Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Nanorod-based InGaN/GaN core-shell nanoLEDs
(USC Thesis Other)
Nanorod-based InGaN/GaN core-shell nanoLEDs
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
i
University of Southern California
Viterbi School of Engineering
Nanorod-based InGaN/GaN
core-shell nanoLEDs
Ph.D. Dissertation Submitted to the USC Graduate School
by: Yen-Ting Lin
Advisor: Prof. P. D. Dapkus
Department of Electrical Engineering – Electrophysics
May 2015
ii
Table of Contents
List of Figures ..................................................................................................................... v
List of Tables ...................................................................................................................... x
Acknowledgements ............................................................................................................ xi
Abstract .................................................................................................................... xii
1 CHAPTER 1: INTRODUCTION ............................................................. 1
1.1 The GaN semiconductor ............................................................................... 1
1.2 GaN/InGaN based LEDs .............................................................................. 2
1.3 Current challenges for GaN/InGaN-based LED .......................................... 2
1.3.1 Lack of suitable substrate for GaN growth ................................................... 3
1.3.2 Polarization and strain induced piezoelectric fields inside the multiple
quantum wells (MQWs) ................................................................................ 3
1.3.3 Efficiency droop ............................................................................................ 5
1.4 Motivation for nanorod-based LEDs ............................................................ 7
1.4.1 Nanoscale foot print to reduce density of threading dislocation .................. 7
1.4.2 Nonpolar sidewall to eliminate quantum confined Stark effect .................... 9
1.4.3 Tunable surface area of the LEDs ............................................................... 10
1.5 Thesis outline ..............................................................................................11
1.6 Chapter references ...................................................................................... 12
2 CHAPTER 2: SELECTIVE AREA GROWTH .................................... 14
2.1 SAG growth mechanism ............................................................................ 14
2.1.1 The mass transport model ........................................................................... 15
2.1.2 The kinetics model....................................................................................... 18
2.2 Sample preparation for SAG method ......................................................... 22
2.2.1 The Precursors for GaN growth ................................................................. 22
2.2.2 GaN bulk growth ......................................................................................... 23
2.2.3 SAG pattern definition ................................................................................ 23
2.3 GaN nanostructures grown by SAG ........................................................... 24
iii
2.4 Chapter references ...................................................................................... 27
3 CHAPTER 3: MECHANISM OF GAN NANOROD GROWTH BY
PULSED MODE ....................................................................................... 30
3.1 Experiments details .................................................................................... 31
3.2 Results and discussion ................................................................................ 33
3.2.1 Effect of N and Ga interruption duration ................................................... 33
3.2.2 Effect of NH3 flow rate and temperature .................................................... 36
3.2.3 Effect of hydrogen ....................................................................................... 37
3.2.4 Discussion ................................................................................................... 38
3.2.5 The mechanism of the pulsed mode growth ................................................ 41
3.3 Summary .................................................................................................... 43
3.4 Chapter references ...................................................................................... 44
4 CHAPTER 4: MECHANISM OF GAN NANOROD GROWTH BY
CONTINUOUS MODE ........................................................................... 46
4.1 Experiments details for continuous growth mode ...................................... 47
4.2 Results and discussion of mechanistic principles ....................................... 48
4.2.1 Effect of V/III molar ratio on nanostructure formation .............................. 48
4.2.2 Effect of growth temperature ...................................................................... 53
4.2.3 Discussion of the nanorod formation .......................................................... 54
4.2.4 Structure and optical characterization ....................................................... 58
4.3 The NH
3
-pulsed growth mode .................................................................... 60
4.4 Nanorod comparison by different growth mode ........................................ 65
4.5 Summary .................................................................................................... 67
4.6 Chapter references ...................................................................................... 68
5 CHAPTER 5: INGAN/GAN CORE-SHELL NANOLEDS .................. 70
5.1 MQWs on GaN nanorods ........................................................................... 70
5.1.1 Growth condition for MQWs on sidewall ................................................... 71
5.1.2 Indium incorporation on nonpolar and semipolar planes .......................... 73
iv
5.2 The growth of p-type GaN ......................................................................... 78
5.3 Contact for p-type GaN .............................................................................. 80
5.4 The core-shell nanoLEDs ........................................................................... 82
5.5 Summary .................................................................................................... 92
5.6 Chapter references ...................................................................................... 93
6 CHAPTER 6: CONCLUSION AND FUTURE DIRECTIONS ........... 95
6.1 Optimization of nanoLEDs ........................................................................ 96
6.1.1 Insertion of electron block layer ................................................................. 96
6.1.2 Number of MQWs pair and MQWs thickness ............................................. 96
6.1.3 Geometry of nanoLEDs............................................................................... 97
6.2 GaN nanosheet as solid state lighting template .......................................... 97
6.3 Ultra-thin nanorod template ....................................................................... 97
6.4 Chapter references ...................................................................................... 99
Bibliography 101
v
List of Figures
Figure 1-1. Type I heterostructure band alignment ............................................................. 2
Figure 1-2. the strain between GaN/InGaN heterostructure and the corresponding
band bending. Esp and Epe represent spontaneous and piezoelectric
electric field, which is parallel to the structure growing direction (c-
axis). Blue and red objects are electron and hole wavefunctions. .................. 4
Figure 1-3. Phenomenon of efficiency droop ..................................................................... 5
Figure 1-4. (a) Coherent and incoherent heterojunction. (b) Epilayer is compliant
to the substrate due to small contact area ........................................................ 8
Figure 1-5. Threading dislocation penetrates through the nanostructure and
bending toward the sidewall. .......................................................................... 9
Figure 1-6. GaN/InGaN heterojunction band structure for (a) polar and (b) non-
polar plane cross section. C-plane (0001) is polar plane and M-plane
{1-100} is nonpolar plane. E represents polarization electric field
(Esp + Epe). Esp and Epe represent spontaneous and piezoelectric
field, respectively. ......................................................................................... 10
Figure 1-7. Schematic diagram of 3D GaN core-shell nanoLED for one unit cell. ..........11
Figure 2-1. Gas-phase reactant concentration (normalized by the inlet
concentration) as a function of height above the disk for differing
value of the surface reaction probability γ[1]. .............................................. 17
Figure 2-2. Schematic of the computational domain and boundary conditions used
in the 2-D diffusion calculation[1]. ............................................................... 18
Figure 2-3. GaN SAG on (0001) GaN plane resulting in (a) complete hexagonal
pyramid shape as its equilibrium shape and (b) truncated hexagonal
pyramid as non-equilibrium shape. GaN SAG on (c) {11-20} plane
and (d) {1-100} plane showing difference in nanostructure shape
along <0001> and <000-1> directions[24]. .................................................. 21
vi
Figure 2-4. The 2-D v-plot for GaN across (a) (0001) and (b) {11-20} plane. The
Legendre transformation of 2-D v-plot across (c) (0001) and (d) {11-
20} plane resulting in the equilibrium shape as hexagon and arrow
headed, respectively, as indicated in blue outline[24]. ................................. 22
Figure 2-5. Procedure of sample preparation for GaN SAG. (a) SiNx mask and e-
beam resist (PMMA or ZEP) deposition. (b) Pattern defined by EBL.
(c) Pattern transfer by RIE. (d) Resulting GaN nanostructure grown
by SAG. ......................................................................................................... 24
Figure 2-6. Different orientation of GaN low index planes (a) polar c-plane, (b)
nonpolar a-plane, (c) semipolar a-plane, (d) nonpolar m-plane, and (e)
semipolar m-plane[27]. ................................................................................. 25
Figure 2-7. SEM images of samples with different nanopatterns before and after
SAG (a) dot pattern, (b) line pattern, (c) nanopyramid grown on dot
pattern, (d) nanostripe with triangle cross section grown on line
pattern (along [11-20] a-direction), (e) nanorod grown on dot pattern,
and (f) nanosheet grown on line pattern (along [11-20] a-direction). ........... 26
Figure 3-1. Schematic illustration of the steps per cycle in the pulsed mode
process. (a) nitrogen injection period, (b) nitrogen interruption period,
(c) gallium injection period, and (d) gallium interruption period. ................ 32
Figure 3-2. Effect of N interruption duration on morphology of the nanostructure:
(a) 0 s, (b) 1 s, (c) 2 s, and (d) 4 s. ................................................................ 34
Figure 3-3. Effect of Ga interruption duration on morphology of the nanostructure:
(a) 0 s, (b) 2 s, and (c) 8 s. (d) Lateral (m-plane) and vertical (c-plane)
growth as a function of Ga interruption duration. ......................................... 35
Figure 3-4. Rod morphology dependence on (a) NH3 flow rate and (b)
temperature. .................................................................................................. 37
Figure 3-5. Nanorod morphology evolution as a function of hydrogen injection
rate (a) 500, (b) 1000, (c) 1500, and (d) 2500 sccm. .................................... 38
Figure 3-6. Calculated Ga adatom desorption rate on c-plane and m-plane as a
function of temperature based on Choi’s study[18]. ..................................... 40
vii
Figure 3-7. Schematic diagram of the GaN nanorod growth process during each
step: (a) nitrogen source injection period, (b) nitrogen source
interruption period, (c) gallium source injection period, and (d)
gallium source interruption period. Large blue balls are Ga adatom
and small yellow balls are N adatom (color on-line). ................................... 43
Figure 4-1. Experiment set A: Nanostructure evolution with respect to NH3 flow
rate a) 150; b) 25; c) 10; and d) 5 sccm, which are taken under a
magnification of 50K. Inset figures are the lower magnification
images of the arrays at the corresponding NH3 flow rate. ........................... 49
Figure 4-2. Exp. set B: Rod morphology with V/III molar ratio equal to a) 12.6; b)
8.4; c) 4.2; and d) 2.1. (Figure 2c is taken at a 20 degree bird’s eye
view while the rest are at 30 degrees.) .......................................................... 51
Figure 4-3. Histogram of 70 nanorods’ profile distribution of a) normalized height;
and b) normalized width. c) SEM image of patterned substrate and d)
its opening size distribution. ......................................................................... 52
Figure 4-4. Bird’s eye view (20 degree) SEM image of Nanorod array grown by
continuous growth mode under different growth temperature a) 1050;
b) 1085; c) 1125 and (d) 1175°C. The nanorod profile evolution as a
function of growth temperature can be seen in the inset figures. ................. 54
Figure 4-5. TEM image of nanorod grown by continuous growth mode. The
nanorod is taken along <11-20> zone axies (Figure 4-5b), and the
diffraction pattern indicates a wurtzite structure (Figure 4-5c). The
nanorod is free from twins, stacking faults and threading dislocations
(Figure 4-5a). ................................................................................................ 58
Figure 4-6. a) CL spectrum of nanorod grown under 2 different V/III ratio (8.4
and 2.1 by changing TMG flow rate, both are un-doped). b) CL
spectrum of nanorod grown with and without Si-doped (V/III ratio is
2.1 in both case). c) CL spectrum of GaN substrate. d) PL spectrum
of continuous mode nanorods with InGaN MQWs. ..................................... 60
Figure 4-7. Growth process of NH3-pulsed mode where only the injection of NH3
is pulsed. ....................................................................................................... 61
viii
Figure 4-8. Nanorod profile as a function of NH3 flow rate and NH3 off time (8
seconds per cycle). ........................................................................................ 62
Figure 4-9. a) Birds-eye view (45 degree) SEM image of nanorod array grown by
NH3-pulsed mode. b) and c) Normalized height/Width histogram of
nanorod grown by NH3-pulsed mode (NP-mode, un-doped) and
continuous mode (C-mode, V/III = 2.1, un-doped). ..................................... 64
Figure 4-10. CL spectrum of nanorods grown by a) pulsed; b) NH3-pulsed; and c)
continuous (V/III = 2.1) mode. ..................................................................... 66
Figure 4-11. PL spectrum of InGaN MQWs grown on different mode nanorod
templates by a) pulsed; b) NH3-pulsed; and c) continuous (V/III = 2.1)
mode. ............................................................................................................. 67
Figure 5-1. QW emission PL spectrum as a function of (a) growth time, (b)
growth temperature, and (c) TMI flow rate. ................................................. 72
Figure 5-2. (a) A single nanorod is confined by the polar c-plane [0001], nonpolar
m-plane {1-100}, and semipolar m-plane {1-101}. (b) c-plane is
eliminated from a nanorod. ........................................................................... 74
Figure 5-3. (a) QWs grown on a nanorod. The white box indicates the mapping
region. (b) The peak intensity and wavelength distribution. (c) CL
signal at various wavelengths (360 – 550 nm). (d) CL signal at
various wavelengths (360 – 550 nm), normalized to each wavelength. ....... 74
Figure 5-4. (a) 6 points on a single nanorod with MQWs are scanned. (b) CL
spectrum of points 1 – 3. (c) CL spectrum of points 4 – 6. The
emission from the pyramid region has a longer peak wavelength due
to a higher indium incorporation rate and much lower intensity due to
lower growth rate, compared with emission from the sidewall. ................... 78
Figure 5-5. Blue LED grown in our reactor with (a) EL at 10 mA current injection,
and (b) its IV curve. ...................................................................................... 79
Figure 5-6. Cross section SEM image of (a) p-type GaN on undoped GaN and (b)
p-type shell on n-type core nanoLED. The p-GaN region is brighter
compared with undoped or n-doped GaN. .................................................... 80
ix
Figure 5-7. The band alignment of the ITO/p-GaN junction (a) without and (b)
with the p+ GaN layer. .................................................................................. 81
Figure 5-8. Cross section illustration of nanoLEDs (a) schematic diagram and (b)
SEM image. The 1st p-GaN layer is not grown in this particular
device shown in (b), and the MQWs cannot be identified at this
magnification. ............................................................................................... 83
Figure 5-9. 3 different structures of p-layer on the nanoLEDs. (a) – (b) sample
5D-1, a single p-GaN layer is grown after the MQW growth. (c) – (d)
sample 5D-2, an unintentionally-doped GaN layer is selectively
grown on the pyramid region by pulsed mode right after MQW
growth. (e) – (f) sample 5D-3, a p-GaN layer is selectively grown on
the pyramid region by pulsed mode right after MQW growth, and
then a second p-layer is grown by continuous mode. ................................... 85
Figure 5-10. (a) IV curve of 3 different sturctures (5D-1 – 5D-3) shown in figure
5-9. (b) EL of 5D-3. ...................................................................................... 86
Figure 5-11. NanoLED insulation test. (a) pyramid region is insulated by 200 nm
SiO2, (b) sidewall is insulated by 100 nm SiNx. (c) IV curves of
sample 5D-1 and 5E-1 – 3. (d) EL of sample 5E-3. ...................................... 88
Figure 5-12. (a) IV curves of samples 5F-1 – 4. (b) – (d) EL image of sample 5F-
1 – 3: (b) 5F-1, (c) 5F-2, and (d) 5F-3. ......................................................... 90
Figure 5-13. (a) EL spectrum of sample 5F-1, (b) CL spectrum of 5F-1, (c) EL
spectrum of sample 5F-3, and (d) CL spectrum of sample 5F-3. For
CL spectrum, 10 different points on the sidewall of a single nanorod
with MQWs are scanned from top to bottom, similar to that in Figure
5-4 (a). Inset of 5-13(b): CL spectrum of pyramid region of sample
5F-1. .............................................................................................................. 91
Figure 5-14. CL spectrum of sample 5G. QWs are grown at 0.5 Å/s with thickness
equal to 5 nm. The FWHM is measured to be 36 nm. .................................. 92
Figure 6-1. Ultra thin GaN core with InGaN buffer layer. ............................................... 99
x
List of Tables
Table 2-1. Growth conditions for GaN bulk template. ..................................................... 23
Table 3-1. Experiment sets and the corresponding controlled parameter. ........................ 32
Table 4-1. Experiments details of experimental sets A, B, and C. .................................... 47
Table 4-2. Statistical data of nanorods in figure 4-2 ......................................................... 50
Table 4-3. Comparison between nanorods grown by different growth modes. ................ 65
Table 5-1. Parameters (temperature, TMI flow rate and QW thickness) of QW
emission via PL study. .................................................................................... 72
Table 5-2. Parameters set for exp. 5E. .............................................................................. 87
Table 5-3. Parameters set for exp. 5F. ............................................................................... 89
xi
Acknowledgements
Many people and organizations got involved to the completion of this dissertation, and I
am truly thankful for all these efforts to accomplish this work.
I would like to express my appreciation to my thesis advisor, Prof. P. D. Dapkus, for his
constant guidance and support, and for providing me the opportunity to study this novel
LED structure which I truly believe has the potential to improve our society.
I would like to thank Prof. Chongwu Zhou, Prof. Aiichiro Nakano, Prof. Michelle
Povinelli and Prof. Wei Wu as my qualification exam and defense committees.
I would like to thank my CSL colleagues, Ting-Wei Yeh, Chun-Yung Chi, Maoqing Yao,
Yoshitake Nakajima, Ashkan Seyedi, Lawrence Stewart, Mitchell Dreiske, who all help
me and give me valuable advice on my thesis. I also thank Mitchell Dreiske and Patrick
Anderson who help me to revise my thesis.
I would like to thank Donghai Zhu for the clean room work and Jenny Lin and Eliza D.
Aceves for administrative supports.
I would like to thank Yuanjie Liang who is always supportive and considerate during my
Ph.D. life.
Finally I would like to show my appreciation to my parents, for their support, encourage,
accompany, and the way they teach me how to be not only a good man, but also a wise
man. You are the best parents.
xii
Abstract
GaN based blue LEDs are the foundation for solid state lighting, which has the potential
to approach the theoretical upper efficacy limit of around 500 lumens per watt. GaN is a
polar III-V semiconductor with a strong polarization field inside the material depending
on the crystal orientation. This strong polarization and piezoelectric field along the polar
direction (c-direction) will bend the band structure of the active regions of LEDs and
enhance the leakage current under forward bias if MQWs are grown on this polar c-plane.
However, the commercial GaN based LED is still grown on the polar c-plane since the
nonpolar substrate is highly cost inefficient due to the technology limitations. The
nanorod-based LED (nanoLED), which relies on polar c-plane, possess nonpolar m-plane
as their sidewalls and therefore are promising to resolve the current difficulties.
In this dissertation, we will discuss in detail the whole procedures involving the nanoLED
fabrication, which include substrate pattern preparation required for selective area growth,
the n-type nanorod templates, the active region (MQWs) growth, the p-type shell growth,
and the 3D p-type contact fabrication. Our prototype nanoLEDs successfully demonstrate
light emission under forward bias, but further optimization is still required to fully
employ the advantages of their nonpolar-based active region.
1
1 CHAPTER 1: INTRODUCTION
1.1 The GaN semiconductor
A Semiconductor is a material whose electrical conductivity can be controlled through
appropriate doping. Semiconductors are the foundation of modern electronics,
including transistors, light-emitting diodes (LEDs), laser diodes (LDs), solar cells,
quantum dots, as well as digital and analog integrated circuits
[1–7]
. Gallium nitride (GaN)
is a binary III/V semiconductor with wurtzite or zinc blende crystal structure and direct
electronic bandgap. Its wide band gap of 3.4 eV affords it special properties for
applications including bright LEDs, high-power and high-frequency devices. Single
crystal GaN film was first reported by Maruska and Tietjen in 1969
[8]
. N-type doping
of GaN is relatively simple and similar to other common semiconductor practices where
Si dopant atoms are used to replace Ga atoms in the crystal. However, people
experienced difficulty growing p-type single crystalline GaN until 1989, when Amano et
al were able to produce high quality p-type GaN film by low-energy electron beam
irradiation treatment on an as-grown p-type GaN layer
[9]
. They also demonstrated the first
p-n homojunction LED emitting in the range of UV and blue radiation in 1992
[10]
. Later
on, Nakamura et al applied a post-anneal treatment on as-grown Mg-doped p-type GaN
and successfully demonstrated efficient blue LEDs and laser diodes based on a
GaN/InGaN heterostructure. Annealing approaches have been widely adopted in the
fabrication of InGaN-based LEDs, both in academic and industrial pursuits.
2
1.2 GaN/InGaN based LEDs
Double-heterostructures serve as the basis for LED active regions due to a variety of
reasons. Firstly, GaN/InGaN heterostructures are classified as a type I (Figure 1-1)
heterojunction, where the material with a lower band gap between the P-N materials
confines the electron-hole carriers inside this region and increases the radiative
recombination rate. Secondly, the In
x
Ga
(1-x)
N alloy has a wide direct band gap range
starting from 0.7 eV (x = 1) and extending to 3.4 eV (x = 0), covering the entire visible
spectrum. As a consequence of these material advantages, GaN/InGaN based LEDs and
LDs have been widely studied and applied to a variety of applications.
Figure 1-1. Type I heterostructure band alignment
1.3 Current challenges for GaN/InGaN-based LED
In principle, it is possible to tailor the emission wavelength through the entire visible
spectrum by controlling the In fraction within In
x
Ga
(1-x)
N quantum wells. In practice,
however, there exist both intrinsic and technical difficulties that impede researchers from
pushing the emission wavelength longer than 550 nm (x = 0.27). Beyond this point, the
3
external quantum efficiency drops dramatically
[11]
. These difficulties will be discussed in
the following sections.
1.3.1 Lack of suitable substrate for GaN growth
Homoepitaxial growth of GaN wafers is still unavailable due to technology limitations.
Instead, foreign materials such as sapphire, silicon carbide (SiC), and silicon (Si) serve as
substrates for GaN bulk template growth. However, the lattice mismatch between these
foreign substrate and GaN (13% for sapphire, 3.4% for SiC and 16.9% for Si) introduce
large dislocation densities up to the order of 10
12
cm
-2
. Several techniques, such as lateral
epitaxial overgrowth (LEO)
[12,13]
, have been shown to reduce the dislocation density to
the order of 10
6
cm
-2
. Although GaN grown using these methods demonstrates better
crystal quality and meet the minimum requirements for laser and LED applications, these
techniques are usually very complicated. Simpler and more general methods are still
under investigation in order to achieve high quality GaN template growth.
1.3.2 Polarization and strain induced piezoelectric fields inside the multiple quantum
wells (MQWs)
In compound semiconductors there is a shift of charge from one atom to another due to
the differences in electron affinity. In unstrained zinc blende structures, the cation and
anion sublattices are arranged in such a way that there is no net polarization within the
material. On the other hand, the arrangement of these cation and anion sublattices in the
wurtzite crystal structure, such as GaN, will have a displacement relative to the ideal
wurtzite lattice position; this relative displacement produces a spontaneous polarization
within the crystal. In addition to spontaneous polarization, strain inside the wurtzite
4
crystal can cause a further shift between the cation and anion sublattices, creating a net
polarization referred to as the piezoelectric effect.
Due to the lattice mismatch between InGaN and GaN, a strain will inevitably be created
when a GaN/ InGaN heterostructure is formed. This strain will introduce a piezoelectric
field pointing toward the c-axis of the GaN, which is the direction normal to the Ga-face
(c-plane). All commercially available, relatively inexpensive GaN templates are grown
along this crystallographic direction. Under forward bias, the piezoelectric field inside the
MQW will bend the band structure and spatially separate the electron and hole
wavefunctions within the MQW (Figure 1-2) and, in turn, decreases the radiative
recombination rate.
Figure 1-2. the strain between GaN/InGaN heterostructure and the corresponding band
bending. Esp and Epe represent spontaneous and piezoelectric electric field, which is
parallel to the structure growing direction (c-axis). Blue and red objects are electron and
hole wavefunctions.
GaN templates grown along nonpolar planes are free from these polarization fields but
are currently very expensive and ill-suited for commercial purposes. Consequently, very
5
thin (around 3 nm) MQW InGaN structures are grown along the c-plane of GaN in order
to increase the overlapping of the electron and hole wavefunctions and to maintain a
reasonable volume of active region. Even though the MQWs strategy works relatively
well, some associated problems still exist, such as the blue shifting of emission caused by
the quantum confined stark effect under increasing current density. For comparison,
GaN/InGaN MQWs grown along the c-plane demonstrate a 14 nm peak wavelength shift
under PL study at different excitation powers, while MQWs on semi-polar m-plane {10-
11} or non-polar a-plane {11-20} only exhibit a 3 nm peak shift
[14,15]
.
1.3.3 Efficiency droop
For cost-efficient lighting applications, it is desirable to collect as much light as possible
from a small LED chip. In order to do this, we need to drive the LEDs at a relatively high
current. However, for InGaN based LEDs it is widely observed that the external quantum
efficiency (EQE) will first raise, and then drop as the current density is further
increased
[16]
(Figure 1-3). This well-known phenomenon, called “efficiency droop”, has
become one of the most important issues in the LED industry and must be overcome
before InGaN based LEDs can gain further relevance in daily lighting applications.
Figure 1-3. Phenomenon of efficiency droop
6
The origin of efficiency droop has been widely studied, and several mechanisms have
been proposed such as effects belonging to: (1) spontaneous reduction, (2) Auger
recombination, (3) carrier leakage
[16]
, and (4) Photon quenching
[17]
. In general, the total
LED injection current can be grouped into four parts
=
+
+
+
where I
SRH
, I
rad
, I
Auger
and I
leak
represent defect-related Shockley–Read–Hall current,
radiative recombination current, Auger recombination current and leakage current,
respectively. Typically these current components are modeled as an ABC formula, which
are
=,
=
,
=
where A, B, and C are the coefficients of radiative and non-radiative terms which can be
treated as constants at low injection levels. The internal quantum efficiency (IQE) is then
calculated as:
=
+
+
+
=
+
+
+
The external quantum efficiency (EQE) can then be obtained by scaling the IQE by an
optical extraction efficiency (EXE), which typically depends on the shape and the
refractive index of the device.
=
∗
The spontaneous reduction effect is related to the B coefficient such that B is no longer a
constant but approaches B = B
0
/[1+(n/n
0
)] at relatively high carrier injection levels. This
7
effect does not directly cause the efficiency droop but it lowers the barrier for non-
radiative processes to trigger the efficiency reduction. At high levels of current injection,
the Auger non-radiative effect becomes the dominant process since it is proportional to
the cubic power of carrier density. In addition, the fermi level of the quantum well rasies
in energy, offering an electron trapped inside the quantum well a higher probability of
escaping and contributing to leakage current. An electron blocking layer composed of
higher band gap materials such as AlGaN or AlInN is usually employed to reduce such
leakage current. The photon quenching effect recently proposed by Sarkissian et al
[17]
has a similar dependence on carrier density as the Auger recombination process and also
possesses a similar magnitude, making it difficult to distinguish between the two effects.
Therefore, great care must be taken while studying the efficiency droop mechanism and
methods for mitigating its effects.
1.4 Motivation for nanorod-based LEDs
Nanorods based GaN LEDs have gained substantial interest due to the following three
properties: (1) the small foot print can help reduce the threading dislocation density, (2)
the sidewalls composed of nonpolar {1-100} m-planes can reduce the quantum confined
Stark effect, and (3) the tunable surface area can effectively reduce the efficiency droop.
Consequently, such structures are great candidates for mitigating the current difficulties
observed in planar c-plane InGaN based LEDs
[3,5,18]
1.4.1 Nanoscale foot print to reduce density of threading dislocation
When a heterojunction is formed by two materials with different lattice constants, strain
is inevitably introduced. The strain energy will increase as the deposited layer becomes
8
thicker, while the crystal lattice will remain coherent (figure 1-4) until the deposited
thickness exceeds a certain value, called critical thickness. At this point, dislocations will
be introduced since they are now energetically more favorable, and the heterojunction
will no longer be coherent. These dislocations serve as non-radiative recombination sites
and reduce the LED quantum efficiency.
Figure 1-4. (a) Coherent and incoherent heterojunction. (b) Epilayer is compliant to the
substrate due to small contact area
However, it has been experimentally and theoretically shown that by reducing the
heterojunction contact area, the critical thickness of the deposited layer can be increased
due to a comparatively smaller strain volume
[19–21]
(figure 1-4 (b)). In addition, even if
the threading dislocation penetrates through the nanostructures, it will tend to bend
toward the sidewall and terminate
[22]
during the early stage of growth (figure 1-5).
(b)
9
Figure 1-5. Threading dislocation penetrates through the nanostructure and bending
toward the sidewall.
1.4.2 Nonpolar sidewall to eliminate quantum confined Stark effect
It has been shown that the quantum confined Stark effect (QCSE) induced by the
spontaneous and piezoelectric polarization fields will spatially separate the electron and
hole wavefunctions, reducing the radiative recombination efficiency. If we take
advantage of the sidewalls of GaN nanorods, which are composed of nonpolar m-planes
{1-100}, and grow the LED active region along these nonpolar planes, the polarization
field will be parallel to the quantum well planes and release the QCSE (Figure 1-6 (a),
(b)).
10
Figure 1-6. GaN/InGaN heterojunction band structure for (a) polar and (b) non-polar
plane cross section. C-plane (0001) is polar plane and M-plane {1-100} is nonpolar plane.
E represents polarization electric field (Esp + Epe). Esp and Epe represent spontaneous
and piezoelectric field, respectively.
1.4.3 Tunable surface area of the LEDs
Various mechanisms and several strategies have been proposed for resolving the
efficiency droop using electron blocking layers
[23]
or polarization matching methods
[24]
.
Another way to reduce this effect is to lower the injection current density by increasing
the LED active area. In this case, the 3D LED structure must be adopted in order to
increase the surface to volume ratio. The schematic diagram of one unit cell of this 3D
core-shell LED is shown in figure 1-7. In this structure, the MQWs are grown along the
n-type GaN nanorods, and the p-type GaN is grown over the MQWs to form a P-N
junction. The position of the contacts are illustrated within the figure.
(a) (b)
11
Figure 1-7. Schematic diagram of 3D GaN core-shell nanoLED for one unit cell.
Assuming the radius and height of the nanoLED are R and H, respectively, and the unit
cell of the nanoLED array is a square with pitch P (Figure 1-7), the active area ratio of 3D
LED to the planar LED is 2RH/P
2
. It is clear from this result that the active areas ratio is
proportional to R and H. For example, assuming that P = 750 nm and R = 200 nm, we can
grow a 4 um long rod to obtain a 10 time increase in active region area.
1.5 Thesis outline
The dissertation is organized in the following manner: in chapter II we first give a brief
introduction about the selective area growth (SAG), which is the main growth technique
adopted throughout the entire study. In this section we will also introduce the procedures
followed for sample preparation and characterization. In chapters III and IV we will
discuss the growth of n-type GaN nanorods, which serves as the basis for our nanoLED
structure. The growth conditions of GaN nanorods deviate significantly from those
typical of GaN bulk growth, and therefore, the growth mechanisms will be discussed in
12
detail. In chapter V we will present the growth of MQWs on the nanorod arrays and the
relations between the emission spectrum and growth parameters (including temperature
and well thickness). The performance of nanoLEDs will also be investigated and
compared to that of planar LEDs. Due to the unique geometry of nanoLEDs, some
additional issues will be encountered, such as large spectral blue shifts at different
excitation levels. Finally, there are several incomplete works that have been conducted
but were then suspended due to unexpected difficulties. These incomplete works are
documented in the last chapter as a reference for the next generation of researchers and
will hopefully lead to future progress in the field.
1.6 Chapter references
[1] Z. Zhong, F. Qian, D. Wang, C. Lieber, Nano Lett. 3, 343.
[2] S. D. Hersee, M. Fairchild, A. K. Rishinaramangalam, M. S. Ferdous, P. M. V . L.
Zhang, B. S. Swartzentruber, A. A. Talin, Electron. Lett. 2009, 45.
[3] T.-W. Yeh, Y .-T. Lin, L. S. Stewart, P. D. Dapkus, R. Sarkissian, J. D. O’Brien, B.
Ahn, S. R. Nutt, Nano Lett. 2012, 12, 3257.
[4] H. Sekiguchi, K. Kishino, A. Kikuchi, Appl. Phys. Lett. 2010, 96, 231104.
[5] Y . J. Hong, C.-H. Lee, A. Yoon, M. Kim, H.-K. Seong, H. J. Chung, C. Sone, Y . J.
Park, G.-C. Yi, Adv. Mater. 2011, 23, 3284.
[6] S. Gradečak, F. Qian, Y . Li, H.-G. Park, C. M. Lieber, Appl. Phys. Lett. 2005, 87,
173111.
[7] Y . Huang, X. Duan, Y . Cui, C. M. Lieber, Nano Lett. 2002, 2, 101.
[8] H. P. Maruska, Appl. Phys. Lett. 1969, 15, 327.
[9] H. AMANO, M. KITO, K. HIRAMATSU, I. AKASAKI, JAP ANESE J. Appl. Phys.
P ART 2-LETTERS 1989, 28, L2112.
13
[10] I. Akasaki, H. Amano, K. Itoh, N. Koide, K. Manabe, GaAs Relat. Compd. Conf.
1992, 129, 851.
[11] M. R. Krames, O. B. Shchekin, R. Mueller-Mach, G. O. Mueller, L. Zhou, G.
Harbers, M. G. Craford, J. Disp. Technol. 2007, 3, 160.
[12] X. Zhang, P. D. Dapkus, D. H. Rich, Appl. Phys. Lett. 2000, 77, 1496.
[13] B. a Haskell, F. Wu, M. D. Craven, S. Matsuda, P. T. Fini, T. Fujii, K. Fujito, S. P.
DenBaars, J. S. Speck, S. Nakamura, Appl. Phys. Lett. 2003, 83, 644.
[14] H. Yu, L. K. Lee, T. Jung, P. C. Ku, Appl. Phys. Lett. 2007, 90, 141906.
[15] D. F. Feezell, M. C. Schmidt, S. P. DenBaars, S. Nakamura, MRS Bull. 2011, 34,
318.
[16] J. Piprek, Phys. Status Solidi 2010, 207, 2217.
[17] R. Sarkissian, S. T. Roberts, T.-W. Yeh, S. Das, S. E. Bradforth, J. O’Brien, P.
Daniel Dapkus, Appl. Phys. Lett. 2013, 103, 041123.
[18] Y . Ra, R. Navamathavan, J. Park, C. Lee, Nano Lett. 2013, 13, 3506.
[19] S. Luryi, E. Suhir, Appl. Phys. Lett. 1986, 49, 140.
[20] R. People, J. C. Bean, Appl. Phys. Lett. 1985, 47, 322.
[21] H. Ye, P. Lu, Z. Yu, Y . Song, D. Wang, S. Wang, Nano Lett. 2009, 9, 1921.
[22] S. D. Hersee, A. K. Rishinaramangalam, M. N. Fairchild, L. Zhang, P. Varangis, J.
Mater. Res. 2011, 26, 2293.
[23] H. Zhao, G. Liu, R. A. Arif, N. Tansu, Solid. State. Electron. 2010, 54, 1119.
[24] J. Xu, M. F. Schubert, A. N. Noemaun, D. Zhu, J. K. Kim, E. F. Schubert, M. H.
Kim, H. J. Chung, S. Yoon, C. Sone, Y . Park, Appl. Phys. Lett. 2009, 94.
14
2 CHAPTER 2: SELECTIVE AREA GROWTH
While conventionally solid state device applications such as light emitting diodes (LEDs),
laser diodes (LDs) and solar cells (SCs) are grown monoepitaxtially or heteroepitaxially
on planar substrates, selective area growth (SAG) techniques have been adopted for more
advanced applications attempting to push forward the limitation of current technologies
over the last decade. SAGallows the crystal growth to take place only in selected regions,
by blocking the rest of the substrate surface via certain mask which is often composed of
amorphous and nonconductive material such as SiO
X
or SiN
X
. The SAG method allows
the formation of some structures with special geometry which are inaccessible in
conventional planar growth, such as nanorods, nanosheets, and high quality GaN with
very low dislocation density (~10
7
cm
-2
) via lateral epitaxial overgrowth (LEO). In this
chapter, we will discuss the SAG growth mechanism, the procedure of sample
preparation for SAG, and the nanostructures grown by this technique.
2.1 SAG growth mechanism
Both mass transport and kinetic models which are typically employed to explain the
planar growth situation are also adopted to describe SAG behavior. While the former is
usually viewed in a larger scale (micron size) and gives the estimation of the relative
growth trend, the later is responsible for the local scale (submicron – nanometer size) and
is used to determine the profile of the nanostructure.
15
2.1.1 The mass transport model
The rotating-disk reactor (RDR) is often used in metalorganic chemical vapor deposition
(MOCVD) of compound semiconductors. In a RDR, dilute reactants are introduced
through the showerhead at the top of the reactor at near room temperature by the carrier
gas which is usually N
2
or H
2
. The reactants are then drawn toward the surface by the
rotational motion of the heated susceptor. The incoming gas fluid obtains circumferential
and radial velocity components and the axial velocity drops from its original value to zero
at the surface as the nearby effects of the disk begin to dominate. The gas is heated due to
convective heat transfer from the disk. Depending on the chemical system, gas-phase
chemistry may take place either in the relatively cool incoming gas stream, or in the
heated region near the disk. The formation of the material at the disk surface removes
reactant species from the gas, setting up a concentration gradient between the incoming
gas stream and the disk surface, causing diffusive transport of reactant species toward the
surface. The changes in velocity, temperature, and concentration fields described above
all occur within a thin region just above the disk denoted the “boundary layer”.
Rigorously, each of these three fields has a different boundary layer thickness, but in
practice they are very similar.
In the typical MOCVD epitaxial growth, it is often the case that the growth rate is limited
by the supply of one of the reagent materials; any other reagents are available in much
greater excess. One can usually calculate reasonably accurate growth rates by treating the
growth as a first-order reaction of the limiting species that reacts at the surface with an
empirical “sticking coefficient” (probability), γ. The boundary condition on the reactant
species concentration at the surface is
[1]
16
=
=
= γ
π
, (1)
where the parameters are
G: molar growth rate
k: rate constant for the first order heterogeneous surface reaction
Cs: gas phase concentration of the limiting reactant species at the interface with the
surface
D: diffusion constant
C: gas-phase concentration of the reactant species
z: the height above the surface
R: gas constant
T: temperature in Kelvin
W: reactant species molecular weight
Transport through the concentration boundary layer is illustrated in figure 2-1
[1]
.
17
Figure 2-1. Gas-phase reactant concentration (normalized by the inlet concentration) as a
function of height above the disk for differing value of the surface reaction probability
γ
[1]
.
Eq. (1) discussed above only describes the growth behavior for case of a uniform thin
film, while in a typical SAG, portions of the surface are covered with a patterned mask
and the reaction can only take place within the designed opening region. Since there is no
reactant consumption on the mask region, an extra supply of precursors is available for
growth on the exposed zones, and this is the well-known growth-rate enhancement
phenomenon on the exposed areas. A schematic diagram describing the 2-D gas phase
diffusion model as well as the boundary condition is shown in figure 2-2, where J is the
source flux, C is the source concentration, δ is the height where C starts to drop linearly
(around 0.7 cm in figure 2-1), x
max
is the susceptor radius, and z
max
is the distance
between showerhead and the substrate.
18
Figure 2-2. Schematic of the computational domain and boundary conditions used in the
2-D diffusion calculation
[1]
.
The mass transport model has been applied to predict the growth behavior by SAG and
matches well with the experimental results qualitatively
[1–8]
. On the other hand, some
studies reveal that the growth of GaN stripes by SAG technique is almost independent of
growth conditions such as growth temperature, reactor pressure and a large range of V/III
molar ration
[9]
. In order to explain the growth behavior of GaN nanostructures in a more
general way, the kinetics model must be included.
2.1.2 The kinetics model
While mass transport model is used to describe the flux of source in gas phase and the
rate they impinge on the growing surface due to the concentration gradient, kinetics
model depicts a more localized view of the epitaxial crystal growth. The kinetics model is
based on the effect of physical and chemical reactions taken place on the vicinal growth
front as well as the surface reconstruction which will directly influence the behavior of
the adatoms that land on the crystal surfaces.
19
The crystalline and electronic properties of the material on the boundaries are quite
different from those of the bulk counterparts. The bulk crystal structure is decided by the
internal chemical energy of the atoms forming the crystal with a certain number of the
nearest neighbors, second nearest neighbors, etc. At the surface, the number of neighbors
is suddenly altered. Thus the spatial geometries which were providing the lowest energy
configuration in the bulk may not provide the lowest energy configuration at the surface.
Therefore there is a re-adjustment or “reconstruction” of the surface bonds to minimize
the surface energy
[10]
.
The most energetic favorable surface reconstruction pattern changes from plane to plane,
and usually differs under different circumstances and growth conditions. For example, on
a Ga-polar (0001) surface of GaN, 2x2 N adatom (H3) reconstruction is the most
favorable pattern under N-rich growth condition, on the other hand a Ga bilayer
reconstruction pattern dominates the surface under Ga-rich growth condition
[11]
. The
prediction of the surface reconstruction pattern is usually performed by ab-initio or first-
principle total energy calculations
[11–19]
, and the results are compared with the
experimental examination by reflection high energy electron diffraction (RHEED), low
energy electron diffraction (LEED), and scanning tunneling microscopy (STM) to verify
these assumptions and the accuracy
[15,17,20,21]
. Based on the surface reconstruction studies,
we are able to determine the favored diffusion path of adatoms on different surfaces, the
surface morphology and roughness, and the relative growth rate under certain growth
conditions.
For a 3-D crystal growth via SAG method, competition for reactant species will
inevitably take place among various planes, and therefore introduces anisotropic growth
20
behavior. From a thermodynamic point of view, the most stabilized profile of the crystal
formation will be the one which possesses the minimum total surface free energy
(assuming a constant volume). This shape can be expressed through a polar diagram in
which the length of the radius vector is proportional to the surface free energy and its
orientation is that of the surface normal. This theory is called the Wulff’s theory and the
corresponding plot is called the “Wulff’s plot” or “γ-plot”
[22–26]
. During the process of
crystal growth, thermodynamic equilibrium is usually not achieved and the “kinetic
Wulff’s plot” or “v-plot” are used to replace the Wulff’s plot or γ-plot, where the length
of the radius vector is now proportional to the growth rate of that facet instead of its
surface energy. Mostly these two quantities are related to each other. After performing a
Legendre transformation, by drawing a line through the point present on the v-plot and
perpendicular to the radius composed of that point and the origin, the inner-most region
that is bounded by all the drawn lines will be the final equilibrium structure under that
particular growth condition. The study of the v-plot is still an open issue due to the
complexity of the growth behavior, and therefore these plots are often times based on the
empirical results. Figure 2-3 show the GaN nanostructures grown by SAG method and
figure 2-4 show their corresponding v-plot and the Legendre transformation of the plot.
21
Figure 2-3. GaN SAG on (0001) GaN plane resulting in (a) complete hexagonal pyramid
shape as its equilibrium shape and (b) truncated hexagonal pyramid as non-equilibrium
shape. GaN SAG on (c) {11-20} plane and (d) {1-100} plane showing difference in
nanostructure shape along <0001> and <000-1> directions
[24]
.
22
Figure 2-4. The 2-D v-plot for GaN across (a) (0001) and (b) {11-20} plane. The
Legendre transformation of 2-D v-plot across (c) (0001) and (d) {11-20} plane resulting
in the equilibrium shape as hexagon and arrow headed, respectively, as indicated in blue
outline
[24]
.
2.2 Sample preparation for SAG method
2.2.1 The Precursors for GaN growth
In this study, trimethylgallium (TMG), triethylgallium (TEG), trimethylindium (TMI),
ammonia (NH
3
), disilane (Si
2
H
6
), and bis(cyclopentadienyl)magnesium (Cp
2
Mg) are
applied as the sources for Ga, In, N, n-type dopant, and p-type dopant, respectively.
23
2.2.2 GaN bulk growth
Due to technology limitations the GaN layers are usually grown on foreign substrate such
as silicon carbide (SiC) or sapphire (α-Al
2
O
3
). In this dissertation, sapphire is adopted as
the substrate for GaN layer growth. We employ a typical 2-step procedure including a
low temperature (LT) buffer layer and a high temperature (HT) layer to ensure high
quality of GaN layers that are then used as a template for the growth of nanostructures.
The growth conditions are listed in table 2-1. After the GaN template growth, its (002)
X-ray diffraction rocking curve is measured to determinee the crystal quality. Our GaN
template typically exhibits a 288 arcsec FWHM, which indicates that the GaN as well as
the reactor is in good condition. For n-type GaN, the molar ratio of n-type dopant to Ga is
kept at 0.31% and the concentration, mobility, and resistivity are typically 5.0×10
18
cm
-3
,
200 cm
2
/s V , and 0.00584 ohm-cm, respectively.
Layers Temperature
(setting, °C)
Pressure
(torr)
V/III molar
ratio
Thickness (nm)
LT 585 200 2800 25
HT 1160 200 1500 2000
Table 2-1. Growth conditions for GaN bulk template.
2.2.3 SAG pattern definition
After GaN template growth, the template is coated with a thin layer of SiN
X
(25~50 nm)
by PECVD at a deposition rate 10 nm/min, which serves as the mask for SAG as well as
the insulator for optical devices. ZEP or PMMA electron beam resists are then deposited
on the SiN
X
and electron beam lithography (EBL) is conducted to define the nanosize
pattern. After development, the pattern defined by EBL is transferred to SiN
X
mask by
24
reactive ion etching (RIE) at an etching rate of 120 nm/min. The pattern preparation
procedure is schematically shown in figure 2-5.
Figure 2-5. Procedure of sample preparation for GaN SAG. (a) SiNx mask and e-beam
resist (PMMA or ZEP) deposition. (b) Pattern defined by EBL. (c) Pattern transfer by
RIE. (d) Resulting GaN nanostructure grown by SAG.
2.3 GaN nanostructures grown by SAG
The samples prepared in 2.2.3 are ready for SAG. Patterns with different design (dot or
line pattern) and different orientation (along [11-20] or [1-100] direction) will influence
the final shape of GaN growth due to the different growth behavior of different planes, as
discussed in 2.1.2. In general, the low index planes such as the polar c-plane (Ga-face
(0001) or N-face(000-1)), nonpolar a-plane {11-20}, semipolar a-plane {11-22}, nonpolar
m-plane {1-100} and semipolar m-plane {1-101} are the most energetic favorable planes
that will be present at the end of SAG. The orientations of these planes are shown in
figure 2-6.
25
Figure 2-6. Different orientation of GaN low index planes (a) polar c-plane, (b) nonpolar
a-plane, (c) semipolar a-plane, (d) nonpolar m-plane, and (e) semipolar m-plane
[27]
.
It has been studied that by changing the growth temperature and pressure, the relative
growth rates among semipolar a-plane {11-22}, nonpolar a-plane {11-20} and polar c-
plane (0001) are tunable
[9]
. Therefore by manipulating the conditions such as temperature
and V/III molar ratio during the growth, we are able to control the growth rate of each
plane (with the same miller index), and hence determine the final shape of the
nanostructure required for special applications. For example, lateral epitaxial overgrowth
(LEO) of GaN nanostructure on line pattern along m-direction <1-100> via SAG is
widely adopted to produce high quality GaN with very low dislocation density (at a range
of 10
7
cm
-2
) which is suitable for GaN based laser applications
[28–32]
.
Conventionally GaN is grown under high V/III molar ratio (over 1000), and under such
growth conditions, the semipolar m-planes {1-101} have the slowest growth rate and will
typically define the shape of GaN nanostructures by confining them into pyramids on a
dot pattern or triangular stripes on line pattern (along a-direction [11-20]). However, if
GaN nanostructures are grown under low V/III molar ratio (around 10), the growth rate of
semipolar m-planes {1-101} will increase significantly and transform the shape of these
nanostructures from a pyramid to a rod (Figure. 2-7). The mechanism for this increase in
26
growth rate is the release of surface passivation effect resulting from H adsorption on the
semipolar m-planes {1-101} which prohibits their growth under high V/III molar ratio
growth conditions
[33,34]
. The nanorod growth mechanism will be discussed in the next
two chapters. SEM images of samples with different nanopatterns before and after SAG
are shown in figure 2-7 (a) dot pattern, (b) line pattern, (c) nanopyramid grown on dot
pattern, (d) nanostripe with triangle cross section grown on line pattern (along a-direction
<11-20>), (e) nanorod grown on dot pattern, and (f) nanosheet grown on line pattern
(along a-direction <11-20>).
Figure 2-7. SEM images of samples with different nanopatterns before and after SAG (a)
dot pattern, (b) line pattern, (c) nanopyramid grown on dot pattern, (d) nanostripe with
triangle cross section grown on line pattern (along [11-20] a-direction), (e) nanorod
grown on dot pattern, and (f) nanosheet grown on line pattern (along [11-20] a-direction).
(a) (b)
(c) (d)
27
Figure 2-7. Continued.
2.4 Chapter references
[1] M. E. Coltrin, C. C. Mitchell, J. Cryst. Growth 2003, 254, 35.
[2] T. Shioda, Y . Tomita, M. Sugiyama, Y . Shimogaki, Y . Nakano, Jpn. J. Appl. Phys. 2007, 46,
L1045.
[3] N. Dupuis, J. Décobert, P.-Y . Lagrée, N. Lagay, D. Carpentier, F. Alexandre, J. Cryst.
Growth 2008, 310, 4795.
[4] T. Shioda, M. Sugiyama, Y . Shimogaki, Y . Nakano, Phys. Status Solidi 2010, 207, 1375.
[5] H.-J. Chu, T.-W. Yeh, L. Stewart, P. D. Dapkus, Phys. Status Solidi 2010, 7, 2494.
[6] K. Choi, M. Arita, Y . Arakawa, J. Cryst. Growth 2012, 357, 58.
[7] T. Akasaka, Y . Kobayashi, S. Ando, N. Kobayashi, M. Kumagai, J. Cryst. Growth 1998,
189, 72.
[8] S.-M. Ko, J.-H. Kim, Y .-H. Ko, Y . H. Chang, Y .-H. Kim, J. Yoon, J. Y . Lee, Y .-H. Cho,
Cryst. Growth Des. 2012, 12, 3838.
[9] H. Miyake, A. MOTOGAITO, K. HIRAMATSU, Jpn. J. Appl. Phys. 1999, 38, 1000.
[10] J. Singh, Electronic and Optoelectronic properties of Semiconductor structures.pdf 2003.
[11] D. Segev, C. G. Van de Walle, Surf. Sci. 2007, 601, L15.
(e) (f)
28
[12] Y . Kangawa, T. Ito, a. Taguchi, K. Shiraishi, T. Ohachi, Surf. Sci. 2001, 493, 178.
[13] T. Akiyama, T. Yamashita, K. Nakamura, T. Ito, Jpn. J. Appl. Phys. 2009, 48, 120218.
[14] T. Ito, T. Akiyama, K. Nakamura, Semicond. Sci. Technol. 2012, 27, 024010.
[15] C. D. Lee, R. M. Feenstra, J. E. Northrup, L. Lymperakis, J. Neugebauer, Appl. Phys. Lett.
2003, 82, 1793.
[16] T. Yamashita, T. Akiyama, K. Nakamura, T. Ito, Jpn. J. Appl. Phys. 2009, 48, 120201.
[17] a. R. Smith, R. M. Feenstra, D. W. Greve, M.-S. Shin, M. Skowronski, J. Neugebauer, J. E.
Northrup, Appl. Phys. Lett. 1998, 72, 2114.
[18] J. E. Northrup, Appl. Phys. Lett. 2005, 86, 122108.
[19] J. Neugebauer, T. Zywietz, M. Scheffler, J. Northrup, H. Chen, R. Feenstra, Phys. Rev. Lett.
2003, 90, 056101.
[20] C. Cobet, T. Schmidtling, M. Drago, N. Wollschläger, N. Esser, W. Richter, R. M.
Feenstra, T. U. Kampen, J. Appl. Phys. 2003, 94, 6997.
[21] J. S. Brown, G. Koblmüller, F. Wu, R. Averbeck, H. Riechert, J. S. Speck, J. Appl. Phys.
2006, 99, 074902.
[22] D. Du, D. Srolovitz, M. Coltrin, C. Mitchell, Phys. Rev. Lett. 2005, 95, 155503.
[23] Q. Sun, C. D. Yerino, T. S. Ko, Y . S. Cho, I.-H. Lee, J. Han, M. E. Coltrin, J. Appl. Phys.
2008, 104, 093523.
[24] V . Jindal, F. Shahedipour-Sandvik, J. Appl. Phys. 2009, 106, 083115.
[25] Q. Sun, B. H. Kong, C. D. Yerino, T.-S. Ko, B. Leung, H. K. Cho, J. Han, J. Appl. Phys.
2009, 106, 123519.
[26] Y . Seo, C. Kim, Appl. Phys. Lett. 2010, 97, 101902.
[27] J. S. Speck, S. F. Chichibu, Nonpolar and Semipolar Group III Nitride-Based Materials.
MRS Bull. 2009, 34, 304–312.
29
[28] C. Liu, P. a. Shields, S. Denchitcharoen, S. Stepanov, a. Gott, W. N. Wang, J. Cryst.
Growth 2007, 300, 104.
[29] B. M. Imer, F. Wu, S. P. Denbaars, J. S. Speck, Improved quality (112??0) a-plane GaN
with sidewall lateral epitaxial overgrowth. Appl. Phys. Lett. 2006, 88.
[30] B. A. Haskell, T. J. Baker, M. B. McLaurin, F. Wu, P. T. Fini, S. P. Denbaars, J. S. Speck, S.
Nakamura, Appl. Phys. Lett. 2005, 86, 1.
[31] T. Koida, S. F. Chichibu, T. Sota, M. D. Craven, B. a Haskell, J. S. Speck, S. P. DenBaars,
S. Nakamura, Appl. Phys. Lett. 2004, 84, 3768.
[32] X. Zhang, P. D. Dapkus, D. H. Rich, Appl. Phys. Lett. 2000, 77, 1496.
[33] R. M. Feenstra, Y . Dong, C. D. Lee, J. E. Northrup, J. Vac. Sci. Technol. B 2005, 23, 1174.
[34] T. Akiyama, D. Ammi, K. Nakamura, T. Ito, Phys. Rev. B 2010, 81, 245317.
30
3 CHAPTER 3: MECHANISM OF GAN NANOROD
GROWTH BY PULSED MODE
As described in chapter I, n-type GaN nanorods will serve as the core of the nanoLED
structures to be studie3d in this dissertation. In order to achieve reproducible, highly
controlled nanorod arrays, it is necessary to understand in detail the growth mechanism
involved.
GaN nanorod arrays are typically grown by vapor-liquid-solid (VLS) growth
mechanism
[1–3]
, molecular beam epitaxy (MBE)
[4]
and metalorganic chemical vapor
deposition
[5–8]
. In the case of catalyst free metalorganic chemical vapor deposition
selective area growth
[9–11]
, GaN nanorods have been successfully grown on sapphire, N-
polar GaN
[6,8]
and Ga-polar GaN
[5]
substrates. It is known that the conventional growth
mode, the continuous source injection mode, or continuous mode, in which the nitrogen
and gallium precursors are injected at the same time typically forms a pyramid structure
bounded by the {1-101} semi-polar m-planes on the Ga-polar substrate. Hersee et al
[5]
introduced a pulsed source injection mode (pulsed mode) and successfully formed a rod
shape GaN nanostructure on the Ga-polar GaN substrate with {1-100} non-polar m-
planes forming the vertical sidewalls of the structure. The strong effect of the growth
mode on the structure morphology suggests that surface kinetics play an important role in
the growth on different GaN surfaces, and that an understanding of the growth
mechanism is important to control the growth of nanorod arrays. In this chapter we will
discuss the nanorod growth mechanism by such a pulsed mode in detail. We first present
experiments that show explicitly the dependence of the growth morphology on the
31
various portions of the growth cycle specifically focusing on the gallium and the nitrogen
interruption periods, the NH
3
partial pressure, the growth temperature, and the effect of
hydrogen. These experimental results are then discussed to provide a more quantitative
description of the pulsed growth process and the mechanisms that occur during each
phase of the growth cycle. Finally, conclusions are drawn from this discussion.
3.1 Experiments details
The GaN/Al
2
O
3
wafer is patterned with a hexagonal array of circular holes using electron
beam lithography and amorphous SiN
X
formed by plasma enhanced chemical vapor
deposition as a masking material. The nominal thickness of the SiN
x
mask and the
circular hole diameter are 25 nm and 100 nm, respectively. Detailed sample preparation
procedure is presented in chapter II. The GaN/Al
2
O
3
substrate is then loaded into a close-
coupled showerhead MOCVD system for GaN nanorod growth. Trimethylgallium (TMG)
and ammonia (NH
3
) are applied as the precursors for the nanostructure growth. The TMG
flow rate is maintained at 17.6 µmol/min and the NH
3
flow rate is varied from 67.0
mmol/min (1500 sccm) to 13.4 mmol/min (300 sccm) depending on the experiment type.
Nitrogen is used as the carrier gas. After growth, the morphology of the as-grown GaN
nanorods is studied by a high-resolution field-emission scanning electron microscope
(FE-SEM). The pulsed mode growth procedure follows Hersee’s original procedure
[5]
, as
schematically shown in figure 3-1: (a) nitrogen injection period, (b) nitrogen interruption
period, (c) gallium injection period, and (d) gallium interruption period. A total number of
120 cycles are applied in all of the experiment sets.
32
Figure 3-1. Schematic illustration of the steps per cycle in the pulsed mode process. (a)
nitrogen injection period, (b) nitrogen interruption period, (c) gallium injection period,
and (d) gallium interruption period.
We design five sets of experiments to investigate the influence of growth parameters and
each period of the pulsed mode growth. In each experiment set, one parameter is
manipulated while maintaining other parameters to be the same. In experiment sets I and
II, the nitrogen (N) and gallium (Ga) interruption duration are modified, respectively. In
experiment set III and IV , NH
3
flow rate and growth temperature are modified,
respectively. And in the last experiment, set V , the hydrogen injection rate is controlled.
The experiments and the corresponding controlled parameters are summarized in table 3-
1. Based on the result of these experiments, we will propose a growth mechanism in the
discussion section.
Exp. set Controlled parameter
I
N interruption duration (s)
0 ~ 4
II
Ga interruption duration (s)
0 ~ 8
Table 3-1. Experiment sets and the corresponding controlled parameter.
33
III
NH
3
flow rate (sccm)
225 ~ 1500
IV
Temperature (°C)
995 ~ 1070
V
H2 flow rate (sccm)
500 ~ 2500
Table 3-1. Continued.
3.2 Results and discussion
3.2.1 Effect of N and Ga interruption duration
Experiment set I is designed to study the influence of the N interruption duration on the
morphology of the nanostructure. The morphology of the nanostructures grown are
shown in figure 3-2 for N interruption times of (a) 0 s, (b) 1 s, (c) 2 s, and (d) 4 s, with
the Ga interruption duration maintained at 2 seconds. The temperature and the pressure
are set to be 1025°C and 250 torr in both experiment sets I and II. It can be seen that
without N interruption, the nanostructure that formed tends to be a pyramid-like structure,
the typical morphology formed under the continuous growth mode. As the N interruption
time is increased, growth along the [1-101] direction increases while the growth on the
m-plane is suppressed, and the nanorod morphology emerges as the dominant
nanostructure. When the N interruption duration is longer than 4 seconds, the
morphology of the nanorod has effectively no change up to 15 seconds, depending on
other parameters. Thermal decomposition effect is observed on GaN nanostructures at an
even longer interruption duration (not shown).
34
Figure 3-2. Effect of N interruption duration on morphology of the nanostructure: (a) 0 s,
(b) 1 s, (c) 2 s, and (d) 4 s.
Experiment set II studies the effect of Ga interruption duration on the morphology of the
nanostructure. The results of this study are shown in figure 3-2 (a) 0 s, (b) 2 s, and (c) 8 s.
The N interruption time is maintained at 4 seconds. Figure 3-2 (d) shows the relation
between the lateral (vertical) growth rates and the Ga interruption duration. (The extent of
the lateral growth is determined by subtracting the pattern diameter, observed by
fracturing the nanorods from the substrate, from the measured nanorod diameter.) The
data show that without a Ga interruption, strong lateral growth on the m-planes occurs.
Comparing the height of the nanorods in figures 3-3 (a)-(c) and the data of figure 3-3 (d),
it can be seen that the vertical growth increases when the Ga on the surface is allowed to
redistribute during a Ga interruption periods indicating that Ga is being supplied to the
surfaces by residual TMG in the reactor or that Ga inter-facet diffusion processes are
(a) (b)
(c) (d)
35
occurring. After 2 seconds of interruption the desorption of Ga from the c-plane begins to
overcome these inter-facet and / or residual gas supply processes and the vertical growth
declines slowly for longer Ga interruption times. The interruption time is essential,
however, to suppress lateral growth by allowing the Ga adatoms on the m-plane to desorb
and redistribute. Owing primarily to the occurrence of the desorption processes there is
an optimal Ga interruption time that clears the m-plane but does not suppress vertical
growth. The N and Ga interruption period seem to provide similar effects on the
morphology evolution of the GaN nanostructure for the first few seconds, however, their
mechanism are different in terms of the growth kinetics. This will be discussed later in
this section.
Figure 3-3. Effect of Ga interruption duration on morphology of the nanostructure: (a) 0 s,
(b) 2 s, and (c) 8 s. (d) Lateral (m-plane) and vertical (c-plane) growth as a function of Ga
interruption duration.
(d)
(a) (b)
(c)
36
3.2.2 Effect of NH3 flow rate and temperature
In order to investigate the N and Ga kinetics and precursor mass transport phenomenon,
experiment sets III and IV are designed to evaluate the influence of NH
3
partial pressure
and growth temperature on the morphology change of GaN nanostructure. For both
experiment sets, the N and Ga interruption period are maintained at 4 seconds and 1
second, respectively. The temperature is set to be 1025°C (thermocouple reading) as the
NH
3
flow is varied and the NH
3
flow rate is set at 1500 sccm as the temperature is varied.
Figure 3-4 (a) shows the effect of NH
3
flow rate on the aspect ratio of nanorod. As the
NH
3
flow rate decreases from 1500 sccm to 1000 sccm, the vertical-to-lateral aspect ratio
increases, indicating a longer diffusion length of Ga adatoms on the surface or a more
rapid relative desorption from the m-plane, reducing the probability of Ga adatoms to
react with an N adatom. This phenomenon is also consistent with the result discussed
elsewhere
[12]
. Due to the fact that NH
3
has a very low homogeneous thermal
decomposition efficiency and surface reaction rate, the growth rate of the c-plane starts to
decline significantly when the NH
3
flow rate is further reduced to 300 sccm.
From the thermodynamic point of view, the temperature is expected to have a strong
effect on the chemical and physical kinetics that affect the growth rate of each facet
which is beyond the scope of this study. Temperature influences surface chemical
reaction rates and both the adatom adsorption/desorption and diffusion behavior. Figure 4
(b) plots the vertical and lateral growth as a function of temperature. The data show that
the vertical growth rate is enhanced and the lateral growth rate is suppressed with
increasing temperature.
37
Figure 3-4. Rod morphology dependence on (a) NH3 flow rate and (b) temperature.
3.2.3 Effect of hydrogen
In the last set of experiment, the temperature, NH
3
flow rate, and source (both TMG and
NH
3
) injection duration are set to be 1040°C, 500 sccm and 4 second, respectively. The
hydrogen is injecting without interruption. Since it is known that hydrogen reacts with
GaN at typical GaN growth temperatures without NH
3
injection
[13]
, both Ga and N
interruption period are removed to shorten the duration when GaN is directly exposed to
hydrogen under growth temperature. The results are shown in figure 3-5, where the
hydrogen injection rates are (a) 500 sccm, (b) 1000 sccm, (c) 1500 sccm, and (d) 2500
sccm, respectively. As the hydrogen partial pressure increases (figure 3-5 (a)-(c)), the
aspect ratio of the nanostructure increases as well. This result is consistent with the work
of Bergbauer
[6]
which states that the presence of hydrogen can suppress the growth of the
m-plane. However when the hydrogen partial pressure is too high (figure 3-5 (d)), the
morphology of the nanorod evolves backward to the condition where the hydrogen partial
pressure is low. If an appropriate hydrogen partial pressure is maintained, the source
interruption period can be removed to simplify the pulsed mode process compared with
the regular pulsed mode procedure described earlier.
(b) (a)
38
Figure 3-5. Nanorod morphology evolution as a function of hydrogen injection rate (a)
500, (b) 1000, (c) 1500, and (d) 2500 sccm.
3.2.4 Discussion
Experiment sets I and II both indicate that the appropriate duration of N and Ga
interruption are essential to create conditions that favor a dominant nanorod morphology
during the pulsed mode growth. Though both experiments show a similar effect on the
morphology evolution of the nanostructure, their mechanisms are basically different. It is
well known that the vapor pressure of the nitrogen adatom is much higher than that of the
Ga adatom. This phenomenon can be confirmed by the fact that, ignoring the thermal
decomposition effect, the GaN nanorod morphology ceases to evolve for a N interruption
duration longer than 4 seconds; while it continuously changes with an even longer Ga
interruption duration. This implies that the mechanism operative during the N
(a) (b)
(d) (c)
39
interruption period is to flush the excess NH
3
inside the reactor before the injection of
TMG, avoiding the overlap of the NH
3
and TMG source, which contributes to a
continuous growth mode. In the continuous growth mode, the empirically determined
growth rates of possible low index planes can be represented by the kinetic Wulff’s plot
[14,15]
. Such plots indicate that over a wide range of growth conditions, the semi-polar m-
plane {1-101}, has the slowest growth rate and limits the shape of the structure to form a
hexagonal pyramid. The slow growth rate of these planes has been attributed to a
hydrogen-passivation effect, discussed elsewhere
[16,17]
, in which the hydrogen atom,
decomposed from NH
3
, bonds to the nitrogen atom on the semi-polar plane surface,
insulating it from the Ga adatom and hence suppressing its growth. Therefore, to prevent
the occurrence of the hydrogen-passivation effect, a minimum N interruption duration
will be required in a pulsed mode growth.
The presence of the N interruption period does not necessarily guarantee the nanorod
formation. As can be seen in figure 3-3 (a), with a 4 second N interruption duration, the
lateral growth of the nanostructure is still significant if there is no appropriate Ga
interruption duration. The vertical-to-lateral aspect ratio increases with the Ga
interruption duration, indicating an inter-plane diffusion of the Ga adatoms, and a longer
lifetime of these adatoms on the c-plane than on the m-plane. Choi et al
[18]
have studied
Ga adsorption and desorption from various GaN planes. In this work, they have shown
that Ga adatom adsorption/desorption behavior is considerably different between the c-
plane and m-plane. Figure 3-6 shows the calculated Ga adatom desorption rate on c-plane
and m-plane as a function of temperature based on their study. The desorption rate of Ga
adatoms on m-plane is about twice of that on c-plane. Bertness et al
[19]
studied the
40
mechanism of GaN nanowire growth in an MBE system and have reached a similar
conclusion - that the sticking coefficient of Ga adatoms on the c-plane is higher than that
on the m-plane.
Figure 3-6. Calculated Ga adatom desorption rate on c-plane and m-plane as a function of
temperature based on Choi’s study
[18]
.
In experiment sets III and IV , the N interruption duration is set to be long enough to
guarantee the absence of NH
3
inside the reactor. Therefore, the change of the NH
3
partial
pressure and growth temperature should directly relate to the Ga adatom kinetics. This
assumption is confirmed by both sets of experimental results. An increased vertical-to-
lateral aspect ratio of the nanorod grown at a lower NH
3
partial pressure is the evidence
of an increased diffusion rate of Ga adatoms
[12]
. As the temperature increases, though the
desorption rate of Ga adatom on both m-planes and the c-plane increase, the diffusion
process is accelerated as well. Due to the higher sticking coefficient of Ga adatoms on the
c-plane than on the m-plane, Ga adatoms that diffuse from m-plane to c-plane reside a
longer time, thus overcoming the effect of increased desorption rate. As a result, the net
vertical growth rate is enhanced.
41
In the last experiment, set V , the results show that if the presence of hydrogen is
maintained at a certain level, GaN nanostructure can still be formed into nanorod even if
the Ga and N interruption period are not introduced. The exact mechanism of how
hydrogen interacts with GaN growth is still unclear in this study, but its presence exhibits
similar effect of increasing temperature and lowering NH
3
during the nanorods growth
process to some extent. The H–H bonding in hydrogen (104 kcal/mol) is stronger than N–
H bonding in NH
3
(93.3 kcal/mol), hence it is believed that the hydrogen-passivation that
prohibits the nanorod formation under a growth condition with high V/III molar ratio
discussed earlier is caused by the hydrogen atom that released from NH
3
decomposition,
instead of hydrogen
[20]
. Besides, it has been reported that injection of hydrogen can
suppress the lateral growth of GaN nanorods grown on N-polar GaN substrate
[21]
.
Therefore, the presence of hydrogen at a certain level during the nanorod formation is
necessary to facilitate their vertical growth. On the other hand, if the hydrogen partial
pressure is too high, it might contribute to hydrogen-passivation effect on the semipolar
m-plane and therefore these planes emerge, as shown in figure 3-5.
3.2.5 The mechanism of the pulsed mode growth
Combining the analysis of the results of these experiments, we now propose the
mechanism for each step of the pulsed mode growth.
In the N injection period, NH
3
will react with Ga species on the various facets left from
the previous cycle and contribute to growth. To form a nanorod, it is necessary to operate
under conditions in which Ga adatoms are present on the c-plane but no longer reside on
m-plane in order to suppress lateral growth on the m-planes.
42
If TMG is injected immediately after the N injection period, continuous growth
conditions will occur owing to the residual NH
3
in the reactor that results from high
partial pressure of NH
3
typically applied in the GaN crystal growth. To avoid the
continuous growth mode, a minimum N interruption duration will be required to evacuate
the residue of NH
3
left from previous exposure.
After NH
3
is fully flushed, TMG is injected into the reactor. During this exposure process,
atomic Ga formed from homogeneous decomposition of TMG will adsorb on all exposed
planes
[22]
. If NH
3
is injected immediately after TMG injection, lateral growth will be
non-negligible owing to the presence of Ga adatoms on the m-plane. As a result, a Ga
interruption period is required to allow the selective desorption of Ga from the m-plane.
During the Ga interruption period, Ga adatoms that originally adsorbed on the m-plane
may diffuse to the c-plane or desorb from the m-plane facets, while those adsorbed on the
c-plane are expected to reside longer. In general, Ga has a longer residence time on the c-
plane than on the m-plane. Moreover, references
[16,17]
also indicate that under Ga rich
conditions, the Ga atom is able to replace the hydrogen atom that passivates the {1-101}
plane and create growth on that plane as well. The Ga interruption period is thus required
to redistribute the Ga adatoms that originally attach on the various planes expressed in the
nanostructure. If the appropriate interruption duration is applied, the growth on the m-
plane can be suppressed while the growth on the c-plane and the semi-polar planes
remain active, allowing the formation of the rod to be achieved.
Figure 3-7 is a schematic diagram of the GaN nanorod growth process during each step:
(a) nitrogen source injection period, (b) nitrogen source interruption period, (c) gallium
source injection period, and (d) gallium source interruption period.
43
Figure 3-7. Schematic diagram of the GaN nanorod growth process during each step: (a)
nitrogen source injection period, (b) nitrogen source interruption period, (c) gallium
source injection period, and (d) gallium source interruption period. Large blue balls are
Ga adatom and small yellow balls are N adatom (color on-line).
3.3 Summary
We have discussed the growth mechanism for the formation of GaN nanorods by a pulsed
mode MOCVD growth in this chapter. Experiments are also performed to support the
proposed mechanism. The suppression of lateral growth while maintaining the vertical
growth are necessary for nanorod formation. These two conditions can be achieved by an
44
appropriately controlled pulsed mode procedure. Due to the higher sticking coefficient of
Ga adatom on the c-plane than that on the m-plane, Ga adatoms possess a longer residual
time on the c-plane, promoting the vertical growth. The nanostructure morphology
evolution with respect to the change of the Ga interruption duration, NH
3
flow rate, and
the growth temperature indicate that the growth of nanostructure is mainly governed by
Ga adatom kinetic behavior. Since the kinetic behavior of the adatom varies under
different growth conditions, an appropriate Ga interruption duration corresponding to a
particular growth condition is essential to optimize the nanorod growth.
3.4 Chapter references
[1] X. Duan, C. Lieber, J. Am. Chem. Soc. 122, 188.
[2] T. Kuykendall, P. Pauzauskie, S. Lee, Y . Zhang, J. Goldberger, P. Yang, Nano Lett.
3, 1063.
[3] X. Zhou, J. Chesin, S. Crawford, S. Gradečak, Nanotechnology 2012, 23, 285603.
[4] M. Sanchez-Garcia, E. Calleja, E. Monroy, F. Sanchez, F. Calle, E. Munoz, R.
Beresford, J. Cryst. Growth 183, 23.
[5] S. D. Hersee, X. Sun, X. Wang, Nano Lett. 6, 1808.
[6] W. Bergbauer, M. Strassburg, C. Koelper, N. Linder, C. Roder, J. Laehnemann, A.
Trampert, S. Fuendling, S. F. Li, H.-H. Wehmann, A. Waag, Nanotechnology 21.
[7] R. Koester, J. S. Hwang, C. Durand, D. L. S. Dang, J. Eymery, Nanotechnology 21.
[8] S. F. Li, S. Fuendling, X. Wang, S. Merzsch, M. a. M. Al-Suleiman, J. D. Wei, H.-
H. Wehmann, A. Waag, W. Bergbauer, M. Strassburg, Cryst. Growth Des. 2011, 11,
1573.
[9] M. E. Coltrin, C. C. Mitchell, J. Cryst. Growth 254, 35.
[10] M. Akabori, J. Takeda, J. Motohisa, T. Fukui, Nanotechnology 14, 1071.
45
[11] H.-J. Chu, T.-W. Yeh, L. Stewart, P. D. Dapkus, Phys. Status Solidi 2010, 7, 2494.
[12] B. Imer, F. Wu, S. DenBaars, J. Speck, Appl. Phys. Lett. 88.
[13] E. V . Yakovlev, R. a. Talalaev, a. S. Segal, a. V . Lobanova, W. V . Lundin, E. E.
Zavarin, M. a. Sinitsyn, a. F. Tsatsulnikov, a. E. Nikolaev, J. Cryst. Growth 2008,
310, 4862.
[14] D. Du, D. Srolovitz, M. Coltrin, C. Mitchell, Phys. Rev. Lett. 95.
[15] V . Jindal, F. Shahedipour-Sandvik, J. Appl. Phys. 2009, 106, 083115.
[16] R. M. Feenstra, Y . Dong, C. D. Lee, J. E. Northrup, J. Vac. Sci. Technol. B 2005,
23, 1174.
[17] T. Akiyama, D. Ammi, K. Nakamura, T. Ito, Phys. Rev. B 2010, 81, 245317.
[18] S. Choi, T.-H. Kim, H. O. Everitt, A. Brown, M. Losurdo, G. Bruno, A. Moto, J.
Vac. Sci. Technol. B Microelectron. Nanom. Struct. 2007, 25, 969.
[19] K. a. Bertness, a. Roshko, L. M. Mansfield, T. E. Harvey, N. a. Sanford, J. Cryst.
Growth 2008, 310, 3154.
[20] E.-H. Park, J.-S. Park, T.-K. Yoo, J. Cryst. Growth 2004, 272, 426.
[21] W. Bergbauer, M. Strassburg, C. Kölper, N. Linder, C. Roder, J. Lähnemann, a
Trampert, S. Fündling, S. F. Li, H.-H. Wehmann, a Waag, Nanotechnology 2010,
21, 305201.
[22] Q. Chen, P. Dapkus, J. Electrochem. Soc. 138, 2821.
46
4 CHAPTER 4: MECHANISM OF GAN NANOROD
GROWTH BY CONTINUOUS MODE
In the previous chapter we have presented the formation of GaN nanorods by a pulsed
mode as well as their growth mechanism. Nevertheless, there are two difficulties
associated with such a pulsed process. First, there are in general four independent periods
in a single pulsed mode cycle (i.e. N (and Ga) injection and interruption period), which
significantly raise the complexity of the growth process. Second, it would be harder to
evacuate the NH
3
and TMG source during their interruption period required for nanorod
formation in a commercial MOCVD reactor with a large chamber size. Therefore, a
simpler process with a stable source injection control is preferred for GaN nanorod
growth.
In this chapter, GaN nanorod growth by a continuous growth mode as well as the
principles behind this formation will be discussed. In continuous mode, the N and Ga
sources are injected simultaneously and therefore avoid the difficulties encountered in
pulsed mode. In addition, a comparison will be made between nanorods grown under
different growth modes. This chapter is organized in the following manner: We first
designed a series of experiments to study GaN nanostructure formation and growth
behavior by manipulating the V/III molar ratio and the growth temperature. We then
investigate their formation mechanism and characterize the as-grown nanorods’ optical
response by cathodoluminescence (CL). Based on the characteristics of the pulsed growth
and continuous growth modes, we investigate a new growth mode, the NH
3
-pulsed
47
growth mode, to improve nanorods. Finally, a comparison of these different growth
modes and the properties of materials grown by each are discussed.
4.1 Experiments details for continuous growth mode
In the continuous growth experiments described here, we apply MOCVD growth using a
variety of growth conditions to selected area growth (SAG) on a nanopatterned substrate.
The sample preparation and the experimental procedures can be found in previous
chapters. Based on our observations in studying pulsed mode processes, we extended our
growth method toward continuous mode by decreasing the V/III molar ratio and
increasing the growth temperature compared to those we applied in the pulsed mode
growth studies. We designed 3 sets of experiments summarized in Table 4-1. The total
pressure is set to be 250 torr and nitrogen is used as the carrier gas. In all cases note that
the relative partial pressure of the V/III precursors is lower than typically employed in
MOCVD growth and in our previous pulsed mode growth studies. After growth, the as-
grown GaN nanorods are studied by a high-resolution field-emission scanning electron
microscope (FE-SEM) for their profile, transmission electron microscopy (TEM) for their
crystal structure, and cathodoluminescence (CL) for their optical properties.
Exp. set Temperature(°C) TMG carrier gas flow rate
(sccm
*1
)
NH
3
flow rate
(sccm
*2
)
A 1125 24 5 – 150
B 1125 8 – 48 5
C 1050 – 1175 36 5
*1
1 sccm of TMG carrier gas flow results in a delivery of 2.2 µmol/min. of TMG under
the conditions employed.
*2
1 sccm of NH
3
flow equals 44.66 µmol/min.
Table 4-1. Experiments details of experimental sets A, B, and C.
48
4.2 Results and discussion of mechanistic principles
4.2.1 Effect of V/III molar ratio on nanostructure formation
In experiment set A, the temperature and TMG flow rate are set at 1125°C and 24 sccm
respectively, while the NH
3
flow rate is varied from 150 sccm to 5 sccm. The
nanostructure morphology evolution as a function of NH
3
flow rate can be seen in
Figure 4-1 a) 150; b) 25; c) 10; and d) 5 sccm. The result shows that the nanopyramid
structure defined by {1-101} semipolar planes is formed when the NH
3
partial pressure is
high (Figure 1a)). The size of the nanopyramids is comparable to the original opening
diameter, indicating that the growth rate of semipolar m-plane {1-101} is extremely low
and limits the formation of nanostructures such as nanopyramids. The slow growth rate of
these planes has been attributed to a hydrogen-passivation effect,
[1,2]
where the
decomposition of NH
3
acts as the source of hydrogen atoms that attach to the surface
bonding sites. When the NH
3
injection rate is decreased to 25 sccm, we see that the
pyramids become larger and less symmetric, and some of the nanostructures become
tetrahedral in shape. This trend indicates that the {1-101} planes start to become unstable
at lower NH
3
partial pressure since the concentration of hydrogen surface species that
passivate the surface has been replaced by Ga and the growth of these planes is triggered.
At still lower NH
3
flow rates (10 sccm), the growth of the {1-101} planes increases and
rod formation is achieved. Finally, when the NH
3
flow rate is only 5 sccm, the rods
become thinner and shorter indicating an insufficiency of NH
3
. Meanwhile, the top of the
rod becomes flat indicating that the relative growth rate of {1-101} plane is even higher
than that of c-plane.
49
Figure 4-1. Experiment set A: Nanostructure evolution with respect to NH3 flow rate a)
150; b) 25; c) 10; and d) 5 sccm, which are taken under a magnification of 50K. Inset
figures are the lower magnification images of the arrays at the corresponding NH3 flow
rate.
In experiment set A, we found that a lower NH
3
injection rate not only promotes nanorod
formation but improves the uniformity of the nanostructure arrays as well. To further
investigate the effect of V/III molar ratio on the nanorod profiles, we maintain the NH
3
flow rate at 5 sccm and adjust the TMG flow rate from 8 sccm to 48 sccm during the
growth, as presented in experiment set B. Each sample is grown for a different duration,
scaling the TMG flow rate to achieve the same growth rate and a nominal height of 1 um.
After growth, the as-grown nanorod arrays are examined by FE-SEM and the rod profile
is characterized by image processing software to obtain the size of individual nanorods.
The growth time versus TMG injection rate and the statistical data on the physical
(a) (b)
(d) (c)
1 um 1 um
1 um 1 um
20 um 20 um
20 um 20 um
50
dimension of the nanorods (70 nanorods’ profiles are collected) is summarized in table 4-
2, the SEM images are shown in Figure 2 with a V/III molar ratio a) 12.6; b) 8.4; c) 4.2;
and d) 2.1 respectively. The histograms of the nanorod profile distributions are shown in
Figure 4-2 a) height and b) width. The SEM image of the patterned substrate and its
opening size distribution are also shown in Figure 3c and 3d.
TMG flow
rate (sccm)
V/III molar
ratio
Growth
time
(seconds)
Measured average
nanorod height (nm) /
standard deviation
Measured average
nanorod width (nm) /
standard deviation
8 12.6 1800 948 / 58.6 179 / 22.2
12 8.4 1200 874 / 43.6 166 / 21.0
24 4.2 300 1077 / 23.9 146 / 5.9
48 2.1 150 943 / 15.7 130 / 7.3
Table 4-2. Statistical data of nanorods in figure 4-2
51
Figure 4-2. Exp. set B: Rod morphology with V/III molar ratio equal to a) 12.6; b) 8.4; c)
4.2; and d) 2.1. (Figure 2c is taken at a 20 degree bird’s eye view while the rest are at 30
degrees.)
(a)
(c) (d)
(b)
2 um 2 um
2 um 2 um
Figure 4-3. Histogram of 70 nanorods’ profile distribution of a) normalized
normalized width. c) SEM image of patterned substrate and d) its opening size
distribution.
From the nanorod profile distribution shown in Figure 4
nanorod array uniformity improves with a lower V/III molar ratio. The nanorod array
uniformity is limited by the size distribution of the openings on the patterned substr
and therefore the array uniformity does not improve at an even lower V/III molar ratio
(a)
(c)
2 um
(c)
Histogram of 70 nanorods’ profile distribution of a) normalized height; and b)
normalized width. c) SEM image of patterned substrate and d) its opening size
From the nanorod profile distribution shown in Figure 4-3 we can clearly see that the
nanorod array uniformity improves with a lower V/III molar ratio. The nanorod array
uniformity is limited by the size distribution of the openings on the patterned substr
and therefore the array uniformity does not improve at an even lower V/III molar ratio
(b)
um
(d)
52
height; and b)
normalized width. c) SEM image of patterned substrate and d) its opening size
3 we can clearly see that the
nanorod array uniformity improves with a lower V/III molar ratio. The nanorod array
uniformity is limited by the size distribution of the openings on the patterned substrate
and therefore the array uniformity does not improve at an even lower V/III molar ratio
53
(2.1). Besides, a reduction of average nanorod width at a lower V/III molar ratio is
observed. This indicates that a lower V/III molar ratio promotes vertical growth of
nanorods and is consistent with the phenomenon observed in previous studies of the
pulsed growth mode.
[3]
4.2.2 Effect of growth temperature
It has been shown that the V/III molar ratio plays a critical role in nanorod formation in
the continuous growth mode. In exp. C we will study the effect of growth temperature.
The nanorod profile evolution as a function of growth temperature is shown in Figure 4-4
a) 1050; b) 1085; c) 1125; and d) 1175°C. When the temperature is low (Figure 4a),
though the nanostructure forms as a nanorod, the uniformity of the nanorod array is poor
with severe parasitic growth. In addition, the sharpness of the nanorod top indicates that
the growth rate of {1-101} is smaller than that of the c-plane, even though the V/III molar
ratio is sufficiently low to allow nanorod formation. As the growth temperature increases
(Figure 4-4b), the vertical-to-lateral aspect ratio of the nanorods becomes larger, the
nanorod array uniformity improves, and the amount of parasitic growth is reduced as well.
When the growth temperature reaches 1125°C (Figure 4-4c), we see a fairly uniform
nanorod array free of parasitic growth, and the flat top of the nanorod indicates at this
temperature, the growth rate of {1-101} has exceeded that of c-plane. Finally, when the
temperature is further increased to 1175°C (Figure 4-4d), thermal etching and Ga
desorption effects become much more significant and the nanostructure formation is
unstable under such growth conditions. Our results agree with previous experimental
reports,
[3–5]
where a lower growth rate of m-plane is observed at a higher growth
temperature.
54
Figure 4-4. Bird’s eye view (20 degree) SEM image of Nanorod array grown by
continuous growth mode under different growth temperature a) 1050; b) 1085; c) 1125
and (d) 1175°C. The nanorod profile evolution as a function of growth temperature can
be seen in the inset figures.
4.2.3 Discussion of the nanorod formation
As presented in exp. set A and B, two phenomena are observed in the nanostructure
evolution with respect to V/III molar ratio: the nanorod formation (as a consequence of
growth rate variation among the various planes) and the improvement of array uniformity.
As shown in Figure 4-1, the shape of the nanostructure evolves from a nanopyramid to a
nanorod, and the uniformity of the nanostructure array improves when the NH
3
flow rate
decreases from 150 to 5 sccm. These facts indicate that the formation of the nanostructure
is the result of not only the relative growth rate between various planes, but their absolute
growth rate as well. When NH
3
injection rate is high (150 sccm), only c-plane growth is
55
observed, which causes the structure to self-limit owing to the low growth rate of the {1-
101} planes, and the volume of the nanostructure is much smaller than that grown under
a much lower NH
3
flow rate (10 sccm). This phenomenon reflects the fact that the growth
of {1-101} plane is not only relatively low, but also absolutely low. The low growth rate
can be attributed to the so-called “hydrogen-passivation” effect
[1,2]
where the hydrogen
atoms tend to bond with the nitrogen atoms on the N-terminated surface, such as {1-101}
and (000-1) planes, therefore insulating these surfaces from growth ambient and in turn
slowing down and even stopping the growth of them. This is always noted in kinetic
studies of facet formation by a deep notch that always manifests itself in a kinetic Wulff’s
plot.
[4,6]
It is well known that the thermal decomposition process of NH
3
will release
hydrogen atoms that limit the growth rate on the {1-101} planes. Therefore, in order to
alleviate this effect, a very low V/III ratio must be applied. The alleviation of hydrogen-
passivation effect can be clearly observed if we compare the volume of the nanostructures
in Figure 4-1. The Ga incorporation rate under a 150 sccm injection of NH
3
is only a
small fraction (~ 4.3% in volume ratio) of that under a 10 sccm NH
3
injection. As the
NH
3
injection rate decreases from 150 to 25 sccm, the nanopyramid becomes larger and
the m-plane appears, which indicate that the growth rate of {1-101} plane increases due
to the reduced NH
3
flow. At this NH
3
injection rate we can see that the corresponding
nanostructure has a pyramid-like but irregular shape (Figure 4-1b). Also from the inset
figure, parasitic growth (which is absent under 150 sccm NH
3
flow rate, as shown in the
inset of Figure 4-1a) with a much larger size than that of the nanostructure itself can be
observed throughout the growth area. Since a reduced NH
3
flow rate will release parts of
the semipolar planes from hydrogen-passivation effect, these nanostructures grow much
faster than those with passivated planes and therefore the uniformity of the nanostructure
56
array becomes poor. As the NH
3
flow is further decreased from 25 to 10 sccm, we see
nanorod formation is achieved, which implies that the growth rate of {1-101} planes
becomes much faster than that of m-plane and is comparable to that of the c-plane.
Parasitic growth becomes denser and with smaller volume compared to those in Figure 4-
1b, which also indicates that growth of nanorods at growth sites becomes favorable and
the uniformity of the nanorod array becomes better as well. Finally, at a 5 sccm NH
3
injection rate, the hydrogen-passivation effect almost disappears due to a very low V/III
ratio, and the {1-101} planes become growth favorable planes so that their growth rate is
even higher than that of c-plane, as implied by the fact that the top of the nanorod
becomes flat. In addition, the field of nanorod array possesses a much better uniformity
as well as being free of parasitic growth. The control of the NH
3
flow rate not only
changes the nanostructure profile, but also influences their overall uniformity. Since at a
higher NH
3
flow rate (> 10 sccm), the hydrogen-passivation effect plays a dominant role
in the nanostructure formation, only part of the nanostructures are able to grow. The
random distribution of these favorable and unfavorable nanostructures therefore renders
poor uniformity to the overall array. When the NH
3
flow rate is sufficiently low (< 10
sccm) such that the nanostructures are totally free from the growth limitations on the {1-
101} planes imposed by the hydrogen-passivation effect, the growth of the nanostructure
array becomes mass transport-limited. Since the diffusion length of the mass transport
region (tens of microns
[7,8]
or even higher at higher temperature and lower V/III molar
ratio in our case) is much larger than the size of the nanorod and the pitch of the array, the
local nanorod array will have a much better uniformity than those grown under a kinetic-
limited condition. In short, a mass transport-limited condition, which can be achieved by
57
applying lower V/III molar ratio, is able to produce a nanorod array with better
uniformity.
Exp. set B provides a clearer view of how V/III molar ratio affects the adatom kinetics
during the growth and therefore the nanorod array uniformity. As discussed above, at a
lower V/III molar ratio the nanorods are grown under a situation away from the kinetic-
limited condition and the growth of {1-101} plane becomes highly favorable as this
crystallographic surface has much lower surface formation energy than {1-100} planes
and even the (0001) plane.
[2,9]
The high growth rate of {1-101} and (0001) planes not
only makes the growth conditions strongly mass-transport limited and therefore improves
the uniformity of nanorod array, but increases the vertical-to-lateral growth ratio of
nanorods as well.
The kinetics of growth were investigated by Lymperakis and Neugebauer
[10]
who
performed calculations that showed a large anisotropic diffusion path barrier for Ga
adatoms on the GaN m-plane. That is, the diffusion barrier along the <0001> direction is
much larger than that toward <11-20> direction. Therefore, some of the Ga adatoms
landing on the sidewalls of the nanostructures are unable to reach the top of them at lower
growth temperatures, which not only reduces the vertical growth but contributes to lateral
growth as well. Besides, as suggested by Sawicka et al,
[11]
a N-rich condition will be
required in order to form a smooth m-plane, while a Ga-rich condition must be applied
for nanorod formation under continuous growth mode. In this case, the undesorbed Ga
species—which are unable to reach the top region of the nanorods—cannot contribute to
smooth lateral growth and possibly contribute to significant parasitic growth, as can be
clearly seen in Figure 4-4a and 4-4b.
58
4.2.4 Structure and optical characterization
The as-grown continuous mode nanorod is examined by transmission electron
microscopy (TEM) and cathodoluminescence (CL) for its structural and optical
characterization. The TEM image of nanorod grown by continuous growth mode is
shown in Figure 4-5, which is taken along the <11-20> zone axis, where the diffraction
pattern (Figure 4-5c) indicates the wurtzite structure of the nanorod. No stacking faults or
dislocations are found within the nanorod, as is the case for nanorod grown under pulsed
growth mode.
[12]
Figure 4-5. TEM image of nanorod grown by continuous growth mode. The nanorod is
taken along <11-20> zone axies (Figure 4-5b), and the diffraction pattern indicates a
wurtzite structure (Figure 4-5c). The nanorod is free from twins, stacking faults and
threading dislocations (Figure 4-5a).
For CL characterization, we compare the spectra of undoped nanorods grown under
different V/III ratios (2.1 and 8.4) as well as un-doped and Si-doped nanorods (both rods
are grown under V/III equal to 2.1), as shown in Figure 6a and 6b. It can be seen that the
yellow-luminescence (YL) to band-edge emission intensity ratio is much larger for
nanorods grown at a lower V/III ratio than at a higher V/III ratio condition. The latter
59
spectrum is similar to a CL spectrum measured for bulk epitaxial GaN layers grown by
MOCVD (Figure 4-6c). There is almost no difference in the emission spectrum between
the rods grown with and without Si-doping (Figure 4-6b). The origin of YL has been
attributed to multiple physical origins.
[13]
Carbon impurities on nitrogen sites have been
implicated in a number of different studies.
[14–17]
The dominance of the YL under low
V/III ratios is consistent with such a model since incomplete removal of C from precursor
fragments and the occurrence of a higher density of N vacancies would be enhanced by a
low V/III ratio. More work is required to investigate the origin of this large YL-to-band-
edge emission intensity ratio in our samples and its effects on the performance of devices
that incorporate GaN nanorods grown under such conditions, which beyond the scope of
this paper.
Though the continuous mode nanorods exhibit an abnormal luminescence spectrum under
CL measurement, the nanorod templates with InGaN multiple quantum wells (MQWs)
grown on them do not shown this strong yellow luminescence under photoluminescence
(PL) measurement, as shown in Figure 4-6d. The emission peak from the MQWs is
designed to be centered around 410 nm to prevent overlapping with the YL emission. The
MQWs growth condition can be found elsewhere.
[12]
We speculate that suppression of YL
band luminescence in the PL spectra of nanorods with MQW shells arises because the
inner core of the structure is not excited by the incident laser light. The high absorption
coefficient of the InGaN/GaN MQW at the wavelength of the excitation laser may
preclude the core from being excited. Also the small size of the nanorod may allow the
carriers that are created there to diffuse to the MQWs region before they are trapped by
60
the defects inside the nanorod templates. More work is required to clarify these
speculations.
Figure 4-6. a) CL spectrum of nanorod grown under 2 different V/III ratio (8.4 and 2.1 by
changing TMG flow rate, both are un-doped). b) CL spectrum of nanorod grown with and
without Si-doped (V/III ratio is 2.1 in both case). c) CL spectrum of GaN substrate. d) PL
spectrum of continuous mode nanorods with InGaN MQWs.
4.3 The NH
3
-pulsed growth mode
From the previous discussions it is shown that a Ga-rich and mass transport-limited
growth condition is required for GaN nanorod formation. These conditions are achieved
by injecting the group V and group III sources separately in the pulsed growth mode, and
by applying a very low V/III molar ratio in the continuous growth mode. Though both
growth modes can successfully achieve nanorod formation, they have corresponding
problems.
(a) (b)
(d)
(c)
61
The continuous mode employs a much simpler growth process and has relatively high
growth rate, however the nanorods grown under continuous mode exhibit an abnormal
emission spectrum, which will be discussed in the next section. Though the pulsed mode
is able to produce GaN nanorods with reduced YL band emission, the growth process is
complicated since many parameters are involved. Here we introduce a growth mode that
we will refer to as the NH
3
-pulsed mode, which adopts a simpler growth process than
pulsed mode (As shown in Figure 4-7) and is able to achieve GaN nanorods with reduced
YL band emission. In this growth mode, only NH
3
flow is pulsed while the injection of
TMG is continuous resulting in better utilization of TMG than in the pulsed mode.
Figure 4-7. Growth process of NH
3
-pulsed mode where only the injection of NH
3
is
pulsed.
To achieve the nanorod formation and good array uniformity, the NH
3
flow rate and NH3
off time must be carefully designed. In exp. set D we change the NH
3
flow rate (15 – 50
sccm) and adjust the NH
3
off time (1 – 3 seconds, 8 seconds for a cycle), while
maintaining the temperature at 1125 °C (the same as continuous growth mode) and TMG
flow rate at 8 sccm. The morphology of the nanostructure as a function of NH
3
flow rate
and off time is shown in Figure 4-8.
62
Figure 4-8. Nanorod profile as a function of NH
3
flow rate and NH
3
off time (8 seconds
per cycle).
The nanorod profile results in exp. set D can be roughly grouped into 3 regions (I), (II)
and (III) as indicated in Figure 4-8a – b, c – e, and f – g, respectively. In region (I) where
NH
3
injection rate is still high and NH
3
off period is short, the overall growth condition is
still not favorable for nanorod formation. On the other hand in region (III), the NH
3
injection rate becomes too low and the NH
3
off period is too long. In this region an
unusual phenomenon of nanorod structure, a “step” like formation, can be observed on
the top part of the nanorod—that will be discussed later. Only under the conditions of
region (II) where a moderate NH
3
flow rate and NH
3
off time are applied can the
formation of the desired nanorod structure be achieved. At this particular growth
temperature and these cycle durations, the locally optimized growth condition for
nanorod (Figure 4-8d) in our system is set at a NH
3
injection rate equal 25 sccm and off
63
time equal 2 seconds. The SEM image of nanorod array grown by NH
3
-pulsed mode as
well as nanorod array uniformity histogram (compared with the case under continuous
growth mode at V/III equal 2.1) are shown in Figure 4-9a – c, respectively. The pulsed
NH3 mode clearly generates high aspect ratio uniform nanorods by controlling the Ga
surface kinetics.
64
Figure 4-9. a) Birds-eye view (45 degree) SEM image of nanorod array grown by NH
3
-
pulsed mode. b) and c) Normalized height/Width histogram of nanorod grown by NH
3
-
pulsed mode (NP-mode, un-doped) and continuous mode (C-mode, V/III = 2.1, un-
doped).
The step-like growth on the top of the nanorods, shown in region (III), is similar to a
phenomena observed by Galopin et al.
[18]
In their study, a step-like growth was attributed
(a)
(b)
(c)
2 um
65
to GaN nucleation at the junction of the nanorod base region and the substrate, rather than
the nucleation directly on the sidewall of the nanorod. The step growth is believed to start
from the bottom of the nanorod and is parallel to the nanorod growth direction. In our
case, the steps seem to nucleate from the top of the nanorod and then grow toward the
bottom of the nanorod. The origin of the step growth, which might provide a clue for a
more detailed nanorod formation mechanism, is still unclear but it is obviously related to
the NH
3
injection condition, which controls the growth kinetics of various planes of the
nanorod. More work is required to clarify this phenomenon.
4.4 Nanorod comparison by different growth mode
In the previous sections we have discussed and characterized nanorods grown by the
continuous mode and NH
3
-pulsed mode. In this section, we provide a comparison
between the growth rates, uniformities, and emission spectra of optimized nanorods
grown under the various modes. (Table 4-3) The CL spectra of nanorods are shown in
Figure 4-10 a) pulsed; b) NH
3
-pulsed; and c) continuous mode. Note that only the
continuous mode nanorods exhibit a strong YL. The PL spectra of InGaN MQWs grown
on each mode nanorod template are also shown in Figure 4-11 a) pulsed; b) NH
3
-pulsed;
and c) continuous mode.
Growth
mode
Growth rate
(nm/hour)
Array
uniformity
CL spectrum
Pulsed 1 um Good Comparable band-edge and YL emission
intensity
Table 4-3. Comparison between nanorods grown by different growth modes.
66
NH
3
-pulsed 6 um Good Large band-edge to YL emission
intensity ratio
Continuous 4 – 20 um
*1
Good*
2
Large YL to band-edge emission
intensity ratio
*
1
The growth rate of nanorods by continuous mode can be enhanced by increasing TMG
injection rate while it is more complicated in pulsed and NH
3
-pulsed mode.
*
2
Over 24 sccm TMG injection rate (V/III molar ratio equal 4.2) is required to obtain
good uniformity of the nanorod array in continuous mode.
Table 4-3. Continued.
Figure 4-10. CL spectrum of nanorods grown by a) pulsed; b) NH
3
-pulsed; and c)
continuous (V/III = 2.1) mode.
(a) (b)
(c)
67
Figure 4-11. PL spectrum of InGaN MQWs grown on different mode nanorod templates
by a) pulsed; b) NH
3
-pulsed; and c) continuous (V/III = 2.1) mode.
4.5 Summary
In this chapter, we achieve GaN nanorod growth on Ga-polar substrates by a continuous
growth mode and discuss the operable growth mechanisms in their formation. A low V/III
molar ratio and high growth temperature favor the growth of {1-101} planes therefore are
critical for nanorod formation as well as achieving good array uniformity. The continuous
mode is simpler to implement than the pulsed mode, but the nanorods produced by
continuous mode exhibit an abnormal emission spectrum with strong YL band emission
(a) (b)
(c)
68
which may result from impurity induced emission. The NH
3
-pulsed mode combines some
advantages of both the continuous and pulsed growth mode was discussed. This mode
can be used to grow nanorods with a relatively high growth rate while achieving low
deep state luminescence emission. We have also compared the different growth modes
with respect to the nanorod profile, luminescence property as well as the emission of
InGaN MQWs that utilize these nanorod templates. Based on the various advantages of
these growth modes, we can now adopt the appropriate growth mode for different
nanotechnology applications.
4.6 Chapter references
[1] R. M. Feenstra, Y . Dong, C. D. Lee, J. E. Northrup, J. Vac. Sci. Technol. B 2005,
23, 1174.
[2] T. Akiyama, D. Ammi, K. Nakamura, T. Ito, Phys. Rev. B 2010, 81, 245317.
[3] Y .-T. Lin, T.-W. Yeh, P. D. Dapkus, Nanotechnology 2012, 23, 465601.
[4] V . Jindal, F. Shahedipour-Sandvik, J. Appl. Phys. 2009, 106, 083115.
[5] T. Akasaka, Y . Kobayashi, S. Ando, N. Kobayashi, M. Kumagai, J. Cryst. Growth
1998, 189, 72.
[6] Q. Sun, C. D. Yerino, T. S. Ko, Y . S. Cho, I.-H. Lee, J. Han, M. E. Coltrin, J. Appl.
Phys. 2008, 104, 093523.
[7] M. E. Coltrin, C. C. Mitchell, J. Cryst. Growth 2003, 254, 35.
[8] T. Shioda, Y . Tomita, M. Sugiyama, Y . Shimogaki, Y . Nakano, Jpn. J. Appl. Phys.
2007, 46, L1045.
[9] D. Segev, C. G. Van de Walle, Surf. Sci. 2007, 601, L15.
[10] L. Lymperakis, J. Neugebauer, Phys. Rev. B 2009, 79, 241308.
[11] M. Sawicka, H. Turski, M. Siekacz, J. Smalc-Koziorowska, M. Kryśko, I.
69
Dzięcielewski, I. Grzegory, C. Skierbiszewski, Phys. Rev. B 2011, 83, 245434.
[12] T.-W. Yeh, Y .-T. Lin, L. S. Stewart, P. D. Dapkus, R. Sarkissian, J. D. O’Brien, B.
Ahn, S. R. Nutt, Nano Lett. 2012, 12, 3257.
[13] M. a. Reshchikov, H. Morkoç, J. Appl. Phys. 2005, 97, 061301.
[14] D. S. Green, U. K. Mishra, J. S. Speck, J. Appl. Phys. 2004, 95, 8456.
[15] C. G. Van de Walle, J. Neugebauer, J. Appl. Phys. 2004, 95, 3851.
[16] K. Laaksonen, M. G. Ganchenkova, R. M. Nieminen, J. Phys. Condens. Matter
2009, 21, 015803.
[17] J. L. Lyons, A. Janotti, C. G. Van de Walle, Appl. Phys. Lett. 2010, 97, 152108.
[18] E. Galopin, L. Largeau, G. Patriarche, L. Travers, F. Glas, J. C. Harmand,
Nanotechnology 2011, 22, 245606.
70
5 CHAPTER 5: INGAN/GAN CORE-SHELL NANOLEDS
The growth of GaN nanorods has been presented in previous chapters, and the subsequent
multiple quantum wells (MQWs, active region) as well as p-type GaN layers will be
grown on these nanorod templates to form a LED structure. Due to the 3-dimensional
structure of nanoLEDs, both nonpolar m-plane and semipolar m-plane coexist on a single
nanorod. These different planes in general have different growth behavior including the
growth rate, indium incorporation rate, and p-type doping efficiency, and therefore the
optical-excited and electrical-excited emission will be different. In this chapter, we will
discuss the growth of MQWs and p-type GaN on the sidewall. After the growth,
photoluminescence (PL), cathodoluminescent (CL), and electroluminescence (EL) will be
conducted to analyze the emission behavior of the device.
5.1 MQWs on GaN nanorods
InGaN MQWs grown on planar GaN substrates have been widely studied, and
commercial efficient blue – green LEDs and laser diodes (LDs) are already available
based on this research. Recently MQWs grown on GaN nanorod structures also received
intensive investigation
[1–3]
due to their potential to provide a better LEDs with less
efficiency droop and higher power density. However difficulties associated with this 3-D
and nonpolar-based nanorod template emerge as the research progresses. The advantages
as well as the difficulties encountered of nanorod-based MQWs will be discussed in
detail in the following sections.
71
5.1.1 Growth condition for MQWs on sidewall
One of the most important features of LEDs is the emission wavelength, which is
controlled by the indium concentration within the In
X
Ga
1-X
N/GaN MQWs. In general, the
indium concentration can be adjusted by the growth conditions such as temperature,
pressure, and TMI flow rate. The quantum well thickness will also affect the emission
wavelength due to a delay of indium incorporation in the MQWs and, due to the quantum
confined effect, especially when the thickness is only a few nanometers. In this section
we will experimentally investigate the emission wavelength of the MQWs on the
sidewalls as a function of temperature, TMI flow rate, and quantum well thickness (which
is proportional to quantum well growth time). For these sets of experiments, the growth
pressure, TEG flow rate, and NH3 flow rate are kept at 300 torr, 20 µmol/min, and
223 mmol/min (V/III molar ratio equals around 10000). The nanorods are 1 µm in height
and are separated by 1 µm center-to-center (1 µm pitch). Table 5-1 summarizes the
growth parameters, the peak wavelength, and the full width half maximum (FWHM) of
each spectrum. The PL spectrums as a function of these parameters are shown in
Figure 5-1: (a) QW growth time (20 – 70 s), (b) temperature (775 – 820 °C), and (c) TMI
flow (4.25 – 7.65 µmol/min).
72
Exp. set Temperature
(°C)
TMI flow
(µmol/min)
QW growth
Time (s)
PL peak
Position (nm)
FWHM
(nm)
5A-1 790 4.25 20 417 40
5A-2 790 4.25 50 437 40
5A-3 790 4.25 70 464 60
Table 5-1. Parameters (temperature, TMI flow rate and QW thickness) of QW emission
via PL study.
5B-1 775 4.25 50 484 68
5B-2 790 4.25 50 437 40
5B-3 820 4.25 50 400 25
5C-1 820 4.25 50 398 30
5C-2 820 5.95 50 420 38
5C-3 820 7.65 50 430 33
Table 5-1. Continued.
Figure 5-1. QW emission PL spectrum as a function of (a) growth time, (b) growth
temperature, and (c) TMI flow rate.
(a) (b)
73
Figure5-1. Continued.
If we compare these spectra, we see that the QWs give a longer emission wavelength
with larger width, higher TMI flow rate, and lower temperature during growth in general.
This is because the indium incorporation rate is higher at lower growth temperature (due
to its high partial pressure compared with gallium) and higher TMI flow, and there is less
quantum confinement with larger QW thickness. Besides, thicker QWs will tend to have
a higher indium incorporation rate because of a delay in the indium incorporation at the
barrier/well interface
[4,5]
. In addition, QWs with a higher indium concentration may have
severe composition fluctuations
[6,7]
, which will in turn cause spectrum broadening and
exhibit a larger FWHM.
5.1.2 Indium incorporation on nonpolar and semipolar planes
A nanorod is defined by a polar c-plane [0001], nonpolar m-planes {1-100}, and
semipolar m-planes {1-101} (Figure 5-2), and therefore all of these planes must be taken
into consideration when MQWs are grown based on the nanorod template. To simplify
the multi-plane study, we apply a continuous growth mode at a high V/III molar ratio
right after the nanorod growth step to eliminate the c-plane, as shown in Figure 5-2(b).
(c)
74
Figure 5-2. (a) A single nanorod is confined by the polar c-plane [0001], nonpolar m-
plane {1-100}, and semipolar m-plane {1-101}. (b) c-plane is eliminated from a nanorod.
After c-plane elimination, the m-plane and semipolar m-plane still exist. To investigate
the emission of MQWs on these planes, we take a CL mapping on a single nanorod with
MQWs, as shown in Figure 5-3: (a) the region of nanorod that is mapped, (b) the peak
intensity and wavelength distribution, (c) the original emission at various wavelength
(360 – 550 nm), and (d) the emission intensity that is normalized at that particular
wavelength.
Figure 5-3. (a) QWs grown on a nanorod. The white box indicates the mapping region. (b)
The peak intensity and wavelength distribution. (c) CL signal at various wavelengths
(360 – 550 nm). (d) CL signal at various wavelengths (360 – 550 nm), normalized to each
wavelength.
(a)
(b)
(a)
(b)
75
Figure 5-3. Continued.
(c)
76
Figure 5-3. Continued.
(d)
77
From the CL mapping we see that the dominant emission located at 420 nm is mainly
from sidewall, as expected. However, another peak located around 480 – 500 nm with a
relatively low intensity is also observed. To further study their emission spectrums, we
scan several points on the nanorod and the spectrums are shown in Figure 5-4: (a) points
scanned (1 – 6), (b) spectrum of points 1 – 3, and (c) points 4 – 6. As indicated in the
figure, the peak emission from pyramid (peak 2, 490 nm) is longer than that from
sidewall (peak 1, 420 nm) for around 70 nm, due to different indium incorporation
efficiency on different planes. Wernicke et al have extensively studied the indium
incorporation on each GaN crystal plane under various growth conditions
[8]
, and they
found that the semipolar m-plane is able to capture more indium than the m-plane. In
their work, MQWs grown on semipolar m-planes exhibit a longer wavelength emission,
ranging from 50 – 70 nm. Compared with the nonpolar m-plane, in all tested conditions,
this value is consistent with our findings. The CL signal with relatively low intensity
from the pyramid region originated from very thin MQWs since the semipolar m-plane
has a much lower growth rate compared to the nonpolar m-plane. MQWs on both
semipolar and nonpolar m-planes will come into play in the final core-shell nanoLEDs
under forward bias, which will be discussed in the later sections in detail.
78
Figure 5-4. (a) 6 points on a single nanorod with MQWs are scanned. (b) CL spectrum of
points 1 – 3. (c) CL spectrum of points 4 – 6. The emission from the pyramid region has a
longer peak wavelength due to a higher indium incorporation rate and much lower
intensity due to lower growth rate, compared with emission from the sidewall.
5.2 The growth of p-type GaN
The p-type GaN plays an important role in GaN-based LED structures, and a good p-GaN
layer is in general more difficult to achieve compared with a n-GaN layer. Magnesium
has been used as a typical acceptor dopant in GaN grown by MOCVD because it has the
smallest ionization energy among the known acceptor donors
[9]
. The highest reported p-
type doping level is only 8.5E17 cm
-3
due to a self-compensation effect
[10]
, and therefore
Peak 1
Peak 2
(a)
(b)
(c)
79
this requires a high quality MOCVD system since the unintentionally-doped GaN is
already a n-type material with a doping level around mid-10
17
cm
-3
. The as-grown p-GaN
requires thermal annealing to activate the p-type dopant
[11]
. In our case, the p-GaN is
annealed in a furnace at 760 °C for 20 minutes in a N
2
ambient. The concentration,
mobility, and resistivity of p-GaN grown in our MOCVD reactor are 3.4E17 cm
-3
,
16.2 cm
2
/s, and 2.39 ohm-cm, respectively. Figure 5-5 shows a planar blue LED grown in
our reactor: (a) EL and (b) IV curve. The detailed growth conditions can be found in our
previous work
[12]
.
Figure 5-5. Blue LED grown in our reactor with (a) EL at 10 mA current injection, and (b)
its IV curve.
The p-GaN condition and doping level can be confirmed by the Hall effect measurement
in the planar case, however in the situation of the 3D nanoLED, this task becomes
nontrivial. Spatially resolved Hall effect measurements have been employed to
investigate the material properties of a single p-type shell and n-type core GaN
nanorod
[13]
. For simplicity, we adopt SEM and examine the cross section of a single core-
shell nanoLED. By differentiating the color contrast shown in the SEM image, we are
able to qualitatively determine if a region is p-type. This technique has been applied to
examine the doping type and doping level of a material to a wide extent, based on
(a)
(b)
80
different secondary electron collection efficiency by the SEM detector
[14–17]
. Figure 5-6
shows the cross section SEM image of (a) p-type GaN (nominal 400 nm) grown on
undoped GaN and (b) p-type shell and n-type core nanoLED structure.
Figure 5-6. Cross section SEM image of (a) p-type GaN on undoped GaN and (b) p-type
shell on n-type core nanoLED. The p-GaN region is brighter compared with undoped or
n-doped GaN.
5.3 Contact for p-type GaN
Indium tin oxide (ITO) has been widely employed as a transparent ohmic electrode on p-
GaN in GaN-based LEDs due to its superior optical transparency and electrical
conductivity
[18–22]
. For the case of 3D nanoLEDs, ITO is sputtered on the surface to form
a conformal thin layer, and the sputtered power during the deposition plays an important
role. It has been reported that high energy particles hitting on the p-GaN surface will
damage the GaN crystal and create N vacancies
[23]
, which will compensate the acceptor
concentration. Therefore, the sputtering power is maintained at 120 – 200 watts, typically.
A lower sputtering power can provide a better conformal ITO film, but the resistivity will
be higher on the other hand. The resistivity of ITO film sputtered at 200, 150, and
120 watts is measured to be 1E-4, 1.59E-4, and 4.66E-4 ohm-cm, respectively, which is
Undoped GaN
p-GaN
sapphire
(a)
(b)
81
around 50 – 230 times higher than gold (2E-6 ohm-cm). We deposit ITO on our device at
150 watts since it provides both good conformity and conductivity.
ITO does not directly form an ohmic contact on p-type GaN since its vacuum level is
around 4.8 eV
[22]
and the valence band maximum of GaN is around 7.5 eV
[24]
. Therefore,
a thin p+ layer several nanometers thick is usually applied to improve the ohmic behavior
of the ITO/p-GaN junction
[25]
. The band alignment of the ITO/p-GaN junction is shown
in Figure 5-7, (a) without and (b) with the p+ GaN layer. After ITO sputtering, the device
is annealed in a furnace at 600 °C for 10 minutes in an O
2
ambient to further decrease the
contact resistance
[23]
.
Figure 5-7. The band alignment of the ITO/p-GaN junction (a) without and (b) with the
p+ GaN layer.
(a)
(b)
82
5.4 The core-shell nanoLEDs
The core-shell nanoLED is grown by the following procedure, and a cross section
illustration of this device is show in Figure 5-8: (a) schematic diagram and (b) SEM
image.
1. N-type nanorods are grown as nanoLED templates by pulsed growth mode
(chapter 3) on a pre-patterned GaN/sapphire substrate.
2. Apply continuous growth mode to terminate the c-plane of the nanorods.
3. 3 (or 6) pairs of MQWs are grown on the n-type nanorods.
4. (Optional) 1
st
p-type layer is grown by pulsed mode for experimental purposes.
This p-type layer will mostly grow on the pyramid region of the nanorods.
5. 2
nd
p-type shell is grown by continuous mode. This p-type layer will mostly grow
on the sidewall of the nanorods.
6. The device is annealed at 760 °C for 20 minutes in a N
2
ambient for p-type dopant
activation.
7. (Optional) SiO
2
is deposited via e-beam evaporation to selectively cover the
pyramid region to avoid the current flow into the MQWs on it.
8. A conformal ITO layer is sputtered (150 watt for 150 minutes, 150 – 200 nm on
the sidewall) on the nanoLEDs.
9. The device is annealed at 600 °C for 10 minutes in an O
2
ambient to further
reduce the contact resistance between ITO and p-GaN.
83
Figure 5-8. Cross section illustration of nanoLEDs (a) schematic diagram and (b) SEM
image. The 1st p-GaN layer is not grown in this particular device shown in (b), and the
MQWs cannot be identified at this magnification.
It has been discussed previously that the growth rate of semi-polar m-plane is very low,
and therefore the p-GaN layer on the pyramid region of the nanoLEDs will be very thin.
To study how this semi-polar p-layer influences the performance of nanoLEDs, we apply
3 different structures of p-layer on the nanoLEDs. For the first structure (sample 5D-1),
we directly grow a p-GaN layer after the MQWs growth. For the second structure
(sample 5D-2), an unintentionally-doped GaN layer is selectively grown on the pyramid
region by pulsed mode right after MQW growth, expected to block the current flow into
the MQWs on semi-polar planes. For the third structure (sample 5D-3), a p-GaN layer is
selectively grown on the pyramid region by pulsed mode right after MQW growth, and
then a second p-layer is grown by continuous mode. These 3 structures are shown
schematically, with cross section SEM image, in Figure 5-9 (a) – (f). Their IV
characteristic curves are shown in Figure 5-10 (a). It can be seen that if we directly apply
a p-layer on the MQWs, the p-GaN layer on top of the pyramid region is so thin (smaller
(a)
(b)
84
than 50 nm) that there is non-negligible reverse current (21.3 mA) at -5V . Besides, this
device does not emit light under forward bias. However, if an undoped GaN region is
applied on the pyramid region by pulsed mode, the IV curve is almost linear and the
device acts as a resistor. Even though this semi-polar layer is grown unintentionally
doped, the pulsed mode nature which requires an effective low V/III molar ratio will
introduce a significant amount of nitrogen vacancies and turn this layer into n-type. This
n-type layer directly links the n-type core and the ITO layer and therefore shorts the
device. For the last structure, since both the sidewall and pyramid region have a relatively
thick p-GaN layer (>100 nm), this device emits light (figure 5-10b) and the reverse
current is reduced to around 10 mA at -5V .
85
Figure 5-9. 3 different structures of p-layer on the nanoLEDs. (a) – (b) sample 5D-1, a
single p-GaN layer is grown after the MQW growth. (c) – (d) sample 5D-2, an
unintentionally-doped GaN layer is selectively grown on the pyramid region by pulsed
mode right after MQW growth. (e) – (f) sample 5D-3, a p-GaN layer is selectively grown
on the pyramid region by pulsed mode right after MQW growth, and then a second p-
layer is grown by continuous mode.
500 nm
500 nm
500 nm
(a)
(b)
(c)
(d)
(e)
(f)
86
Figure 5-10. (a) IV curve of 3 different sturctures (5D-1 – 5D-3) shown in figure 5-9. (b)
EL of 5D-3.
As described in section 5.1, MQWs grown on the sidewall and pyramid of the nanorods
will emit light with different peak wavelengths. And since nanoLEDs depend on the
active region on the sidewall, it is desirable to block the current from flowing into the
MQWs on the pyramid region. A simple way to implement this attempt is to selectively
deposit a dielectric mask on the pyramid before ITO sputtering. To do this, we apply e-
beam evaporated SiO2 (nominal thickness: 200 nm) to insulate the ITO and pyramid
region (called pyramid insulation later on). The SEM image of the nanoLEDs after SiO2
is shown in Figure 5-11 (a). Even though SiO2 deposited by e-beam evaporation is very
directional, there is still inevitably some residue absorbed on the sidewall. This residue
will block the sidewall from sputtered ITO to some extent and therefore reduce the
current flow. In experiments 5E-1 and 5E-2 (nanoLEDs structure of both 5E-1 – 2 are the
same as that of 5D-1), we deposit two different SiO2 thicknesses (70 and 200 nm) on
nanoLEDs. As a comparison, we also insulate the sidewall of the nanoLEDs (Figure 5-
11(b)) to allow the current only flowing via the pyramid region, as in sample 5E-3 (5E-3
has structure the same as 5D-3). Table 5-2 shows a detailed description of nanoLED
(a)
(b)
87
structure and the insulation condition used in sample 5E-1 – 3. For the IV curve study as
shown in Figure 5-11(c), we see that after pyramid insulation, the reverse current at -5V
drops from 21.3 mA to 4.92 mA (70 nm SiO2) and 4.55 mA (200 nm SiO2), and the
forward current for sample 5E-1 is much higher than sample 5E-2. Therefore we
conclude that SiO2 deposition with 70 nm is sufficient to block the current from the
pyramid region while minimizing the insulation between the sidewall and ITO. Note that
the remaining reverse current in 5E-1 and 5E-2 is not coming from the pyramid region
but from the sidewall, which will be confirmed later this section. For the case of sample
5E-3, even though its IV curves show typical PN jucntion behavior with relatively a low
reverse bias current (0.19 mA) at -5V , it barely emits light (Figure 5-11(d), 20 mA current
injection), and the device fails at 8 – 10V forward voltage.
sample Surface insulation / thickness nanoLED structure used
5E-1 Pyramid / 70 nm 5D-1
5E-2 Pyramid / 200 nm 5D-1
5E-3 Sidewall / 100 nm 5D-3
Table 5-2. Parameters set for exp. 5E.
88
Figure 5-11. NanoLED insulation test. (a) pyramid region is insulated by 200 nm SiO2,
(b) sidewall is insulated by 100 nm SiNx. (c) IV curves of sample 5D-1 and 5E-1 – 3. (d)
EL of sample 5E-3.
In experiments 5E-1 and 5E-2, there is still non-negligible reverse current (~4.5 mA) at -
5V after pyramid insulation. To investigate the origin of this leakage current, we design a
set of experiments 5F, where the MQWs and p-GaN thickness is varied, and a SiO2
insulating layer (70 nm thickness) is deposited on these samples. Table 5-3 summarizes
the parameters involved in these experiments.
(a)
(b)
(c)
(d)
89
sample QWs growth time (s) p-GaN thickness (nm)
5F-1 15 200
5F-2 30 200
5F-3 60 200
5F-4 60 100
Table 5-3. Parameters set for exp. 5F.
Figure 5-12 shows the IV curve and the EL image of experiment set 5F. The reverse bias
current is only 0.67 mA at -5V for sample 5F-1. However, it increases as quantum wells
become thicker or p-GaN becomes thinner. A non-negligible reverse bias current
indicates the poor quality of the PN junction, which could originate from poor QW
crystal quality. To confirm the quality of QWs, EL (20 – 100 mA current injection) and
10-point CL spectum (10 different points on the sidewall of a single nanorod with MQWs
are scanned from top to bottom, similar to that in Figure 5-4 (a)) of sample 5F-1 and 5F-3
are further investigated, which is shown in Figure 5-13 (a) – (d). For sample 5F-1, we see
that the emission for both EL and CL is almost single peak (FWHM = 42 nm), and their
peaks are very close (422 nm for EL and 405 nm for CL). This differece can be attributed
to run-to-run error or the exsiting internal electric field during the EL measurement. No
blue shift is observed during the EL measurement. However, in the case of 5F-3, both the
EL and CL spectra have a relatively large line width (FWHM = 110 nm), and multiple
peaks can be observed. Besides, the main peak shown in CL (450 nm) does not exhibit in
EL, but the second minor peak (545 nm) coincidently matches the peak location in EL
(550 nm). These results indicate a poor QW crystal quality of the nanoLEDs, which may
in turn degrade the quality of PN junction, leading to a significant reverse bias current, as
well as a non-decent EL.
90
In addition to the main peak located at 422 nm, there is a minor broad peak centered
around 575 nm, which is believed to be emitted from the pyramid region. Note that even
though the pyramid region has been insulated by SiO
2
, a small amount of current can still
flow into the MQWs on the pyramids from the sidewall. The CL spectrum of the pyramid
region of sample 5F-1 is shown in the inset of Figure 5-13(b), and a peak centered at
575 nm is indeed observed. This position also coincides with the typical peak of yellow
luminescence of GaN. More work is required to confirm the origin of this emission,
which occurrs in all nanoLEDs samples.
Figure 5-12. (a) IV curves of samples 5F-1 – 4. (b) – (d) EL image of sample 5F-1 – 3:
(b) 5F-1, (c) 5F-2, and (d) 5F-3.
(c) (d)
(a)
(b)
91
Figure 5-13. (a) EL spectrum of sample 5F-1, (b) CL spectrum of 5F-1, (c) EL spectrum
of sample 5F-3, and (d) CL spectrum of sample 5F-3. For CL spectrum, 10 different
points on the sidewall of a single nanorod with MQWs are scanned from top to bottom,
similar to that in Figure 5-4 (a). Inset of 5-13(b): CL spectrum of pyramid region of
sample 5F-1.
It has been reported that MQWs grown at a lower rate (0.1 – 0.3 Å/s) can help improve
their crystal quality
[26,27]
. To see how this strategy can help improve the performance of
nanoLEDs, we grow MQWs (sample 5G) with a growth rate decreasing from 1 Å/s to
0.5 Å/s. The nominal well thickness is 5 nm. After the growth, a 10-point CL spectrum is
taken on the sidewall, as shown in Figure 5-14. If we compare sample 5G with 5F-3, the
former exhibits a CL spectrum with finer line width (FWHM 36 nm) and less multi-peak
top
bottom
(a) (b)
(c) (d)
inset
92
behavior. Our results are consistent with reported work, and a further decrease of MQWs
growtt rate toward 0.1 – 0.3 Å/s may improve the nanoLED performance even more.
Figure 5-14. CL spectrum of sample 5G. QWs are grown at 0.5 Å/s with thickness equal
to 5 nm. The FWHM is measured to be 36 nm.
5.5 Summary
We have studied in detail the structure and fabrication of InGaN/GaN core-shell
nanoLEDs in this chapter. The emission wavelength of nanoLEDs is controlled by the
growth condition of MQWs, and their crystal quality is important in order to obtain
highly efficient LEDs. High quality MQWs can be formed at a relatively slow growth
rate (0.1 – 0.3 Å/s). Different orientations of planes will have different emission
behaviors, but this influence can be minimized by introducing a pyramid insulation layer
so only the MQWs on the sidewall will be excited. A minimum p-GaN layer thickness
around 200 nm is also required to reduce the reverse bias current as well as the leakage
current. For the ITO contact, a moderate sputtering power (150W) will be beneficial for
both contact conformity and film conductivity. Further optimization of growth conditions
and nanoLEDs structure (insertion of an electron block layer, number of QW pairs, etc)
can help improve the performance of nanoLEDs, which is out of the scope of this work.
top
bottom
93
5.6 Chapter references
[1] H.-S. Chen, Y .-F. Yao, C.-H. Liao, C.-G. Tu, C.-Y . Su, W.-M. Chang, Y .-W. Kiang,
C. C. Yang, Opt. Lett. 2013, 38, 3370.
[2] Y . J. Hong, C.-H. Lee, A. Yoon, M. Kim, H.-K. Seong, H. J. Chung, C. Sone, Y . J.
Park, G.-C. Yi, Adv. Mater. 2011, 23, 3284.
[3] T.-W. Yeh, Y .-T. Lin, L. S. Stewart, P. D. Dapkus, R. Sarkissian, J. D. O’Brien, B.
Ahn, S. R. Nutt, Nano Lett. 2012, 12, 3257.
[4] V . Hoffmann, C. Netzel, U. Zeimer, A. Knauer, S. Einfeldt, F. Bertram, J. Christen,
M. Weyers, G. Trnkle, M. Kneissl, J. Cryst. Growth 2010, 312, 3428.
[5] Y . Ra, R. Navamathavan, J. Park, C. Lee, Nano Lett. 2013, 13, 3506.
[6] Y . Y . Chung, Y . S. Lin, S. W. Feng, Y . C. Cheng, E. C. Lin, C. C. Yang, K. J. Ma, C.
Hsu, H. W. Chuang, C. T. Kuo, J. S. Tsang, J. Appl. Phys. 2003, 93, 9693.
[7] M. Rao, D. Kim, S. Mahajan, Appl. Phys. Lett. 2004, 85, 1961.
[8] T. Wernicke, L. Schade, C. Netzel, J. Rass, V . Hoffmann, S. Ploch, A. Knauer, M.
Weyers, U. Schwarz, M. Kneissl, Semicond. Sci. Technol. 2012, 27, 024014.
[9] K. S. Kim, G. M. Yang, H. J. Lee, Solid. State. Electron. 1999, 43, 1807.
[10] H. Obloh, K. . Bachem, U. Kaufmann, M. Kunzer, M. Maier, A. Ramakrishnan, P.
Schlotter, Self-compensation in Mg doped p-type GaN grown by MOCVD. J.
Cryst. Growth 1998, 195, 270–273.
[11] S. Nakamura, N. Iwasa, M. Senoh, T. Mukai, Japanese J. Appl. Physics, Part 1
Regul. Pap. Short Notes Rev. Pap. 1992, 31, 1258.
[12] T.-W. Yeh, Dissertation 2013.
[13] K. Storm, F. Halvardsson, M. Heurlin, D. Lindgren, A. Gustafsson, P. M. Wu, B.
Monemar, L. Samuelson, Nat. Nanotechnol. 2012, 7, 718.
[14] I. V olotsenko, M. Molotskii, Z. Barkay, J. Marczewski, P. Grabiec, B. Jaroszewicz,
G. Meshulam, E. Grunbaum, Y . Rosenwaks, J. Appl. Phys. 2010, 107.
94
[15] S. Chung, V . Wheeler, R. Myers-Ward, L. O. Nyakiti, C. R. Eddy, D. K. Gaskill, M.
Skowronski, Y . N. Picard, J. Appl. Phys. 2011, 110, 014902.
[16] A. K. W. Chee, R. F. Broom, C. J. Humphreys, E. G. T. Bosch, J. Appl. Phys. 2011,
109, 013109.
[17] D. Tsurumi, K. Hamada, Anal. Technol. 2013, 1, 141.
[18] J.-H. Choi, S.-H. Jang, J.-S. Jang, Electron. Mater. Lett. 2013, 9, 425.
[19] K.-K. Kim, H. Kim, S.-N. Lee, S. Cho, Electron. Mater. Lett. 2011, 7, 145.
[20] D. W. Kim, Y . J. Sung, J. W. Park, G. Y . Yeom, In Thin Solid Films; 2001; V ol.
398-399, pp. 87–92.
[21] Y . C. Lin, S. J. Chang, Y . K. Su, T. Y . Tsai, C. S. Chang, S. C. Shei, C. W. Kuo, S.
C. Chen, Solid. State. Electron. 2003, 47, 849.
[22] J. O. Song, J.-S. H. J.-S. Ha, T.-Y . S. T.-Y . Seong, IEEE Trans. Electron Devices
2010, 57, 42.
[23] S. J. Chang, C. H. Lan, J. D. Hwang, Y . C. Cheng, W. J. Lin, J. C. Lin, H. Z. Chen,
Sputtered Indium-Tin-Oxide on p-GaN. J. Electrochem. Soc. 2008, 155, H140.
[24] J.-K. Ho, C.-S. Jong, C. C. Chiu, C.-N. Huang, K.-K. Shih, L.-C. Chen, F.-R. Chen,
J.-J. Kai, J. Appl. Phys. 1999, 86, 4491.
[25] Y . Chen, H. Wu, G. Yue, Z. Chen, Z. Zheng, Z. Wu, G. Wang, H. Jiang, Appl. Phys.
Express 2013, 6.
[26] T. Hikosaka, T. Shioda, Y. Harada, K. Tachibana, N. Sugiyama, S. Nunoue, Phys.
Status Solidi 2011, 8, 2016.
[27] K. Jacobs, B. Van Daele, M. R. Leys, I. Moerman, G. Van Tendeloo, In Journal of
Crystal Growth; 2003; V ol. 248, pp. 498–502.
95
6 CHAPTER 6: CONCLUSION AND FUTURE
DIRECTIONS
In this dissertation we have discussed all the procedures involved in nanoLED fabrication.
An e-beam lithography-defined pattern is needed for selective area growth to form
nanorod arrays. These nanorod templates can either be grown by pulsed mode or
continuous mode. A Ga-rich growth condition is required in either case to assure the
formation of nanorods. An appropriate growth condition is also necessary to have a
uniform nanorod array in order to fabricate nanoLEDs with good performance. MQWs
grown on these nanorods will exhibit different emission behavior due to the different
surfaces that define the 3D nanorod. To prevent the emission from MQWs on the pyramid,
a dielectric mask (70 nm SiO
2
) is deposited on the pyramid region (before p-type contact
deposition) to only allow the current flowing into the MQWs on the sidewalls. MQW
crystal quality can be determined by a CL spectrum, and a lower growth rate (0.1 – 0.3
Å/s) can help improve MQW quality and make the luminescence more uniform and
efficient. A minimum thickness (100 – 200 nm) of the p-GaN layer is required to have a
low reverse bias leakage current and to produce a reasonable amount of light. Finally,
ITO is sputtered at a moderate power (150W) as a 3D p-type contact to assure both high
conductivity and good continuity. A grid design of a metal layer (Ti/Au) is also applied to
help the current spreading and to reduce the contact resistance between the probe and the
device.
96
The following sections provide some further directions to help improve the performance
of nanoLEDs, and some promising findings during the work of this dissertation are also
listed.
6.1 Optimization of nanoLEDs
The 3D nanoLED is a relatively new structure
[1–3]
and therefore its performance requires
further optimization. Several methodologies already applied to optimize the planar LEDs
can be adopted for the case of 3D nanoLEDs, as discussed here:
6.1.1 Insertion of electron block layer
It is well known that the mobility of electrons is much higher than that of holes in GaN,
and therefore an electron blocking layer (EBL) is typically inserted before the growth off
the p-GaN layer to prevent the electrons from leaking into the p-GaN layer directly.
AlGaN or AlInN are typical materials used for the EBL, and it is already reported that by
insertion of EBL, the LED performance can be improve by 32 – 50%
[4]
.
6.1.2 Number of MQWs pair and MQWs thickness
MQWs are the active regions for the LED and therefore if their volume is too small, the
electron-hole pairs may not be combined radiatively efficiently. Besides, the effective
carrier density (carriers per QW volume) will be higher as well, which will in turn
enhance the efficiency droop effect. In the case of planar GaN-based LEDs, 5 – 10 pairs
of MQWs are usually adopted with a quantum well thickness of 2 nm. But in the case of
a single nanoLED, due to its unique nanorod shape and the slower growth rate of the
semi-polar m-plane, the sidewall will become shorter as the number of MQW pairs
97
increase. In addition, if the quantum well thickness or number of pairs of MQWs is
increased, the induced strain will be increased correspondingly, and defects or
dislocations will be generated if the effective critical thickness of the InGaN layer is
exceeded. A compromise must be made between the thickness of the QWs and the
number of MQWs in order to reach the optimized LED efficiency.
6.1.3 Geometry of nanoLEDs
NanoLEDs are based on a nanorod array template, and therefore the pitch and the size of
the nanorod array, the dimension of MQWs adopted in the nanoLEDs, and the thickness
of the p-GaN and ITO layers will all influence their total active area and emission
efficiency. A computer-aid simulation is suggested to calculate the design geometry in
order to optimize the emission and the efficiency of nanoLEDs.
6.2 GaN nanosheet as solid state lighting template
Growth of GaN nanosheets and their potential as nanoLED templates has been published
in our group
[5]
. Their stripe-like structure makes them a potential candidate for further
applications such as for lasers and tri-gates. However, their growth window is much
smaller than that of nanorods, probably due to the growth kinetic limitation of the m-
plane and semi-polar m-planes, which is still not fully understood. More work is required
to study their growth mechanism in order to grow a uniform nanosheet array.
6.3 Ultra-thin nanorod template
It is known that if we grow a thick InGaN layer on a GaN nanorod with a thickness
exceeding the critical thickness (usually smaller than 10 nm) corresponding to the indium
composition, a dislocation will be formed which might serve as nonradiative
98
recombination site for carriers. On the other hand, if the GaN core itself is thin enough,
the strain induced by the lattice mismatch between GaN and InGaN will be transferred
from the InGaN shell to the GaN core gradually as InGaN growth proceeds, and beyond a
certain point the InGaN layer will be free of strain regardless its thickness
[6,7]
. For
example, a dislocation-free In
0.1
Ga
0.9
N shell (thickness over 20 nm) will require a GaN
core to be as thin as 20 nm in diameter
[6]
. Given this InGaN/GaN structure, other InGaN
QWs with even higher indium compositions can then be grown on it and help push the
emission wavelength even longer. A schematic structure and band gap diagram of this
multilayer GaN/InGaN nanostructure is shown in Figure 6-1. This strain-transferred
strategy is promising to resolve the current bottleneck of LED industry – the “green gap”,
where the efficiency of LED drops dramatically when its emission wavelength
approaches 550 nm
[8]
. However, this strategy requires SAG with a very small mask
opening (20 nm) and it is currently limited by the resolution of lithography technologies.
Post-growth wet etching
[3,9,10]
and the growth of nanorockets
[11]
might be candidates for
forming this ultra-thin nanorod.
99
Figure 6-1. Ultra thin GaN core with InGaN buffer layer.
6.4 Chapter references
[1] Y . Ra, R. Navamathavan, J. Park, C. Lee, Nano Lett. 2013, 13, 3506.
[2] H.-S. Chen, Y .-F. Yao, C.-H. Liao, C.-G. Tu, C.-Y . Su, W.-M. Chang, Y .-W. Kiang,
C. C. Yang, Opt. Lett. 2013, 38, 3370.
[3] G. T. Wang, Q. Li, J. J. Wierer, D. D. Koleske, J. J. Figiel, Phys. Status Solidi 2014,
211, 748.
[4] S. Choi, H. J. Kim, S. S. Kim, J. Liu, J. Kim, J. H. Ryou, R. D. Dupuis, A. M.
Fischer, F. A. Ponce, Appl. Phys. Lett. 2010, 96.
[5] T.-W. Yeh, Y .-T. Lin, B. Ahn, L. S. Stewart, P. Daniel Dapkus, S. R. Nutt, Appl.
Phys. Lett. 2012, 100, 033119.
[6] S. Raychaudhuri, E. T. Yu, J. Appl. Phys. 2006, 99.
100
[7] Y . Liang, W. D. Nix, P. B. Griffin, J. D. Plummer, J. Appl. Phys. 2005, 97.
[8] M. R. Krames, O. B. Shchekin, R. Mueller-Mach, G. O. Mueller, L. Zhou, G.
Harbers, M. G. Craford, J. Disp. Technol. 2007, 3, 160.
[9] Y .-J. Liu, C.-H. Yen, K.-H. Yu, T.-P. Chen, L.-Y . Chen, T.-H. Tsai, W.-C. Liu, Jpn.
J. Appl. Phys. 2009, 48, 082104.
[10] M. Itoh, T. Kinoshita, C. Koike, M. Takeuchi, K. Kawasaki, Y . Aoyagi, Jpn. J.
Appl. Phys. 2006, 45, 3988.
[11] X. J. Chen, B. Gayral, D. Sam-Giao, C. Bougerol, C. Durand, J. Eymery, Appl.
Phys. Lett. 2011, 99, 251910.
101
Bibliography
Akabori, M., J. Takeda, J. Motohisa, T. Fukui, Nanotechnology 14, 1071.
Akasaka, T., Y . Kobayashi, S. Ando, N. Kobayashi, M. Kumagai, J. Cryst. Growth 1998,
189, 72.
Akasaki, I., H. Amano, K. Itoh, N. Koide, K. Manabe, GaAs Relat. Compd. Conf. 1992,
129, 851.
Akiyama, T., D. Ammi, K. Nakamura, T. Ito, Phys. Rev. B 2010, 81, 245317.
Akiyama, T., D. Ammi, K. Nakamura, T. Ito, Phys. Rev. B 2010, 81, 245317.
Akiyama, T., T. Yamashita, K. Nakamura, T. Ito, Jpn. J. Appl. Phys. 2009, 48, 120218.
AMANO, H., M. KITO, K. HIRAMATSU, I. AKASAKI, JAPANESE J. Appl. Phys.
PART 2-LETTERS 1989, 28, L2112.
Bergbauer, W., M. Strassburg, C. Koelper, N. Linder, C. Roder, J. Laehnemann, A.
Trampert, S. Fuendling, S. F. Li, H.-H. Wehmann, A. Waag, Nanotechnology 21.
Bergbauer, W., M. Strassburg, C. Kölper, N. Linder, C. Roder, J. Lähnemann, a Trampert,
S. Fündling, S. F. Li, H.-H. Wehmann, a Waag, Nanotechnology 2010, 21, 305201.
Bertness, K. a., a. Roshko, L. M. Mansfield, T. E. Harvey, N. a. Sanford, J. Cryst. Growth
2008, 310, 3154.
Brown, J. S., G. Koblmüller, F. Wu, R. Averbeck, H. Riechert, J. S. Speck, J. Appl. Phys.
2006, 99, 074902.
Chang, S. J., C. H. Lan, J. D. Hwang, Y . C. Cheng, W. J. Lin, J. C. Lin, H. Z. Chen,
Sputtered Indium-Tin-Oxide on p-GaN. J. Electrochem. Soc. 2008, 155, H140.
Chee, A. K. W., R. F. Broom, C. J. Humphreys, E. G. T. Bosch, J. Appl. Phys. 2011, 109,
013109.
Chen, H.-S., Y .-F. Yao, C.-H. Liao, C.-G. Tu, C.-Y. Su, W.-M. Chang, Y .-W. Kiang, C. C.
Yang, Opt. Lett. 2013, 38, 3370.
102
Chen, Q., P. Dapkus, J. Electrochem. Soc. 138, 2821.
Chen, X. J., B. Gayral, D. Sam-Giao, C. Bougerol, C. Durand, J. Eymery, Appl. Phys.
Lett. 2011, 99, 251910.
Chen, Y ., H. Wu, G. Yue, Z. Chen, Z. Zheng, Z. Wu, G. Wang, H. Jiang, Appl. Phys.
Express 2013, 6.
Choi, J.-H., S.-H. Jang, J.-S. Jang, Electron. Mater. Lett. 2013, 9, 425.
Choi, K., M. Arita, Y . Arakawa, J. Cryst. Growth 2012, 357, 58.
Choi, S., H. J. Kim, S. S. Kim, J. Liu, J. Kim, J. H. Ryou, R. D. Dupuis, A. M. Fischer, F.
A. Ponce, Appl. Phys. Lett. 2010, 96.
Choi, S., T.-H. Kim, H. O. Everitt, A. Brown, M. Losurdo, G. Bruno, A. Moto, J. Vac. Sci.
Technol. B Microelectron. Nanom. Struct. 2007, 25, 969.
Chu, H.-J., T.-W. Yeh, L. Stewart, P. D. Dapkus, Phys. Status Solidi 2010, 7, 2494.
Chung, Y . Y ., Y . S. Lin, S. W. Feng, Y . C. Cheng, E. C. Lin, C. C. Yang, K. J. Ma, C. Hsu,
H. W. Chuang, C. T. Kuo, J. S. Tsang, J. Appl. Phys. 2003, 93, 9693.
Cobet, C., T. Schmidtling, M. Drago, N. Wollschläger, N. Esser, W. Richter, R. M.
Feenstra, T. U. Kampen, J. Appl. Phys. 2003, 94, 6997.
Coltrin, M. E., C. C. Mitchell, J. Cryst. Growth 2003, 254, 35.
Coltrin, M. E., C. C. Mitchell, J. Cryst. Growth 254, 35.
Du, D., D. Srolovitz, M. Coltrin, C. Mitchell, Phys. Rev. Lett. 2005, 95, 155503.
Du, D., D. Srolovitz, M. Coltrin, C. Mitchell, Phys. Rev. Lett. 95.
Duan, X., C. Lieber, J. Am. Chem. Soc. 122, 188.
Dupuis, N., J. Décobert, P.-Y . Lagrée, N. Lagay, D. Carpentier, F. Alexandre, J. Cryst.
Growth 2008, 310, 4795.
Feenstra, R. M., Y . Dong, C. D. Lee, J. E. Northrup, J. Vac. Sci. Technol. B 2005, 23,
1174.
103
Feezell, D. F., M. C. Schmidt, S. P. DenBaars, S. Nakamura, MRS Bull. 2011, 34, 318.
Galopin, E., L. Largeau, G. Patriarche, L. Travers, F. Glas, J. C. Harmand,
Nanotechnology 2011, 22, 245606.
Gradečak, S., F. Qian, Y . Li, H.-G. Park, C. M. Lieber, Appl. Phys. Lett. 2005, 87, 173111.
Green, D. S., U. K. Mishra, J. S. Speck, J. Appl. Phys. 2004, 95, 8456.
Haskell, B. a, F. Wu, M. D. Craven, S. Matsuda, P. T. Fini, T. Fujii, K. Fujito, S. P.
DenBaars, J. S. Speck, S. Nakamura, Appl. Phys. Lett. 2003, 83, 644.
Haskell, B. A., T. J. Baker, M. B. McLaurin, F. Wu, P. T. Fini, S. P. Denbaars, J. S. Speck,
S. Nakamura, Appl. Phys. Lett. 2005, 86, 1.
Hersee, S. D., A. K. Rishinaramangalam, M. N. Fairchild, L. Zhang, P. Varangis, J. Mater.
Res. 2011, 26, 2293.
Hersee, S. D., M. Fairchild, A. K. Rishinaramangalam, M. S. Ferdous, P. M. V . L. Zhang,
B. S. Swartzentruber, A. A. Talin, Electron. Lett. 2009, 45.
Hersee, S. D., X. Sun, X. Wang, Nano Lett. 6, 1808.
Hikosaka, T., T. Shioda, Y . Harada, K. Tachibana, N. Sugiyama, S. Nunoue, Phys. Status
Solidi 2011, 8, 2016.
Ho, J.-K., C.-S. Jong, C. C. Chiu, C.-N. Huang, K.-K. Shih, L.-C. Chen, F.-R. Chen, J.-J.
Kai, J. Appl. Phys. 1999, 86, 4491.
Hoffmann, V ., C. Netzel, U. Zeimer, A. Knauer, S. Einfeldt, F. Bertram, J. Christen, M.
Weyers, G. Trnkle, M. Kneissl, J. Cryst. Growth 2010, 312, 3428.
Hong, Y . J., C.-H. Lee, A. Yoon, M. Kim, H.-K. Seong, H. J. Chung, C. Sone, Y . J. Park,
G.-C. Yi, Adv. Mater. 2011, 23, 3284.
Huang, Y ., X. Duan, Y . Cui, C. M. Lieber, Nano Lett. 2002, 2, 101.
Imer, B. M., F. Wu, S. P. Denbaars, J. S. Speck, Improved quality (112??0) a-plane GaN
with sidewall lateral epitaxial overgrowth. Appl. Phys. Lett. 2006, 88.
Imer, B., F. Wu, S. DenBaars, J. Speck, Appl. Phys. Lett. 88.
104
Ito, T., T. Akiyama, K. Nakamura, Semicond. Sci. Technol. 2012, 27, 024010.
Itoh, M., T. Kinoshita, C. Koike, M. Takeuchi, K. Kawasaki, Y . Aoyagi, Jpn. J. Appl.
Phys. 2006, 45, 3988.
Jacobs, K., B. Van Daele, M. R. Leys, I. Moerman, G. Van Tendeloo, In Journal of
Crystal Growth; 2003; Vol. 248, pp. 498–502.
Jindal, V ., F. Shahedipour-Sandvik, J. Appl. Phys. 2009, 106, 083115.
Kangawa, Y ., T. Ito, a. Taguchi, K. Shiraishi, T. Ohachi, Surf. Sci. 2001, 493, 178.
Kim, D. W., Y . J. Sung, J. W. Park, G. Y . Yeom, In Thin Solid Films; 2001; V ol. 398-399,
pp. 87–92.
Kim, K. S., G. M. Yang, H. J. Lee, Solid. State. Electron. 1999, 43, 1807.
Kim, K.-K., H. Kim, S.-N. Lee, S. Cho, Electron. Mater. Lett. 2011, 7, 145.
Ko, S.-M., J.-H. Kim, Y .-H. Ko, Y . H. Chang, Y .-H. Kim, J. Yoon, J. Y . Lee, Y .-H. Cho,
Cryst. Growth Des. 2012, 12, 3838.
Koester, R., J. S. Hwang, C. Durand, D. L. S. Dang, J. Eymery, Nanotechnology 21.
Koida, T., S. F. Chichibu, T. Sota, M. D. Craven, B. a Haskell, J. S. Speck, S. P. DenBaars,
S. Nakamura, Appl. Phys. Lett. 2004, 84, 3768.
Krames, M. R., O. B. Shchekin, R. Mueller-Mach, G. O. Mueller, L. Zhou, G. Harbers,
M. G. Craford, J. Disp. Technol. 2007, 3, 160.
Kuykendall, T., P. Pauzauskie, S. Lee, Y . Zhang, J. Goldberger, P. Yang, Nano Lett. 3,
1063.
Laaksonen, K., M. G. Ganchenkova, R. M. Nieminen, J. Phys. Condens. Matter 2009, 21,
015803.
Lee, C. D., R. M. Feenstra, J. E. Northrup, L. Lymperakis, J. Neugebauer, Appl. Phys.
Lett. 2003, 82, 1793.
Li, S. F., S. Fuendling, X. Wang, S. Merzsch, M. a. M. Al-Suleiman, J. D. Wei, H.-H.
Wehmann, A. Waag, W. Bergbauer, M. Strassburg, Cryst. Growth Des. 2011, 11, 1573.
105
Liang, Y ., W. D. Nix, P. B. Griffin, J. D. Plummer, J. Appl. Phys. 2005, 97.
Lin, Y . C., S. J. Chang, Y . K. Su, T. Y . Tsai, C. S. Chang, S. C. Shei, C. W. Kuo, S. C.
Chen, Solid. State. Electron. 2003, 47, 849.
Lin, Y .-T., T.-W. Yeh, P. D. Dapkus, Nanotechnology 2012, 23, 465601.
Liu, C., P. a. Shields, S. Denchitcharoen, S. Stepanov, a. Gott, W. N. Wang, J. Cryst.
Growth 2007, 300, 104.
Liu, Y .-J., C.-H. Yen, K.-H. Yu, T.-P. Chen, L.-Y . Chen, T.-H. Tsai, W.-C. Liu, Jpn. J. Appl.
Phys. 2009, 48, 082104.
Luryi, S., E. Suhir, Appl. Phys. Lett. 1986, 49, 140.
Lymperakis, L., J. Neugebauer, Phys. Rev. B 2009, 79, 241308.
Lyons, J. L., A. Janotti, C. G. Van de Walle, Appl. Phys. Lett. 2010, 97, 152108.
Maruska, H. P., Appl. Phys. Lett. 1969, 15, 327.
Miyake, H., A. MOTOGAITO, K. HIRAMATSU, Jpn. J. Appl. Phys. 1999, 38, 1000.
Nakamura, S., N. Iwasa, M. Senoh, T. Mukai, Japanese J. Appl. Physics, Part 1 Regul.
Pap. Short Notes Rev. Pap. 1992, 31, 1258.
Neugebauer, J., T. Zywietz, M. Scheffler, J. Northrup, H. Chen, R. Feenstra, Phys. Rev.
Lett. 2003, 90, 056101.
Northrup, J. E., Appl. Phys. Lett. 2005, 86, 122108.
Obloh, H., K. . Bachem, U. Kaufmann, M. Kunzer, M. Maier, A. Ramakrishnan, P.
Schlotter, Self-compensation in Mg doped p-type GaN grown by MOCVD. J. Cryst.
Growth 1998, 195, 270–273.
Park, E.-H., J.-S. Park, T.-K. Yoo, J. Cryst. Growth 2004, 272, 426.
People, R., J. C. Bean, Appl. Phys. Lett. 1985, 47, 322.
Piprek, J., Phys. Status Solidi 2010, 207, 2217.
106
Ra, Y ., R. Navamathavan, J. Park, C. Lee, Nano Lett. 2013, 13, 3506.
Rao, M., D. Kim, S. Mahajan, Appl. Phys. Lett. 2004, 85, 1961.
Raychaudhuri, S., E. T. Yu, J. Appl. Phys. 2006, 99.
Reshchikov, M. a., H. Morkoç, J. Appl. Phys. 2005, 97, 061301.
S. Chung, V . Wheeler, R. Myers-Ward, L. O. Nyakiti, C. R. Eddy, D. K. Gaskill, M.
Skowronski, Y . N. Picard, J. Appl. Phys. 2011, 110, 014902.
Sanchez-Garcia, M., E. Calleja, E. Monroy, F. Sanchez, F. Calle, E. Munoz, R. Beresford,
J. Cryst. Growth 183, 23.
Sarkissian, R., S. T. Roberts, T.-W. Yeh, S. Das, S. E. Bradforth, J. O’Brien, P. Daniel
Dapkus, Appl. Phys. Lett. 2013, 103, 041123.
Sawicka, M., H. Turski, M. Siekacz, J. Smalc-Koziorowska, M. Kryśko, I. Dzięcielewski,
I. Grzegory, C. Skierbiszewski, Phys. Rev. B 2011, 83, 245434.
Segev, D., C. G. Van de Walle, Surf. Sci. 2007, 601, L15.
Sekiguchi, H., K. Kishino, A. Kikuchi, Appl. Phys. Lett. 2010, 96, 231104.
Seo, Y ., C. Kim, Appl. Phys. Lett. 2010, 97, 101902.
Shioda, T., M. Sugiyama, Y . Shimogaki, Y . Nakano, Phys. Status Solidi 2010, 207, 1375.
Shioda, T., Y . Tomita, M. Sugiyama, Y . Shimogaki, Y . Nakano, Jpn. J. Appl. Phys. 2007,
46, L1045.
Singh, J., Electronic and Optoelectronic properties of Semiconductor structures.pdf 2003.
Smith, a. R., R. M. Feenstra, D. W. Greve, M.-S. Shin, M. Skowronski, J. Neugebauer, J.
E. Northrup, Appl. Phys. Lett. 1998, 72, 2114.
Song, J. O., J.-S. H. J.-S. Ha, T.-Y . S. T.-Y . Seong, IEEE Trans. Electron Devices 2010,
57, 42.
Speck, J. S., S. F. Chichibu, Nonpolar and Semipolar Group III Nitride-Based Materials.
MRS Bull. 2009, 34, 304–312.
107
Storm, K., F. Halvardsson, M. Heurlin, D. Lindgren, A. Gustafsson, P. M. Wu, B.
Monemar, L. Samuelson, Nat. Nanotechnol. 2012, 7, 718.
Sun, Q., B. H. Kong, C. D. Yerino, T.-S. Ko, B. Leung, H. K. Cho, J. Han, J. Appl. Phys.
2009, 106, 123519.
Sun, Q., C. D. Yerino, T. S. Ko, Y . S. Cho, I.-H. Lee, J. Han, M. E. Coltrin, J. Appl. Phys.
2008, 104, 093523.
Tsurumi, D., K. Hamada, Anal. Technol. 2013, 1, 141.
Van de Walle, C. G., J. Neugebauer, J. Appl. Phys. 2004, 95, 3851.
V olotsenko, I., M. Molotskii, Z. Barkay, J. Marczewski, P. Grabiec, B. Jaroszewicz, G.
Meshulam, E. Grunbaum, Y . Rosenwaks, J. Appl. Phys. 2010, 107.
Wang, G. T., Q. Li, J. J. Wierer, D. D. Koleske, J. J. Figiel, Phys. Status Solidi 2014, 211,
748.
Wernicke, T., L. Schade, C. Netzel, J. Rass, V . Hoffmann, S. Ploch, A. Knauer, M.
Weyers, U. Schwarz, M. Kneissl, Semicond. Sci. Technol. 2012, 27, 024014.
X. Zhou, J. Chesin, S. Crawford, S. Gradečak, Nanotechnology 2012, 23, 285603.
Xu, J., M. F. Schubert, A. N. Noemaun, D. Zhu, J. K. Kim, E. F. Schubert, M. H. Kim, H.
J. Chung, S. Yoon, C. Sone, Y . Park, Appl. Phys. Lett. 2009, 94.
Yakovlev, E. V ., R. a. Talalaev, a. S. Segal, a. V . Lobanova, W. V . Lundin, E. E. Zavarin,
M. a. Sinitsyn, a. F. Tsatsulnikov, a. E. Nikolaev, J. Cryst. Growth 2008, 310, 4862.
Yamashita, T., T. Akiyama, K. Nakamura, T. Ito, Jpn. J. Appl. Phys. 2009, 48, 120201.
Ye, H., P. Lu, Z. Yu, Y . Song, D. Wang, S. Wang, Nano Lett. 2009, 9, 1921.
Yeh, T.-W., Dissertation 2013.
Yeh, T.-W., Y .-T. Lin, B. Ahn, L. S. Stewart, P. Daniel Dapkus, S. R. Nutt, Appl. Phys.
Lett. 2012, 100, 033119.
Yeh, T.-W., Y .-T. Lin, L. S. Stewart, P. D. Dapkus, R. Sarkissian, J. D. O’Brien, B. Ahn, S.
R. Nutt, Nano Lett. 2012, 12, 3257.
108
Yu, H., L. K. Lee, T. Jung, P. C. Ku, Appl. Phys. Lett. 2007, 90, 141906.
Zhang, X., P. D. Dapkus, D. H. Rich, Appl. Phys. Lett. 2000, 77, 1496.
Zhao, H., G. Liu, R. A. Arif, N. Tansu, Solid. State. Electron. 2010, 54, 1119.
Zhong, Z., F. Qian, D. Wang, C. Lieber, Nano Lett. 3, 343.
Abstract (if available)
Abstract
GaN based blue LEDs are the foundation for solid state lighting, which has the potential to approach the theoretical upper efficacy limit of around 500 lumens per watt. GaN is a polar III-V semiconductor with a strong polarization field inside the material depending on the crystal orientation. This strong polarization and piezoelectric field along the polar direction (c-direction) will bend the band structure of the active regions of LEDs and enhance the leakage current under forward bias if MQWs are grown on this polar c-plane. However, the commercial GaN based LED is still grown on the polar c-plane since the nonpolar substrate is highly cost inefficient due to the technology limitations. The nanorod-based LED (nanoLED), which relies on polar c-plane, possess nonpolar m-plane as their sidewalls and therefore are promising to resolve the current difficulties. ❧ In this dissertation, we will discuss in detail the whole procedures involving the nanoLED fabrication, which include substrate pattern preparation required for selective area growth, the n-type nanorod templates, the active region (MQWs) growth, the p-type shell growth, and the 3D p-type contact fabrication. Our prototype nanoLEDs successfully demonstrate light emission under forward bias, but further optimization is still required to fully employ the advantages of their nonpolar-based active region.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
GaN nanostructures grown by selective area growth for solid-state lighting
PDF
Efficient yellow and green emitting InGaN/GaN nanostructured QW materials and LEDs
PDF
GaN power devices with innovative structures and great performance
PDF
Nanophotonic light management in thin film silicon photovoltaics
PDF
GaAs nanowire optoelectronic and carbon nanotube electronic device applications
PDF
Growth, characterization of gallium arsenide based nanowires and application in photovoltaic cells
PDF
Efficiency droop in indium gallium nitride light emitters: an introduction to photon quenching processes
PDF
Light management in nanostructures: nanowire solar cells and optical epitaxial growth
PDF
Nanostructure electronic band modeling for solar cell and lighting applications
PDF
Application of optical forces in microphotonic systems
PDF
Integrating material growth and device physics: building blocks for cost effective emerging electronics and photonics devices
PDF
Optimizing nanoemitters using quasi-aperiodicity and photoelectrochemical etching
PDF
Photodetector: devices for optical data communication
PDF
Multilayer grown ultrathin nanostructured GaAs solar cells towards high-efficiency, cost-competitive III-V photovoltaics
PDF
Nano-engineered devices for optical application, biomedical detection and circuit accelerator
PDF
Nano-fabricated devices in electrochemistry and cancer therapy
PDF
Synthesis, characterization, and device application of two-dimensional materials beyond graphene
PDF
Carbon material-based nanoelectronics
PDF
The selective area growth and coalescence of indium phosphide nanostripe arrays on silicon through MOCVD for NIR monolithic integration
PDF
Energy proportional computing for multi-core and many-core servers
Asset Metadata
Creator
Lin, Yen-Ting
(author)
Core Title
Nanorod-based InGaN/GaN core-shell nanoLEDs
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Electrical Engineering
Publication Date
02/13/2015
Defense Date
12/03/2014
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
GaN,LEDs,nanoLEDs,nanorods,OAI-PMH Harvest,SAG
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Dapkus, Paul Daniel (
committee chair
), Nakano, Aiichiro (
committee member
), Povinelli, Michelle L. (
committee member
), Wu , Wei (
committee member
)
Creator Email
b91901142@gmail.com,linyenti@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-532332
Unique identifier
UC11297801
Identifier
etd-LinYenTing-3189.pdf (filename),usctheses-c3-532332 (legacy record id)
Legacy Identifier
etd-LinYenTing-3189.pdf
Dmrecord
532332
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Lin, Yen-Ting
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
GaN
LEDs
nanoLEDs
nanorods
SAG