation (abbr. sublimation) technique, one kind o f PVT
growth method, is the most successful bulk AIN single crystal growth method thus far. In
this method, AIN source decomposes to gaseous Al and nitrogen at elevated temperature
and then crystallizes at the colder end o f the crucible as single crystals. To effectively
enhance and control the growth rate, nitrogen is used as the process gas during growth.
Though the growth rate o f the sublimation process is generally lower than that in the
vaporization and solution routes, sublimation has advantages o f long-term growth
capability and easy implementation o f the process. To date, the largest AIN single crystal
boules 15 mm in diameter grown by sublimation were prepared by Schowalter and
coworkers [73-76] with FWHM 100 arcsec 1122
V
X-ray diffraction rocking curves and
/
density o f dislocation lower than lO^'cw'^ [77].
3.2. Crystal Growth by Sublimation
The steps in sublimation crystal growth include sublimation o f the source
materials; mass transport in the bulk gas phase; adsorption on the growth surface; surface
diffusion; and surface desorption [78]. The sublimation and recondensation o f AIN
follow the reaction:
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
^ / W ( s ) « J / ( g ) + iw ,( « )
(l)
Though the sublimation process seems simple, the actual growth process is quite
complicated. Various factors, including temperature, pressure, mass transport, growth
kinetics, impurity incorporation, and thermodynamics, are all o f great importance. The
two main challenges in AIN sublimation growth are the absence o f appropriate
substrate/seeding materials, and ehemically/physieally stable crucible materials in the
AIN crystal growth environment.
3.2.1. Temperature and Pressure
Temperature is a critical process parameter to control the crystal growth habit.
Sublimation o f AIN is feasible over a wide range o f temperatures starting from 1850 °C
[79], but stable growth o f well-faceted crystals is possible only at temperatures exceeding
2100 °C [80]. However, according to thermodynamic calculations, the growth
temperature should not exceed 2493 °C, where the formation o f liquid aluminum on
nucleation surface is inevitable [69, 80]. Tanaka et al. [81] found that the shapes o f AIN
crystals are temperature dependent: needles with rectangular cross section, needles with
hexagonal cross section, and flat plates were obtained at relatively lower (1900 °C),
medium (2000 °C) and higher temperatures (2100 °C), respectively [81]. The latter had a
(0002) rocking curve FWHM as low as 12 arcsec, indicating better crystalline quality at
high process temperature. Most recently, a similar temperature dependent crystal growth
was reported by Epelbaum et al. [82]. With increasing temperatures, the AIN crystals
change from needle-like to prismatic forms and then turn to thick, asymmetric platelets
[82].
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Pressure is another adjustable parameter in sublimation growth. In general, lower
pressure leads to higher sublimation rate. At extreme condition, e.g. ultra high vacuum,
the mass transport o f reactive species switches from diffusion to drift, resulting in several
orders o f magnitude higher growth rate [83]. However, under supersaturation conditions
(low pressure and high temperature), the relatively high Al vapor mole fraction will
degrade the performance o f furnace fixtures due to the volatile nature o f aluminum vapor.
Moreover, the system pressure has a profound effect on the growth mechanism [84, 85].
3.2.2. Growth Modeling
A comprehensive understanding o f the sublimation growth mechanism by process
modeling is a key for the successful development o f larger AIN crystals [8 6 ].
Dryburgh [87] developed a model to predict the growth rate o f AIN sublimation
system using thermodynamic analysis and concluded that the maximum growth rate
occurs at the stoichiometric vapor phase composition. However, the author [87] did not
validate his model with experiments. Segal et al. [84] proposed a model that takes into
account both diffusive and convective transport o f gaseous Al and nitrogen as well as the
kinetic limitation o f nitrogen adsorption/desorption on AIN surfaces. Based on the model
predictions, the authors claimed that the growth rate reaches a maximum under somewhat
nitrogen rich conditions, and is much lower under Al-rich conditions [84]. Under
stoichiometric and slightly nitrogen rich conditions, the low sticking coefficient o f
nitrogen on the growth surface, estimated to be 10'"^ at 1800 °C [88-90], suppresses the
overall growth rate [83, 84]. The predictions o f the latter model agreed well with the
experimental data [91, 92].
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Liu and Edgar [78] originally modeled their sublimation system by considering
the mass transport in the gas phase only. The calculated growth rates generated from
experimental data for system pressure over
1 0 0
torr were successfully predicted by the
model, indicating the dominant effect o f mass transport under those conditions. However,
the growth rate at lower pressure
( < 1 0 0
torr) was inconsistent with model predictions.
This implied that at low pressure, i.e. slightly supersaturated condition, mass transport is
not the rate-limiting step anymore. To compensate for this difference, Liu and Edgar [85]
further modified their model by introducing factor o f surface kinetics. The modified
global growth rate model can successfully predict the growth rates in all pressure ranges
[85].
Most recently, Noveski et al. [93] proposed a one dimensional mass transfer
model based on equilibrium sublimation and gas phase diffusion. Based on the
observation that the predictions from this model were in good agreement with
experimental values, the authors concluded that, at a pressure o f 600 torr, the growth rate
of AIN sublimation is controlled by the diffusion o f Al species to the growth surface [93].
3.2.3. Self Seeding
Thus far, the best quality bulk AIN crystals have been produced by self-seeding, a
process in which no seed is employed. The souree materials sublime at elevated
temperature, transport: to the eooler end o f the erueibles, recondense, and erystallize as
single crystals. Ideally, only one nucleus forms at the coldest end o f the crucible in the
initial stage, providing a perfect nucleation site for subsequent growth. In practice,
however, numerous nuclei form and each o f them leads to single crystal grain at the
initial stage o f nucleation. Some o f these grains may merge together and form
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polycrystalline AIN, while other grains remain as single crystals as growth proceeds. The
single crystal grains eventually begin to dominate and the polycrystalline regions
decrease as the boule grows [70].
Compared to seeded growth on SiC or sapphire, AIN crystals grown hy selfseeding method have higher quality since the process avoids the stress-causing lattice and
coefficient o f thermal expansion mismatch, plus the inheritance o f dislocations from the
substrates. The growth temperature, and thus the growth rate, can be higher than those in
seeded growth too, as substrate decomposition is not a concern in self-seeding processes.
A polished AIN wafer produced at KSU, about 1 mm thick, cut from a
polycrystalline AIN boule grown by sublimation directly on a polycrystalline tungsten lid
is shown in Fig. 3.1. Single crystalline grains up to 3 x 3 mm^ are clearly seen. The top
left region is polycrystalline AIN competing with the single crystalline grains. The dark
spots are probably due to incomplete polishing, and the cracks are possibly caused by the
Figure 3.1 Polished AIN wafer cut from a poly crystalline boule, the square is 1 mm^
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
differences in lattice constants and thermal expansion coefficients between AIN and
tungsten. An unpolished AIN wafer having single crystalline grains up to 5 x 5 mm^ was
recently reported by Bickermann et al. [94, 95].
3.2.4. Seeded Growth
Achieving seeded growth will immensely improve the crystal quality as the seed
provides high quality nucleation sites. The crystal orientation and polarity are more easily
controlled compared to self-seeding growth. Furthermore, scaling up o f crystal size is no
longer a problem as long as the decomposition o f substrate can be effectively avoided at
the stage of initial growth.
SiC is a good substrate for AIN seeded growth due to its relatively small lattice
mismatch with AIN (0.9%). Balkas et al. [96] grew AIN single crystalline platelets on
1 0
x 1 0 mm^ SiC substrates in a SiC-coated graphite crucible and obtained discrete
hexagonal AIN crystals with dimensions o f approximately 2 x 2 mm^. A growth rate o f
up to 0.5 mm/hr was demonstrated in a temperature range o f 2150 to 2250 °C.
Contamination o f Si, C and O and high density o f screw dislocations and cracks due to
the stress caused by thermal expansion mismatch between AIN and SiC were observed
[96]. A similar observation o f cracking was reported in [97]. Edgar and coworkers [78,
85, 98-103] designed a novel sublimation sandwich technique and utilized a tungsten
resistively-heating reactor to grow AIN on SiC. A MOCVD layer o f AIN and AlN/SiC
alloy were deposited on SiC substrate prior to pure AIN growth to compensate the
thermal expansion and lattice mismatch. Though cracks and stress can not be completely
eliminated, a free standing AIN crystal ( 4 x 6 mm^) was obtained after 100 hour growth
[99]. The growth temperature, and therefore the growth rate, was kept low since silicon
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from SiC decomposition will form a low melting point euteetie alloy with tungsten and
ruin the furnace fixtures. Tungsten heater degradation when SiC is used as seeds was also
reported by Epelbaum et a/. [104].
Liu and Edgar [105] summarized three major problems with seeding AIN on SiC:
(1) cracks in AIN film are inevitable due to the higher thermal expansion coefficient of
AIN compared to SiC and the low critical shear stress o f SiC; (2) the decomposition o f
SiC at AIN growth condition introduces Si and C into the grown AIN films, or even
worse forming an AIN and SiC alloy instead o f pure AIN; and (3) achieving two
dimensional growth on Si face (0001) is difficult since the growth is prompted by steps
associated with screw dislocations. Epelbaum et al. [104] reported similar observations of
three dimensional growth o f AIN on Si-face on-axis (0001) 6H-S1C. However, they found
continuous growth o f AIN can be realized by employing a slightly off-axis n-orientated
substrate, where the step-flow growth mode was achieved [104].
Using AIN single crystals grown by self-seeding instead o f SiC is an alternative
approach for seeded growth. However, since self-seeding crystals are usually small,
sealing up the crystal size is always a eoneem. Furthermore, random nucleation near the
small seeds may occur, leading to polyerystalline AIN under high growth rate condition.
Most recently, a low oxygen content (100 ppm) polyerystalline AIN wafer, 0.5-1
mm thick and 12.5 mm in diameter, was reported by Bickermarm et al. [94, 95]. The
wafers consisted o f single crystalline areas up to 5 x 5 mm^. This offers the opportunity
o f growing even larger AIN single crystals by repeated growth. However, Noveski et al.
[106, 107] found that repeated homoepitaxial growth without treatment/optimization of
the source material, growth surface, and growth procedure usually results in
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polyerystalline AIN growth, i.e. the consecutive growth after the exposure o f wafer to air
restarts nucleation with small grains. This phenomenon, called secondary nucleation, is
believed to be caused by surface oxides, impurities from source material and growth
ambient and low temperature deposition during temperature ramp-up [106]. By sintering
the source material at elevated temperature before growth and temporarily inverting the
temperature gradient during ramp-up, the authors [106, 107] suppressed the secondary
nucleation and observed a gradual grain size expansion. Besides sintering the source
materials before growth, adding hydrogen during growth might be another approach to
reduce oxygen in the sublimation system. Karpov et al. [108] found that AIN crystal
growth in hydrogen containing ambient, with hydrogen partial pressure as high as
p
—^
= 0 .8 , has a comparable crystal growth rate to that in pure nitrogen.
^total
3.2.5. Chemical Stability of the Crucible
The sublimation growth o f AIN requires a process temperature higher than 2100
°C in order to obtain a viable growth rate and well faceted crystals [78, 80]. At such an
extreme temperature, the furnace fixture can react with Al or nitrogen, sublime or even
melt. Keeping the furnace fixture from damage will significantly reduce the crystal
growth process cost. Good stability o f the crucible material is also required to ensure low
impurity incorporation, which has a profound effect on defect generation, nucleation and
crystal morphology.
A direct solution o f this problem is to use a chemically and physically inert
material as the crucible. Boron nitride (BN) and the nitrides/carbides o f refractory
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transition metals (W, Ta, Nb, Zr) are all promising candidates since they have high
melting points [109].
BN has been widely used as the crucible material in AIN sublimation growth [7172, 110]. The crystals grown in BN were colorless and transparent, and the growth rate
was anisotropic along different crystalline directions [110, 111]. Depending on growth
temperature, crystals habits include whiskers, hexagonal platelets, and prisms. Some
crystals have striations running along the growth direction, which are probably caused by
boron incorporation. Compositional analysis by glow-diseharge mass spectrometry
(GDMS) shows about 100 ppm boron in grown crystals compared to only 2.5 ppm in the
sintered AIN powder source [111].
The feasibility o f using refractory metals and their carbides/nitrides as crucibles is
still under investigation. Preliminary results show that the growth is isotropic and the
crystals are well-faceted [111]. AIN nucleates in a much higher density on tantalum
carbide (TaC) compared with BN, rapidly forming a continuous deposition [110].
0.4
0.3
0,2
Distance (pm)
Figure 3.2 SAM results showing compositions o f each element at AIN and TaC interface
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Prismatic needles and hexagonal hillocks were obtained in TaC and niobium earbide
(NbC) coated graphite crucible [110]. However, erueibles made o f nitrides or carbides
suffer severe cracking, which is believed to be eaused by the diffusion o f aluminum and
nitrogen along the grain boundaries and/or mismatch o f thermal expansion coefficients.
An example o f aluminum and nitrogen diffusion into TaC is shown in Fig. 3.2, a
scanning auger microseopy (SAM) profile showing the eomposition ehange o f eaeh
element at the interfaee o f AIN crystal and TaC foil.
Crystals grown in tungsten crucible usually have an amber eolor, presumably due
to nitrogen vacaneies or aluminum interstitials [69]. Though it was reported that tungsten
is prone to grain bomidary attack by aluminum [69], the tungsten erueibles used in our
group remains good shape after long growth times (>100 hours). No Al was detected by
SAM in a piece o f cmshed tungsten crucible [112]. Nevertheless, seeded growth on SiC
substrate is not viable in a tungsten heating-element fumaee, since the sublimed silieon
and/or carbon will form low melting point eutectic alloy or react with tungsten and ruin
the whole reactor setup [104].
Base on above discussion, we conclude that, in view o f its lifetime, tungsten
might be an ideal crucible material for AIN sublimation growth in a carbon free
environment. A detailed investigation regarding crucible material selection can be found
in [ 1 1 2 ].
An indirect approach to prevent crucible from damage is to use an inverted
temperature profile, i.e. the source materials inside crucible are heated to high
temperature to achieve AIN sublimation growth, while the crucible is maintained in a
relatively low temperature range. AIN crystal growth in such an inverted temperature
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
profile was recently demonstrated by Zhuang et al. [113] using microwaves as heat
source. Microwave heating occurs in the body o f ceramic materials with energy being
lost from the surface by radiation, conduction, and convection. The temperature gradient
is thus opposite to traditional heating methods, in which the surface o f a body is heated
and the interior is heated by conduction. Temperature readings from two infrared
pyrometers confirmed that the source temperature was higher than that on crucible. AIN
single crystal platelets up to 2 x 3 mm^ and needles 1 mm in diameter and 3 mm in length
were successfully grown by directly heating the source materials with microwaves.
Raman spectra, synchrotron white beam X-ray topography (SWBXT) and the etch pit
density (EPD), confirmed that these crystals have good structural quality, with
dislocation density as low as 1 0 ^ [ 1 1 3 , 114].
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4
Experimental and Characterization
4.1. Experimental
4.1.1. Sublimation Systems
Two sublimation systems were employed in this study.
Figure 4.1 Photograph o f the microwave reactor
The photograph o f the microwave growth reactor, developed by Micramics, Inc.,
is shown in Fig. 4.1. The schematic o f this system is shown in Fig. 4.2. Two separate 3
kW variable power supplies provide the microwaves for the sublimation process. The
AIN source, with approximately 0.6 wt% oxygen as the main impurity (analyzed by a
standard inert gas fusion method, TC-136, Leco Co.), was contained in a cylindrical
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pyrolytic boron nitride (pBN) crucible. This crucible was packed in hexagonal BN
powder contained inside a pBN retort. The pBN containers were surrounded by alumina
insulation, and the entire arrangement was held in a stainless steel chamber. The alumina
insulation is transparent to microwaves, but is thermally insulating { k = 0.39 W I m - K at
1650 °C). The BN powder acted as additional insulation and helped to absorb the
microwave energy. A computer controlled tuning plate at the bottom o f chamber was
adjusted to minimize microwave reflection by moving up and down. The source
temperature and crucible sidewall temperature were measured by two color infrared
pyrometers from the top and side viewpoint, respectively. The growth time was varied
from 12 hours to 24 hours and the chamber pressure was kept constant at 910 torr.
to vacuum
1.
2.
3.
4.
Source MW Generator
Seed MW Generator
Top View Point
Top Plate
5.
6.
7.
8.
Chamber
Insulator
Source Wave-guide
Seed Wave-guide
9.
10.
11.
12.
Large Crucible
BN Powder
Small Crucible
AIN Source
Figure 4.2 Schematic view o f the microwave reactor
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The second growth reactor was a commercial furnace manufactured by
CENTORR, as shown in Fig. 4.3. The reactor is resistively heated by two wire mesh
tungsten heating elements and is capable o f reaching 2400 °C at a pressure o f one
atmosphere. The temperature gradient was estimated to be 3 -5 °C/mm. The tungsten
crucible was included inside a tungsten retort to prevent any escape o f Al species during
growth. The temperature at retort lid was measured by an infrared pyrometer.
Figure 4.3 Photograph o f the tungsten resistively heated reactor
4.1.2. Etching Experiment Setup
Aqueous and molten KOH etching was undertaken to identify the crystal polarity,
the types o f defects present, and to estimate the defect density. Before etching, all AIN
samples were cleaned with hydrochloric acid (HCl) for ten minutes to remove any
impurities on the surface. This does not etch AIN at all. The etching parameters in
aqueous KOH were set as 10 minutes at 60 °C by calculating the etch rates o f a
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polyerystalline AIN at the same etching condition. The best etching condition for
KOH/NaOH eutectic alloy was determined as 2 -6 minutes at 350-380 °C for Al polarity
and 1-3 minutes at 300 °C for nitrogen polarity, respectively. The molten eutectic was
contained in a platinum crucible held in an aluminum plate. A quartz cover was used to
stabilize the temperature. After etching, all samples were rinsed in 38 wt% HCl solution
for 5 minutes to neutralize the KOH residues.
Scanning electron microscopy (SEM) was used to determine the etch pit density
(EPD). Synchrotron v^hite beam X-ray topography (SWBXT) was also employed as an
effective tool for dislocation decoration. See the next section for details.
4.2.
Characterization
4.2.1. Wet Etching
Wet etching is one o f the oldest methods used to characterize single crystalline
materials. Defect-selective etching produces etch pits or hillocks on a semiconductor
surface due to the inhomogeneous nature o f defects (incomplete chemical bonds or higher
impurity concentrations) compared with the crystal matrix. When a new etchant or a new
etching system has been carefully calibrated with other more sophisticated methods, e.g.
transmission electron microscopy (TEM), atomic force microscopy (ATM) and/or X-ray
topography (XRT), one can estimate the defect density in single crystalline materials by
measuring the EPD. Estimation o f dislocation density by defect-selective etching has
advantages o f low cost, simple experimental procedure, capability o f examining large
sample areas, and no requirement o f sample geometry. In addition, etching o f group III
nitrides in aqueous KOH solution was foimd effective in identifying crystal polarity.
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Details o f etching characterization o f AIN arc included in the attached paper: “Wet
Etching o f GaN and AIN: A Review”.
4.2.2. SEM
Scanning electron microscopy (SEM) is the best known and most widely-used
surface analytical techniques. The primary electron beam bombards the sample surface
and generates many low energy secondary electrons. The intensity o f these secondary
electrons is largely governed by the surface topography o f the sample; thus, the surface
morphology can be revealed by analyzing the intensity o f secondary electrons as a
function of position o f the scanning primary electron beam. A Hitachi 4700-S SEM was
used in this study to picture the etching features.
4.2.3. SWBXT
Synchrotron white beam X-ray topography (SWBXT), a family member o f X-ray
topography, is a non-destructive technique that can be used to rapidly characterize single
crystals with low defect density (< 1 0 ^cw“^ ) , providing both the character and
distribution o f crystallographic defects [115]. SWBXT was employed in this study to
validate the reliability o f defect-selective etching o f AIN crystals. The synchrotron
topography experiments were carried out at the Synchrotron Radiation Topography
experimental station 2.2 at the Stanford Synchrotron Radiation Laboratory, Stanford, CA.
Transmission topographs were taken using Laue diffraction technique and white beam
radiation.
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.4. Photoluminescence
Photoluminescence (PL) is a nondestructive technique for the determination of
certain impurities that produce radiative recombination processes in the semiconductors
[116]. When a semiconductor is excited by an optical source - a laser with energy higher
than the semiconductor bandgap - for example, electron hole pairs (EHPs) are generated.
By measuring the intensities of photons emitted by the radiative recombination o f EHPs,
the presence o f specific impurities is determined. Though identifying impurities in the
semiconductors by PL is simple, the measurement o f the impurity density is more
difficult, and the samples need to be cooled to temperatures near liquid helium (4.2 K) to
minimize the thermally activated nonradiative recombinations. Deep UV (down to 196
nm) PL spectroscopy was employed in this study to detect the presence o f impurities in
the AIN crystals.
4.2.5. Raman Spectroscopy
Raman spectroscopy is an effective tool for structure characterization of
semiconductors. It is sensitive to strain and crystal structure changes, allowing it to be
used to detect stress and to characterize the crystalline quality. In addition, Raman
spectroscopy is capable o f determining the sample composition and free carrier
concentration. When a light is scattered from the surface o f a sample, the scattered light
is found to contain mainly wavelengths that were incident on the sample. However,
scattered lights with different wavelengths, representing the interaction o f incident light
with optical phonons (Raman scattering), is also detectable, though their intensity is low.
By measuring the intensities o f such Raman scattered light, a Raman spectrum is
generated. A detailed discussion o f Raman spectroscopy can be found in [117].
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Generally in Raman spectra o f AIN, the width (FWHM) and phonon frequency of
the E2 (high) Raman peak are sensitive to the crystalline quality and stress, respectively.
In this study, Raman spectroscopy was employed to characterize the stress and crystalline
quality o f grown AIN crystals. The Raman spectra were recorded from the sample using a
Renishaw micro-Raman system with the 488 nm line o f an Ar^ ion laser as excitation
sources.
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reference:
1. International Technology Roadmap for Semiconductor, Semiconductor Industry
Association, (1997)
2. J. B. Gunn, IBM J. Res. & Dev., 8, 141 (1964)
3. H. Rupprecht, J. M. Woodall, K. Konnerth and D. G. Pettit, Appl. Phys. Lett., 9, 221
(1966)
4.
D. K. Hillman and G. E. Smith, IEEE
Spectrum 5, 62 (1968)
5.
J. H. Edgar, (Editor), Properties o f
Group III Nitrides, Electronic Materials
Information Service (EMIS), London, (1994)
6
. H. Okano, Y. Takahashi, T. Tanaka, K. Shibata, and S. Nakano, Jpn. J. Appl. Phys.,
31, 3446 (1992)
7. K. Tsubouchi, andN . Mikoshiba, Ultrason. Symp. Pro. IEEE, 299 (1983)
8
. Y. Nanisbi, Y. Saito, and T. Yamagucbi, Vacuum Science and Technology: Nitrides
as Seen by the Technology, (Eds.) T. Paskova and B. Monemar, 401, Research
Signpost, Trivandrum, India (2002);
S.-H . Wei, X. Nie, I. G. Batyrev,and S. B.
Zhang, Phys. Rev. B, 67, 165209 (2003)
9. I. Groczyca, N. E. Christensen, P. Perlin, I. Grzegory, J. Jun, and M. Bockowski,
Solid State Commun., 79, 1033 (1991)
10.
O. Madelung, (Editor), Semiconductors: Group IV E lem ents and III-V Compounds,
Springer-Verlag, Berlin (1991)
11.
O. Madelung, Semiconductors-Basic Data, 2"‘*ed.. Springer-Verlag, Berlin (1996)
12.
M. Leszczynski, T. Suski, P. Perlin, H. Teisseyre, I. Grzegory, M. Bockowski, J.
Jim, S. Porowski, and J. Major, J. Phys.: D, 28, A149 (1995)
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13. http://www.ioffe.rssi.ru/SVA/NSM/Semicond/
14. http://www.onr.navv.mil/sci tech/information/312 electronics/ncsr/properties.asp
15. S. Keller, B. P. Keller, Y. F. Wu, B. Heying, D. Kapolnek, J. S. Speck, U. K.
Mishra, S. P. Denbaars,
Phys. Lett.,
6 8
, 1525 (1996)
16. A. Bykhovski, B. Gelmont, M. Shur, J. Appl. Phys., 74, 6734 (1993)
17. M. Seelmann-Eggebert, J. L. Weyher, H. Obloh, H. Zimmermann, A. Rar, and S.
Porowski, ylpp/. Phys. Lett., 71, 2635 (1997)
18. J. L. Rouviere, J. L. Weyher, M. Seelmann-Eggebert, and S. Porowski, Appl. Phys.
Lett., 73,
6 6 8
(1998)
19. M. Sumiya, K. Yoshimura, K. Obtsuka, and S. Puke, Appl. Phys. Lett., 76, 2098
(2000)
20. E. C. Piquette, P. M. Bridget, Z. Z. Bandic, and T. C. McGill, J. Vac. Sci. Technol.
B, 17, 1740 (2000)
21. B. Daudin, J. L. Rouviere, and M. Arlery, Appl. Phys. Lett., 69, 2480 (1996)
22. L. K. Li, M. J. Jurkovic, W. I. Wang, J. M. Van Hove, and P. P. Chow, Appl. Phys.
Lett., 76, 1740 (2000)
23. S. F. Chicbibu, A. Setoguchi, A. Uedono, K. Yoshimura, and M. Sumiya, Appl.
Phys. Lett., 78, 28 (2001)
24. G. A. Martin, A. Botchkarev, A. Rockett, and H. Morko 9 , Appl. Phys. Lett.,
2541 (1996)
25. U. Karrer, O. Ambacher, and M. Stutzmann, Appl. Phys. Lett., 7 7 , 2012 (2000)
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 8
,
26. H. Morko?, R. Cingolani, and B. Gil, Solid-State Electron., 43, 1909 (1999)
27. V. Kirilyuk, A. R. A. Zauner, P. C. M. Christianen, J. L. Weyher, P. R. Hageman,
and P. K. Larsen, Appl. Phys. Lett., 76, 2355 (2000)
28. I. Grzegory, and S. Porowski, Thin Solid Films, 367, 281 (2000)
29. S. Porowski, J. Cryst. Growth, 166, 583 (1996)
30. S. Porowski, J. Cryst. Growth, 189/190, 153 (1998)
31. D. Kapolnek, X. H. Wu, B. Heying, S. Keller, B. P. Keller, U. K. Mishra, S. P.
DenBaars, and J. S. Speck,
Phys. Lett., 67, 1541 (1995)
32. F. A. Ponce, B. S. Krusor, J. S. M. Jr., W. E. Plano, and D. F. Welch, Appl. Phys.
Lett., 67, 410 (1995)
33. S. Mahajan, Acta Mater., 48, 137 (2000)
34. T. Nishida, T. Makimoto, H. Saito, and T. Ban, Appl. Phys. Lett., 84, 1002 (2004)
35. G. Tamulaitis, I. Yilmaz, M. S. Shur, R. Gaska, C. Chen, J. Yang, E. Kuokstis, A.
Kan, S. B. Schujman, and L. J. Schowalter, y4/7J!7/. Phys. Lett., 83, 3507 (2003)
36. G. Tamulaitis, I. Yilmaz, M. S. Shur, R. Gaska, C. Chen, J. Yang, E. Kuokstis, A.
Khan, J. C. Rojo, and L. J. Schowalter, Mat. Res. Soc. Symp. Proc., 743, L3.34
(2003)
37. X. Hu, J. Deng, N. Pala, R. Gaska, M. S. Shur, C. Q. Chen, J. Yang, G. Simin, M.
A. Khan, J. C. Rojo, L. J. Schowalter,^/?/?/. Phys. Lett., 82, 1299 (2003)
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38. J. C. Rojo, L. J. Schowalter, G. Slack, K. Morgan, J. Barani, S. Schujman, S.
Biswas, B. Raghothamachar, M. Dudley, M. Shur, R. Gaska, N. M. Johnson, and
M. Kneissl, Mat. Res. Soc. Symp. Proc., 112 , K l . l (2002)
39.
X. Hu, R. Gaska, C. Chen, J. Yang, E. Kuokstis, A. Khan, G. Tamulaitis, I. Yilmaz,
M. S. Shur, J. C. Rojo, L. J. Schowalter, Mat. Res. Soc. Symp. Proc., 743, L6.30
(2003)
40.
R. Gaska, C. Chen, J. Yang, E. Kuokstis, A. Khan, G. Tamulaitis, I. Yilmaz, M. S.
Shur, J. C. Rojo, and L. J. Schowalter,^/?/?/. Phys. Lett., 8 1 , 4658 (2002)
41. R. Gaska, Q. Fareed, G. Tamulaitis, I. Yilmaz, M. S. Shur, C. Chen, Y. Yang, E.
Kuokstis, A. Khan, J. C. Rojo, and L. J. Schowalter, Mat. Res. Soc. Symp. Proc.,
764, C6.9 (2003)
42.
J. Edwards, K. Kawab , G. Stevens, and R. H. Tredgold, Solid State Comm., 3, 99
(1965)
43.
G. A. Cox, D. O. Cummins, K. Kawabe, and R. H. Tredgold, J. Phys. Chem. Solids,
28, 543 (1967)
44.
S. Yoshida, S. Misawa, Y. Fujii, S. Takada, H. Hayakawa, S. Gonda, and A. Itoh, J.
Vac. Sci. Technol, 16, 990 (1979)
45.
W. M. Yim, E. J. Stofko, P. J. Zanzucchi, J. I. Pankove, M. Ettenberg, and S. L.
Gilbert, J. A ppl Phys., 44, 292 (1973)
46.
http://www.netl.doe.gov/ssl/materials.html
47.
http://www.tdii.com/news/TruebulkGaN.htm
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48. T. L. Chu, D. W. Ing, and A. J. Noreika, Electrochem. Technol, 6, 56 (1968)
49. W. M. Yim, E. J. Stofko, P. J. Zanzucchi, J. I. Pankove, M. Ettenberg, and S. L.
Gilbert, J. A ppl Phys., 44, 292 (1973)
50. T. L. Cbu, and R. W. Kelm, Jr., J. Electrochem. Soc., 122, 995 (1975)
51. A. Saxler, P. Kung, C. J. Sun, E. Bigan, and M. Razegbi, Appl. Phys. Lett., 64, 339
(1994)
52. P. Kung, A. Saxler, X. Zbang, D. Walker, T. C. Wang, 1. Ferguson, and M.
Razegbi,^/?/?/. Phys. Lett., 66, 2958 (1995)
53. J. D. Hartman, A. M. Roskowski, Z. J. Reitmeier, K. M. Tracy, R. F. Davis, and R.
J. Nemanicb, J. Vac. S ci Technol, A, 21, 394 (2003)
54. H. M. Liaw, R. Doyle, P. L. Fejes, S. Zollner, A. Konkar, K. J. Lintbicum, T.
Gebrke, R. F. Davis, Solid-State Electron., 44, 747 (2000)
55. S. Yosbida, S. Misawa, and A. Itob, A ppl Phys. Lett., 26, 461 (1975)
56. S. Winsztal, B. Wauk, H. Majewska-Minor, and T. Niemyski, Thin Solid Films, 32,
251(1976)
57. H. U. Baier, and W. Mdncb, J. Appl. Phys.,
6 8
, 586 (1990)
58. S. A. Nikisbin, N. N. Faleev, V. G. Antipov, S. Francoeur, L. Grave de Peralta, G.
A. Seryogin, M. Holtz, T. I. Prokofyeva, S. N. G. Cbu, A. S. Zubrilov, V. A.
Elyukbin, 1. P. Nikitina, A. Nikolaev, Y. Melnik, V. Dmitriev, and H. Temkin, Mat.
Res. Soc. Symp. Proc., 595 , W8.3 (1999)
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59.
M. Albrecht, I. P. Nikitina, A. E. Nikolaev, Y. V. Melnik, V. A. Dmitriev, and H. P.
Strunk, Phys. Stat. Sol, A, 176, 453 (1999)
60.
A. Nikolaev, I. Nikitina, A. Zubrilov, M. Mynbaeva, Y. Melnik, and V. Dmitriev,
Mat. Res. Soc. Symp. Proc., 595, W6.5 (2000)
61.
A. M. Tsaregorodtsev, and A. N. Efimov, Tech. Phys. Lett., 22, 130 (1996)
62.
Y. Melnik, A. Nikolaev, S. Stepanov, I. Kikitina, K. Vassilevski, A. Ankudinov, Y.
Musikhin, and V. A. Dmitriev, Mater. Sci. Forum, 264-268, 1121 (1998)
63.
T. Kunisato, Y. Nomura, H. Ohbo, T. Kano, N. Hayashi, M. Hata, T. Yamaguchi,
M. Shono, M. Sawada, and A. Ibaraki, Phys. Stat. So l C, 0(7), 2063 (2003)
64.
T. Shibata, K. Asai, S. Sumiya, M. Mouri, M. Tanaka, O. Oda, H. Katsukawa, H.
Miyake, and K. Hiramatsu, Phys. Stat. S o l C, 0(7), 2023 (2003)
65.
Y. Wu, A. Hanlon, J. F. Kaeding, R. Sharma, P. T. Fini, S. Nakamura, and J. S.
Speck,
6 6
.
67.
Phys. Lett., 84, 912 (2004)
J. Pastmak, and L. Roscovkova, Phys. Stat. S o l, 7, 331 (1964)
C. O. Dugger, Mat. Res. Bull, 9, 331 (1974)
.
M.
Bockowski, Cryst. Res. Technol, 36, 771 (2001)
69.
G.
A. Slack, and T. F. McNelly, J. Cryst. Growth, 34, 263(1976)
70.
G.
A. Slack, and T. F. McNelly, J. Cryst. Growth, 42, 560 (1977)
71.
R. Schlesser, R. Dalmau, R. Yakimova, and Z. Sitar, Mat. Res. Soc. Symp. Proc.,
6 8
693,19.4 (2002)
72.
R. Schlesser, and Z. Sitar, J. Cryst. Growth, 234, 349 (2002)
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73. L. J. Schowalter, Y. Shusterman, R. Wang, I. Bhat, G. Arunmozhi, and G.A. Slack,
M RS Internet J. Nitride Semicond. Res., 4S1, G3.76 (1999)
74. L. J. Schowalter, Y. Shusterman, R. Wang, I. Bhat, G. Arunmozhi, and G.A. Slack,
Appl. Phys. Lett., 76, 985 (2000)
75. L. J. Schowalter, J. C. Rojo, G. A. Slack, Y. Shusterman, R. Wang, I. Bhat, and G.
Arunmozhi, J. Cryst. Growth, 211, 78 (2000)
76. L. J. Schowalter, J. C. Rojo, N. Yakolai, Y. Shusterman, K. Dovidenko, R. Wang,
and G. A. Slack, Internet J. Nitride Semicond. Res., 581, W6.7 (2000)
77. J. C. Rojo, G. A. Slack, K. Morgan, L. J. Schowalter, and M. Dudley, Mat. Res.
Soc. Symp. Proc., 639, Gl. lO (2001)
78.
L. Liu, and J. H. Edgar, J. Cryst. Growth, 220, 243 (2000)
79.
L. Liu, B. Liu, Y. Shi, and J. H. Edgar, M RS Internet J. Nitride Semicond. Res., 6 , 7
(2001)
80. B. M. Epelbaum, M. Bickermann, and A. Winnacker, private communication
81.
M. Tanaka, S. Nakahata, K. Sogabe, H. Nakata, and M. Tobioka, Jpn. J. Appl.
Phys., 36, L1062 (1997)
82.
B. M. Epelbaum, C. Seitz, A. Magerl, M. Bickermann, and A. Winnacker, in press
83.
S. Yu. Karpov, D.V. Zimina, Yu. N. Makarov, E. N. Mokhov, A. D. Roenkov, M.
G. Ramm, and Yu. A. Vodakov, Phys. Stat. Sol. A, 176, 435 (1999)
84.
A. S. Segal, S. Yu. Karpov, Yu. N. Makarov, E. N. Mokhov, A. D. Roenkov, M. G.
Ramm, and Yu. A. Vodakov, J. Cryst. Growth, 211,
6 8
(2000)
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85.
8 6
.
L. Liu, and J. H. Edgar, J. Electrochem. Soc., 149, G12 (2002)
M. V. Bogdanov, S. E. Demina, S. Yu. Karpov, A. V. Kulik, M. S. Ramm, and Yu.
N. Makarov, Cryst. Res. Technol, 38, 237 (2003)
87.
8 8
.
P. M. Dryburgh, J. Cryst. Growth, 125, 65 (1992)
M. V. Averyanova, S. Yu. Karpov, Yu. N. Makarov, I. N. Przhevalskii, M. S.
Ramm, and R. A. Talalaev, M RS Internet J. Nitride Semicond. Res., 1, 31 (1996)
89.
M. V. Averyanova, I. N. Przhevalsky, S. Yu. Karpov, Yu. N. Makarov, M. S.
Ramm, and R. A. Talalaev, Mater. Sci. Eng., B, 43, 167 (1997)
90.
S. Yu. Karpov, Yu. N. Makarov, and M. S. Ramm, M RS Internet J. Nitride
Semicond. Res., 2, 45 (1997)
91.
S. Yu. Karpov, A. V. Kulik, M. S. Ramm, E. N. Mokhov, A. D. Roenkov, Yu. A.
Vodakov, and Yu. N. Makarov, Mater. Sci. Forum, 353-356, 779 (2001)
92.
E. N. Mokhov, A. D. Roenkov, Yu. A. Vodakov, S. Yu. Karpov, M. S. Ramm, A.
S. Segal, Yu. A. Makarov, and H. Helava, Mater. Sci. Forum, 433-436, 979 (2003)
93.
V. Noveski, R. Schlesser, S. Mahajan, S. Beaudoin, and Z. Sitar, J. Cryst. Growth,
264, 369 (2004)
94.
M. Bickermann, B. M. Epelbaum, and A. Winnacker, private commimication
95.
M. Bickermann, B. M. Epelbaum, and A. Winnacker, Phys. Stat. S o l, C, (0)7, 1993
(2003)
96.
C. M. Balkas, Z. Sitar, T. Zheleva, L. Bergman, R. Nemanich, and R. F. Davis, J.
Cryst. Growth, 179, 363 (1997)
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97.
W. L. Samey, L. Salamanca-Riba, T. Hossain, P. Zhou, H. N. Jayatirtha, H. H.
Kang, R. D. Vispute, M. Spencer, and K. A. Jones, M RS Internet J. Nitride
Semicond. Res., 5S1, W5.5 (2000)
98.
B. Liu, Y. Shi, L. Liu, and J. H. Edgar, Mat. Res. Soc. Symp. Proc., 639, G3.13
(2000)
99.
Y. Shi, B. Liu, L. Liu, J. H. Edgar, E. A. Payzant, J. M. Hayes, and M. Kuball, MRS
Internet J. Nitride Semicond. Res.,
6
, 5 (2001)
100. L. Liu, B. Liu, Y. Shi, and J. H. Edgar, M RS Internet J. Nitride Semicond. Res.,
6
,7
(2001)
101. Y. Shi, Z. Y. Xie, L. H. Liu, B. Liu, J. H. Edgar, and M. Kuball, J. Cryst. Growth,
233, 177 (2001)
102. Y. Shi, B. Liu, L. Liu, J. H. Edgar, H. M. Meyer III, E. A. Payzant, L. R. Walker,
N. D. Evans, J. G. Swadener, J. Chaudhuri, and Joy Chaudhuri, Phys. Stat. Sol, A,
188, 757(2001)
103. J. H. Edgar, L. Liu, B. Liu, D. Zhuang, J. Chaudhuri, M. Kuball, and S.
Rajasingam, J. Cryst. Growth, 246, 187 (2002)
104. B. M. Epelbaum, M. Bickermann, and A. Winnacker, Mater. Sci. Forum, 433-436,
983(2003)
105. L. Liu, and J. H. Edgar, Mater. Sci. Eng., R, 37, 61 (2002)
106. V. Noveski, R. Schlesser, B. Raghothamachar, M. Dudley, S. Mahajan, S.
Beaudoin, and Z. Sitar, J. Cryst. Growth, 264, 369 (2004)
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107. V. Noveski, R. Schlesser, J. A. Freitas Jr, S. Mahajan, S. Beaudoin, and Z. Sitar,
Mat. Res. Soc. Symp. Proc., 798, Y2.8 (2004)
108. S. Yu. Karpov, A. V. Kulik, A. S. Segal, M. S. Ramm, and Yu. N. Markarov, Phys.
Stat. Sol, A, 188, 763 (2001)
109. http://www.webelements.com/
110. B. Liu, D. Zhuang, and J. H. Edgar, Vacuum Science and Technology: Nitrides as
Seen by the Technology, (Eds.) T. Paskova and B. Monemar, 43, Research Signpost,
Trivandrum, India (2002)
111. R. Dalmau, B. Raghothamachar, M. Dudley, R. Schlesser, and Z. Sitar, Mat. Res.
Soc. Symp. Proc., 798, Y2.9 (2004)
112. J. H. Edgar, D. Zhuang, and Z. Gu, unpublished
113. D. Zhuang, J. H. Edgar, B. Liu, H. E. Huey, H. X. Jiang, J. Y. Lin, M. Kuball, F.
Mogal, J. Chaudhuri, and Z. Rek, J. Cryst. Growth, 262, 168 (2004)
114. D. Zhuang, J. H. Edgar, B. Strojek, J. Chaudhuri, and Z. Rek, J. Cryst. Growth,
262, 89(2004)
115. M. Dudley, Encyclopedia o f Advances Materials, (Eds.) D. Bloor, R. Brook, M.
Flemings, S. Mahajan, and R. Cahn, 4, 2950, Pergamon Press, New York (1994)
116. H. B. Bebb, and E. W. Williams, Semiconductors and Semimetals, (Eds.) R. K.
Willardson, and A. C. Beer, 8,181, Academic Press, New York (1972)
117. 1. R. Lewis, and H. G. M. Edwards, Handbook o f Raman Spectroscopy : From the
Research Laboratory to the Process Line, Marcel Dekker Inc. (2001)
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bulk AIN Crystal Growth by Direct Heating of the Source Using
Microwaves
D. Zhuang*, J. H. Edgar**, B. Liu*, H. B. Huey^, H. X. Jiang^, J. Y. Lin^,
M. Kuball'*, F. Mogal^, J. Chaudhuri^, and Z. Rek^
*Department o f Chemical Engineering, Kansas State University, Manhattan, KS 66506,
USA
^Mieramies, Ine., 945 Berryessa Rd. #5, San Jose, CA 95133, USA
^Department o f Physics, Kansas State University, Manhattan, KS 66506, USA
'*H. H. Wills Physics Laboratory, University o f Bristol, Bristol BS 8 ITL, UK
^Department o f Mechanical Engineering, Wichita State University, Wichita, KS 67260,
USA
^Stanford Linear Acelerator Center, 2575 Sand Hill Road, Mail Stop 69, Menlo Park, CA
94025, USA
Abstract
AIN single crystal platelets up to 2 x 3 mm^ and needles 1 mm in diameter and 3
mm in length were successfully grown by directly heating the source materials with
microwaves. The process temperature was over 2000 °C and the pressure was kept
constant at 910 ton*. The growth rate was typically 300 pm/hr in the c-direction. An
Corresponding Author Tel.: 785-5324320; fax: 785-5327372; email: edgarih@ksu.edu: mail address:
Room 105 Durland H all, Manhattan, KS 66506-5102
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
emission around 5.5 eV was observed in the photo luminescence (PL) spectrum, probably
caused by magnesium impurity. The dislocation density was low, bxlO ^cw "^, as
determined by both synchrotron white beam X-ray topography (SWBXT) and etching in
molten potassium hydroxide-sodium hydroxide eutectic alloy. Etching produced
hexagonal pits and hexagonal hillocks on the Al-polar and N-polar surfaces, respectively.
Raman spectra. X-ray topography, and etch pit densities demonstrate that the crystals
have good structural quality.
FACS: 61.72, 78.30, 78.55, 81.10
Keywords: A l. Characterization, A l. Etching, A l. X-ray topography, A2. Single Crystal
Growth, B l. Nitrides, B2. Semiconducting Aluminum Compounds
1. Introduction
Significant progress has been made with group III nitride semiconductor devices
in the past decade. Blue and green LEDs and blue laser diodes fabricated using
heteroepitaxial group III nitride films are commercially available, and electronic devices,
such as power transistors, have been successfully demonstrated [I, 2]. However, high
dislocation densities caused by the lattice constant mismatch between group III nitrides
and 6 H-SiC and/or sapphire substrates remains an obstacle to the further development of
these devices, both in aspects o f device lifetime and performance. Since AIN has a small
lattice constant mismatch (-2.4% in o-axis [3]) and an almost identical thermal expansion
coefficient with GaN, bulk AIN single crystalline substrates for epitaxial GaN film
growth should reduce its defects density immensely, triggering an expeditious increase in
group III nitride device quality. In addition, the high thermal conductivity (340
W I m - K ) o f AIN makes it suitable for high power applications, where the heat
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
generated by devices must be efficiently dissipated [4]. AIN is also a perfect material for
high power microwave devices and UV detectors, due to its high resistivity (10’~10*^
Q -c m ) and very wide energy bandgap (6.2 eV).
Though intensive research efforts have been devoted to AIN single crystal growth
via sublimation-recondensation method, first validated by Slack and McNelly in the
1970s [5], no bulk AIN substrates with high crystalline quality have yet been
commercialized. According to model calculations [6 ], the main challenge is that the
sublimation technique requires a process temperature o f more than 2100 °C in order to
obtain a viable fast growth rate (> 100 pm/hr). At such an extreme temperatures, furnace
fixtures can react with Al or N i, sublime or even melt. Keeping furnace fixtures at low
temperature should reduce production cost o f AIN single crystal growth by allowing the
crucibles to last longer.
The temperature o f the furnace fixtures can be minimized by directly heating the
AIN source using microwaves. The advantages o f microwave heating to sinter ceramic
materials have been recognized for many years [7]. These include efficient use o f energy,
rapid and more uniform heating. Microwave heating occurs in the body o f the ceramic
materials with energy being lost from the surface by radiation, conduction, and
convection. The temperature gradient is thus opposite to that o f traditional heating
methods, in which the surface o f a body is heated and the interior is heated by
conduction. The mechanism o f microwave heating o f AIN varies with temperature. At
room temperature, the free electron concentration in AIN is very low hence its electrical
conductivity is very low (on the order o f 10’’* Q"'
-cw"'
). At room temperature,
microwave heating occurs by reorientation o f the dipoles in the AIN, which is
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subsequently randomized as heat. As the temperature and therefore the carrier
concentration o f the AIN increases [8 ], the heating mechanism changes to more efficient
oscillation o f conduction electrons. Heating and sintering o f AIN ceramics by
microwaves has been demonstrated [9].
Oxygen contamination has a vital role when evaluating AIN crystal quality since a
high oxygen concentration decreases the thermal conductivity o f AIN [10, 11] and can
cause crystal defects. Slack et al. [5] speculated that the peaks at 2.86 eV in AIN PL
spectra was due to the combination o f oxygen impurities and nitrogen vacancies. Harris
and Youngman [10] reported a continuous shift o f an oxygen-related PL peak to a lower
energy level (4.0 to 3.3 eV) as the oxygen concentration increases to a critical
concentration o f about 0.75 at.%. Due to the limitation o f material quality, few research
groups reported near-band-edge emissions, a characteristic o f low impurity level and low
defect density, in bulk AIN luminescence studies so far. Kuokstis et al. [12] reported their
PL spectra o f bulk AIN crystals, showing free excitation emission at 5.94 eV and 5.95 eV
for a- and c-planes, respectively. One AIN sample grown by Slack et al. [5, 13] had a
peak position at 4.03 eV and a measured room temperature thermal conductivity o f 275
W ! m -K . Nevertheless, some groups have successfully demonstrated near-band-edge
emission in heteroepitaxial AIN films. In 1998, Tang et al. [14] first reported
cathodoluminescence (CL) spectra with free excitons or excitons bound to shallow donor
or acceptor impurities at 6.11 eV, from an undoped AIN sample grown on sapphire
substrate. More recently, bandedge emission lines at 5.96 eV and 6.015 eV have been
observed in AIN epilayers by deep UV (down to 196 nm) PL spectroscopy measurement
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[15, 16]. Mg doped AIN epilayers have also been studied and an activation energy of
about 0.51 eV for Mg acceptors in AIN has been determined [17].
In this communication, we present the development o f AIN single crystals growth
by a microwave-heated furnace employing the sublimation-recondensation method, and
the characterization o f these crystals.
2. Experimental
Fig. 1 shows a cross section view o f the growth reactor used in this experiment,
developed by Micramics, Inc. Two separate 3 kW variable power supplies provide the
microwaves for the sublimation process. The AIN source is contained in a cylindrical
pyrolytic boron nitride (pBN) crucible. This crucible is packed in hexagonal BN powder
contained inside a pBN retort. The pBN containers are surrounded by alumina insulation,
and the entire arrangement is held in a stainless steel chamber. The alumina insulation is
transparent to microwaves, but is thermally insulating {k = 0.39 W I m -K at 1650 °C).
The BN powder acts as additional insulation and helps to absorb the microwave energy.
A computer controlled tuning plate at the bottom o f chamber was adjusted to minimize
microwave reflection by moving up and down. The source temperature and crucible
sidewall temperature were measured by two color infrared pyrometers from the top and
side viewpoint, respectively. The growth time was varied from 12 hrs to 24 hrs and the
chamber pressure was kept constant at 910 torr nitrogen.
Defect levels and crystalline quality were analyzed by PL and Raman spectrum,
respectively. In the PL study, the AIN samples were excited at a wavelength o f 196 nm.
Raman spectra were recorded from the samples using a Renishaw micro-Raman system
with the 488 nm line o f an Ar^ ion laser as excitation sources. The synchrotron
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
topography experiments were carried out at the Synchrotron Radiation Topography
experimental station 2.2 at the Stanford Synchrotron Radiation Laboratory, Stanford, CA.
Transmission topographs were taken using Laue diffraction technique and white beam
radiation. The dislocation density was estimated by etching in molten eutectic
KOH/NaOH and SWBXT. A series o f eutectic etchings were performed to optimize
etching parameters. The final etching parameters for Al-polar crystal were at temperature
o f 380 °C and for three minutes.
3.
Results and Discussions
An optical image o f self-seeded AIN crystals after 10 hour growth at 2080 °C and
910 torr is shown in Fig. 2. These needle and plate-like crystals were colorless and
transparent. The growth rate was approximately 300 pm/hr in the c-direetion.
3.1. Photoluminescence
PL spectra for the AIN crystals along with the spectra from AIN crystals produced
in a tungsten crucible and a pBN erueible (for comparison) are shown in Fig. 3. Sample A
was a polyerystalline AIN crystal grown in a resistively heated tungsten furnace
employing self-seeding mechanism; sample B (single crystal) was produced in a pBN
erueible in the microwave furnace described above; sample C (single crystal) was grown
in pBN crucibles using graphite heating-clement reactor; and sample D was original
sintered AIN (raw source materials). All samples other than sample D were nominally
pure AIN crystals, except for residual impurities from the source and impurities
introduced from the growth ambient. According to reference [15, 16], sample A has a
near-band-edge emission o f 5.95 eV (at room temperature), which is believed to be due to
free excitons or excitons bond to shallow impurities (presumably tungsten). This near-
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
band-edge emission disappeared in PL o f sample B and C. Instead PL peaks at 5.50 eV
are clearly visible. This might indicate the presence o f Mg impurities in the AIN
following the result o f Nam et al. [17] on Mg-doped AIN. In agreement with this
conclusion composition analysis using laser ablation mass spectrometry finds the
presence o f Mg in the crystals. Its results are shown in Table 1. Meanwhile, considering
the fact that the PL spectrum o f raw source materials (spectrum D in Fig. 3) also has a
peak at 5.50 eV, we believe Mg impurity presumably comes from the source materials,
from which the crystals were grown. Further work is needed to obtain better
understanding o f the origins o f the peaks at 4.10 eV, 3.90 eV, 3.70 eV and 2.80 eV.
3.2. Raman Analysis
Generally in Raman spectra o f AIN, the width (FWHM) o f the Ea (high) Raman
peak reflects the material’s crystalline quality, and the Ea (high) phonon frequency is
sensitive to stress [18]. Fig. 4 shows the Raman spectrum from an AIN crystal produced
in the microwave furnace. Strongly visible are the symmetry-allowed Ea (high) and Ai
(TO) Raman modes in the used x(yy)-x scattering geometry. The Ea (high) peak position
is 656.0 cm'* and the FWHM is
6 .6
cm'*. This FWHM value is slightly improved with
respect to other bulk AIN crystals reported in the literature [19], i.e., indicating
improvements in crystalline quality with the new technique. The Ea (high) phonon
frequency is similar to values reported for stress-free AIN [20, 21].
3.3. Eutectic Etch and SWBXT
Etching o f AIN in molten euteetic potassium hydroxide (59 wt%) and sodium
hydroxide (41 wt%) is an effective method for studying dislocation types, densities, and
distributions, just as it is for GaN [22-24]. The shape o f etching features formed in this
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
study is similarly dependent on the erystal polarity due to AlN’s isomorphie strueture and
chemical bonding with GaN (wurtzite).
Since AIN heteroepitaxial film always has Al-polar when growing on Si face SiC
substrate [25], the etching condition for such sample was used to determine etching
parameters o f Al-polar microwave generated crystals. The optimized etching condition
for heteroepitaxial AIN [26] is 2 minutes at 350 °C in eutectic KOH/NaOH. It is expected
that the microwave generated AIN erystal has a better quality than heteroepitaxial
crystals, thus we etched such crystals at slight higher temperature (380 °C) and for longer
time (3 minutes). The results were shown in Fig. 5. The calculated defect density is
6.06 X10^ cm~^. The SWBXT pattern (defect density = 5.99 x 10^ cm~^ shown in Fig. 6 )
o f AIN erystal from the same run, confirmed that defect density o f microwave generated
AIN crystal on Al-polar is on the order o f magnitude o f 6x10^ cm~^, which was
immensely reduced compared to bulk AIN seeded grown on
6
H-SiC substrate (10^
[26, 27]. Details o f euteetic etching o f AIN crystals will be reported elsewhere
[27].
4. Concluding Remarks and Expectation
A new technique for bulk AIN single erystal growth employing microwaves as the
heat source was successfully demonstrated. The crystals have good structural quality, as
indicated by Raman spectra, etch pit density and SWBXT studies. It is believed that this
technique is advanced in preventing damages o f the fiimace fixtures since the source
materials can be heated to elevated temperature (>2000 °C), while the crucible
temperature is kept in a relatively low range. The future work will focus on scaling up
crystal size and seeded growth for effective control o f crystalline orientation.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
Support from the Office o f Naval Research through awards number NOOO14-02-10290 and N00014-99-1-0536 and NSF EPSCoR through award number EPS-9977776 is
greatly appreciated. The work in Bristol was in part supported by EPSRC and Renishaw
pic. This work was partially done at SSRL, Stanford, CA, which is operated by the
Department o f Energy , Office o f Basic Energy Sciences.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reference:
1. T. Mukai, S. Nagahama, N. Iwasa, N. Senoh, T. Yamada, J. Phys. Condes. Matter.,
13, 7089 (2001)
2.
O. Ambacher, J. Phys. D: Appl. Phys.
31, 2653 (1998)
3.
J. H. Edgar, (Editor), Properties
o f Group III Nitrides, Electronic Materials
Information Service (EMIS), London, (1994)
4. M. Kuball, J. M. Hayes, M. J. Uren, T. Martin, J. C. H. Birbeck, R. S. Balmer, B. T.
Hughes, IEEE Electron Dev. Lett., 23, 7 (2002)
5. G. A. Slack, T. F. McNelly, J. Crystal Growth, 34, 263 (1976)
6
. L. Liu, J. H. Edgar, J. Crystal Growth, 220, 243 (2000)
7. J.D.Katz,
8
Rev. Mater. Sci., 22 , 152 (1992)
. R.W. Francis, W.L. Worrell, J. Electrochem. Soc., 123, 430 (1976).
9. H. E. Huey, Q. S. Wang et. a l. Ceramic Engineering & Science Proc., 21(4), 599
(2000)
10.
J .H. Harris, R. A. Youngman, R. G. Teller, J. Mater. Res., 5, 1763 (1990)
11.
L. E. McNeil, M. Grimsditch, R. H. French, J. Am. Ceram. Soc., 76, 1132 (1993)
12.
E. Kuokstis, J. Zhang, Q. Farced, J.W. Yang, G. Simin, M. A. Khan,R. Gaska, M.
Shur, C. Rojo, L. Schowalter,Phys. Lett., 81(15), 2755 (2002)
13.
G. A. Slack, R. A. Tanzilli, R. O. Pohl, and J. W. Vandersande, J. Phys. Chem.
Solids, 48, 641 (1987)
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14.
X. Tang, F. Hossain, K. Wongchotigul, M. G. Spencer, Appl. Phys. Lett., 72(12),
1501(1998)
15.
J. Li, K. B. Nam, M. L. Nakarmi, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett.,
81(18), 3365 (2002)
16.
K. B. Nam, J. Li, M. L. Nakarmi, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett.,
82(11), 1694 (2002)
17.
K. B. Nam, M. L. Nakarmi, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett., 83(5), 878
(2002)
18.
M. Kuball, Surf. Interface A n a l, 31, 987 (2001)
19. M. Kuball, J. M. Hayes, Y. Shi, J. H. Edgar, A p p l Phys. Lett., 77,1958 (2000)
20. J. M. Hayes, M. Kuball, Y. Shi, J. H. Edgar, Jpn. J. A ppl Phys., 39, L710 (2000)
21. V. Y. Davydov, Y. E. Kitaev, 1. N. Goncharuk, A. N. Smirnov, J. Graul, O.
Semchinova, D. Uffmann, M. B. Smirnov, A. P. Mirgorodsky, R. A. Evarestov,
Phys. Rev. B, 58, 12899 (1998)
22.
J. L. Weyher, P. D. Brown, J. L. Rouviere, T. Wosinski, A. R. A. Zauner, 1.
Grzegory, y. Crystal Growth, 2 \^ , 151 (2000)
23.
G. Kamler, J. L. Weyher, I. Grzegory, E. Jezierska, and T. Wosinski, J. Crystal
Growth, 246, 21 (2002)
24.
J. L. Weyher, L. Macht, G. Kamler, J. Borysiuk, and I. Grzegory, Phys. Stat. Sol. C,
0(3), 821 (2003)
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25. D. Zhuang, J. H. Edgar, L. Liu, B. Liu, L. Walker, M RS Int. J. Nitride Semicond.
Res., 7, 4 (2002)
26. Y. Shi, B. Liu, L. Liu, J. H. Edgar, E. A. Payzant, J. M. Hayes, M. Kuball, MRS
Internet J. Nitride Semicond. Res., 6 , 5 (2001)
27. D. Zhuang, J. H. Edgar, B. Strojek, J. Chaudhuri, and Z. Rek, J. Cryst. Growth,
262, 89(2004)
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1 Compositional Analysis, ppm
Sample
B
Crucible
pBN
pBN
Carbon
10920
6561
Magnesium
2772
2985
Silicon
2003
1968
Iron
521
858
.W v W I
to vacuum
1.
2.
3.
4.
Source MW Generator
Seed MW Generator
Top View Point
Top Plate
5.
6.
7.
8.
Chamber
Insulator
Source Wave-guide
Seed Wave-guide
9.
10 .
11 .
12 .
Large Crucible
BN Powder
Small Crucible
AIN Source
Figure 1 Cross section view o f growth reactor
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.«! t n m ^
Figure 2 Self-seeded AIN crystals after 10 hours growth at 2080 °C
and 910 torr
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
'I a r\
T-300K
-\j\
3 -5 0 e \ ^
r g c pW
5 .9 5 e V
A IN b u lk
6
A-Tungsten ^
fumace
/
3
■
*
\
0
B-Microwave
fumace
6
^
5 50 cV
____
3 .9 0 e V
o
3
2 .8 0 e
V
/
1 4 .1 0 e V
^
-
I _
_
^
xri
C
(U
«
0
20
C-Graphite
fumace
ft
J
%
\
- .'
•m
/
t
10
h \
0
JV
A
\
________ ^
D-Sintered AIN,
15-
10
_J
5
0’
/
V
3
4
5
6
E(eV)
Figure 3 PL spectra of AIN crystals produced in Tungsten
furnace (A), microwave fumace (B), graphite furnace (C) and
raw source materials (D)
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.
E^Chigh):
656.Ocm'^
6.6cm ' FWHM
:3
A,(TO):
CO
is*
CO
c
0)
c
£I
609.6cm''
7.2cm ' FWHM
CO
E,(TO):
CO
669.3cm'
6.7cm ' FWHM
E
ex:
500
550
600
650
700
750
800
Raman Shift [cm'^]
Figure 4 Raman spectrum o f AIN crystal produced in microwave
furnace, recorded in x(yy)-x scattering geometry
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
07-Oct-02
WD18.4mm
OOkV
k SOO
Figure 5 SEM image o f Al-polar AIN crystal produced in microwave
fumace after three minutes etch at 380 °C in eutectic KOH/NaOH alloy
00 .- 1]
[ 1- 1 .0]
^ il
6 21
uin
g /
Figure 6 SWBXT in transmission from the sample # 04-08-02, A=dislocation slip line,
g=038 reflection, wavelength X = 0.05049 nm. Dislocation density = 5.99x10^ cm~^
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wet Etching of GaN and AIN: A Review
D. Zhuang and J. H. Edgar*
Department o f Chemical Engineering, Kansas State University, Manhattan, KS 66506,
USA
Abstract
The wet etching o f GaN and AIN is reviewed including conventional etching in
aqueous solutions, electrochemical etching in electrolytes and defect selective chemical
etching in molten salts. The mechanism o f each etching process is discussed. Etching
parameters leading to highly anisotropic, dopant-type/bandgap selective, defect-selective,
as well as smooth surfaces are discussed. The controversy concerning the origin o f etch
pits on Ga-polar GaN is discussed and interpreted. The applications o f wet etching
techniques to characterize group III nitride crystal polarity and defects are reviewed.
Additional applications o f wet etching for device fabrication, such as producing
crystallographic etch profiles, are also reviewed.
PACS: 61.72, 81.65
Keywords: A l. Defect, A l. Etching, B l. Nitrides
1. Introduction
Single crystalline group III nitride semiconductors have attracted huge research
interest during last two decades due to their unique properties and potential applications
Corresponding author Tel.: 785-532-4320; fax: 785-532-7372; email: edgarih@ksu.edu: mail address:
Room 105, Durland H all, Kansas State Univeristy, Manhattan, KS 66506-5102 U S A
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in short-wavelength light source/detector and high temperature/frequency devices.
Taking advantage o f the direct wide bandgap o f GaN (3.44 eV), blue and green light
emitting diodes (LEDs) have been commercialized [1]. The ability o f GaN to form solid
solutions with AIN and InN, making the bandgap engineering possible, is essential for
defining the emission wavelength o f the LEDs. Ultra violet (UV), high brightness, and
long-life LEDs with potential to replace incandescent bulbs are under development. The
high thermal conductivities of GaN (210 W I m - K ) and AIN (340 WI m - K ) [2] make
them suitable for high power applications, where the heat generated by devices must be
efficiently dissipated. Furthermore, one can utilize the piezoelectric properties o f GaN
and AIN to fabricate high frequency devices, such as surface acoustic wave (SAW)
devices. To date, the vast majority o f group III nitride semiconductors have been
deposited on foreign substrates, such as 6H-SiC and sapphire, due to the lack o f bulk
group III nitride single crystals. Thus, most wet etching studies have been made on such
thin heteroepitaxial films. The large lattice mismatch between the epilayer and substrate
results in a high dislocation density in heteroepitaxial GaN, typically in the range of
10’ - 1 0 " cm-^ [3,4].
Many studies o f group III nitride wet etching for material characterization and
device fabrication have been conducted. Different etching mechanisms and various
etchants, such as aqueous mineral acid and base solutions and molten salts have been
investigated. Since the optimal etching parameters are highly dependent on material
quality and properties, a systematic review is imperative in order to help researchers
select/optimize an appropriate etching process for specific etching purposes.
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wet etches may remove material uniformly or non-uniformly with slower or
faster rates at defects. Defect-selective etching produces etch pits or hillocks on a
semiconductor surface due to the inhomogeneous nature o f defects (incomplete chemical
bonds or higher impurity concentrations) compared with the crystal matrix. When a new
etchant or a new etching system has been carefully calibrated with other more
sophisticated methods, e.g. transmission electron microscopy (TEM), atomic force
microscopy (AFM) and/or X-ray topography (XRT), one can estimate the defect density
in single crystalline materials by measuring the etch pit density (EPD). TEM and AFM
are established methods for determining the dislocation density in single crystals, but
TEM requires arduous sample preparation and AFM requires a relative large and smooth
sample surface. Moreover, for materials with relatively low dislocation densities, locating
a dislocation using TEM or AFM is difficult. When the dislocation density is
around 10'^ cm~^, for example, theoretically there is only one dislocation presenting in a
100x100 pm^ area. Thanks to its distinguished advantages, such as low cost, simple
experimental procedure and no required sample geometry, wet etching is widely used for
the defect evaluation purpose, once suitable calibration is established.
In addition to evaluating the types and densities o f defects, the crystal polarity
may also be evaluated by the outcome o f wet etching. The group III nitride
semiconductors have the hexagonal wmrtzite structure consisting o f alternating layers of
III-N pairs, stacked along the [0001] direction in an ABABAB sequence. Thus the basal
plane (i.e. the (0001) plane) can be either N- or group III element polar. The polarity of
(0001) planes is conventionally defined as follows. As the crystal surface is approached
from the bulk along the c-direction, if the long bond goes from the nitrogen atom to the
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
group III atom, the crystal is nitrogen polar. If instead the long bond goes from the group
111 atom toward the nitrogen atom, the crystal is group 111 polar. The polarity o f the group
111 nitrides has significant effects on its surface and bulk properties [5-12], electrical and
optical properties [13-17], oxidation rate, and impurity incorporation rate during epitaxy.
Compared to other polarity identification methods, such as TEM and reflection high
energy electron diffraction (RHEED), wet etching has advantages o f lower process cost
and simpler procedures.
Lastly, wet etching is widely used in device fabrication processes, as an important
complement to dry etching techniques. Generally, dry etching processes - reactive ion
etching (RIE) for example - can be highly anisotropic, an ideal characteristic for
producing vertical profiles. However, due to the strong physical component in dry
etching, it has low etch selectivity between materials and can cause severe subsurface
damage by ion bombardment [18, 19]. In contrast, wet etching, produces negligible
damage, can be highly selective, is relatively inexpensive, and can be done with simple
equipment. In addition, wet etching eliminates the risks o f hydrogen incorporation from
hydrogen containing chemistries, and along with it, the associated changes in the
conductivity o f group 111 nitrides layers [20].
By distinguishing the etching mechanisms, wet etching o f semiconductor can be
generally divided into two categories, namely: electrochemical etching including anodic
etching, electroless etching, and photoelectrochemical (PEC) etching, and chemical
etching including conventional etching in aqueous etchants and defect-selective etching
in molten salts.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In anodic etching, the semiconductor and an inert electrode are attached to the
positive and negative terminals o f a direct voltage source, respectively. Both electrodes
are put inside an electrolyte, e.g. aqueous potassium hydroxide (KOH). The
semiconductor is oxidized by removal o f bonding electrons (injecting holes) from the
surface bonds via an external voltage source. The resulting oxides subsequently dissolve
into the electrolyte.
For electroless etching, neither an external voltage nor electrical contact to the
samples is required. The oxidation o f the semiconductor is driven by the potential o f an
oxidizing agent in the electrolyte, which depletes valenee-band electrons in the
semiconductor and thus supplying holes. Electroless etching is thermodynamically
possible only if the redox potential is higher than the potential o f the semiconductor solid
in equilibrium with its ions in the solution. The etch rate o f electroless etching depends
on the location o f the energy band o f the semiconductor in relation to the energy levels o f
the redox couple in the solution. Besides etching in the dark, some oxidizing agents are
capable o f etching semiconductors under above bandgap illuminations. The mechanism
o f such photo-assisted electroless etching is schematically shown in Fig. 1. Electron-hole
pairs are generated by photons from an illumination source with energy equal to or
greater than the band gap energy o f the semiconductor. The photogenerated holes assist
in the oxidation of the semiconductor surface and the excess electrons are consumed by
the reduction o f oxidizing agent in the electrol)de. Increasing absorption o f incident
optical radiation with energy greater than bandgap energy increases the supply o f holes at
the surface, thereby enhancing the etch rates. A competing process is the recombination
o f electrons and holes.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
If the excess electrons in photo-assisted electroless etching are consumed by the
reduction reaction taking place on the counter electrode instead o f reducing the oxidizing
agents, the etching process is called photoelectrochemieal (PEC) etching. Note the
electrical contact and counter electrode(s) are required in PEC etching, but are not
necessary for photo-assisted electroless etching. Thus, for simplicity and comparison
reasons, the photo-assisted electroless etching is also called contactless etching. Both
PEC etching and contactless etching belong to photo-assisted etching since they all
utilize above bandgap UV illumination to generate eleetron-hole pairs, which are
indispensable for etching to occur.
Furthermore, there are some combinations o f the aforementioned etching
processes. For example, when the anodic etching o f «-type semiconductor occurs under
illuminations, it is called photo-assisted anodic etching.
Finally, chemical etching has a completely different etching mechanism compared
to the eleetroehemieal etching processes, in that no free carriers or electrolyte are
involved, thus the etching process is not affected by an external potential. The reactive
molecules from the etchant break the bonds at the semiconductor surface and form oxides
which are subsequently dissolved in the etchant. Defect-selective etching in molten salts
and conventional etching in aqueous solution have been widely used for eharaeterizing
the defect density/distribution in single crystalline materials and for polarity identification
and semiconductor patterning, respectively.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Generally speaking, GaN crystals are usually etched by photo-assisted etching
and chemical etching and AIN crystals are etched by chemical etching. In some cases, ptype GaN is etched by anodic etching.
2. Conventional Etching
2.1. Conventional Etching of Epitaxial GaN
The group III nitrides are notable for their excellent chemical stability, which
consequently makes them invulnerable to wet etching. Some early reports showed that
GaN can be etched in sodium hydroxide (NaOH) solution. However, etching ceased upon
the formation o f an insoluble coating o f presumably gallium hydroxide (Ga(0H)3) [21,
22]. The coating had to be removed by continual jet action for further etching to take
place [22]. In ref. [23-25], various aqueous acid and base solutions were tested for
etching o f GaN and AIN at temperatures up to 75 °C (see Table 1). No wet etch solutions
appreciably etched GaN films grown by metal organic molecular beam epitaxy
(MOMBE). In contrast, Carosella et al. [26] reported AZ-400K photo resist (active
ingredient being KOH) etched MBE-grown GaN at 80 °C, but did not etch GaN grown by
metal organic chemical vapor deposition (MOCVD) or hydride vapor phase epitaxy
(HVPE) [26]. The discrepancy between [23-25] and [26] could have been caused by the
polarity variation from sample to sample.
The difference in the etching characteristics o f Ga- and N-polar materials has
been examined in a number o f studies. Palacios et al. [27] reported up to 500 A/min etch
rate for etching o f a MBE grown nitrogen face GaN sample in KOH solution in the
temperature range o f 26-80 °C and obtained triangular shape pyramids limited by
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 1 2 l| planes, as shown in Fig. 2. That no etching o f Ga-polar crystals occurred
demonstrates its higher chemical stability [27]. A clear SEM image depicting the surface
morphologies o f a GaN sample with alternative polarities after etching in aqueous KOH
is shown in Fig. 3 [28-30]. After etching, the N-polar area was completely covered by
hexagonal hillocks with facets o f G O lU planes but the morphology o f the Ga-polar
areas remained unchanged. Similar polarity dependent etching was also reported in [3135]. The N-face GaN films etched quickly in hot phosphoric acid, resulting in either
complete film removal or a drastic change in the surface morphology. In contrast, the
etchant attacks only defect sites in Ga-polar films, producing hexagonal etch pits but
leaving the defect-free GaN intact and the morphology unchanged. Similarly, Shimizu et
al. [36] observed severe morphology degradation for presumably N-polar MBE grown
GaN etched in phosphoric acid at temperatures higher than 220 °C. All o f the
aforementioned experiments suggested that GaN epitaxial films having different polarity
show different etching behavior in aqueous etchants.
Heilman [37] summarized the various polarity identification techniques and
concluded that the Ga-face o f GaN is chemically more stable than the N-face. Thus, one
can distinguish crystal polarities o f GaN by simply observing the surface morphology
after etching. The mechanism o f such polarity selective etching was interpreted by Li et
al. [38, 39], who employed X-ray photoelectron spectroscopy (XPS) to study the change
o f surface chemistries before and after etching in aqueous KOH solutions for both Gapolar and N-polar GaN. They concluded the different etching characteristics o f Ga-polar
and N-polar crystals are due to the different states o f surface bonding and are only
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dependent on polarities, not on the surface morphology or growth methods. The
mechanism o f etching o f N-polar GaN surface is schematically shown in Fig. 4 [39]. The
hydroxide ions {OH^ ) are first adsorbed to the sample surface (Fig. 4b) and subsequently
react with Ga atoms following the reaction:
2GaN + S/Zj O - 1 2 ^ Ga^ O, + 2N H ,
(l)
KOH works as a catalyst and is also a solvent for the resulting Ga2 0 3 (Fig. 4d).
As the stages o f (a) to (d) in Fig. 4 repeat, the N-polar GaN can be etched. Note it does
not matter which atoms form the surface termination layer. If the surface is Gaterminated, the etching can be initialized by stage (c). In contrast, the inertness o f Gapolar GaN is ascribed to the repulsion between
and three occupied dangling
bonds o f nitrogen, which prevent the hydroxide ions from attacking the Ga atoms.
The same polarity dependent etching behaviors were observed for bulk GaN
crystals as well [8 ]. That N-face GaN can be etched in KOH solution at low temperature
has been utilized to obtain flat surfaces o f N-face bulk GaN by chemical mechanical
polishing (CMP) [7, 40].
Unlike N-polar, Ga-polar GaN epitaxial films (OGGI) are selectively attacked and
show hexagonal etch pits after etching in hot phosphoric acid [41-44, 47]. Morimoto [41]
and Shintani et al. [42] reported a similar etch rate, about 1 pm/min, for MOCVD grown
GaN etched in phosphoric acid at a temperature around 2GG °C. Etching produced
hexagonal etch pits reflecting the crystal symmetry o f GaN [42]. Kim et al. [43] reported
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the same shape pits after etching in phosphoric acid at a temperature around 215 °C. No
significant etching was observed at 160 °C and below.
Ideally, one would expect that a specific etch pit shape and size would be
associated with specific defect. However, the origin o f the hexagonal etch pits is
disputed. Hong et al. [44] examined the same sample from [43] and argued these
hexagonal etch pits are formed on the nanopipes, i.e. open core screw dislocations [45],
but not on the threading dislocations. In contrast, dislocation related hexagonal etch pits
were reported in [32, 35, 46] after etching Ga-polar GaN in phosphoric acid. Further
details o f this issue will be presented in subsection o f 5.1. Another difficulty in evaluating
defect density by EPD is that under certain etching conditions, some terraces may
develop on the inside wall o f the etch pits, as shown in Fig. 5 [47], causing the pits to
merge, thus leading to an underestimation o f the EPD. Introducing metallic ions in
phosphoric acids may help to obtain etch pits with smooth side wall and prevent merging
of etch pits [47].
Specific non c planes can be exposed by crystallographic etching, a technique
capable o f producing crystal planes with different orientation [48, 49]. The etching was
done in two steps. To expose planes other than (0001) plane, first, an etch depth was
produced by dry etching, PEC etching or cleaving. Next, samples were immersing into
phosphoric acid, molten KOH or KOH dissolved in ethylene glycol to test their ability to
produce crystallographie etehing. Three planes inclined to the c-plane ( H
O ll l jk, H
jlO
| l 012>,
and ( l O n j ) and one plane perpendicular to the c-plane ( j l O l o j ) were variously and in
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
some cases, simultaneously produced in phosphoric acid, molten KOH, or hot 10%-50%
KOH in ethylene glycol, as shown in Fig.
6
[48]. O f the etches studied, the latter had the
most practical use, e.g. InGaN/GaN lasers were successfully fabricated using such
crystallographic etching process [50-52].
Besides polarity identification, defect revelation and material patterning, other
applications of conventional etching o f epitaxial GaN include improving the optical
qualities [27, 53] and surface cleaning, which subsequently alters the electrical properties
[43, 54-55]. Palacios et al. [27] found that the photoluminescence (PL) band o f N-polar
GaN shifted to higher energy level and the intensity was significantly enhanced when the
samples were etched in an aqueous KOH solution. This improvement o f optical quality
did not degrade after long periods o f exposure to the ambient atmosphere, eliminating the
possibility that such improvement was caused by removing an oxide layer. The energy
shift and intensity enhancement were tentatively ascribed to the relaxation o f tensile
strain during the formation o f the etching pyramids. These observations contrast to PL
measurement o f GaN films after dry etching, where PL signal was degraded [27]. The PL
spectra o f GaN after etching in
H 3P O 4
were studied by Reshchikov et al. [53] as well.
That the peaks at 3.30 eV and 3.42 eV disappeared after etching enabled the authors to
correlate these peaks to defects in the surface layer o f GaN. Surface defects removal by
conventional etching was reported by Kim et al. [43], who found the etch rates on the
defects were higher than that on the defect-free area. Thus, surface defects, e.g. nitrogen
vacancies, making ohmic contacts difficult, can be preferentially removed. An enhanced
I-V characteristic o f metal contacts on /?-GaN after etching in
H 3P O 4
is shown in Fig. 7
[43]. The metal contact to the etched surface showed better ohmic characteristics.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tripathy et al. [54, 55] reported similar reduction o f metal contact resistivity on j?-GaN
after treating the GaN samples in KOH solution.
2.2. Conventional Etching of Epitaxial AIN
Conventional etching o f epitaxial AIN grown by MOMBE in mineral acid & base
solutions were investigated in [23-25]. Only KOH and NaOH containing solutions etched
AIN at temperatures below 80 °C. However, low quality (amorphous/polycrystalline)
AIN films deposited on Si by plasma enhanced chemical vapor deposition were etched in
hot phosphoric acid at 60 °C [56]. Other etchants, such as hot HF/H 2 O [57-59], HF/HNO 3
[60] and NaOH [61] can etch sputtered or reactively evaporated amorphous AIN. Vartuli
et al. [62] correlated the etch rates o f reactively sputtered AIN samples etched in
AZ400K photoresist developer as a function o f annealing temperature. As shown in Fig.
8
[62], the etch rate dropped significantly as the sample annealing temperature increased,
indicating a higher crystal quality after a high temperature atmeal. The calculated
activation energy for etching was about 2.0 ±0.5 kcal m ol'^ which is in the diffusionlimited range (1-6 kcal mol’*) [63-66]. The characteristics o f diffusion controlled etching
implied a poor crystal quality; etching is so rapid that the solution became depleted of
reactants near the sample surface [62]. Mileham et al. [67] reported a similar relationship
between the etch rate and crystal quality by examining AIN crystals grown by MOMBE.
The activation energy was determined as 15.510.4 kcal mof*. The etch rate was not
sensitive to agitation, but did depend on the etchant concentration. Thus, etching was
reaction-limited, indicating that MOMBE grown AIN has a higher crystalline quality than
AIN prepared by reactive sputtering.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Differences in the etch behavior due to the crystal polarity were reported by
Jasinski et al. [6 8 ] for an AIN epitaxial layer grown on sapphire substrate. The etch rate
o f the Al-polar crystal was much slower than that o f the N-polarity crystal in a 50 wt%
aqueous KOH solutions at 80 °C [6 8 ].
Taking advantage o f the high etching selectivity o f AIN over GaN and AI2 O 3 ,
Mileham et a l [67, 69] successfully developed a wet etching process for the fabrication
o f a GaN/InGaN/AlN micro-disk laser structure. The AIN layer was substantially
undercut after etching in AZ400K for 15-30 minutes at temperature o f 85 °C. Ide et al.
[70] selectively etched AIN over GaN in hot phosphoric acid and subsequently fabricated
a MISFET device.
2.3. Conventional Etching of Bulk AIN
Conventional etching o f bulk AIN single crystals grown by sublimationrecondensation technique has also been investigated [71, 72]. Schowaiter et a l [71]
reported different etching behaviors for the two sides o f a vicinal c-face AIN substrate
(cut 20° off axis). A phosphoric/sulfuric acid mixture produced a rough surface
morphology on the nitrogen face, but did not etch the Al face. A 1:2.5 KOH:water (by
weight) solution produced same etching results, though the etch rate was slower than that
in phosphoric acid. After etching, AFM revealed that the nitrogen polarity has a surface
consisting o f pyramids 0.5 pm and 1.5 pm high. Zhuang et al. [72] found that for bulk
AIN, the nitrogen polarity (OOOl) basal plane initially etched rapidly, while the aluminum
polarity basal plane and prismatic
f ~
1100
\
planes were not etched, as shown in Fig. 9
V
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[72], The etch rate o f the nitrogen polarity basal plane eventually decreased to zero, as
the surface became completely covered with hexagonal hillocks, presumably bounded by
j l l O l j planes, see Fig. 10 [72]. The
are chemically stable because of a
smaller number o f bonds through these planes compared to other plane families [73].
Because o f the isomorphic crystal structure o f wurtzite GaN and AIN, the planes limiting
the pyramids formed on N-polar GaN surface reported in [28-30] and the hillocks formed
on N-polar AIN have the same orientation (the plane families
equivalent with eaeh other). Note neither the hexagonal pyramids formed on N-polar
GaN nor those formed on N-polar AIN are associated with the struetural defects.
Zhuang et al. [72] concluded that freely nucleated AIN crystals predominately
have the aluminum to nitrogen direction pointing out from the nueleation surface, i.e. the
ends o f the AIN crystals facing the source have aliuninum polarity. This conclusion is
useful for obtaining a better understanding o f the sublimation-reeondensation process,
and subsequently helpful for scaling up the size and improving the quality o f AIN crystals.
3. Photo-assisted Wet Etching
An advantage o f photo-assisted etching over conventional etching is its ability to
etch GaN at room temperature. In addition, there is no significant difference in the Gapolar and N-polar GaN surface morphologies after photo-assisted wet etching at similar
conditions; i.e. the PEC etehing is independent o f crystal polarity [74]. However, photo­
assisted etching o f AIN should require UV illumination by light wavelengths shorter than
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200 nm since AIN has relatively larger bandgap (6.2 eV). To date, no photo-assisted wet
etching o f AIN has been reported.
3.1.
PEC Etching and Contactless Etching of GaN
The first PEC etching o f unintentionally doped «-type GaN was demonstrated by
Minsky et al. [75] using a He-Cd laser with a wavelength o f 325 nm and light intensities
o f -570 mW/cm^ in aqueous KOH and HCl solutions. The etch rates were about 400
nm/min and 40 nm/min in KOH and HCl solutions, respectively. No etching was
observed without illumination or under sub-bandgap energy illumination. By comparing
to the mechanism o f GaAs PEC etching, the authors proposed that, in PEC etching o f
GaN, the photogenerated holes assist in the oxidation o f Ga atoms, and the resulting
oxides are subsequent dissolved in the solution. The excess electrons are consumed by
the reduction reaction taking place on the cathode. An anisotropic etch profile with a
rough surface was obtained, as shown in Fig. 11 [75]. Youtsey et al. [76] performed PEC
etching o f «-GaN in a similar electrochemical cell, shown in Fig. 12, but using
illuminations from a Hg arc lamp. The system cathode was a Pt wire and a thin layer of
Ti (100 nm) was used as a mask and electrical contact to the samples. No external bias
was applied during etching. The etch rates were proportional to the light intensities and
varied from 50 to 300 nm/min for light intensities between 10 and 50 mW/cm^ in a well
stirred KOH solution. They further validated the GaN PEC etching mechanism proposed
by Minsky et al. [75] by calculating the total charges that passed through the
electrochemical cell during the etching process and postulated that the following
oxidation reaction is responsible for the decomposition o f GaN:
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2GaN + 6h^ ^ 2Ga^* +
(2)
PEC etchings o f n-GaN in electrolytes other than KOH and HCl were performed
by Seo et al. [77]. No etching was observed using HNO 3 and CH 3 COOH solutions
without external bias. The rates o f PEC etching in KOH were twice higher than that in
H 3 PO 4 .
Contactless etching o f GaN in a solution containing K 2 S2 O 8 and KOH was
investigated by several researeh groups [78-82, 85]. This etehing technique does not
require an electrical contact to the samples nor system eathodes; is potentially suitable for
device fabrication on insulating substrates; and can in principle be sealed up for bateh
processing. As in PEC etehing, photogenerated holes are involved in oxidation o f Ga
atoms. However, the exeess electrons are eonsumed by reduction o f sulfate ion radieals
(
A
SO ,
A
OH
and/or hydroxyl radieals
in the electrolyte, instead o f reactions on
V
cathodes. These radicals are formed by the photolysis o f peroxydisulfate ions
),
which occurs under UV illumination (T <310nw ) [79, 80]. The sulfate ion radieals are
eatalytieally produced at the Pt mask, whieh subsequently inereases the etch rate. The
etch rates dropped about an order o f magnitude when an inert mask, e.g. Si0 2 , was used
[79, 80]. As a function o f pH value o f the electrolyte, etch rates up to 50 nm/min were
observed using a catalytic mask [78]. The etch rates o f contactless etehing are typically
lower than that obtained in PEC etehing, presumably due to the slower rate o f eleetron
consumption and/or the eleetrolyte eomposition variations. Maher et al. [81, 82] reported
the main drawbaeks o f contactless etching is that the etch rate saturates after a eertain
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
time etching (10 minutes in their case) when the pH value o f etchants was less than 12.0.
Thus, to remove thick layers, the etching solution must be continually replenished to
compensate the pH drift. The variation o f pH value was believed to be due to the
photolysis o f peroxydisulfate ion (<S'2 0 g^~) under UV exposure, which leads to the
formation o f monopersulfuric (H 2 SO 5 ) [83] and sulfuric (H 2 SO 4 ) [84] acids. The pH
dependent etch rates were also reported by Parish et al. [85] who further related the etch
rates to surface roughness after etching.
The effects o f pH on the etch rates were investigated in anodic [8 6 ] and PEC
etching [87,
8 8
] o f GaN. Lu et al. [8 6 ] reported on the photo-assisted anodic etching o f n-
GaN in solutions o f tartaric acid/ethylene glycol at room temperature. Etch rates as high
as 160 nm/min were obtained for Hg arc lamp illumination. The maximum etch rate
occurred at pH=7 and dropped quickly as the pH value increased or decreased. Peng et al.
[87] and Ko et al. [8 8 ] also reported pH dependent etch rate for PEC etching o f «-type
GaN. The peak etch rates o f 125 and 90 nm/min occur in the H 3 PO 4 and KOH solution at
pH=0.75 and 14.25, respectively. A possible explanation o f this relationship between etch
rate and pH value is that the oxidized layer (Ga2 0 3 ) only dissolves in bases or acids o f
suitable pH concentration [89].
3.2. Anisotropic Etch Proftle and Surface Roughness
Several studies have focused on obtaining anisotropic etching with smooth etched
surface, as these characteristics are required for device fabrication. Youtsey et al. [76, 90]
achieved highly anisotropic etching o f n-type GaN using a PEC etching technique, as
shown in Fig. 13 [76]. The sample was etched to a depth o f -3 .5 pm, through the entire
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
thickness o f the GaN epilayer. The rough surface and striations on the sidewall were
believed to be due to a high dislocation density in the film ( 1 0 * ~
1 0
'“cw"^), which
generally propagate along the growth direction [76]. Anisotropic etch profiles with very
smooth surfaees were also obtained under eonditions o f very low KOH concentrations
(<0.01 M) and high light intensities [91]. Such conditions confine the reaetion kinetics to
the diffusion limited regime. The surface roughness on an etehed sample (1.5 nm) was
eomparable to that o f the unetched surface (0.3 nm), see Fig. 14 [91]. The surface
roughness of the area adjaeent to etching mask regions, however, may be high due to the
enhanced eteh rates, a consequenee o f loeal variations o f solution concentration, as
shovra in Fig. 15 [91]. Harush et al. [92] reported similar etch rate variations near the
mask region and concluded that mass transport from the bulk o f the solution to the
sample surface is rate limiting, and is responsible for the enhaneed etch rate occurring
adjacent to the mask.
From Table 2, whieh summarizes the effeets o f etehing on surface roughness, we
infer that the surfaee morphology is elosely related to the etching kinetics, i.e. lower etch
rates in the diffusion limited regime generally lead to smoother etched surfaces. In
diffusion-controlled regime, the etch rates on the peaks may be higher than that at the
valleys sinee the etehants are more likely to be depleted by the reaetion with peaks before
they diffuse to the valleys. Thus as etching proceeds, the surface is smoothed out. As
noted previously [75-76, 78, 87-88, 93], etching in the reaction limited regime results in
rough surfaces. In addition to etching condition optimization, post-etching methods have
also been developed to reduee surfaee roughness [94, 95]. Shelton et al. [94], for
example, showed that treating the etched sample by sonication could reduce the surface
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
roughness. In [95], a post etching in a hot KOH solution proved effective in removing
whisker shape etching residues, thus smoothing out the etched surface. The variations of
PEC etching techniques, such as current-controlled PEC etching [96, 97] and PEC with
applied external bias [98] were also investigated for surface roughness reduction. Rotter
et al. [96, 97] reported current-controlled PEC etching o f n-GaN in KOH solutions under
He-Cd laser illumination with etch rates o f up to -1 3 0 nm/min and a smooth “mirror­
like” etched surface. Stocker et al. [98] claimed that decreasing etchant (KOH)
concentration and increasing bias voltage decreases the roughness o f etched surface. The
RMS surface roughness decreased from approximately 1700 nm to a minimum o f 20 nm
as the bias voltage was increased from zero to 2.5 V, and as the KOH concentration was
decreased from 1.0 to 0.01 M.
3.3. Dopant-type/bandgap Selective Etching
Selective etching o f different materials o f different compositions or conductivity
type is quite useful for device fabrication and has been investigated for GaN. In p-type
materials, photogenerated holes are swept away from the depletion region near the
surface into the bulk, thus photo-assisted etching is impossible due to the absence o f the
holes. Consequently, no PEC etching o f />-GaN has been reported [82, 90, 99]. PEC
etching is not appreciable in intrinsic GaN either, because o f its high resistivity. There is
no conductive path between the etched region and the electrical contact to the sample
[99]. Thus excellent PEC etching selectivity for «-GaN over intrinsic GaN and />-GaN is
possible [99, 100], as shown in Fig. 16 and Fig. 17 [99]. The 300 nm thick «-type GaN
layer in Fig. 16 was completely etched and the etching stopped on the intrinsic GaN layer
[99]. Fig. 17 shows a nearly free-standing structure formed by undercutting the /?-GaN
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
layer over «-GaN for at least 10 (rm [99]. Stonas et al. [100] selectively PEC etehed nGaN under illumination from the back side o f the substrate (sapphire), leaving the top pGaN intact. The dopant-type selective etching was also observed in contactless etching.
No etching o f the carbon doped (p-type) GaN layer was observed [79, 80].
Highly selective etching can also be achieved by selecting a wider bandgap
material as an etching stop. Stonas et a l [101] selectively etched InxGai.xN (lower
bandgap energy with respect to GaN) layer over GaN using a GaN filter to restrict the
illumination wavelength. This process is potentially useful for substrate lift-off and
undercut structure formation. Similar bandgap selective etching for InGaN ternary alloy
was reported by Cho et al. [102]. Furthermore, Visconti et a l [103] found that AlxGai.xN
could be etched only if its bandgap is smaller than the excitation photon energy. A very
thin layer (5 nm) AIN was effective in completely stopping PEC etching.
3.4. Applications of Photo-assisted Etching in Device Fabrication Processes
Photo-assisted etching has been applied to fabricate the GaN devices, such as
MODFETs [104], AlGaN/GaN HFETs [81-82,105-107], GaN MESFETs [108,109], and
GaN Schottky rectifiers [110]. GaN micro-eleetromechanical system (MEMS) [111] and
improved ohmic contacts [85] on etched surfaces have also been reported.
3.5. Defect Evaluation by PEC Etching
In addition to material removal/patteming, PEC etching is capable o f revealing
the defects and their distribution. Youtsey et a l [112] first reported whisker shape
etehing features formed after PEC etching in a narrow KOH concentration range
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(0.01-0.04 M) under unstirred conditions, as shown in Fig. 18 [112]. Etchant with
concentration lower than 0.01 M confines the reaction kinetics to the diffusion-limited
regime and leads to a smooth etched surface [91]. Above 0.04 M, reaction-limited etching
resulted in highly anisotropic etching profile with rough surface [76, 90]. The
concentration window between reaction-limited and diffusion-limited regimes can be
extended by agitation during etching but using more dilute solution (0.001-0.01 M KOH)
[113]. In addition, the etching kinetics can also been controlled by varying the
illumination flux [103].
The “whisker” formed after PEC etching is believed to be due to the nonradiative
and recombinative electrical properties o f dislocations [114-117]. The photogenerated
carriers recombine at dislocations, and thus PEC etching is inhibited. This was confirmed
by Chen et al. [118], who reported the PEC etching ceases at the dislocations. The
dislocation-free GaN between dislocations is etched away, leaving the dislocations intact,
thus forming the whiskers that remain. Cross-seetional TEM images confirmed that these
whiskers indeed correspond to pure edge or mixed dislocations, as shown in Fig. 19
[112]. Both whiskers and dislocations in the unetched GaN are shown in Fig. 19a. Under
high magnification (Fig. 19b), one can see that pure edge and mixed dislocations
propagate from the bulk into the whiskers.
Since the diameters o f whisker feature are typically in the range o f 50-100 nm, in
low magnification SEM images, they appear as spots or dots, forming a “star map”
depicting the distribution o f dislocations. By counting the number o f spots in a fix area,
one can estimate the dislocation density. In [32-33, 103, 113, 119], the authors claimed
that the calculated whisker densities were in excellent agreement with the dislocation
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
densities obtained by both TEM and AFM measurements. However, the PEC etching
behaviors o f other defects, such as inversion domains and nanopipes are not yet fully
understood [74]. Thus far, only PEC etching o f heteroepitaxial GaN films, but not bulk
GaN, has been reported. Presumably the low defect densities (<10^ caw~^) in bulk GaN
would make it difficult to use.
4. Anodic Etching of GaN
Several other GaN etching methods have also been developed in order to avoid
some difficulties encountered in photo-assisted etching. For example, the photo-assisted
etching process can only etch moderately doped «-GaN, but not heavily n doped nor ptype doped GaN, due to minority carrier tunneling phenomena and surface band bending,
respectively. According to the relationship n ~ \ ! W j ,\ \ \ q space charge layer width fV^.^
becomes very small in heavily «-type doped materials, causing tunneling of
photogenerated minority carriers, which subsequently inhibits photo-assisted etching.
O et al.
[120]
demonstrated p-GaN
and InGaN etching using pulsed
electrochemical methods (anodic etching, where the semiconductor was positively
biased) in H 3 P 0 4 /ethylene glycol/HiO solutions. However, Yang et al. [121] etched a ptype GaN sample by applying negative voltage to the sample in PEC etching process.
They claimed that the surface band was bent upward upon applying negative bias to the
substrate and the etch rate increased as the voltage became more negative. This
contradicts the conventional understanding that the surface band o f /?-type semiconductor
bends downward upon negative bias [ 1 2 2 ].
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nowak et al. [123] reported an anodic etching process for a highly conductive
GaN single crystal after CMP. At a relatively high applied potential (2.8 V), the etching
occurred preferentially on defect-free areas. At low voltage, however, the etch rate did
not depend on the presence of crystallographic defects. In other words, anodic etching
with low applied potential could be used to remove scratches formed during CMP
process.
Yoshida [124] found high quality GaN with mirror like surface could not be PEC
etched in a dilute KOH solution (2.4 wt%), though similar samples with rough surfaces
were etched at the same conditions. The author thus etched the high quality GaN samples
by photo-assisted anodic etching, producing higher etch rates than that in anodic etching
without illumination. A photo-assisted anodic etching process in NaOH with added
chloride ions was developed by Ohkubo [125]. The presence o f chloride ions in the
electrolyte accelerated the rate o f photo-assisted anodic etching by reducing the extent of
the formation o f gallium oxide layer on the GaN surface.
5. Defect-selective Chemical Etching
The main purpose o f defect-selective etching is to reveal defects in the crystals
including dislocations, nanopipes and inversion domains as well as their distributions.
These defects, especially dislocations, are known to affect both the electrical and optical
properties o f the materials [115, 126]. Though we have seen previously that PEC etching
is an excellent means for dislocation density estimation, this technique is not applicable
to intrinsic, highly resistive or heavily n-doped samples.
5.1. Defect-selective Etching of Epitaxial GaN
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As noted in subsection 2.1, hot phosphoric acid can selectively etch Ga-polar GaN
epilayers and produce hexagonal etch pits on defects [31-35, 42-44, 46-47]. Liquid
H 2 SO 4 /H 3 PO 4 mixtures [127, 128] and molten KOH [32, 35, 74, 129-130] are also
capable o f producing hexagonal pits on MOCVD grown GaN. A problem o f defectselective etching using molten KOH is that the etching temperature is limited by the
melting point o f pure KOH (360 °C). This difficulty is circumvented by adding 41 wt%
NaOH into KOH to form a eutectic alloy (denoted as E), which has a melting point of
170 °C. See Fig. 20 [131] for the phase diagram o f KOH/NaOH eutectic alloy.
The origin o f the hexagonal etch pits formed on GaN epilayers after etching is
controversial. On the one hand, several research groups found that the etch pits varied in
size and the EPD was generally lower than the dislocation density evaluated by TEM
[129, 132-133]. Kozawa et al. [129] ascribed the etch pits formed in GaN by molten
KOH etching to dislocations, though the EPD (2 x l0 ’ c/n“^) was one order o f magnitude
). Ono et al. [132] reported a reduction of
lower than that measured by TEM (2x
EPD from 4 x l 0 ’ cw “^ to 6xl0® cw “^ by inserting a thin InGaN buffer layer. However,
according to TEM measurement, the dislocation density remained unchanged (lO^cw"^).
Yamamoto et al. [133] observed three different size etch pits with EPD in the range from
10®
to lO^cw”^ after molten KOH etching. Based on these observations and by
examination o f TEM images o f etch pits, Hong et al. [44, 134] concluded that etch pits
were formed on nanopipes, which have larger Burgers vectors (thus higher energy) than
dislocations, but not on threading dislocations. On the other hand, Visconti et al. [32, 35]
and Xu et al. [46] reported that the EPD obtained by etching o f Ga-polar GaN in hot
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
phosphoric acid was consistent with dislocation densities found by TEM, AFM and PEC
etching. Moreover, etching a GaN sample at 360° C in molten KOH, Shiojima [135]
argued that the pits are formed on mixed screw-edge threading dislocations after
characterization o f the etch pits by AFM and cross sectional TEM. No etch pits
associated with nanopipes were observed. The TEM images o f such sample are shown in
Fig. 21 [135], where a sharp V-shape cross section o f etch pits containing the bottom,
which terminated at a mixed dislocation can be seen.
By careful consideration o f the growth conditions and etching parameters o f the
aforementioned samples [44, 134-135], we propose two possible causes to explain the
discrepancy between [134] and [135]. First, the sample examined in [135] may have a
very low density o f nanopipes or even be nanopipes free due to the growth condition
variation. Under such circumstances, it is possible that no nanopipes were included in the
AFM and/or TEM examining area (up to 15 pm^). Second, considering that nanopipes
have larger Burgers vector (higher energy) than dislocations and the sample in [135] was
etched under more vigorous conditions (360 °C) than that from [44] and [134] (215 °C),
there might exist an intermediate etching temperature where only nanopipes but no
dislocations can be etched. In other words, at a higher etching temperature, both
dislocations and nanopipes can be etched on a same sample. This situation was
demonstrated by W eyher et al. [74] who etched a MOCVD grown GaN epilayer in
molten E. In the TEM image, shown in Fig. 22 [74], etch pits formed on nanopipes
(marked as 1) and mixed dislocations (marked as 2) were clearly seen. No etch pits were
developed on pure edge dislocations (marked as 3) since it has even lower energy than
mixed dislocations according to [134]. Note the intermediate etching conditions are
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
probably solely determined by etching temperature but not the amount o f time, since in
[44], no pits were formed on dislocations after prolonged etching and only the sizes o f the
pits on nanopipes were found to be proportional to the etching time.
To avoid the influence o f other defects when evaluating the dislocation densities
in epitaxial films, careful optimization o f the etching parameters are required.
Furthermore, over etching should be avoided as it can lead to an underestimation o f the
dislocation density. An example o f over etching is shown in Fig. 23 and Fig. 24 [32]. Fig.
23a and 23b are AFM images o f GaN sample after etching in molten KOH and hot
phosphoric acids with EPDs consistent with each others, about
cm~^. Fig. 24 shows
an AFM image o f the same sample etched in phosphoric acid for a longer time. Two
types o f etch pits with different size were seen. The EPD is one order o f magnitude lower
than that in Fig. 23, indicating the sample was over etched.
5.2. Defect-selective Etching of Bulk GaN
Defect-selective etching o f bulk GaN samples in E was demonstrated by Weyher
et a l [74, 130] and Kamler et a l [136]. The optimal etching temperatures are polarity
dependent [130]: the best temperatures are 350-450 °C and 200-250 °C for Ga-polar and
N-polar surface, respectively. The appropriate etching time depends on defect
density/type and is in the range o f 1-5 minutes. As shovm in Fig. 25, etching produced
hexagonal etch pits on Ga-polar [130] and circular terraced etch pits on N-polar surface
[74, 130], respectively. The hexagonal etch pits on Ga-polar reflect the crystallographic
symmetry o f the GaN lattice and often form linear arrays. Randomly distributed etch pits,
however, were also observed [130]. According to TEM measurements and indentation
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
calibrations - a technique which is capable o f intentionally introducing dislocations at
relative lower temperature - the authors confirmed that these etch pits were formed on
dislocations [74, 130].
A modified etching system (denoted as E+M) for bulk plate-like GaN was
introduced by Kamler et a l [136] to overcome the difficulty o f over etching N-polar
surface while the etching temperature was set for Ga-polar etching. The viscosity o f the
etchant (E) can be signifieantly increased by adding 10 wt% magnesium oxide. Thus, a
sticky droplet o f etehant can be applied to only one side o f sample, i.e. the Ga-polar
surface. The over-flowed etchant readily reacts with the aluminum plate placed
underneath the samples. No significant difference o f etching results between E and E+M
were seen [136].
5.3. Defect-selective Etching of AIN
Very recently, Zhuang et a l [137] reported the results o f defect-selective etching
in E and E+M on epitaxial AIN film as well as bulk AIN crystals. They proposed the
etching occurred by the following reaetion:
A IN + 6 K 0 H -> A I { 0 H \ i +NH, t +3K^O i
(3)
The SEM images o f etching pattem produced in E and E+M for an Al-polar [72]
epitaxial AIN film [138] are shown in Fig. 26 and Fig. 27. A linear array o f hexagonal
pits is visible in Fig. 26, indicating these defects belong to the same grain. Randomly
distributed pits are also seen. All hexagonal pits had the same azimuthal orientation and
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were approximately the same size. The calculated EPDs from Fig. 26 (lOVm ^) and Fig.
27 (9 . 4 X10®
) are almost identical.
The etching results o f a bulk Al-polar AIN sample grown by directly heating of
the source materials by microwaves [139] is shown in Fig. 28 [137]. A hexagonal etch pit
is clearly visible and the correspondent EPD is about bxlO^cm"^. This value is in
agreement with the dislocation density (bxlO^cw”^) determined from the synchrotron
white beam X-ray topography (SWBXT) pattem (shovm in Fig. 29 [137]) taken from the
same sample. Thus, the authors concluded that the etch pits on Al-polar surface do form
on dislocations, i.e. the evaluation o f dislocation density for Al-polar AIN crystals by
defect-selective etching in eutectic E or E-i-M are effective and reliable.
Etching in E produced hexagonal hillocks on N-polar AIN, as shown in Fig. 30
[137]. The sample was first etched in 45 wt% KOH in order to identify area having
nitrogen polarity (Fig. 30a) [72]. After etching in E (Fig. 30b), the hillocks became larger
in size and had lower density. No evidence has been found to support the argument that
these hillocks were associated with dislocations.
Although GaN and AIN have isomorphic lattice stmcture (wurtzite), etching in E
produced hexagonal hillocks on N-polar AIN [137], while rounded pits were observed on
N-polar GaN [74, 130]. Further investigation is needed to give a reasonable explanation
of this difference.
6. Summary and Future Research
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The excellent ehemieal stability o f GaN and AIN makes their group III element
polar (0001) planes highly resistant to conventional wet etching. At moderate
temperature, no aqueous etchants have been found to etch (0001) GaN, and only KOH
and NaOH aqueous solutions are capable o f etching AIN (0001) epilayers. Etching is
polarity dependent, i.e. N-polar is more reactive than Ga(Al) polar Ga(Al)N. Such
polarity dependent etehing is believed to be caused by different surface banding states.
Thus, one can identify different polarities by observing the surfaee morphology changes
after simple wet etching processes. With six fold symmetry, the plane families defining
the hexagonal hillocks formed by etching in KOH solution on both N-polar GaN and Npolar AIN are equivalent. Etching o f Ga-polar GaN in hot phosphoric acid can reveal
micro-defects, such as nanopipes and dislocations; however, careful attention to the
process and post-etehing calibration are essential to obtain reasonable estimation of
defect densities. Crystallographic etching o f non c-plane GaN has been demonstrated [48,
49], and similar process for AIN etehing is needed to accelerate development o f device
fabrication processes.
Highly anisotropic, dopant-type/bandgap selective, and diffusion-limited PEC
etehing processes resulting in smooth surfaee morphologies have been successfully
developed. Etching in diffusion limited regime generally results in a smooth etched
surfaee. Moreover, some post-etehing treatments have proven effective at surface
cleaning. PEC etching is capable o f revealing dislocation density in a moderate doped nGaN epilayer grown on conductive substrate. However, the etehing behaviors of
inversion domains and nanopipes in PEC etehing require further investigation. A
contactless etching, whieh does not require counter electrode or electrical contact to the
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
samples, has also been developed. However, due to the concentration and pH variation,
the etchant needs to be replenished periodically when a thick layer o f material is to be
etched.
Optimization o f etching conditions is required for defect-selective chemical
etching o f group III nitride epitaxial films due to the presence o f nanopipes, which have
larger Burgers vector (higher energy) than dislocations. Defect-selective etching o f bulk
GaN and Al-polar AIN crystals in KOH/NaOH eutectic alloy has been confirmed to be
effective and reliable for dislocation density estimation by indentation, TEM and
SWBXT calibrations. The optimal etching condition depends on the crystal polarity,
defect type and distributions. The details o f dislocation microstructures, e.g. pure edge,
pure screw or mixed edge-screw dislocations, leading to hexagonal etch pits formation
are undetermined. The mechanism o f forming hexagonal hillocks on N-polar AIN and
whether these etching features correspond to dislocations is another interesting topic for
further investigation.
Acknowledgement
Support for this research from the Office o f Naval Research, Award No. N0001402-1-0290 is greatly appreciated.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reference
1. 0 . Ambacher, J. Phys. D, 31, 2653 (1998)
2. J. H. Edgar, (Editor), Properties o f Group III Nitrides, Electronic Materials
Information Service (EMIS), London, (1994)
3. D. Kapolnek, X. H. Wu, B. Keying, S. Keller, B. P. Keller, U. K.Mishra,
S. P.
DenBaars, and J. S. ^^Qck, Appl. Phys. Lett., 67, 1541 (1995)
4.
F. A. Ponce, B. S. Krusor, J. S. M. Jr., W. E. Plano, and D. F. Welch, Appl. Phys.
Lett., 67,410(1995)
5.
S. Keller, B. P. Keller, Y. F. Wu, B. Keying, D. Kapolnek, J. S. Speck, U. K. Mishra,
S. P. Denbaars, Appl. Phys. Lett.,
6
.
7.
.
M. Seelmann-Eggebert, J. L. Weyher, K. Obloh, K. Zimmermann, A. Rar, and S.
Phys. Lett., 71, 2635 (1997)
J. L. Rouviere, J. L. Weyher, M. Seelmann-Eggebert, and S. Porowski, Appl. Phys.
Lett., 73,
9.
, 1525 (1996)
A. Bykhovski, B. Gelmont, M. Shur, J. Appl. Phys., 74, 6734 (1993)
Porowski,
8
6 8
6 6 8
(1998)
M. Sumiya, K. Yoshimura, K. Ohtsuka, and S. Fuke, Appl. Phys. Lett., 76, 2098
(2000)
10. E. C. Piquette, P. M. Bridger, Z. Z. Bandic, and T. C. McGill, J. Vac. Sci. Technol.
B, 17, 1740 (2000)
11. B. Daudin, J. L. Rouviere, and M. Arlery, Appl. Phys. Lett., 69, 2480 (1996)
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12. L. K. Li, M. J. Jurkovic, W. I. Wang, J. M. Van Hove, and P. P. Chow, Appl. Phys.
Lett., 76, 1740 (2000)
13. S. F. Chichibu, A. Setoguchi, A. Uedono, K. Yoshimura, and M. Sumiya, Appl.
Phys. Lett., 78, 28 (2001)
14. G. A. Martin, A. Botchkarev, A. Rockett, and H. Morko?, Appl. Phys. Lett., 68, 2541
(1996)
15. U. Karrer, O. Ambacher, and M. Stutzmann, Appl. Phys. Lett., 77, 2012 (2000)
16. H. Morko 9 , R. Cingolani, and B. Gil, Solid-State Electron., 43, 1909 (1999)
17. V. Kirilyuk, A. R. A. Zauner, P. C. M. Christianen, J. L. Weyher, P. R. Hageman,
and P. K. Larsen, Appl. Phys. Lett., 76, 2355 (2000)
18. E. D. Haberer, C. H. Chen, A. Abare, M. Hansen, S. Denbaars, L. Coldren, U.
Mishra, and E. L. Hu, Appl. Phys. Lett., 76, 3941 (2000)
19. R. Dimitrov, V. Tilak, W. Yeo, B. Green, H. Kim, J. Smart, E. Chumbes, J. R.
Shealy, W. Schaff, L. F. Eastman, C. Miskys, O. Ambacher, and M. Stutzmann,
Solid-State Electron., 44, 1361 (2000)
20. S. J. Pearton, C. R. Abernathy, C. B. Yartuli, J. D. Mackenzie, R. J. Sbul, R. G.
Wilson, and J. M. Zavada, Electron. Lett., 31, 836 (1995)
21. T. L. Chu, J Electrochem. Soc., 118,1200 (1971)
22. J. I. Pankove, J! Electrochem. Soc., 119, 1118 (1972)
23. H. Cho, D. C. Hays, C. B. Yartuli, S. J. Pearton, C. R. Abernathy, J. D. MacKenzie,
F. Ren, and J. C. Zolper, Mat. Res. Soc. Symp. Proc., 483, 265 (1998)
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24. C. B. Vartuli, S. J. Pearton, C. R. Abernathy, J. D. MaeKenzie, F. Ren, J. C. Zolper,
and R. J. Shul, Solid-State Electron., 41(12), 1947 (1998)
25. S. J. Pearton and R. J. Shul in Gallium Nitride I, J. Pankove and T. D. Moustakas
Eds., Semiconductor and Semimetals Series, 50, 103, Aeademie Press, New York,
NY (1998)
26. C. A. Carosella, B. Molnar, S. Sehiestel, and J. A. Sprague, M RS Internet J. Nitride
Semicond. Res. SSI, W 11.70 (2000)
27. T. Palaeios, F. Calle, M. Varela, C. Ballesteros, E. Monroy, F. B. Naranjo, M. A.
Sanehez-Garcia, E. Calleja, and E. Munoz, Semicond. Sci. Technol, 15, 996 (2000)
28. H. M. Ng, N. G. Weimann, and A. Chowdhury, J. A ppl Phys., 94, 650 (2003)
29. H. M. Ng, A. Chowdhury, W. Parz, and N. G. Weimann, Proceedings o f
Electrochemical Society, (State-of-the-Art Program oh Compound Semiconductors
XX XIX and Nitride and Wide Bandgap Semiconductors fo r Sensors, Photonics, and
Electronics IV), 2003-11, 3 (2003)
30. H. M. Ng, W. Parz, N. G. Weimann, and A. Chowdhury, Jpn. J. Appl. Phys. Part 2,
42, L1405 (2003)
31. D. Huang, P. Visconti, K. M. Jones, M. A. Reshchikov, F. Yun, A. A. Baski, T.
King, and H. Moxkof;, Appl. Phys. Lett., 78, 4145 (2001)
32. P. Visconti, K. M. Jones, M. A. Reshchikov, R. Cingolani, H. Morko?, and R. J.
Molnar, Appl. Phys. Lett., 77, 3532 (2001)
33. P. Visconti, D. Huang, M. A. Reshchikov, F. Yun, T. King, A. A. Baski, R.
Cingolani, C. W. Littion, J. Jasinski, Z. Liliental-Weber, and H. Morko?, Phys. Stat.
Sol B, 228(2), 513 (2001)
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34. P. Visconti, D. Huang, F. Yun, M. A. Reshchikov, T. King, R. Cingolani, J. Jasinski,
Z. Liliental-Weber, and H. Morko 9 , Phys. Stat. Sol. A, 190(1), 5 (2002)
35. P. Visconti, D. Huang, M. A. Reshchikov, F. Yun, R. Cingolani, D. J. Smith, J.
Jasinski, W. Swider, Z. Liliental-Weber, and H. Morko?, Mater. Sci. Eng. B, 93, 229
(2002)
36. M. Shimizu, A. Suzuki, M. Watanabe, J. Shirakashi, K. Balakrishan, and H.
Okumura, Blue Laser and Light Emitting Diodes II, [International Symposium on
Blue Laser and Light Emitting Diodes], 2nd, Chiba, Japan, Sept. 29-Oct. 2, 723,
(1998)
37. E. S. Heilman, M RS Internet J. Nitrides Semicond. Res., 3, 11 (1998)
38. D. Li, M. Sumiya, K. Yoshimura, Y. Suzuki, Y. Fukuda, and S. Fuke, Phys. Stat. Sol.
A, 180, 357 (2000)
39. D. Li, M. Sumiya, S. Fuke, D. Yang, D. Que, Y. Suzuki, and Y. Fukuda, J. Appl.
Phys., 90, 4219 (2001)
40. J. L. Weyher, S. Miiller, 1. Grzegory, and S. Porowski, J. Cryst. Growth, 182, 17
(1997)
41. Y. Morimoto, J. Electrochem. Soc., 121, 1383 (1974)
42. A. Shintani, and S. Minagawa, J. Electrochem. Soc., 123, 706 (1976)
43. B. J. Kim, J. W. Lee, H. S. Park, Y. Park, and T. I. Kim, J. Electron. Mater., 27(5),
L 3 2 (1998)
44. S. K. Hong, B. J. Kim, H. S. Park, Y. Park, S. Y. Yoon, and T. I. Kim, J. Cryst.
Growth, 191, 275 (1998)
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45. W. Qian, M. Skowronski, K. Doverspike, L. B. Rowland, and D. K. Gaskill, J. Cryst.
Growth, 151, 396 (1995)
46. X. Xu, R. P. Vaudo, J. Flynn, and G. R. Brandes, J. Electron. Mater., 31, 402 (2002)
47. M. G. Mynbaeva, Yu. V. Melnik, A. K. Kryganovskii, and K. D. Mynbaev,
Electrochem. Solid-State Lett., 2(8), 404 (1999)
48. D. A. Stocker, E. F. Schubert, and J. M. Redwing, Appl. Phys. Lett., 73, 2654 (1998)
49. D. A. Stocker, I. D. Goepfert, E. F. Schubert, K. S. Boutros, and J. M. Redwing, J.
Electrochem. Soc., 147, 763 (2000)
50. D. A. Stocker, E. F. Schubert, and J. M. Redwing, Appl. Phys. Lett., 7 7 , 4253 (2000)
51. D. A. Stocker, E. F. Schubert, K. S. Boutros, and J. M. Redwing, M RS Internet J.
Nitride Semicond. Res., 4S1, G7.5 (1999)
52. T. Bottcher, Ch. Zellweger, S. Figge, R. Kroger, Ch. Fetter, H. -J . Biihlmann, M.
Ilegems, P. L. Ryder, and D. Hommel, Phys. Stat. Sol. A, 191, R3 (2002)
53. M. A. Reshchikov, D. Huang, and H. Morko 9 , Mat. Res. Soc. Symp. Proc., 743,
L11.3 (2003)
54. S. Tripathy, S. J. Chua, and A. Ramam, J. Phys.: Condens. Matter, 14, 4461 (2002)
55. S. Tripathy, S. J. Chua, A. Ramam, E. K. Sia, J. S. Pan, R. Lim, G. Yu, and Z. X.
Shen, J. Appl. Phys., 91, 3398 (2002)
56. T, Y, Sheng, Z. Q. Yu, and G. J. Collins, Appl. Phys. Lett., 52, 576 (1988)
57. K. M. Taylor, and C. Lenie, J. Electrochem. Soc., 107, 308 (1960)
58. G. Long, and L. M. Foster, J. Am. Ceram. Soc., 42, 53 (1959)
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59. N. J. Barrett, J. D. Grange, B. J. Sealy, and K. G. Stephens, J. Appl. Phys., 57, 5470
(1985)
60. C. R. Aita, and C. J. Gawlak, J. Vac. Sci. Technol. A, 1, 403 (1983)
61. G. R. Kline, and K. M. Lakin, Appl. Phys. Lett., 43, 750 (1983)
62. C. B. Vartuli, S. J. Pearton, J. W. Lee, C. R. Abernathy, J. D. MacKenzie, J. C.
Zolper, R. J. Shul, and F. Ren, J. Electrochem. Soc., 143, 3681 (1996)
63. S. S. Tan, M. Ye, and A. G. Milnes, Solid-State Electron., 38, 17 (1995)
64. S. Adachi, and K. Oe, J. Electrochem. Soc., 130, 2427 (1983)
65. S. K. Ghandi, VLSI Fabrication Principles, Wiley, New York, 1994
66. W. M. Moreau, Semiconductor Lithography: Principles, Practices and Materials,
Plenum, New York, 1988
67. J. R. Mileham, S. J. Pearton, C. R. Abernathy, J. D. MacKenzie, R. J. Shul, and S. P.
Kilcoyne,^/?/?/. Phys. Lett., 67(8), 1119 (1995)
68. J. Jasinski, Z. Liliental-Weber, Q. S. Paduano, and D. W. Weybume, Appl. Phys.
Lett., 83, 2811 (2003)
69. J. R. Mileham, S. J. Pearton, C. R. Abernathy, J. D. MacKenzie, R. J. Shul, and S. P.
K ilcoyne,Z Vac. Sci. Technol. A, 14(3), 836 (1996)
70. I . Ide, M. Shimizu, A. Suzuki, X. Q. Shen, H. Okumura, and T. Nemoto, Jpn. J.
Appl. Phys., 40, 4785 (2001)
71. L. J. Schowalter, J. C. Rojo, N. Yakolev, Y. Shusterman, K. Dovidenko, R. Wang, I.
Bhat, and G. Slack, M RS Internet J. Nitride Semicond. Res., 5S1, W6.7 (2000)
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72. D. Zhuang, J. H. Edgar, L. Liu, B. Liu, and L. Walker, M RS Internet J. Nitride
Semicond. Res. 7, 4 (2002)
73. S. Kitamura, K. Hiramatsu, N. Sawaki, Jpn. J. Appl. Phys., 24, L1184 (1995)
74. J. L. Weyher, P. D. Brown, J. L. Rouviere, T. Wosinski, A. R. A. Zauner, and I.
Grzegory, J. Cryt. Growth, 210, 151 (2000)
75. M. S. Minsky, M. White, and E. L. \{\x,Appl. Phys. Lett., 68, 1531 (1996)
16. C. Youtsey, I. Adesida, and G. BvXmsxx, Appl. Phys. Lett., 71, 2151 (1997)
77. J. W. Seo, C. S. Oh, J. W. Yang, G. M. Yang, K. Y. Lim, C. J. Yoon, and H. J. Lee,
Phys. Stat. Sol. A, 188, 403 (2001)
78. J. A. Bardwell, 1. G. Foulds, J. B. Webb, H. Tang, J. Fraser, S. Mosia, and S. J.
Rolfe, J. Electron. Mater., 28, L24 (1999)
79. J. A. Bardwell, J. B. Webb, H. Tang, J. Fraser, and S. Moisa, Proceedings o f
Electrochemical Society, 200I-I(III-Nitride Based Semiconductor Electronics and
Optical
Devices
and
Thirty-Fourth
State-of-the-Art-Program
on
Compound
Semiconductors), 193-203 (2001)
80. J. A. Barwell, J. B. Webb, H. Tang, J. Fraser, and S. Moisa, J. Appl. Phys., 89, 4142
(2001)
81. H. Maher, D. DiSanto, G. Soerensen, M. W. Dvorak, T. W. MacElwee, J. B. Webb,
and C. R. Bolognesi, Proceedings o f IEEE/Cornell Conference on High Performance
Devices, Ithaca, NY, United States, Aug. 7-9, 192-198 (2000)
82. H. Maher, D. W. DiSanto, G. Soerensen, C. R. Bolognesi, H. Tang, and J. B. Webb,
Appl. Phys. Lett., 77, 3833 (2000)
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83. L. Dogliotti, and E. Hayon, J. Phys. Chem., 71, 2511 (1967)
84. I. M. Kolthoff, and I. K. Miller, J. Am. Chem. Soc., 73, 3055 (1951)
85. G. Parish, P. A. Scali, S. M. R. Spaargaren, B. D. Nener, Proceedings o f SPIE-The
International
Society fo r
Optical
Engineering,
4592(Device
and
Process
Technologies fo r MEMS and Microelectronics II), 104-110 (2001)
Lu, Z. Wu, and I. Bhat, J. Electrochem. Soc., 144, L8 (1997)
86.
H.
87.
L.-H . Peng, C. -W . Chuang, J. -K . Ho, , C. -N . Huang, and C. -Y . Chen, Appl.
Phys. Lett., 72, 939 (1998)
88. C. H. Ko, Y. K. Su, S. J. Chang, W. H. Lan, J. Webb, M. C. Tu, and Y. T. Chemg,
Mater. Sci. Eng. B, 96, 43 (2002)
89. C. Vanleugenhaghe, N. de Zoubov, and M. Pourbaix, Atlas o f Electrochemical
Equilibria in Aqueous Solutions, p. 428, National Association o f Corrosion
Engineers, Houston (1974)
Youtsey, 1. Adesida, and G. Bulman, Electron. Lett., 33, 245 (1997)
90.
C.
91.
C.Youtsey, I. Adesida, L. T. Romano, and G. Bulman, Appl. Phys.Lett.,
72, 560
(1998)
92. E. Harush, S. Brandon, J. Salzman, and Y. Paz, Semicond. Sci. Technol, 17, 510
(2002)
93. B. M. Kinder, and T. L. Tansley, Proc. Conference on Optoelectronics and
Microelectronics Materials and Devices (COMMAD), 195 (1999)
94. B. S. Shelton, T. G. Zhu, M. M. Wong, H. K. Kwon, C. J. Eiting, D. J. H. Lambert,
S. P. Turini, and R. D. Dupuis, Electrochem. Solid-State Lett., 3, 87 (2000)
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95. J. Skxiniarova, A. van der Hart, H. P. Bochem, A. Fox, and P. Kordos, Mater. Sci.
Eng. B, 91-92, 298 (2002)
96. T. Rotter, D. Uffmann, J. Ackermann, J. Aderhold, J. Stemmer, J. Graul, Mat. Res.
Soc. Symp. Proc., 482, 1003 (1998)
97. T. Rotter, J. Aderhold, D. Mistele, O. Semchinova, J. Stemmer, D. Uffmann, and J.
Graul, Mater. Sci. Eng. B, 59, 350 (1999)
98. D. A. Stocker, and E. P. Schubert, J. Electrochem. Soc., 146, 2702 (1999)
99. C. Youtsey, G. Bulman, and I. Adesida, J. Electron. Mater., 27, 282 (1998)
100. A. R. Stonas, P. Kozodoy, H. Marchand, P. Fini, S. P. DenBaars, U. K. Mishra, E.
L. Wu,Appl. Phys. Lett., 77, 2610 (2000)
101. A. R. Stonas, T. Margalith, S. P. DenBaars, L. A. Coldren, and E. L. Hu, Appl.
Phys. Lett., 78, 1945 (2001)
102. H. Cho, S. M. Donovan, C. R. Abernathy, S. J. Pearton, F. Ren, J. Han, and R. J.
Shul, M RS Internet J. Nitride Semicon. Res., 481, G6.40 (1999)
103. P. Visconti, M. A. Reshchikov, K. M. Jones, F. Yun, D. F. Wang, R. Cingolani, and
H. Morko?, Mat. Res. Soc. Symp. Proc., 639, G3.14 (2001)
104. J. A. Bardwell, I. Foulds, B. Lamontagne, H. Tang, J. B. Webb, P. Marshall, S. J.
Rolfe, J. Stapledon, and T. W. MacElwee, J. Vac. Sci. Technol. A, 18, 750 (2000)
105. H. Maher, D. W. DiSanto, M. W. Dvorak, G. Soerensen, C. R. Bolognesi, J. A.
Bardwell, H. Tang, and J. B. Webb, Electron. Lett., 36, 1969 (2000)
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106. J. S. Lee, J. W. Kim, D. C. Jung, C. S. Kim, W. S. Lee, J. H. Lee, J. H. Shin, M. W.
Shin, J. E. Oh, and J. H. Lee, Jpn. J. Appl. Phys., 40, L198 (2001)
107. D. W. DiSanto, A. C. Kwan, and C. R. Bolognesi, Electron. Lett., 38, 921 (2002)
108. A. T. Ping, D. Selvanathan, C. Youtsey, E. Finer, J. Redwing, and 1. Adesida,
Electron. Lett., 35, 2140 (1999)
109. W. S. Lee, Y. H. Choi, K. W. Chung, D. C. Moon, and M. W. Shin, Electron. Lett.,
36, 265 (2000)
110. T. G. Zhu, D. J. H. Lambert, B. S. Shelton, M. M. Wong, U. Chowdhury, and R. D.
'Dupuis, Appl. Phys. Lett., 77,2918 (2000)
111. A. R. Stonas, N. C. MacDonald, K. L. Turner, S. P. DenBaars, and E. L. Hu, J. Vac.
Sci. Technol. B, 19, 2838 (2001)
112. C. Youtsey, L. T. Romano, and I. Adesida, Appl. Phys. Lett., 73, 797 (1998)
113. C. Youtsey, L. T. Romano, and I. Adesida, Appl. Phys. Lett., 74, 3537 (1999)
114. N. G. Weimann, L. F. Eastman, D. Doppalapudi, H. M. Ng, and T. D. Moustakas, J.
Appl. Phys., 83, 3656 (1998)
115. S. J. Rosner, E. C. Carr, M. J. Ludowise, G. Girolami, and H. I. Erikson, Appl.
Phys. Lett., 70, 420 (1997)
116. T. Sugahara, H. Sato, M. Hao, Y. Naoi, S. Kurai, S. Tattori, K. Yamashita, K.
Nishino, L. T. Romano, S. Sakai, Jpn. J. Appl. Phys., 37, L398 (1998)
117. F. Bertram, J. Christen, M. Schmidt, M. Topf, S. Koymov, S. Fischer, B. Meyer,
Mater. Sci. Eng. B, 50, 165 (1997)
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118. K. L. Chen, R. Zhang, S. L. Gu, D. Q. Lu, X. Q. Xiu, H. Q. Yu, B. Shen, Y. Shi,
and Y. D. Zheng, Proceedings o f International Conference on Solid-State and
Integrated Circuit Technology, 6th, Shanghai, China, Oct. 22-25, 2, 1217 (2001)
119. J. L. Weyher, L. Maeht, F. D. Tiehelaar, H. W. Zandbergen, P. R. Hageman, and P.
K. Larsen, Mater. Sci. Eng. B, 91/92, 280 (2002)
120. J. O, P. S. Zory, and D. P. Hour, SPIEProc., 3002, 11 (1997)
121. J. W. Yang, B. M. Kim, C. J. Yoon, G. M. Yang, and H. J. Lee, Electron. Lett., 36,
88 (2000)
122. R. D. Rauh in Electrochemistry o f Semiconductors and Electronics, J. McHardy and
F. Ludwing Eds., p. 177, Noyes Publications, NJ (1992)
123. G. Nowak, X. H. Xia, J. J. Kelly, J. L. Weyher, and S. Porowski, J. Cryst. Growth,
222, 735 (2001)
124. S. Yoshida, J. Cryst. Growth, 181, 293 (1997)
125. M. Ohkubo, Mater. Sci. Eng. B, 59, 355 (1999)
126. B. Garni, J. Ma, N. Perkins, J. Liu, T. F. Kuech, and M. G. Lagally, Appl. Phys.
Lett., 68, 1380 (1996)
127. T .C. Wen, W. I. Lee, J. K. Sheu, and G. C. Chi, Solid-State Electron., 46, 555
(2002)
128. T. C. Wen, S. C. Lee, H. S. Chuang, C. H. Chiou, and W. I. Lee, Mat. Res. Soc.
Symp. Proc., 639, G3.56 (2001)
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129. T. Kozawa, T. Kachi, T. Ohwaki, Y. Taga, N. Koide, and M. Koike, J.
Electrochem. Sac., 143, L17 (1996)
130. J. L. Weyher, L. Maeht, G. Kamler, J. Borysiuk, and I. Grzegory, Phys. Stat. Sol. C,
0(3), 821 (2003)
131. httn://www.crct.polvmtl.ca/fact/phase diagram.php?file=PhasKOH-NaOH.ipg
132. Y. Ono, Y. lyechika, T. Takada, K. Inui, and T. Matsye, J. Cryst. Growth, 189/190,
133 (1998)
133. K. Yamamoto, H. Ishikawa, T. Egawa, T. Jimbo, and M. Umeno, J. Cryst. Growth,
189/190,575 (1998)
134. S. K. Hong, T. Yao, B. J. Kim, S. Y. Yoon, and T. I. Kim, Appl. Phys. Lett., 77, 82
(2000)
135. K. Shiojima, J. Vac. Sci. Technol, B, 18, 37 (2000)
136. G. Kamler, J. L. Weyher, I. Grzegory, E. Jezierska, and T. Wosinski, J. Cryst.
Growth, 246, 21 (2002)
137. D. Zhuang, J. H. Edgar, B. Strojek, J. Chaudhuri, and Z. Rek, J. Cryst. Growth,
262, 89(2004)
138. Y. Shi, B. Liu, L. Liu, J. Edgar, E. Payzant, J. Hayes, and M. Kuball, M RS Internet
J. Nitride Semicond. Res. 6, 5(2001)
139. D. Zhuang, J. H. Edgar, B. Liu, H. E. Huey, H. X. Jiang, J. Y. Lin, M.Kuball, F.
Mogal, J. Chaudhuri, and Z. Rek, J. Cryst. Growth, 262, 168 (2004)
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O ....
=555555555555555^ 555^5555^
Reduction
Oxidation
Figure 1 Etching mechanism o f photo-assisted etching
Figure 2 (a) Cross-sectional TEM multi-beam image o f a KOH-etehed sample
(30 min, [KOH] = 7.1 M at 40 °C). Note that the pyramidal nanostructures are not
associated with structural defects, (b) High-resolution eross-seetional TEM image of
the same sample (after [27])
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3 SEM image o f GaN sample taken after etching in 2 M KOH at 90 °C for 45
min. Hexagonal pyramids were formed in the N-polar region while the surface o f the
Ga-polar region remained smooth and intact (after [28])
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ncpt^ivdy diaj»sd
liT C I^ d a i ^ m | rbcm fi
ilfan
1'
psoil
#c
m
Cteimkd aslsotpdonofOH*
ir
m
P n im iitls M J o f O a ,C J |_ .j, l i w l w t v e n j
OijO,, (snhel ia «liaili)
OH-
w
Rammtofa)
(d)
Figure 4 Schematic diagrams o f the cross sectional GaN film viewed along the
1120
direction for N-polar GaN to explain the mechanism o f the polarity selective
etching, (a) nitrogen terminated layer with one negatively charged dangling bond on
each nitrogen atom (b) adsorption o f hydroxide ions (c) formation o f oxides ( d )
d i s s o l v i n g t h e o x i d e s ( a f t e r [ 39 ] )
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5 SEM image o f a wall o f a pit obtained after etching: (a) in pure hot H 3 P O 4
and (h) in hot H 3 P O 4 with added A1 ions. In the latter case, no terraces can he seen
(after [47])
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure
6
1
(a) 1013
SEM images o f crystallographic surfaces o f GaN made by wet etching.
i plane etched by H PO . (b) Undercut i 1012 i plane etched by H PO . (c)
3
4
3
Vertical jlO lO j plane etched by KOH in ethylene glycol, (d) Undercut
4
plane
etched by molten KOH (after [48])
6.0x1O'*
<
0.0
I
-e
-
2
0
2
Voltage(V)
Figure 7 The comparison o f I-V characteristics o f metal contacts on p-GaN.
Dashed line: metal contact on the as-annealed surface o f p-GaN. Solid line:
metal contact on the etched surface o f annealed p-GaN (after [43])
106
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
•
■
f
♦
•
2 .1
2J
3J
5 .1
12
3J
wt dapasltefl
7OT “G
MO’C
HSW'C
14
3 .S
I M M i K '}
Figure 8 Etch rate o f AIN in KOH as a function o f annealing temperature (after [62])
Figure 9 Bulk AIN crystals before (a) and after (b,c,d) etching in KOH aqueous
solution, (a) before etching; (b) after etching; (c) enlarged image o f circle in (a) after
etching; and (d) enlarged image o f circle in (b) after etching (after [72])
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 10 Nitrogen polar AIN etched in 45% KOH for 10 minutes (after
[72])
Figure 11 GaN sample etched by PEC in 45% KGHiHiG (1:3) for 5 minutes. Total
etched depth is ~2 pm. The thin upper layer is the Ti mask (after [75])
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UV l l i J t M i M i k m
T r f li
\
M t|A t«kilii
KOM»luli«
Figure 12 Schematic o f PEC etching apparatus (after [76])
-r-:
Fig.ure 13 Highly anisotropic GaN etch profile. Etch depth is 3.5 pm (after
[76])
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 14 SEM micrograph o f GaN features smoothly etched to a depth o f 400 nm
(after [91])
Figure 15 SEM micrograph o f etched surface adjacent to mesa region. A transition is
observed between smooth and rough surface morphologies due to local variations in
solution concentration (after [91])
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 16 SEM micrographs o f etched n-on-i GaN sample by dopant-type selective
PEC etching (after [99])
Figure 17 (a) p-on-n GaN sample following CAIBE etching o f 4 pm diameter column,
(b) undercutting o f ;?-type layer by PEC etching (after [99])
111
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 18 Scanning electron micrograph o f a PEC etched GaN surface after 15 min of
etching. The distribution o f whiskers in this image appear to form a cellular structure
(after [112])
Figure 19 (a) Low-magnification cross-sectional transmission electron micrograph of
the etched GaN film. Visible in this image are the etched whiskers as well as the
dislocations in the underlying, unetched GaN film, (b) High-magnification crosssectional transmission electron micrograph showing the propagation o f dislocations
from the unetched GaN film into the etched whiskers. Both mixed (m) and edge (e)
dislocations are associated with whisker formation in this image (after [112])
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
400
U
H
A04
320v
Liquid
3 00
^ ^ \ k O H ) + Liquid
YT^ll.a
200
,37.3
(KOH) + Liquid
Liquid + NaOH(s)
Jl.j
17q"
(KOH]+ NaOH(s)
100
0
10
20
30
40
60
60
70
80
90
100
Mole % NaOH
Figure 20 Phase diagram o f KOH/NaOH eutectic alloy (after
[131])
Figure 21 Cross-sectional TEM images near the j^lOl 0
zone axis: (a) bright field
image with g= <0000> (b) dark field image with g= <0002> , and (c) dark field image
with g =< 1120 > . The etch pit originates from a dislocation (after [135])
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m
- \
Figure 22 TEM weak beam plan-view tilted image (g, 6g) with g = 1010
o fa G a N
heteroepitaxial layer etched in molten B at 200 °C for 3 minutes. (1) Eteh pit formed
on a nanopipe, (2) etch pit on a mixed-type dislocation (Burgers vector b component
perpendicular to c), (3) edge-type dislocation {b=a) on which no eteh pit has been
formed (after [74])
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m
Figure 23 AFM images (2x2 |rm^) o f the GaN samples etched by wet etching, (a)
Surface morphology o f the GaN sample after etching by molten KOH for 2 min at 210
°C. Etch pits are revealed on the surface with a density o f lO’ cm”^ . (b) Surface
morphology o f the GaN sample after etching by H 3 PO 4 for 6 min at 160 °C. The EPD
is the same found for the KOH-etched sample. The vertical scale ranges from 0 to 10
nm (after [32])
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IS
i
^
. «
t
•
15 |jm
i_
Figure 24 AFM image (15x15 |im^) of the GaN sample etched for 10 min at 200 °C
using H 3 PO 4 . Two different types o f etch pits with different sizes are revealed on the
etched surface. Altogether, we estimated the EPD to be
The vertical scale
ranges from 0 to 450 nm (after [32])
'
■20jJiP
,
..
.
'
•
■,».i
r-'jii
: iOfc-
Figure 25 DIG (a and b) and SEM (c and d) images o f dislocations revealed by E+M
etch on N-polar (a and c) and Ga-polar (b and d) surfaces o f GaN single crystal (after
[130])
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WD 6 . 7nun 5 - O O k V
x5 . Ok
lO u m
Figure 26 SEM image o f Al-polar AIN single erystal grown on Si faee 6H-SiC
substrate after etching in E for 2 minutes at 350 °C. Dislocation density = \ ( f cm~^
(after [137])
SX -19207 S.OkVI 3.8mm xlO.OkSE(M) 12/10/2002
'
'
'
' 5.(30um '
Figure 27 SEM image o f Al-polar AIN single crystal grown on Si face 6H-SiC
substrate after etching in E+M for 6 minutes at 350 °C. Dislocation
density= 9.43xl0^cw “^ (after [137])
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WD18 .4mm S.O O kV
k 8 00
Figure 28 SEM image o f Al-polar AIN single crystal grown in a microwave furnace
after etching in E for 3 minutes at 380 °C. Dislocation density = 6.06 x 10^ cm~^ (after
[137])
[00 .- 1]
[ 1- 1 .0 ]
g/
621 urn
Figure 29 SWBXT in transmission from the microwave sample, A=dislocation slip
line, g=038 reflection, wavelength=0.05049nm. Dislocation density = 5.99x lO^cw"^
(after [137])
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1. h i m
2 . O VkV xbOO
Figure 30 SEM images o f (a) nitrogen polar AIN crystal after aqueous KOH etching
and (b) nitrogen polar AIN crystal after etching in E for 1 minute at 300 °C (after
[137])
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1 GaN and AIN etching results in acid and base solutions. Etching was conducted
at room temperature (25 °C) unless otherwise noted, (after [23])
Etching solutions
Citric acid
Succinic acid
Oxalic acids
Nitric acid
Phosphoric acid
Hydrochloric acid
Hydrofluoric acid
Hydroiodic acid
Sulfuric acid
Hydrogen peroxide
Potassium iodide
2% Bromine/methanol
N-methyl-2-pyrrolidonone
Sodium hydroxide
Potassium hydroxide
AZ400K photoresist developer
Hydroiodic acid/hydrogen peroxide
Hydrochloric acid/hydrogen
peroxide
Potassium triphosphate
Nitric acid/potassium triphosphate
Hydrochloric acid/potassium
triphosphate
Boric acid
Nitric/boric acid
Nitric/boric/hydrogen peroxide
HCI/H 2 O 2 /HNO 3
Potassium tetraborate
Sodium tetraborate
Sodium tetraborate/hydrogen
peroxide
Potassium triphosphate
Potassium triphosphate/hydrogen
peroxide
GaN etch rate
nm/min
0 (75 °C)
0 (75 °C)
0 (75 °C)
0(85 °C)
0 (82 °C)
0 (80 °C)
AIN etch rate
nm/min
0 (75 °C)
0
0
0
0
0 (85 °C)
0 (82 °C)
0 (80 °C)
0
0
0 (82 °C)
0
0
0
0
0
0
0
0
0
0
50 (75 °C)
2,265
- 6 -1 , 0 0 0
0
0
0
0
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0 (75 °C)
0
0
0
0
0
0
0 (75 °C)
0 (75 °C)
oxide removal
0 (75 °C)
0
0
0 (75 °C)
0 (75 °C)
0
0
120
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Table 2 GaN surface roughness after etching performed by various groups
No.
RMS before
etching, nm
RMS after
etching, nm
Etch rate,
nm/min
Note
Ref.
1
1 0
60
50
contactless
[78]
2
29
27
13
contactless, far
from the
catalytic mask
[79]
13
contactless, far
from the
catalytic mask
[80]
2
contactless
[81,82]
3
29
2 0
4
5.3
6 .2
5
0.3
1.5
50
PEC
[90]
4.21
0.97
—
ultrasonic
treatment
[94]
—
—
—
post etching in
hot KOH
[95]
8
-3.5
3.5
up to 133
current
controlled PEC
[96, 97]
9
1700
2 0
400
external bias
[98]
6
7
Note: For no. 2 & 3, no RMS was mentioned for area near the mask. However, the
authors stated that the etch rates near the masks were about 40-50 nm/min and the rough
surface can be observed by SEM.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wet Chemical Etching of AIN Single Crystals
D. Zhuang', J. H. Edgar^, L. Liu^ B. Liu^ and L. Walker^
^Department o f Chemical Engineering, Kansas State University, Manhattan, KS 66506,
USA
^Oak Ridge National Laboratory, Metals and Ceramics Division, Oak Ridge, TN 37831,
USA
Abstract
Anisotropic chemical etching is an important means for characterizing the polarity
and defect density o f single crystals. In this letter, we present the results o f our studies on
the etching o f bulk AIN crystals in aqueous potassium hydroxide solution. The nitrogen
polarity ( 0 0 0 1 ) basal plane initially etched rapidly, while the aluminum polarity basal
plane and prismatic
/ - \
1100 planes were not etched. The etch rate o f the nitrogen polarity
V
y
basal plane eventually decreased to zero, as the surfaee beeame completely covered with
hexagonal hillocks that were bounded by
planes. The hillock density for the self­
seeded AIN crystals studied was typically in the range o f 5 x lO ’ cm“^ to lO^cw"^. From
our analysis o f etched AIN crystals, we infer that freely nucleated crystals predominately
have the aluminum to nitrogen direction pointing out from the nucleation surface, that is
the ends o f the AIN crystals facing the source are aluminum polarity.
1. Introduction
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
GaN has attracted vast interest due to its unique properties and potential
applications in optoelectronic and microelectronic devices. However, the dislocation
density in GaN heteroepitaxial layers as high as
[1] shortens the lifetime o f GaN
based devices. The chemieal compatibility and lattice/thermal expansion match between
AIN and GaN make bulk AIN single erystals potentially suitable for GaN epitaxial
growth. In addition, the high thermal conduetivity (340 W I m - K ) and high electrical
resistivity make AIN ideal for high power devices [2]. Recently, several groups [3] [4] [5]
have reported growth o f relatively large AIN single crystals, typically at millimeter
square scale, produced by the sublimation process, originating from Slack and McNelly's
work in 1970s [6 ]. Understanding the sublimation growth process and the quality o f the
AIN crystals it produces is a key to producing large, low defect density substrates suitable
for commercial device fabrication.
AIN has a polar wurtzite structure consisting o f closely spaced hexagonal layers,
alternating between cation (Al^"^) and anion (N^‘) layers stacked together along c axis [2].
Thus, the basal plane (i.e. the (0001) plane) can be either N- or Al-polarity. The polarity
o f AIN is important in controlling impurity incorporation and piezoelectric effects in
epitaxial GaN films. To date, few studies o f the polarity o f AIN crystals have been
reported. Polarity studies o f GaN, however, have been investigated for a long time using
many techniques. Heilman [7] summarized the conclusions o f GaN polarity studies from
various groups. Wet chemical etching is the most effective and economical method to
study the polarity o f group III nitrides. Heilman [7] found that the Ga-face o f GaN is
chemically more stable than the N-face, in that KOH solutions will etch the N-face, but
not the Ga-face. In case o f AIN, the same rule applied according to Schowaiter et a /’s
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
observation o f etching AIN in KOH solution [8 ]. Wet chemical etching is also an
effective means for defect study. By observing SEM images after etching, one can
estimate defect density in the crystal [9].
Studying the wet etching o f AIN epitaxial films on sapphire substrates in various
acids and bases, Pearton et al. [10] concluded that only KOH etches AIN appreciably at
room temperature, and the etch rate o f AIN films is a strong function o f crystal quality.
They inferred that armealing improves the quality o f heteroepitaxial AIN films, as their
etch rate was lower than that o f as-deposited films.
In this paper, we report oiu etching study o f AIN single crystals produced by
different sublimation growth methods, both self-seeded and seeded on 6H-S1C (0001)
substrates.
2. Experimental
We examined several crystals grown in different furnaces and crucible materials.
Sample A was prism-shaped needle grown in a graphite heating-clement furnace using a
NbC coated graphite crucible; sample B was a hexagonal platelet grown in the same
furnace with a plain graphite erueible; sample C was grown in a microwave-heated
furnace; and sample D was grown in a tungsten heating element furnace with a tungsten
crucible. Sample A, B, and C employed a self-seeding mechanism, while sample D was a
thick AIN film grown directly on a 6H-S1C (Si-face) substrate. Before etching, all
samples were cleaned by hydrochloric acid for ten minutes to remove any impurities on
the surface. To estimate the appropriate etching time for single crystals, we calculated
etch rate of a polyerystalline AIN sample under stirred condition, as function o f time by
measuring the mass and dimension changes due to etching. From this measurement, a
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
standard etching condition for single crystals was set o f 10 minutes at 60 °C in 45 wt%
KOH solution. After etching, all samples were rinsed in 38 wt% HCl solution for 5
minutes to neutralize the KOH residues.
3. Results and Discussions
The etch rate o f polyerystalline AIN as function o f time is shown in Fig. 1. Within
an hour, the etch rate decreased over 70% from 0.55 pm/min to 0.15 pm/min. Since the
solution was well stirred, we suspect that the decrease o f etch rate was due to the
depletion o f the easiest etched crystal planes, instead o f etchant depletion at the sample
surface.
SEM images o f sample A (before and after etching) are shown in Fig. 2. Clearly,
the planes perpendicular to the basal plane did not etch. Rapid etching was observed on
the basal (0001) plane, leading to the formation o f the hexagonal hillocks. By analogy to
the results reported for GaN [8 , 11], we conclude that this basal plane has a nitrogen
polarity. Etching also occurred on the crystal plane inclined at a smaller angle than 90°
from the basal plane (Fig. 2d). The hillock density for this crystal was approximately
5 XlO^cw”^.
Fig. 3 displays the etching effects o f sample B. Fig. 3a shows the AIN crystal
before etching; the images 3b and 3c are for 10 minutes etch, and all others are for
additional 20 minutes etch. Hexagonal hillocks were observed again on the (0001) basal
plane, shown in 3c and 3d. Fig 3e gives us an overview o f these hillocks at higher
magnification. Hillocks in 3c (10 minutes etching) are approximately 1 pm in diameters
and diameters o f hillocks in 3d (additional 20 minutes etch) are about 2 pm. Considering
the small crystal dimension and self-convection o f solution at elevated temperature (60
125
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
°C), we believe that depletion o f etchant was not significant, i.e., the concentration of
KOH did not change. Therefore, we conclude that the etch rate decreases with time, as
the area o f exposed (0001) planes diminishes toward zero. Another interesting
observation is that one side o f vertically placed (0 0 0 1 ) crystal was unchanged by etching
(Fig. 3f) and is thus most likely o f aluminum polarity, while the other side has hexagonal
hillocks on it (Fig. 3g and 3h) and is hence nitrogen polarity. This confirms that only a
single polarity o f the (0001) orientation is etched. The hillock density o f this crystal was
about lO^cm”^ .
For the crystals grown in the microwave furnace we observed the same etching
effect, as shown in Fig. 4. By measuring the triangle outlined on Fig. 4b, we found that
the angle between the basal plane and tilted plane is about 61.6°, which is close to the
angle between (0001) and
1101 j planes. According to ref. [12], jllO ljp la n e s are
energetically stable because o f smaller number o f bonds through these planes. Thus, we
believed that the hillocks were bonded by
planes. Fig. 5 shows SEM images o f
sample D, which should have Al polarity according to ref. [7]. As we see, no obvious
change was observed before and after etching. This further confirms our conclusion that
Al polarity is more inert than the N polarity.
4. Conclusions
For self-seeded AIN single crystals, the nitrogen polarity (0001) basal plane
initially etched rapidly, while the aluminum polarity basal plane and prismatic
f - \
1100
V
/
planes were not etched. The etch rate o f the nitrogen polarity basal plane eventually
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
decreased to zero, as the surface became completely covered with hexagonal hillocks
which were bounded by j l l O u planes. The hillock density for the self-seeded AIN
crystals studied was typically in the range o f 5 x lO ’ cw “^ to
cm~^. From our analysis
o f etched AIN crystals, we infer that freely nucleated crystals predominately have the
aluminum to nitrogen direction pointing out from the nucleation surfaee, that is the ends
o f the AIN erystals facing the source are aluminum polarity.
Acknowledgment
Supports from the Office o f Naval Research through grants No. NOOO14-02-10290 and research program sponsored by the Assistant Secretary for Energy Effieieney
and Renewable Energy, Office o f Transportation Technologies, as part o f the High
Temperature Materials Laboratory User Program, Oak Ridge National Laboratory,
managed by UT-Battelle, LLC, for the U.S. Department o f Energy under contract number
DE-AC05-00OR22725 are gratefully appreciated.
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reference:
1. Y. Park, J. Korean. Phys. Sac., 34, S I99 (1999)
2. O. Ambacher, J. Phys. D: Appl. Phys. 31, 2653-2710 (1998)
3.
M. Tanaka, S. Nakahata, K. Sogabe, H. Nakata, and M. Tobioka, Jpn. J. Appl. Phys.
36,L1062-L1064(1997)
4. Y. Shi, B. Liu, L. Liu, J. Edgar, E. Payzant, J. Hayes, and M. Kuball, MRS Internet J.
Nitride Semicond. Res. 6, 5(2001)
5. R. Schlesser, R. Dalmau, R. Yakimova, and Z. Sitar, Mat. Res. Sac. Symp. Proc., 693,
19.4.1 (2002)
6
. G. Slack, and T. McNelly, J. Cryst. Growth, 34,263-279 (1976)
7.
8
E. Heilman, M RS In te r n e t! Nitride Semicond. Res. 3, 11 (1998)
. L. Sehowalter, J. Rojo, N. Yakolev, Y. Shusterman, K. Dovidenko, R. Wang, I. Bhat,
and G. Slack, M RS Internet J. Nitride Semicond. Res. 5S1, W6.7 (2000)
9. P. Visconti, K. M. Jones, M. A. Reshchikov, R. Cingolani, H. Morkoc, and R. J.
Molnar,
10.
Phys. Lett., 77, 3532 (2000)
C. Vartuli, S. Pearton, J. Lee, C. Abernathy, and J. Mackenzie, J. Electrochem.
Yoc., 143(11), 3681 (1996)
11.
S. Krukowski, Z. Romanowski, I. Grzegory, and S. Porowski, J. Cryst. Growth,
189-190, 159(1998)
12.
S. Kitamura, K. Hiramatsu, and N. Sawaki, Jpn. J. Appl. Phys., 34, LI 184 (1995)
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.6
©
0.55
S am ple 1
S am ple 2
0.5
0.45
c
E
I
0.4
(]) 0.35
■g
UJ
0.3
0.25
0.2
0
10
20
30
Time, min
40
50
60
Figure 1 Etch rate o f polyerystalline AIN in 45 wt% KOH as function of
time, at 60 °C
f
a
SE
31-Aug-Ol
TO23.0mni 2 ,0 0 k V *120
Figure 2 Sample A before (a) and after (b,c,d) etching, (a) before
etching; (b) after etching; (c) enlarged image o f circle in (a) after
etching; and (d) enlarged image o f circle in (b) after etching
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3 Sample B before and after etching, (a) before etching; (b)
after etching; (c) higher magnification o f circle area in (a); (d) after
additional 20 minute etch; (e) hillock in (d); (f) one side o f
vertically placed crystal; (g) the other side o f vertical crystal; and
(h) higher magnification o f (g)
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4 Crystal produced by microwave, after 10 minute etching
SE
16-Jan-02
™ 4/.0 rt uTi 2 . 0 0 W x i a
2.binin
Figure 5 AIN crystal grown on SiC substrate for 10 minute
etching, (a) before etching; and (b) after etching
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Defect-selective Etching of Bulk AIN Single Crystals in Molten
KOH/NaOH Eutectic Alloy
D. Zhuang*, J. H. Edgar**, B. Strojek^, J. Chaudhuri^, and Z. Rek"*
^Department o f Chemieal Engineering, Kansas State University, Manhattan, KS 66506,
USA
^ Faculty o f Chemistry, Warsaw University o f Technology, Division o f Inorganical
Chemistry, Noakowskiego 3,02-036 Warsaw, Poland
^Department o f Mechanical Engineering, Wichita State University, Wichita, KS 67260,
USA
"^Stanford Linear Acelerator Center, 2575 Sand Hill Road, Mail Stop 69, Menlo Park, CA
94025, USA
Abstract
The effectiveness and reliability o f estimating the dislocation density in GaN thin
films and bulk crystals by defect-selective etching in eutectic KOH/NaOH have already
been successfully demonstrated. In this communication, we report the results o f applying
this technique to bulk AIN crystals. Etching produced hexagonal pits on the Al-polar
(0001) plane, while hexagonal hillocks formed on the nitrogen face. According to
synchrotron white beam X-ray topography (SWBXT) calibration, we believed that the
etching pits at Al polarity form primarily at dislocations. The optimized etching
Corresponding author Tel.; 785-5324320; fax: 785-5327372; email; edgarih@.ksu.edu: mail address;
Room 105, Durland H all, Manhattan, KS66506-5102
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temperature for Al-polarity is in the range o f 350-380 °C, which is typically 50-100 °C
higher than that for nitrogen polarity, indicating higher stability o f Al-polarity. For Alpolarity AIN single crystals grown on Si-face 6H-SiC (0001) substrates, the dislocation
density is about 10’ cm“^ and for self-seeded erystals, the dislocation densities for Alpolarity and the hillock densities for N-polarity are both on the order o f \0^cm~^. As far
as the dislocation density is concerned, self-seeded crystals have a better quality than
crystals grown on Si-face 6H-SiC (0001) substrates.
PACS: 61.72
Keywords: A l. Defect, A l. Etching, A l. X-ray topography, A2. Growth from vapor, B l.
Nitrides
1. Introduction
Bulk AIN single crystals are promising substrates for GaN heteroepitaxial growth,
due to its very good chemical compatibility and minimal lattice/thermal expansion
mismatch to GaN. AIN is also a good substrate for high Al-content AlGaN alloy growth,
where lattice match is a critical factor to avoid dislocations and cracks. In addition, one
can take advantage o f AlN ’s high thermal conductivity (340 W I m - K ) and high
electrical resistivity (>10“^ Q • cw) for high power devices fabrication [1].
A major objective in group III nitride crystal growth is to reduce the dislocation
density, to improve the material’s optical and electrical characteristics and device
performance. This has been most dramatically achieved with thin films by lateral
epitaxial overgrowth [2-4]. It has also driven the need for reliable techniques for
quantifying the dislocation density in GaN and related compounds. Transmission electron
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
microscopy (TEM) and atomic force microscopy (AFM) [5] are the established methods
for determining the dislocation density in single crystalline materials. However, TEM
requires arduous sample preparation and AFM requires a relative large and smooth
sample surface. Another technique for revealing the dislocation density in GaN crystal is
called photoelectrochemical etching (PEC) [6, 7]. Unfortunately, PEC is difficult to apply
to AIN due to its much larger band gap energy (6.28 eV) compared with GaN (3.4 eV).
The photo-generated minority carriers at the AIN surface do not have enough energy to
weaken the surface bond, which will enable dissolution o f the surface atoms by
electrolyte in the solution.
Weyher et al. [8-10] successfully applied orthodox etching in molten bases KOHNaOH eutectic (denoted as E) to estimate the dislocation density in bulk GaN single
crystals and its reliability has been confirmed by TEM calibration. For GaN, hexagonal
shape pits formed on Ga-polar surfaces after 1~5 minutes etching in a temperature range
o f 350~450 °C [9]. Round pits were formed on N-polar surfaces after 1-5 minutes
etching at 200-250 °C [9]. To overcome the difficulty o f over etching N-polarity crystals
when the temperature was set for Ga-polarity etching, Kamler et al. [10] added 10 wt%
MgO (denoted as E+M). This increases the etchant viscosity so it remains on the Gapolar surface and does not flow over the edges to the N-polar surface. The optimal
etching parameters depend on the crystalline state o f the materials, type and density of
defects, and the crystal polarity (group III element or nitrogen). AIN is expected to have
the same etching behavior due to its isomorphic structure (wurtzite) and similar chemical
bonding with GaN.
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In authors’ previous work, we inferred that in the self-seeding process, the crystal
facing the nucleation surface is predominately N-polar and in seeded growth on 6H-SiC
(Si face) substrates, the growing surfaee has Al-polarity [11]. The Al-polarity is more
inert than the N-polaiity in that the Al-polarity remains unchanged after etching in an
aqueous KOH solution (45 wt%) at 60 °C for ten minutes, while N-polarity surface forms
hexagonal hillocks after etching at the same condition; i.e. by etching crystals in aqueous
KOH solution, one can tell Al-polarity apart from N-polarity [11].
In this paper, we will present our etching study o f AIN single erystals in E and
E+M. The crystals were produced by different sublimation growth methods, both self­
seeded and seeded on 6H-S1C (0001) substrates. The goals were to validate the reliability
o f estimating the dislocation density by calculation from the etch pit density and to gain
better understanding o f the dislocation density in both self-seeding and seeded-growth
processes to further improve the crystal quality.
2. Experimental
According to references [9, 10], quite different etching parameters are necessary
to reveal defects on Ga- and N-polar surfaces. Thus, we need to identify the polarity o f
each sample before etching. Prior studies by our group showed that AIN crystals
produced from a tungsten-heating-element furnace by seeded growth on Si face 6H-SiC
(0001) substrate have Al-polarity on the growing surface [11, 12]. For all other samples
with unknown polarity, including self-seeded crystals, the crystals were first etched in an
aqueous KOH solution to determine the crystal’s polarity [11]. Table 1 summarized the
samples we examined and their etching conditions.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Before etching, all samples were cleaned by hydrochloric acid for ten minutes to
remove any impurities on the surface. This did not etch the AIN at all. The etching
parameters for Al-polarity crystals are typically 350-380 °C for 2 -6 minutes, depending
on the sample quality and/or defect distribution. Samples with N-polarity were etched at
300 °C in the time range o f 1-3 minutes. We took several scanning electron microscopy
(SEM) images to observe the features after etching. A synchrotron white beam X-ray
topography (SWBXT) image for sample C was taken to determine the reliability of
estimating dislocation density from the etch pit density. The molten eutectic was
contained in a platinum crucible held in an aluminum plate. The temperature was
stabilized by placing a quartz cover over the crucible. After etching, all samples were
rinsed in 38 wt% HCl solution for 5 minutes to neutralize any alkali residues.
3. Results and Discussions
During etching, bubbles (on the sample surface) and white floccules (in etchant)
were formed. The color o f the etchant changes from yellowish to grey. Therefore, we
presume that the etching occurred by the following reaction formula:
AI N + 6 K 0 H
A I { 0 H \ i +NH, ' t + ' i K ^ O i
(l)
3.1. Etching of Al-polarity
Hexagonal shape etch pits were clearly visible after etching an AlN/6H-SiC
(0001) sample (A) (Fig. 1). A linear array o f hexagonal pits is visible in the left-most
quarter o f the image, indicating these defects belong to the same grain. Randomly
distributed pits are also seen. All hexagonal pits had the same azimuthal orientation and
were approximately the same size. The calculated defect density was about 10’
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.
A second AlN/6H-SiC sample (B) was etched in E+M to avoid over-etching of
the other side o f the sample (N-polarity or the carbon-face o f the SiC substrate) (Fig. 2).
The hexagonal pits are about the same dimension compared with those in Fig. 1. The
defect density is about 9 .4 x l0 ^ c w “^ , which is in good agreement with previous
calculation.
For the self-seeded crystals grown in the microwave furnace (sample C) we
observed the same shape etching pits, as shown in Fig. 3. This sample is believed to have
a lower dislocation density, since self-seeded growth avoids stress-causing lattice and
thermal-expansion mismatch, plus the inheritance o f dislocations from those in the
substrates. Thus, we etched it at relative higher temperature. The etch pit density was
about 6 .0 6 x l0 ^ c ffj"\ To verify that these etching pits correspond to dislocations, a
SWBXT test covering the entire sample was performed, shown in Fig. 4. The dislocation
density data from SWBXT measurement (s.99x lO^cw"^) is in nice agreement with
calculated results from etching data. This indicates that for Al-polarity, the etch features
were primarily formed on dislocation, i.e. assessments o f dislocation density in bulk AIN
single crystals by orthodox etching in F and F+M are reliable and precise.
3,6. Etching of N-polarity
According to reference [11], self-seeded AIN crystals predominately have the Alpolarity pointing out to the source materials. Nevertheless, since nucleation and growth
occur randomly, some [0001] oriented crystals with the N-polarity are also formed. This
was verified by first etching the polyerystalline AIN in an aqueous KOH solution. Fig.
5(a) is SFM image o f one o f these areas after aqueous etching and Fig. 5(b) shows
pattems after etching in F. Crystals after etching in an aqueous KOH solution have
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hexagonal shape hillocks formed and these hillocks became bigger after etching in E.
This contradicts the results o f etching N-face GaN in E, where rounded pits were reported
[8-10]. Future research work is needed to give a reasonable explanation o f this difference.
To verify that the hillocks on sample D after etching in E (Fig. 5 (b)) were not due
to etching in the aqueous KOH solution, we repeated the same etching experiment on
another polyerystalline AIN (sample E). Fig. 6 (a) and (b) show the SEM images of
sample E after etching in aqueous KOH and E, respectively. In Fig. 6 (a), we see that the
left part o f the image was not attacked by aqueous KOH, while in Fig. 6 (b), hexagonal
hillocks were clearly visible in that area.
As expected, another self-seeded crystal (sample F) with N-polarity shows
hexagonal hillocks after etching in E, as shown in Fig. 7. The calculated hillock density is
about 4 .5 x l0 ^ c w ”^.
4. Conclusions
Hexagonal pits formed on Al-polar AIN (0001) crystals after etching in the
eutectic KOH/NaOH alloy. AIN crystals with N-polarity form hexagonal hillocks after
etching. Etching in E and/or E+M has been confirmed to be a reliable means for quick
assessment o f the dislocation density in Al polarity by comparing the calculated results o f
etching pattern with SWBXT scan. The optimal etching parameters depend on polarity;
defect type; density aiad distribution. The etching temperature for Al-polarity is in the
range o f 350-380 °C, which is typically 50-100 °C higher than that for N-polarity,
indicating higher stability o f Al face. From the dislocation density data calculated from
analyzed samples, we concluded that self-seeded crystals have lower dislocation density
than crystals seeded grown on Si-face 6H-SiC (0001) substrates.
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgment
Support from the Office o f Naval Research through grants No. NOOO14-02-10290 is gratefully appreciated.
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reference:
1. O. Ambacher, J Phys. D: Appl. Fhys. 31, 2653-2710 (1998)
2. Z. Yu, M. A. L. Johnson, T. McNulty, J. D. Brown, J. W. Cook, Jr., and J. F.
Schetzina, M RS Internet J. Nitride Semicond. Res., 3, 6 (1998)
3. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, H.
Kiyoku, Y. Sugimoto, T. Kozaki, H. Umemoto, M. Sano, and K. Chocho, Appl. Phys.
L e tt.,1 2 ,2 \ \ (1998)
4. P. Kozodoy, J. P. Ibbetson, H. Marchand, P. T. Fini, S. Keller, J. S. Speak, S.
P.DenBaars, and U. K. Mxsha., Appl. Phys. Lett., 73, 975 (1998)
5. D. Kapolnek, X. H. Wu, B. Keying, S. Keller, U. K. Mishra, S. P. Denbaars, and J. S.
S^Qck,Appl. Phys. Lett., 67, 1541 (1995)
6. C. Youtsey, L. T. Romano, I. Adesida, Appl. Phys. Lett., 73, 797 (1998)
7. C. Youtsey, L. T. Romano, R. J. Molnar, I. Adesida, Appl. Phys. Lett., 74, 3537
(1999)
8. J. L. Weyher, P. D. Brown, J. L. Rouviere, T. Wosinski, A. R. A. Zauner, I.
Grzegory, J. Crystal Growth, 210, 151 (2000)
9. J. L. Weyher, L. Macht, G. Kamler, J. Borysiuk, and I. Grzegory, Phys. Stat. Sol. C,
0(3), 821 (2003)
10. G. Kamler, J. L. Weyher, I. Grzegory, E. Jezierska, and T. Wosinski, J. Crystal
Growth, 246, 21 (2002)
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11. D. Zhuang, J. H. Edgar, L. Liu, B. Liu, L. Walker, M RS Int. J. Nitride Semicond.
Res., 7, 4 (2002)
12. Y. Shi, B. Liu, L. Liu, J. Edgar, E. Payzant, J. Hayes, and M. Kuball, M RS Internet J.
Nitride Semicond. Res. 6, 5(2001)
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1 Etching condition summary
Sample No.
T. °C
Time, min
Etchant
Polarity
Note
A
350
2
E
A1
Seeded growth
B
350
6
E+M
A1
Seeded growth
C
380
3
E
A1
Self-seeding
D
300
1
E
N
Self-seeding
E
300
2
E
N
Self-seeding
F
300
3
E
N
Self-seeding
VID 6 . 7 m m o . O O k V x 5 . Ok
l Oum
Figure 1 SEM image o f Al-polar AIN single crystal (sample A) grown on
Si face 6H-SiC substrate after etching in E for 2 minutes at 350 °C.
Dislocation density=10’ cm~^
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SX-19207 5,0kV13.8mm x10.0k SE(M) 12/10/2002
'
'
'
' S.OOum '
Figure 2 SEM image o f Al-polar AIN single crystal (sample B) grown on
Si face 6H-SiC substrate after etching in E+M for 6 minutes at 350 °C.
Dislocation density= 9.43 x 10^ cm~^
SE
07-O ct-02
W D 1 8 . 4 m m S . O O k V x80 0
50um
Figure 3 SEM image o f Al-polar AIN single crystal (sample C) grown in a
microwave furnace after etching in E for 3 minutes at 380 °C. Dislocation
density= =6.06xl0^cw ^
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[00 .- 1]
[ 1 - 1 .0 ]
'
”
-
621 um
'
g^
Figure 4 SWBXT in transmission from the sample C, A=dislocation slip line, g=038
refleetion, wavelength=0.05049nm. Dislocation density = 5.99xl0^cm~^
Figure 5 SEM images o f (a) nitrogen polar AIN crystal (sample D) after aqueous KOH
etching and (b) nitrogen polar AIN crystal (sample D) after etching in E for 1 minute
at 300 °C
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6 SEM images o f (a) nitrogen polar AIN crystal (sample E) after aqueous
KOH etching and (b) nitrogen polar AIN crystal (sample E) after etching in E for 2
minute at 300 °C
SE
07-Oc t- 02
W D l 9 . 2 m m IS .O kV x500
l OO um
Figure 7 SEM image o f nitrogen polar AIN single crystal (sample F) grown
in a microwave furnace after etching in E for 3 minutes at 300 °C
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
3 756 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа