micro-structure engineering of ingan/gan quantum wells for high brightness light emitting devices
TRANSCRIPT
Micro-structure Engineering of InGaN/GaN Quantum
Wells for High Brightness Light Emitting Devices
Presented by
Chao Shen
沈超
In Partial Fulfillment of the Requirements
For the Degree of
Master of Science
King Abdullah University of Science and Technology,
Thuwal, Kingdom of Saudi Arabia
May 2013
2
The thesis of Chao Shen is approved by the examination committee:
Committee Chairperson: Dr. Boon S. Ooi
Committee Member: Dr. Osman Bakr
Committee Member: Dr. Xixiang Zhang
3
ABSTRACT
Micro-structure Engineering of InGaN/GaN Quantum Wells
for High Brightness Light Emitting Devices
Chao Shen
With experimental realization of micro-structures, the feasibility of achieving high
brightness, low efficiency droop blue LED was implemented based on InGaN/GaN micro-
LED-pillar design. A significantly high current density of 492 A/cm2 in a 20 μm diameter
(D) micro-LED-pillar was achieved, compared to that of a 200 μm diameter LED (20
A/cm2), both at 10 V bias voltage. In addition, an increase in sustained quantum
efficiency from 70.2% to 83.7% at high injection current density (200 A/cm2) was
observed in micro-LED-pillars in conjunction with size reduction from 80 μm to 20 μm. A
correlation between the strain relief and the electrical performance improvement was
established for micro-LED-pillars with D < 50 μm, apart from current spreading effect.
The degree of strain relief and its distribution were further studied in micro-LED-
pillars with D ranging from 1 μm to 15 μm. Significant wavenumbers down-shifts for E2
and A1 Raman peaks, together with the blue shifted PL peak emission, were observed in
as-prepared pillars, reflecting the degree of strain relief. A sharp transition from strained
to relaxed epitaxy region was discernible from the competing E2 phonon peaks at 572
cm-1 and 568 cm-1, which were attributed to strain residue and strain relief, respectively.
A uniform strain relief at the center of micro-pillars was achieved, i.e. merging of the
competing phonon peaks, after Rapid Thermal Annealing (RTA) at 950℃ for 20 seconds,
4
phenomenon of which was observed for the first time. The transition from maximum
strain relief to a uniform strain relief was found along the narrow circumference (< 2.5
μm) of the pillars from the line-map of Raman spectroscopy.
The extent of strain relief is also examined considering the height (L) of micro-LED-
pillars fabricated using FIB micro-machining technique. The significant strain relief of up
to 70% (from -1.4 GPa to -0.37 GPa), with a 71 meV PL peak blue shift, suggested that
micro-LED-pillar with D < 3 µm and L > 3 µm in the array configuration would allow the
building of practical devices. Overall, this work demonstrated a novel top-down
approach to manufacture large effective-area, high brightness emitters for solid-state
lighting applications.
5
摘 要
应用于高亮度发光器件的氮化铟镓/氮化镓多重量子阱
微结构的研究
沈超
本文探讨了在氮化铟镓量子阱微柱的基础上构建高电流下具有较高发光强度和减轻
效率下降的发光器件的可行性。实验表明氮化铟镓量子阱微结构具有卓越的电学性
能：在 10V 电压驱动下，直径为 20 微米的氮化铟镓量子阱微柱能够获得高达 492
A/cm2 的注入电流密度，而直径为 200 微米的大面积器件仅获得 20 A/cm2。此外，
微柱在高注入电流下受量子效率下降效应的影响也有减弱：在 200 A/cm2 的高电流
注入下，直径为 80 至 20 微米的量子阱微柱能够获得 70.2%到 83.7%的峰值效率。
研究表明这些性能的提升源自于微结构中应力的释放以及电流扩散的增强。特别的，
我们指出前者在直径小于 50 微米的微柱中具有重要影响。
本文进一步研究了在较小氮化铟镓量子阱微柱（直径在 1 微米至 15 微米）的结构
中应力释放的机理及其分布。通过显微拉曼光谱中氮化镓 E2 和 A1 声子峰的显著蓝
移以及光致发光光谱中发光峰的蓝移，确认了微柱中应力释放的程度。我们首次观
察到微柱中氮化镓分裂的 E2 声子峰出现在 572 cm-1 和 568 cm-1，它们分别被确认为
源自结构中应力保持区和应力释放区。当经过 20 秒，950℃的高温快速热退火之后，
分裂的 E2 声子峰重新合一。这说明在经过退火的微结构获得了均匀的应力释放。
通过拉曼光谱的线扫描，我们发现氮化铟镓量子阱微柱的边缘部分具有最大程度的
应力释放，微柱中部为应力均匀释放区，而应力的变化则集中在微柱外延较窄的环
状区域（小于 2.5 微米）。随着微柱直径的减小，应力均匀释放区的应力水平逐渐
接近最大程度的应力释放。所以直径小于 3 微米的极小氮化铟镓量子阱微柱可以
获得均匀的应力释放。
除了探究氮化铟镓量子阱微柱大小对应力释放的影响，我们也利用聚焦离子束制作
了不同刻蚀深度（即微结构高度）的微柱来研究高度的影响。结果表明直径小于 3
微米，高度大于 3 微米的氮化铟镓量子阱微柱具有约 70%的显著应力释放（从-1.4
GPa 到 -0.37 GPa），所以由这种微柱组成的微结构阵列是适合构建高亮度发光器
件的结构单元。本文论证了一种至顶向下的制备高亮度发光器件的方法，为生产高
性能固态光源提出了新的思路。
6
ACKNOWLEDGEMENTS
With my earnest thank-you to King Abdullah Bin Abdulaziz Al Saud, and KAUST’s
proceeding and current management who realize his dream and vision. I will always
cherish the tremendous experiences gained in the past two years.
I sincerely thank my supervisor, Prof. Boon S. Ooi, a dedicated scientist and
entrepreneur with great passion and insights. I am always impressed by his charisma
when he presents scientific ideas. His leadership and constant support helped me
accomplished the thesis research. I would also like to express my appreciation to our
laboratory manager, Dr. Tien Khee Ng, who is always helpful and accommodating, and
yet demanding top performance from me as an engineer and scientific worker. He
introduced me to this exciting field and spent so much time on discussing creative
solutions to existing and new problems during my research. Besides, I thank Prof.
Mohammad Alsunaidi (KFUPM) for his suggestions and discussion during my study. The
great people I met at KAUST Photonics Lab would always be in my thank list as well. Dr.
Zhang Yaping, Dr. Adel Najar provided support in my project design. I learnt a lot from
the ZnO work done by Li Qian, “InGaN nanomushroom” discovered by Anwar Gasim and
microspheres investigated by Yasser Khan. I would like to acknowledge many other lab
mates, such as Dr. AbdulMajid Mohammed, Dr. Hind Althib, Ahmed AlJabr, Ahmed Ben
Slimane, Bilal Janjua, Damián San Román, Hala Al-Hashim, Hassan Oubei, Jameelah
Alzahrani, Mohammed Zahed, Pawan Mishra, Rami EiAfandy and Youssef Ibrahim for
sharing various ideas on the study of optoelectronics and making my days here
enjoyable and amusing.
7
Many of the experiments wouldn’t have been accomplished smoothly without the
collaborative effort of scientists and engineers from KAUST’s core-laboratories. I would
like to offer my great acknowledgement to Dr. Yang Yang for interpreting micro-Raman
and PL spectra of thin films; Dr. Wang Xianbin and Dr. Wang Zhihong for the
development of lithography process; Dr. Yue Weisheng and Dr. Zhao Lan for guidance of
SEM imaging; Dr. Dongkyu Cha and Mr. Li Jun for FIB optimization; Dr. Chen Longqing
for wet chemical process; Mr. Basil Chew for mask making; Mr. Wang Qingxiao for TEM
imaging; Mr. Kim Chong Wong for the help in designing thermal annealing process; Mr.
Ahad Syed and Mr. Andrew Rubin for keeping ICP-RIE at tip-top condition; Dr. Zhao
Chao and Dr. Li Liang for discussion of strain effect in semiconductors; Dr. Zhang Qiang
and Dr. Yao Yingbang for the preparation of ITO thin films, and last but not least, Ms.
Razan Aboljadayel for providing further information on the profiler.
The collaborative environment at KAUST allowed me to interact with professors
and researchers from other labs and disciplines. I have benefited from discussions with
Dr. Cheng Yingchun, especially his in-depth understanding of solid-state physics and
experience in stimulating strain effect in large Bandgap nitride materials. I would also
like to thank Prof. Iman Roqan, PI of Semiconductor Spectroscopy and Devices
Laboratory, for her course on Advanced Semiconductor Physics and the suggestions on
spectroscopy techniques. Mr. Fan Yiqiang, Mr. Li Bodong, Mr. Li Fuquan, Mr. Sun Jian in
CEMSE division and Mr. Alfonso Caraveo in Physical Sciences and Engineering division
who shared their experience in device fabrication, which helped me in the process
development in this thesis as well.
8
Finally, I thank my family for their unconditional love and unweaving morale support
and many friends, who had enriched my life experience.
Shen Chao
May,2013
Thuwal, Kingdom of Saudi Arabia
9
To…
The Tao of nature, with a heart of reverence and the earth, a lovely planet embraces
humans.
My beloved parents, who build the concept of home and love, you are always there
offering me unlimited support.
May this work contribute to the development of a better world, and benefits mankind.
10
Table of Contents
ABSTRACT ............................................................................................................................ 3
ACKNOWLEDGEMENTS ....................................................................................................... 6
LIST OF ABBREVIATIONS ................................................................................................... 12
LIST OF FIGURES ................................................................................................................ 13
LIST OF TABLES .................................................................................................................. 15
1 Chapter One: Background ......................................................................................... 16
1.1 Introduction........................................................................................................ 16
1.2 III-nitride materials for solid-state lighting ........................................................ 21
1.3 Lighting efficiency ............................................................................................... 27
1.4 Motivation and significance of this thesis .......................................................... 29
2 Chapter Two: Fabrication of InGaN/GaN MQW micro-LED-pillars ........................... 33
2.1 Process Overview ............................................................................................... 33
2.2 Devices Characterization .................................................................................... 35
2.2.1 Optical Microscopy ..................................................................................... 35
2.2.2 Transmission Electron Microscopy ............................................................. 36
2.2.3 Electroluminescence ................................................................................... 38
2.2.4 Electrical property ....................................................................................... 41
2.2.5 Efficiency Characteristic .............................................................................. 43
2.3 Materials Characterization ................................................................................. 45
2.3.1 Photoluminescence..................................................................................... 45
2.3.2 Raman spectroscopy ................................................................................... 49
3 Chapter Three: Strain Engineering in InGaN/GaN MQW micro-LED-pillars .............. 53
3.1 Introduction........................................................................................................ 53
3.2 Experiments ........................................................................................................ 55
3.3 Results and discussion ........................................................................................ 60
3.4 Summary ............................................................................................................ 71
4 Chapter Four: Building up of micro-LED-pillar arrays ................................................ 72
4.1 Introduction........................................................................................................ 72
4.2 Experiments ........................................................................................................ 73
4.3 Results and discussion ........................................................................................ 75
5 Conclusions and future work ..................................................................................... 77
11
REFERENCES ...................................................................................................................... 80
APPENDICES ...................................................................................................................... 89
Appendix A. Design of mask for large-area LEDs and micro-LED-pillars fabrication .... 89
Appendix B. Fabrication process flow of InGaN/GaN MQW LEDs ................................ 91
Appendix C. Programs for I-V and L-I measurements ................................................... 94
Interface for I-V measurements: ............................................................................... 94
Labview code for I-V measurements (selected): ....................................................... 95
Interface for L-I measurements: ................................................................................ 96
Labview code for I-V measurements (selected): ....................................................... 97
12
LIST OF ABBREVIATIONS
EL … Electroluminescence
EQE … External Quantum Efficiency
FIB … Focused Ion Beam
FWHM … Full Width at Half Maximum
HRTEM … High Resolution Transmission Electron Microscopy
IQE … Internal Quantum Efficiency
ITO … Indium Tin Oxide
LD … Laser Diode
LED … Light Emitting Diode
MBE … Molecular Beam Epitaxy
MOCVD … Metal – Organic Chemical Vapor Deposition
MQW … Multiple Quantum Well
OLED … Organic Light Emitting Diode
µPL … Micro Photoluminescence
PL … Photoluminescence
QCSE … Quantum Confined Stark Effect
QW … Quantum Well
SEM … Scanning Electron Microscope
SSL … Solid State lighting
TEM … Transmission Electron Microscope
TIR … Total Internal Reflection
13
LIST OF FIGURES
Figure 1 Design of electric Lamp in U.S. patent 223898, by Thomas Edison.................... 17
Figure 2 Schematic diagram of the principle of a typical LED. ......................................... 19
Figure 3 Energy vs. crystal momentum for a semiconductor with an indirect band-gap. 21
Figure 4 Bandgap energy and lattice constant of various III-V semiconductors at room temperature. ..................................................................................................................... 22
Figure 5 Bandgap energy and lattice constant of various III-Nitrides at room temperature. ..................................................................................................................... 23
Figure 6 Three dominant ways to produce white light based on LEDs. CRI stands for Color Rendering Index, which describes its ability to reproduce the natural colors [14]. 24
Figure 7 EQE vs. Current injection at three LED devices. A droop of EQE is observed when increasing injection current density [54]. ............................................................... 29
Figure 8 InP nanowires prepared by top-down approach (a) [61] and bottom-up approach (b) [62]. A better control of size, height and shape is observed in (a). ............ 31
Figure 9 Schematic structure of InGaN/GaN LED wafer. .................................................. 34
Figure 10 Optical images of sample surface before (a) and after (b) proper cleaning. Scale bar shows a 100 µm distance. ................................................................................. 35
Figure 11 Optical images of fabricated large-area circular LEDs (a) and square LEDs with micro-LED-pillars (b). ........................................................................................................ 35
Figure 12. Cross-sectional TEM image of InGaN/GaN MQW LED structure. The capping layer is a 500 nm sputtered Al layer for TEM imaging. The scale bar is 0.5 µm. .............. 36
Figure 13. Cross-sectional TEM image of InGaN/GaN MQWs. The 10 nm GaN barrier and 3 nm InGaN quantum well is clearly observed. The scale bar is 10 nm. .......................... 37
Figure 14 (a) Z-contrast TEM image of InGaN/GaN MQWs. (b) EDX of Ga, In and N concentration scanning along with the line in (a). ........................................................... 38
Figure 15. EL, I-V and L-I measurement system set-up. ................................................... 39
Figure 16. EL spectrum from a large-area LED. ................................................................ 40
Figure 17. Light emission observed from microscope when DC bias is applied to an InGaN/GaN MQW large-area LED without ITO current spreading layer (a) and with 170 nm ITO current spreading layer (b). .................................................................................. 40
Figure 18. Current – Voltage measurements showing the calculation of threshold voltage (Vth) (a) and sheet resistance (Rs) (b). ............................................................................. 41
Figure 19. Current Density vs. voltage in various micro-LED-pillars compared with a large-area one. .................................................................................................................. 42
Figure 20. (a) L-I relationship recorded from micro-LED-pillar with D = 50 µm. (b) EQE vs. current density curve calculated from (a). ....................................................................... 43
Figure 21. Nominalized EQE vs. current density in various micro-LED-pillars compared with a large-area one. ....................................................................................................... 44
Figure 22 The photoluminescence process in a direct-gap semiconductor [73]. ............ 46
Figure 23. Integration of 405 nm laser with HR 800 micro-PL system. ............................ 47
Figure 24. PL peak positions as a function of micro-LED-pillar size . ................................ 48
Figure 25. PL integrated intensity as a function of micro-LED-pillar size. ........................ 49
Figure 26. Schematic drawing shows the principles of Raman and Rayleigh scattering. . 50
14
Figure 27. Raman spectrum collected from the GaN based LED epitaxy. ........................ 51
Figure 28. Optical phonon modes in the Wurtzite structure. .......................................... 52
Figure 29. Basic four functions of FIB processing: (a) Milling, (b) Deposition, (c) implantation and (d) imaging. .......................................................................................... 56
Figure 30. SEM images of FIB fabricated micro-LED-pillars with D = 1 µm to 15 µm with untitled (a)and tilted (b) stage. ......................................................................................... 58
Figure 31. SEM images of FIB fabricated micro-LED-pillars with D = 15 µm (a), D = 10 µm (b), D = 5 µm (c), D = 3 µm (d), D = 1 µm (e) identifying the sidewall profiles and (f) surrounding of 3 pillars showed no observable redoposition. ......................................... 59
Figure 32. Room temperature micro-Raman Spectra from micro-LED-pillar and bulk-area LED. ................................................................................................................................... 61
Figure 33. Room temperature micro-PL Spectra of micro-LED-pillar and bulk-area LED with Gaussians fitting. ....................................................................................................... 62
Figure 34. Raman spectra of a micro-LED-pillar with D =1 µm measured when independently irradiated with four different laser sources. ............................................ 63
Figure 35. Line-mapping of fitted Raman peaks across the micro pillar with D = 15 µm. A strain-relief ring region is identifiable at the vicinity of the micro-pillar. ........................ 65
Figure 36. The width of the strain-relief ring region derived from Raman line-mapping as a function of pillar diameter. ............................................................................................ 66
Figure 37 . (a). E2 phonon Raman peak positions at the center of each micro-pillar, compared with that of the bulk area. (b). FWHM of fitted E2 Raman peak as a function of pillar diameter. .................................................................................................................. 68
Figure 38. Stack of Raman spectra from micro-LED-pillars with different sizes. ............. 69
Figure 39. Raman spectra from bulk LED epitaxy and micro-LED-pillars with different micro-machining depth. .................................................................................................... 70
Figure 40. (a), (b) The SEM images of nanopillars fabricated by self-assembly Ni dots lithography. (c) cross-sectional SEM and (d)cross-sectional TEM images of nanopillars prepared by AAO template etching. ................................................................................ 73
Figure 41. Design of micro-LED-pillar array for FIB milling. .............................................. 74
Figure 42. SEM images of FIB fabricated micro-LED-pillar arrays with D = 2 µm (a) and 1 µm (b). ............................................................................................................................... 74
Figure 43. PL of micro-LED-pillar arrays compared with the bulk area. ........................... 75
Figure 44. Raman spectra of micro-LED-pillar arrays compared with the bulk area. ...... 76
Figure 45. Mask design for MESA of large area LEDs and micro-LED-pillars. ................... 89
Figure 46. Mask design for N contact of large area LEDs and micro-LED-pillars. ............. 90
15
LIST OF TABLES
Table 1 Materials Parameters of GaN and InN [23]. ......................................................... 25
Table 2 Physical properties of sapphire, from Ref [32]. ................................................... 26
Table 3 FIB parameters in the fabrication process ........................................................... 57
16
1 Chapter One: Background
1.1 Introduction
Two centuries ago, Humphry Davy succeeded in obtaining light by electrical power
when passing the current through a thin strip of platinum. However, it was Thomas
Edison who eventually the well-known improvement of light bulbs, with his first patent
filed on Oct. 14, 1878, that humankind began to benefit from electric-powered lightings.
His design (see Figure 1) can still be widely seen in our daily life. However, this
technique is of quite low efficiency as most of the energy is converted into heat emitted
in the near infra-red wavelength regime instead of visible light generation. As a result,
the electrical power is wasted and the lifetime is limited by the acceleration of materials
aging. In U.S. alone, about 22 percent of the electricity produced every year is
consumed in lighting, roughly equals $50 billion/year [1]. Solid-state lighting (SSL),
referring to a type of lighting that uses semiconductor light-emitting diodes (LEDs) or
organic light-emitting diodes (OLED) as illumination sources, has the potential to
revolutionize the lighting industry. It was estimated that the energy consumption could
be reduced by a factor of 3 to 6 times by employing solid-state lighting [2]. Compared
with an effective efficiency of 4% from incandescent lamps and that of 17% from
fluorescent lamps, the effective efficiency of white LED has reached over 20% in
commercial products and 35% in labs [3]. Recent research work discussed the feasibility
of “ultra-efficiencies” of 70%, or greater [4]. The limitations in current technology and
the ways to overcome them have attracted great amounts of interests from researchers
all over the world.
17
Recent LED-based SSL technology has its roots in the initial demonstration of a
high performance blue LED-based using GaN in 1993 and a white LED through
combining the blue LED with a YAG phosphor a few years later [5, 6]. Following these
announcements, there has been an explosion of R&D activity worldwide. In EU,
Photoinics21 European Technology Platform was established to coordinate the R&D
activity with a total investment of approximately €110 million till the year of 2012. In the
area of LED-based SSL technology there are a number of projects currently underway,
for example, the EU's Seventh Framework Program for Research (FP7) including
Figure 1 Design of electric Lamp in U.S. patent 223898, by Thomas Edison.
18
FAST2Light, SSL4EU, OLED100.eu, SCOOP, LAMP, IMOLA, CSSL, ENLIGHT, SMASH and
THERMOGRIND [7]. China has identified LED manufacturing as a strategic market and
has provided significant financial incentives, over US$1.7 billion, for companies and
institutions. Thirteen industrial science parks have been established in mainland China
for SSL R&D and Industrial Technology Research Institute (ITRI) was set up in Taiwan.
Patent activity in mainland China has increased significantly in the past few years with
28,912 LED related patents at the end of 2009, including 59% on applications and 13%
on packaging. In South Korea, private sectors such as Samsung electronics, LG
electronics, Hynix and Hyundai, play a key role in promoting R&D of SSL with total
amount of €90-120 million invested.
Despite the rapid pace of its development, SSL has not yet come close to
unleashing its full potential. Significant amount of research and technological
advancement remains to improve its performance and reduce the cost further. As
mentioned, LEDs and OLEDs are two major categories in SSL. LEDs are semi-conducting
devices based on inorganic materials that produce light with external electrical current
injection. OLEDs are made up of organic or carbon-based materials. Unlike LEDs, which
are point sources in most of cases, OLEDs are made in sheets that provide a diffuse-area
light source. However, compared with LED technology, OLED technology is less mature
while their efficiency, lifetime, and light output can hardly compete with those of LEDs.
Targeting at a practical general illumination source, our work will focus on the efficiency
improvement of LEDs.
19
A LED consists of semiconductor materials doped with different kinds of impurities
to create a p-n junction (see Figure 2). Without bias, the Fermi level across the device
lays at a flat level with a built-in field formed the depletion region, stopping the holes
and electrons meeting in the junction. When a positive bias is applied, the electrons and
holes begin to flow into the junction from electrodes. As a result, they recombine and
fall to a lower energy level with extra energy released in the form of a photon. The band
gap energy of the materials forming the junction determined the wavelength of the light
emitted, and thus its color. A few more concepts will be discussed to further clarify the
principle.
Radiative and non-radiative recombination: Electrons and holes in semiconductor
materials recombine either radiatively (i.e. accompanied by the emission of a photon) or
non-radiatively. The former process is favored for LED devices. The recombination rate
is proportional to the product of electron and hole concentrations so a higher current
injection gives rise to an increased light output. However, non-radiative recombination
is impossible to avoid in practical. In non-radiative recombination, the energy is
Figure 2 Schematic diagram of the principle of a typical LED.
20
converted to vibrational energy of lattice (i.e. phonons) instead of photon emission,
which leads to heating. There are several mechanisms for non-radiative recombination
including defects, deep levels and Auger recombination [8]. As native defects could lead
to both increased Auger coefficient and deep levels, the reduction of defects would be
an important issue in minimizing non-radiative recombination.
Direct and indirect band-gaps: There are two types of band structures in
semiconductor materials, direct band gap or indirect band-gap. If the momentum of
electrons and holes stands at the same point in both the conduction band and the
valence band, it is called direct band gap. Thus an electron can directly emit a photon
through radiative recombination. For semiconductor with an indirect gap, the minimal-
energy state in the conduction band and the maximal-energy state in the valence band
do not consist in a certain crystal momentum (see Figure 3). As a result, the electron
must pass through an intermediate state and transfer momentum to the crystal lattice
so that makes the emission of photons difficult. As the radiative recombination is far
slower in indirect band gap materials than direct band gap ones, LEDs and laser diodes
(LDs) are always fabricated with direct band gap materials.
21
Silicon is widely used in CMOS devices but as an indirect band-gap material, LEDs
can hardly be made using Si. III-V semiconductors, such as GaN, InP and GaAs, become
promising materials benefiting from their direct band-gap characteristic for SSL.
1.2 III-nitride materials for solid-state lighting
GaAs and InP based materials, such as the alloy of AlGaInP, are widely studied to
be suited for LEDs operating in the red, orange, yellow, and yellow-green wavelength
range [9-13]. However, blue and green emissions can hardly be obtained due to their
small band gap. Figure 4 shows the band gap energies of various semiconductors. The
selection of materials for GaAs and InP based alloys is quite limited in short wavelength
emissions. Besides, the semiconductors offering emission characteristics in blue region,
like GaP and AlAs turns to in-direct band gap. So AlGaInP material system is only suited
for high-brightness visible-spectrum LEDs emitting in the red, orange, amber, and yellow
wavelength range but not capable in blue and green wavelength range as the radiative
Figure 3 Energy vs. crystal momentum for a semiconductor with an indirect band-gap.
22
efficiency of the AlGaInP system becoming rather low because of the direct - indirect
transition.
Benefiting from its large band-gap, GaN has been regarded as a promising
material for SSL since the early 1990s. In the late 1990s, InGaN LEDs emitting in the blue
and green wavelength range became commercially available. To date, GaN is the
primary material system for high-brightness LEDs for SSL in blue, green and white
applications. The band gap energies of GaN materials system is illustrated in Figure 5.
The wide coverage capability of wavelength from deep UV to IR, together with its high
radiative efficiency makes GaN a promising material for SSL [8].
Figure 4 Bandgap energy and lattice constant of various III-V semiconductors at room temperature.
23
At present, three approaches have been envisioned for white light LEDs (see
Figure 6 [14]), including multi-chip LEDs integrating blue, green and red LEDs, blue LEDs
coated with one or more types of phosphors, and UV LEDs combined with multiple
phosphors. Products utilizing the last two approaches are currently available in market.
The use of multiple LEDs emitting at different colors will likely produce the best white
color, by combining GaN and AlGaInP LEDs in a single package. However, the
integration of LEDs with a required feedback circuit to maintain the desired relative
intensities of the constituent colors contributes to the high complexity and cost.
Figure 5 Bandgap energy and lattice constant of various III-Nitrides at room temperature.
24
Figure 6 Three dominant ways to produce white light based on LEDs. CRI stands for Color Rendering Index, which describes its ability to reproduce the natural colors [14].
Utilizing a blue LED together with one or more phosphors is the second approach
to white LED. The phosphors absorb a portion of the blue light, and then fluoresce at
longer wavelengths such as red and yellow. Therefore, white light can be produced by a
mixture of the native blue light and the fluoresced light. The drawback with this
approach is a relatively poor coloring index (CRI) but can be improved by adding
additional phosphors in more colors. While this approach is the least costly, the
development of phosphors is still under progress for the high quality light used as
general illumination.
Similar to the second way, the third approach is a variation by employing an
ultraviolet (UV) LED to illuminate three or more different colored phosphors. High CRI is
provided by this approach as this scheme depends on all visible light being emitted by
the phosphors. However, the drawbacks to this approach reside with a lower efficiency
25
due to the inherent energy loss in the process of UV absorption and emission of photon
with a lower energy. Besides, the processing and packaging to withstand the high flux of
UV radiation is another issue to deal with.
The different approaches towards white LEDs listed above all will benefit from
improvements in GaN-based materials. So the improvement of electrical properties in
GaN based LEDs becomes a fundamental issue towards producing high brightness white
LEDs for SSL.
The alloy of GaN with InN has attracted numerous attentions since the last decade
[15-22]. InGaN based LEDs are the most widely applied devices in commercial products,
benefiting its emission wavelength coverage discussed previously. The materials
parameters for GaN and InN can be found in Table 1.
PROPERTY / MATERIAL Wurzite GaN Wurzite InN
Lattice Parameter(s) at 300K a0 = 0.3189 nm
c0 = 0.5185 nm
a0 = 0.3544 nm
c0 = 0.5718nm Density at 300K 6.095 g.cm
-3 6.81 g.cm
-3
Linear Thermal Expansion
Coeff. at 300 K 5.59x10
-6 K
-1 4 x 10
-6 K
-1
Dielectric Constant 10.4 15 Refractive Index 2.67 2.9
Nature of Energy Gap Eg Direct Direct Energy Gap Eg at 300 K 3.45 eV 0.7 eV
Table 1 Materials Parameters of GaN and InN [23].
Unlike silicon, gallium nitride bulk crystal is difficult to be obtained. One of the
reasons is that the decomposition pressure at the melting temperature is extremely
high due to the strong covalent bonding of GaN. That is, if the pressure is below the
decomposition pressure (Pm at about 4.5 GPa, according to Ref [24]) , GaN will
decompose instead of melt when temperature increases. Consequently, the standard
26
crystal growth technique for semiconductor wafers cannot be applied to produce GaN.
Ammonothermal growth of GaN is a method developed recently to fabricate GaN bulk
crystal [25]. Although the availability of bulk GaN material is growing in amount, size,
doping and orientation, the cost could hardly be reduced to an acceptable level. So GaN
based LED structures are usually epitaxy on foreign substrates, such as silicon carbide
[26-28], silicon[15, 29] and sapphire[30, 31]. Among those selections, sapphire, which is
a single crystal of Al2O3, has been widely accepted as the substrate due to its stable
physical properties (see Table 2) and a relative low cost.
Table 2 Physical properties of sapphire, from Ref [32].
There are two major heteroepitaxy techniques, Metal Organic Chemical Vapor
Deposition (MOCVD or MOVPE) [19, 33-38] and Molecular Beam Epitaxy (MBE) [39-45]
for fabricating GaN based LEDs. In this study, our devices were grown by MOCVD on a c-
27
plane sapphire substrate. A typical recipe for GaN growth is given as [46]:
Trimethylgallium (TMGa), trimethylaluminum (TMAl), trimethylindium (TMI) and
ammonia used as precursors with nitrogen and hydrogen carrier gases at a reactor
pressure of 76 Torr with silane and bis(cyclopentadienyl) magnesium used as dopant
sources.
1.3 Lighting efficiency
Discussed in previous section, solid-state lighting will provide important
environmental benefits, significant energy savings and dramatically new ways to utilize
and control light [47]. The improvement of lighting efficiency becomes the crucial topic
to achieve these benefits. The output power density is directly related to the external
quantum efficiency, which is defined as [8],
where P is the optical power emitted into the free space.
It can be inferred that both the internal quantum efficiency (ηint) and the
extraction efficiency (ηextraction) determine the external quantum efficiency. The
definitions of internal quantum efficiency and the extraction efficiency are as follows,
where Pint is the optical power emitted from the active region and I is the injection
current, and
28
The low extraction efficiency in current LEDs limits the improvement of external
quantum efficiency. Due to the large difference of the refractive indexes between GaN
(n=2.5) based LED structure and the external medium, the critical angle of total internal
reflection is about 23° according to Snell’s law if the external medium is air (n=1) [48].
Therefore, only a small proportion of emitting light is capable to escape outside the LED.
Many methods have been proposed to increase the extraction efficiency, such as
photonic crystal [49], patterned substrate [50] and surface roughening [51].
The improvement of internal quantum efficiency suffers from several problems in
current LED technology. The influence of the GaN material quality, especially the
structural defects, on LED performance is discussed in [52]. It is demonstrated that a
reduction of defects will lead to higher internal quantum efficiency. Besides, the biaxial
strain resulted from a mismatch of lattice constant and thermal expansion coefficient,
introduces a piezoelectric field inside the active layer. As a result, the band profile is
inclined so that a decrease in oscillator strength of electron–hole pairs would be
expected, which will consequently reduce the internal quantum efficiency [53]. Thus
engineering strain would be effective to improve the device efficiency.
There is another issue limits the maximum light output power in GaN based LEDs
namely efficiency droop effect. Typically, the quantum efficiency reaches its peak value
at low current density and rapidly decreases with increasing injection current under the
29
influence of the effect (see Figure 7) [54]. The issue of efficiency droop has become an
impediment to the development of high power devices in addition to the factors
discussed before. The origin of efficiency droop has been widely discussed, and direct
measurement of the Auger electrons using Faraday cup in an ultra-high vacuum
chamber confirm this as the dominant contributor to the droop effect [55, 56]. So it
turns out to be an equally important issue to alleviate the droop effect when designing
high brightness GaN based LEDs, which sustains high EQE under large drive current
density.
Figure 7 EQE vs. Current injection at three LED devices. A droop of EQE is observed when increasing injection current density [54].
1.4 Motivation and significance of this thesis
InGaN-QW based micro-structure, used in light communication [57] and micro-
display [58], were demonstrated to hold unique electrical characteristics suitable for
high power LED lighting. It was reported that GaN micro-LED-pillars showed a reduced
strain level compared with the bulk region [59]. Similarly, a reduction of strain in GaN
30
based LED structure can be expected. It is also worth mentioning that engineering
micro-structure in InGaN/GaN MQWs brings benefit to the extraction efficiency because
of an improved light escape design. The extraction efficiency is limited by total internal
reflection (TIR) in large-area LED structures. So the possibility of achieving high
efficiency and reduced droop effect in InGaN LED micro-structures and the feasibility of
utilizing InGaN micro-LED-pillar as the building block for high brightness emitters
motivate this research. With proper heat sinking design, the eventual high wall-plug
efficiency may potentially lead to realization of high brightness LEDs beyond >200
lumens per watt for general illumination.
There are two categories of approaches towards micro-structures: the top-down
approach and the bottom-up approach. The former approach is commonly seen in
CMOS IC technology, with the idea of patterning bulk wafer by etching away unwanted
region. On the contrary, micro-structure could also be fabricated or grown based on
self-organized or self-assembled bottom-up approach [60]. Top-down approach is
chosen in our study due to several reasons. Firstly, the preparation process of
InGaN/GaN MQWs bulk wafers is well developed. High epitaxy uniformity between
samples can be expected in top-down approach, which is necessary for device
manufacturing. Secondly, a better control of micro-structures, in terms of size, shape
and height, can be obtained benefiting from the CMOS compatible patterning and
etching process. However, the micro-structure grown by bottom-up approach could
hardly offer a compatible control of morphology. The self-assembled InGaN/GaN micro-
LED-pillar epitaxy process to yield equal height nano-wires is still under investigation,
31
and the research itself is non-trivial. An example of InP nanowires prepared by two
approaches [61, 62] is shown in Figure 8. Thirdly, the top-down approach allows the
integration with other pre-fabricated features while bottom-up growth requires a fresh
prepared starting surface and the high growth temperature of >600 degree Celsius
prevents further integration with CMOS integrated circuits.
Targeting high brightness InGaN based LED, several issues are to be investigated in
this study:
Fabrication of InGaN based large-area LEDs and micro-LED-pillars utilizing
in-house ANIC facilities.
(a)
(b)
Figure 8 InP nanowires prepared by top-down approach (a) [61] and bottom-up approach (b) [62]. A better control of size, height and shape is observed in (a).
32
Determine the improvement of electrical property, efficiency and its droop
characteristics in micro-LED-pillars compared with large-area LEDs (> 200
µm).
Study the mechanism of improvements in micro-LED-pillars.
Implement high performance InGaN micro-LED pillars as the building
blocks for high brightness LEDs benefiting from the scaling-down effect.
The thesis consists of three parts: fabrication of InGaN/GaN MQW micro-LED-
pillars, strain engineering in InGaN/GaN MQW micro-LED-pillars and building micro-LED-
pillar array for high brightness LEDs. Background and motivation of this thesis work is
detailed in chapter 1. The design, fabrication and characterization of InGaN/GaN MQW
LEDs and micro-LED-pillars will be presented in chapter 2 with a detailed analysis of
performance improvement in micro-LED-pillars. Chapter 3 focused on the strain relief in
smaller micro-LED-pillars, which would be expected to act as the building blocks for high
brightness LED. The strain distribution and thermal annealing effect will be discussed.
The design of strain relief InGaN/GaN MQW micro-LED-pillar will be detailed in this
chapter. Fabrication of micro-LED-pillar array is demonstrated in chapter 4, presenting a
possible way of building large effective-area, high brightness emitter. Summary of this
work, including the significance and future work can be found in the final chapter.
33
2 Chapter Two: Fabrication of InGaN/GaN MQW micro-LED-
pillars
2.1 Process Overview
A typical InGaN based LED structure involves multiple quantum wells as the active
layer to achieve better electron confinement [63-65]. Quantum well is the structure that
a thin InGaN (< 5 nm) is sandwiched by GaN barrier (~10 nm). The active layer is
embedded by a top layer of Mg doped P type GaN and a sub layer of Si doped N type
GaN. Undoped GaN grain layer is introduced before the epitaxy was grown on c-plane
sapphire substrate to reduce the strain and defects. The design of blue InGaN LED
involved in this study is illustrated in Figure 9, with a 12 stack of InGaN/GaN MQWs as
the active layer. We use DC sputtered Ni and RF sputtered ITO as the p contact layer, or
the current spreading layer. A post deposition RTP process proved to enhance the
transparence of ITO layer and reduce the resistance [66].
34
Figure 9 Schematic structure of InGaN/GaN LED wafer.
The MESA is defined by the UV lithography and the design of masks is detailed in
Appendix A. The N-contact is then completed by lift-off process using a stack of
Ti/Al/Ti/Au [67-69]. InGaN/GaN MQWs large-area LEDs, both in circular and square
shape, with the diameter (D) of 200 µm to 700 µm and micro-LED-pillars ranging from D
= 20 µm to 80 µm were fabricated. The process flow developed at KAUST nano-fab is
detailed in Appendix B.
The preparation of a clean surface is critical to reduce the surface contamination
and improved current spreading in fabricated devices because particles and organic
clusters may introduce surface damage in sputtering and micro-masks in dry etching.
We found that a cleaning process by dipping in HCl for 1 minute, ultrasonic washing in
IPA for 3 minutes followed by ultrasonic washing in acetone for 3 minutes gives a clean
35
surface for fabrication. The surface of InGaN/GaN MQWs wafer for LED fabrication
before and after cleaning was checked using optical microscopy (see Figure 10) and the
effectiveness of the cleaning steps was confirmed.
2.2 Devices Characterization
2.2.1 Optical Microscopy
The InGaN/GaN MQWs large-area LEDs and micro-LED-pillars fabricated is
examined using optical microscope as shown in Figure 11. The images were collected
using a silicon based Charge-Coupled Device (CCD) camera.
(a) (b)
Figure 11 Optical images of fabricated large-area circular LEDs (a) and square LEDs with micro-LED-pillars (b).
(a) (b)
Figure 10 Optical images of sample surface before (a) and after (b) proper cleaning. Scale bar shows a 100 µm distance.
36
2.2.2 Transmission Electron Microscopy
Cross-sectional Transmission Electron Microscopy (TEM) measurements provide
high-resolution images of layer structures (see Figure 12), by which the thickness of
each layer is confirmed. The sample for TEM imaging was prepared using focused ion
beam (FIB) milling.
The InGaN/GaN MQWs were featured in Figure 13 with a uniform InGaN and GaN
interface found in TEM image. The width of the InGaN well is 3 nm and that of the GaN
barrier is 10 nm.
Figure 12. Cross-sectional TEM image of InGaN/GaN MQW LED structure. The capping layer is a 500 nm sputtered Al layer for TEM imaging. The scale bar is 0.5 µm.
Sapphire
U-GaN
N-GaN
MQWs
P-GaN
Capping
layer
37
Figure 13. Cross-sectional TEM image of InGaN/GaN MQWs. The 10 nm GaN barrier and 3 nm InGaN quantum well is clearly observed. The scale bar is 10 nm.
GaN
barrier
InGaN
QW
38
The Energy Dispersive X-ray (EDX) spectroscopy line-scan across the InGaN/GaN
MQW region showed the 12-stack quantum-well / quantum-barrier with the respective
distribution of In, Ga and N atomic concentrations (Figure 14). The EDX measurements
indicate a uniformity of In concentrations in all twelve InGaN wells. TEM results
confirmed the structure quality of the InGaN MQWs LED wafer.
2.2.3 Electroluminescence
The functionality of LED devices is based on the electroluminescence (EL) process,
in which a material structure emits light in response to the passage of an electric
current. When DC bias is applied between the P and N contacts of an LED, electrons and
(a)
(b)
Figure 14 (a) Z-contrast TEM image of InGaN/GaN MQWs. (b) EDX of Ga, In and N concentration scanning along with the line in (a).
39
holes are injected into the device. As a result, photons are generated from radiative
recombination of electrons and holes. So the EL spectrum is critical to understand the
emission properties of the LED device.
An electrical testing setup is built in KAUST photonics Lab with the ability of EL, I-V
(current vs. voltage) and L-I (light power vs. current injection) measurements (see Figure
15). DC bias is applied utilizing a Keithley 2400 source meter with two probes
connected. EL spectrum was collected by Ocean Optics QE650000 spectrometer using a
fiber based connector. The Newport 2936C power meter and Newport 818UV
photodetector is installed for L-I measurements. The program used was written in
Labview and can be found in Appendix C.
Figure 15. EL, I-V and L-I measurement system set-up.
Optical Power meter
Source meter
Probe stations
Photodetector
Microscope
Sample Stage Spectrometer
40
When DC bias is applied on the LED, a strong blue emission is observed when Vapp
> 2.8 V (see Figure 17). A better current spreading across the entire LED can be
identified when Ni/ITO contact layer is employed.
A typical EL spectrum for large-area LED can be found in Figure 16, collected at
300 ms integrating time and 3 times average. The injection current to the device is 0.1
mA. The emission peak locates at around 450 nm with FWHM (Full width at half
maximum) of 38.7 nm.
Figure 16. EL spectrum from a large-area LED.
(a) (b)
Figure 17. Light emission observed from microscope when DC bias is applied to an InGaN/GaN MQW large-area LED without ITO current spreading layer (a) and with 170 nm ITO current spreading layer (b).
41
2.2.4 Electrical property
The electrical property of LEDs could be evaluated by measuring its I-V
characteristic. Take the micro-LED with diameter (D) of 50 um as an example, its
threshold voltage and sheet resistance can be obtained from the measured I-V curve
(see Figure 18). Compared with the ideal threshold voltage for In0.24Ga0.76N /GaN MQW
LEDs defined by V th the = Eg/e = 2.71 V, we confirmed a good electrical property in
fabricated LEDs. Similar results can be found in all devices fabricated, indicating a proper
contact was formed.
From the I-V measurements, current density in InGaN based LEDs can be
calculated and strong size dependence was observed in Figure 19. A higher current
density of 492 A/cm2 was measured in 20 μm micro-LED-pillar, compared with that of
200 μm LED (20 A/cm2) at a DC bias of 10 V. Therefore, micro-LED-pillar with reduced
size is expected to be able to sustain high density of current injection. The improvement
in electrical properties is essential for achieving high brightness LED, which would run
under high drive current.
(a) (b)
Figure 18. Current – Voltage measurements showing the calculation of threshold voltage (Vth) (a) and sheet resistance (Rs) (b).
42
The improvement of electrical property in smaller micro-LED-pillar is attributed to
an enhanced current spreading [70]. According to the current density distribution
theory [71], the lateral current spreading length, Ls, is:
√( )
which is defined as the length at the point where the lateral current density is reduced
to 1/e of the current density at the pillar edge. , and stands for the sheet
resistance of contact layer, n and p GaN layer, respectively. For our case, Ls is calculated
to be 55 μm, and the fraction of current density at the center of the pillar to that at the
edge are 0.83, 0.63 and 0.48 for devices with D = 20 µm, 50 µm and 80 µm, respectively.
Thus, a uniform current spreading (less current density variation across the device) in
smaller micro-LED-pillar results in the improved electrical property.
Figure 19. Current Density vs. voltage in various micro-LED-pillars compared with a large-area one.
43
2.2.5 Efficiency Characteristic
The lighting efficiency can be tested by measuring the light output power under
different injection current (i.e. L-I measurements). A typical curve of L-I relation is shown
in Figure 20 (a), collected from the micro-LED-pillar with D = 50 µm. However, little
information can be extracted directly from the L-I curve so the external quantum
efficiency (EQE) vs. injected current density is calculated for further analysis (see Figure
20 (b)). The EQE is calculated by ηext = (p/hν) / (I/e). Current density is given by current /
surface area.
From Figure 20 (b), we observed that the EQE reached its peak value at low
current density and rapidly decreased with increasing injection current, which is the
efficiency droop effect. This effect limits the maximum light output power one could
harness for high brightness lighting applications. As the micro-LED-pillars were found to
hold enhanced electrical properties, a reduced efficiency droop effect can hence be
expected in micro-LED-pillars. To prove it is the case, we analyze the EQE from a variety
of micro-LED-pillars.
(a) (b)
Figure 20. (a) L-I relationship recorded from micro-LED-pillar with D = 50 µm. (b) EQE vs. current density curve calculated from (a).
44
It should be noted that the EQE calculated is in arbitrary unit instead of a finite
value. This is because we are unable to collect all the emitted photons in the current
setup without integrating sphere. However, we could still obtain comparable data after
nominalization (see Figure 21).
The strong size-dependence in efficiency droop is evidenced, where the peak EQE
(ηpeak) occurs at a higher injection current density for smaller LED. With increasing
current, the EQE steadily drops beyond ηpeak. To realize a high brightness LED, the ability
to sustain high quantum efficiency under large injection current density is required.
Comparing the ratios of sustained EQE at 200 A/cm2 (η200 / ηpeak), we observed a
reduction of efficiency droop in small micro-LED-pillars. Over 80% ηpeak was sustained in
the 20 μm micro-LED-pillar, compared to only 70% in that of 80 μm µLED-pillar.
Figure 21. Nominalized EQE vs. current density in various micro-LED-pillars compared with a large-area one.
45
The current spreading effect discussed above could contribute to the enhanced
efficiency characteristics and reduced droop effect. The mechanism we attribute to is
that, with a reduced size, the current crowding effect at the active region is less
significant, leading to reduced Auger recombination probability, and local heating [72].
However, in addition to the current spreading effect, our subsequent investigation
reveal that strain relief plays an important part in sustaining EQE at high injection
current as well, notably in small micro-LED-pillars. This finding is detailed in the
following part.
2.3 Materials Characterization
In this section, two widely used materials characterization technique,
photoluminescence (PL) and Raman spectroscopy are discussed for determining the
physical properties of InGaN based MQWs in terms of strain level, defects and radiation
efficiency. The aims lay on identifying the potential contributor to the reduction of
efficiency droop observed in micro-LED-pillars.
2.3.1 Photoluminescence
Luminescence, the spontaneous emission of light from the excited electronic
states in a material can be preceded by a variety of agents. If achieved by the absorption
of light, it is then named photoluminescence.
As a nondestructive characterization method, PL analysis is widely used in
determining various parameters of a semiconductor, including the electronic
structures, doping and defect levels, electron-hole combination dynamics. In a
direct-gap material, such as GaN in our study, the electron is excited by a high energy
46
photon (Ein > Eg) to move from valance band to conduction band and the recombination
of electron-hole pair could then happen as illustrated in Figure 22. The excited electron
transfers its energy by scattering inside the lattice until it relaxes to the bottom of
the conduction band. Finally recombination of electron-hole pair occurs with an
emission of a photon (Eout = Eg).
Figure 22 The photoluminescence process in a direct-gap semiconductor [73].
The PL measurements are carried out in two systems: Aramis system with 40 X UV
lens using a 325 nm He-Cd laser and a HR 800 system with 100 X objective using a self-
built 405 nm semiconductor laser line (see Figure 23). The reason to use a 405 nm laser
is to obtain a smaller spot size, which will offer improved special resolution.
47
The PL signals were gathered from the centers of each device. No observable peak
shift is found in InGaN/GaN MQWs LEDs and micro-LED-pillars with D > 60 μm. However,
we do figure out a blue shift of PL emission peak from micro-LED-pillars with D < 50 μm
(Figure 24).
The blue shift of PL emission peak indicated a relaxation of stress from smaller
micro-LED-pillar [60, 74]. The peak shift trend evidences that strain relief is not
negligible in µLED-pillars with D < 50 µm. As strain effect introducing a piezoelectric
field, lead to the reduction of electrical properties and enhanced efficiency droop [75],
our investigation suggests that µLED-pillars with D < 20 µm, benefitting from both the
uniform current spreading and the strain relief, will have further improvement in
electrical and quantum efficiency characteristics.
Figure 23. Integration of 405 nm laser with HR 800 micro-PL system.
Power supply
405 nm laser
plano convex lens
mirror
Re-collimator
Dichroic mirror
Reflector
48
PL integrating intensity is another factor to evaluate the lighting efficiency. The PL
was collected from the center of each pillar with a spot size of 4 µm. As the laser spot
size is far less than the area of micro-LED-pillar, the PL signal can serve as an effective
probe of the photon radiation process inside the micro-LED-pillar, which is directed
related to its EQE. The increase of PL integrated intensity as illustrated in Figure 25,
suggested a rising of quantum efficiency while shrinking the size of micro-LED-pillar.
Figure 24. PL peak positions as a function of micro-LED-pillar size .
49
2.3.2 Raman spectroscopy
Raman spectroscopy is a technique capable to observe vibrational, rotational, and
other low-frequency modes in a semiconductor material. It relies on inelastic scattering,
or Raman scattering, of monochromatic light, usually from a laser in the visible range. In
the scattering process, most of the photons are elastically scattered while only a very
small proportion of phonon is inelastically scattered. By detecting the scattering modes,
information of vibrational states can be obtained.
As illustrated in Figure 26, the electrons are excited up to a virtual state in
between the ground state and the excited states when a laser interacts with the
material. In most cases, the excited electrons relax back to the original state with an
emission of phonon, thus undergo elastic scattering, or named Rayleigh scattering.
Figure 25. PL integrated intensity as a function of micro-LED-pillar size.
50
Raman scattering occurs when there is an increase or decrease of vibrational
wavelength, named Stokes and anti-Stokes shifts respectively. These vibrational states
represent phonon modes in thin films.
The micro-Raman spectrum was obtained using a backscattering configuration HR
800 system with four semiconductor laser sources: 473 nm, 532 nm, 633 nm and 785
nm. Only Stokes shifts are detected by our system. All the Raman spectra were collected
using 532 nm laser, if not specified in this work. Due to the fact that the photon energy
of a 532 nm laser is far less than the band-gap of InGaN or GaN, the heating effect
caused by photon absorption can be eliminated. A typical Raman spectrum of our
InGaN/GaN MQWs LEDs can be found in Figure 27.
Figure 26. Schematic drawing shows the principles of Raman and Rayleigh scattering.
51
Three Raman phonon structures can be resolved in the spectrum. The 570 cm-1
peak is from E2 mode of the strained GaN. The A1 mode of GaN locates at 735 cm-1.
Besides, the E2 mode of the strained InGaN can be found at 558 cm-1 but it is naturally
weak due to its low volume [76]. Because the hexagonal Wurzite GaN has a crystal
structure of (6-fold rotation axis, with the addition of 6 mirror planes), the zone-
center (Γ point) optical modes are A1+2B1+E1+2E2 (see Figure 28) [77]. The
corresponding selection rules led us to the assignments of the peaks as B modes are
silent while E1 modes are only infrared-active. So only E2 and A1 modes are expected in
our measurements along the c-axis [78]. For GaN based materials, the acoustic modes
are just translations and would not be considered since the wave vector is zero.
Raman spectrum is expected to offer detailed information about the strain level
[59, 79-81] in semiconductor thin films. The stress will affect the inelastic scattering
Figure 27. Raman spectrum collected from the GaN based LED epitaxy.
52
process and then, introduce a wavenumber shift. The effect of stress on the frequencies
of the three optical modes follows the secular equation [82]:
|
( ) ( ) ( )
|
Here p, q and r are the phonon deformation potentials while are the strain
tensor components. So in the presence of stress, the Raman frequency of each mode
will shift:
Thus a compressive strain will introduce a red shift in Raman spectrum, i.e. to
higher wavenumber. Although no significant Raman peak shift was identified here,
considering a PL shift of only 4 meV observed, we suggested that the level of strain relief
was beyond the measurement capability of micro-Raman in our case. But we do expect
an enhanced strain relief if the size of micro-LED-pillar is further reduced, especially
beyond 20 µm.
Figure 28. Optical phonon modes in the Wurtzite structure.
53
3 Chapter Three: Strain Engineering in InGaN/GaN MQW micro-
LED-pillars
3.1 Introduction
In Chapter 2, we observed improvement of electrical property and reduction of
efficiency droop in InGaN/GaN MQW micro-LED-pillars, resulted from a better current
spreading and strain relief. As the current spreading effect become less significant when
the diameter of micro-LED-pillar was reduced to less than 20 µm, it was expected that
strain relief will turn to the dominant factor. Especially, the particular driving force in
our study is achieving high efficiency under high injection, but it becomes an increasing
challenge. A critical factor is the piezoelectric field caused by the strain of MQWs grown
on conventional c-plane sapphire substrate [83]. This polarization field caused by non-
centro-symmetric structure of III-nitrides and the lattice mismatch between epitaxial
layers with the substrate generates a quantum confined Stark effect (QCSE), leading to a
decrease of the efficiency [84]. So it becomes a critical problem to engineer the strain
field in InGaN/GaN MQWs micro-LED-pillars with D < 20 µm, in order to fabricate high
brightness LEDs which requires sustaining high efficiency under large drive current.
By introducing micro/nano-structures, the strain field may be partly relaxed based
on PL [85] and cathodoluminescence (CL) observations [86] recently. Besides, nanotips
[87] and nanocolumns [88] have also been studied for strain relief structures with
enhanced PL emission observed. However, little work on the process of strain relief as
well as its distribution in microstructures has been done. Sacconi F. et,al. numerically
54
studied the strain components in an InGaN quantum disk based nanocolumn p-i-n
structure by TIBER CAD and suggested that a strong relaxation effect in the column
boundaries [89]. Recent experiments observed a blue shift of CL characteristics along
the edge of fabricated micro-LED-pillar, which was attributed to a strain relief [86]. As
the damage and defects introduced by the dry etching process may also affect the
luminance peak position [90], it calls for a further analysis with different technique, such
as Micro-Raman spectroscopy, to investigate the strain field in microstructures with
optimized fabrication process.
The fabrication parameters, such as the height and size of pillar were chosen
randomly and hence the studies lack a systemic understanding of designing low strain
structures targeting general illumination.
The studies in this chapter focus on the strain relief and its distribution in micro-
LED-pillars fabricated by a top-down approach with annealing process. Micro-Raman
and PL measurements of micro-LED-pillars ranging from 1 µm to 15 µm were studied in
detail to evaluate the strain relief according to pillar size, together with the effect of
thermal annealing. Benefiting from the high spatial resolution micro-Raman and PL
setup, a downshift of Raman peaks and blue shift of PL emission was observed,
indicating a strain relief in as-prepared micro-LED-pillars and annealed samples. The
spreading of strain and effect of RTA was evaluated based on the splitting of E2 phonon
in micro-LED-pillars. The distribution of strain relief was further analyzed based on
Raman line-scan, by which enabled us to understand the mechanism of strain relief
55
across the microstructures fabricated. With this information, a systemic evaluation of
micro-LED-pillar height and size were carried out providing design rules for eventual
large scale manufacturing of strain released high brightness LEDs based on, for example,
inductively coupled plasma (ICP) etching.
3.2 Experiments
In this study, micro-LED-pillars with diameter ranging from 1 µm to 15 µm were
fabricated by focused ion beam (FIB) technique. A 400-nm-thick aluminum layer, instead
of ITO used in chapter 2, was deposited at the top of the epitaxy using DC sputter as the
discharging layer before the FIB patterning.
Similar to the scanning electron microscope (SEM), the FIB scans over the sample
surface, typically using Ga+ ions. Several reasons determined the choice of Ga ions in
most commercialized system [91]. Firstly, a compact gun with limited heating can be
constructed owing to the low melting temperature of Ga. Secondly, Ga+ ion is optimal
choice for milling a variety of materials. Thirdly, for analysis of the sample after FIB
processing, the Ga element has low analytical interference with other elements. There
are four functions of FIB processing including milling, deposition, implantation and
imaging (see Figure 29) [92].
56
The unique capabilities of FIB are related to the complicated ion-solid interaction
[93-95]. In our study, only FIB milling process is involved, which aims at removing the
atoms in defined area when energetic ions hit the surface. FIB milling is a promising
technique for the fine patterning of GaN micro- and nanostructures due to its direct-
writing mode which can be concurrently monitored [96-99]. Excellent demonstrations
for FIB patterning capability include fabrication of cavity for GaN-based laser diodes
[100].
The micro-LED-pillars were fabricated in an FEI Quanta 3D dual-beam focused ion
beam (FIB) system. A two-step milling process was developed to achieve a balance
between etching profile and process time (see Table 3).
Figure 29. Basic four functions of FIB processing: (a) Milling, (b) Deposition, (c) implantation and (d) imaging.
57
The topography was examined by SEM with uniform surface and sidewall profiles
of FIB fabricated micro-LED-pillars shown in Figure 30 and Figure 31.
Diameter of
pillar
1st step 2nd step Result diameter
micro-LED-pillar
15 µm 17 µm (r), 800 nm (Z) 6.5
nA, 30 KV
15.5 µm®, 400 nm
(Z)0.92 nA, 30 KV
15.1 µm
10 µm 12 µm (r), 800 nm (Z) 6.5
nA, 30 KV
10.5 µm®, 400 nm
(Z)0.92 nA, 30 KV
10.3 µm
7 µm 9 µm (r), 800 nm (Z) 6.5
nA, 30 KV
7.5 µm®, 400 nm (Z)0.92
nA, 30 KV
7.2 µm
5 µm 7 µm (r), 800 nm (Z) 6.5
nA, 30 KV
5.5 µm®, 400 nm (Z)0.92
nA, 30 KV
5.2 µm
3 µm 4.5 µm (r), 800 nm (Z)
6.5 nA, 30 KV
3.5 µm®, 400 nm (Z)0.92
nA, 30 KV
3.2 µm
1 µm 3 µm (r), 800 nm (Z) 6.5
nA, 30 KV
1.5 µm®, 400 nm (Z)0.92
nA, 30 KV
1.4 µm
Table 3 FIB parameters in the fabrication process
58
(a)
(b)
Figure 30. SEM images of FIB fabricated micro-LED-pillars with D = 1 µm to 15 µm with untitled (a)and tilted (b) stage.
59
(a) (b)
Figure 31. SEM images of FIB fabricated micro-LED-pillars with D = 15 µm (a), D = 10 µm (b), D = 5 µm (c), D = 3 µm (d), D = 1 µm (e) identifying the sidewall profiles and (f) surrounding of 3 pillars showed no observable redoposition.
(d) (c)
(f) (e)
60
After FIB etching, the Al layer was removed in a solution of H3PO4:H2O2:H2O (8:1:1)
at 35℃ for 3 minutes. Rapid thermal annealing was processed in a Jetfirst RTP system in
N2 environment. The RTA lasted for 20 seconds at 950 ℃ .
No significant surface profile change was identified after RTA, which was
confirmed ex-situ by profilometer (ZYGO Newview 7300 optical surface profiler).
µRaman and µPL signals were measured based on a back-scattered geometry using the
Horiba HR800 system described previously. All the PL signals were collected using 405
nm laser, if not specified. The GaN Raman signal (strain-related signal) can be measured
at a sufficiently strong intensity level. As both GaN and InGaN are having similar thermal
mismatch, and both are compressively stressed with respect to sapphire substrate, one
can then infer that the relieve in the GaN barrier strain will result in strain relieve in
InGaN well layer and hence measuring the GaN Raman signal provides a valid and
accurate strain relief measurement.
3.3 Results and discussion
Figure 32 shows the Raman spectra of as-prepared as well as annealed micro pillar
with diameter of 1 µm, and that of the bulk area. Based on the discussion in Chapter 2,
the 570 cm-1 peak is from E2 mode of the strained GaN. The A1 mode of GaN locates at
735 cm-1. As both E2 and A1 peaks shares the same trend in our observation, only E2 is
analyzed. Compared with that of bulk area, the E2 phonon related peak showed a strong
shift to lower wavenumber, from 571.1 cm-1 to 567.8 cm-1. Quantum confined size effect
and strain relaxation are two possible mechanisms that contribute to the phonon peak
shift. The former cause is eliminated in our micro-LED-pillars, since the pillar diameters
61
are too large to result in notable size confinement effect [101]. Thus the down-shift of
the Raman peaks is mainly caused by the reduction of strain after formation of
microstructures because the epitaxial growth InGaN/GaN MQWs on c-plane sapphire
exhibits high degree of strain induced built-in piezoelectric field, leading to the shifts of
the optical phonon frequency of GaN [102].
The strain relief was also evident in the PL excited by 405 nm laser illustrated in
Figure 33. Utilizing Gaussian curve fitting, two and three Gaussians are revealed from
the measured spectrum of bulk-area and micro-LED-pillar, respectively. Both of the two
spectra share two similar Gaussians, denoted as Gaussian 1 (at around 465 nm) and 2
(at around 472 nm). The PL emission at 452.4 nm, named Gaussian 3, comes only from
Figure 32. Room temperature micro-Raman Spectra from micro-LED-pillar and bulk-area LED.
62
the micro-LED-pillar and thus can be attributed to the strain relaxed region. Here we
refer Gaussian 1 to the strained region while the Gaussian 2 would not be further
discussed because of its relatively low PL intensity [103]. A blue shift of 71 meV from PL
analysis indicated a strong strain relief in the micro-LED-pillar compared to the bulk area.
Micro-Raman spectroscopy provides a direct measurement of strain field in GaN
based epitaxy structures according to the crystal symmetry related formalism [76],
( ) √( ) ( )
where σij is the stress tensor in the Wurtzite crystal basis and a, b and c are the stress-
related deformation potentials.
Figure 33. Room temperature micro-PL Spectra of micro-LED-pillar and bulk-area LED with Gaussians fitting.
63
Considering the shear component c is reported to be weak (< 0.3 cm-1/GPa) and
assuming in-plain strain in GaN, σzz = 0, the formula can be simplified as:
( )
With the strain-free Raman shift of E2 phonon peak in GaN taken as 567.25 cm-1,
the strain field can be examined based on the wavenumber shift of E2 phonon. After the
formation of micro-LED-pillars, we found a peak split in E2 phonon signature (see Figure
31). The main peak at 568 cm-1 indicates a strain relief up to -0.4 GPa in the structure.
Here we observed a second “shoulder-like” peak at 572 cm-1, which was the first time
reported. This peak is attributed to residual strain with discussed as follows:
Figure 34. Raman spectra of a micro-LED-pillar with D =1 µm measured when independently irradiated with four different laser sources.
64
Several reasons lead us to this conclusion. Firstly, when micro-machined down to
the bottom of the epitaxy, or the trench area, only single peak at 572 cm-1 is found.
There was no peak splitting or downshifting measured, which will be discussed in detail
later. Secondly, a power dependent measurement showed no difference in peak
positions and intensity ratio, and hence the emerging second peak may not come from
micro-machining damage. Finally, we measured the depth dependent Raman spectra
using four laser sources. Benefiting from the difference in penetration depth, which is
short for 473 nm laser and long for 785 nm, the strain behavior can be further
understood (see Figure 33). It was found that strain-relief related Raman peak
dominated E2 phonon in Raman spectrum from 473 nm excitation source. When the
excitation laser wavelength increased to 785 nm, residual strain related peak became
dominant. Therefore, the residual strain came from the bottom of the epi-layers, which
means the grain layer in this case.
It was notices that the residual peak was merged with the main peak after RTA
(Figure 32), indicating that the enhanced spreading of strain relief. Thus a proper
annealing is necessary towards a uniform strain relieved structure.
Based on the strain profile characterized by Raman analysis, we would be able to
further investigate the strain distribution. Micro-Raman line-scans across the micro-LED-
pillars were conducted. The Raman shifts inside and outside the micro-LED-pillar with D
=15 µm are shown in Figure 34 with the strain field examined based on the wavenumber
shift.
65
Different strain levels can be told from micro-LED-pillar, trench and bulk area
regions (see figure 34). Due to the removal of epitaxy layers, a high level of strain is
expected at the trench area. In the center area of the fabricated micro-LED-pillars, a
uniform strain distribution was obtained. A reduction of strain from -1.4 GPa to -1.0 GPa
(corresponding to E2 phonon peak downshift from 570.6 cm-1 to 569.6 cm-1) was found
when moving from the bulk area to the center area of the micro-LED-pillar. Along the
edge of micro-LED-pillar, the strain level drops, which is a gradual release of strain,
before it arrives at a uniform relief and the ring-like area is thus named the strain-relief
ring region. The outside edge of the ring stands for the maximum strain relief induced
by the patterning, which remains at a fixed value depending only on the process
condition. In our cases, the value was observed at 568.3 cm-1.
Figure 35. Line-mapping of fitted Raman peaks across the micro pillar with D = 15 µm. A strain-relief ring region is identifiable at the vicinity of the micro-pillar.
66
Here we will focus on the strain distribution at the edge of micro-LED-pillars. Line-
scan from all directions showed similar result to a selected micro-LED-pillar, which
confirmed the uniformity of strain vibration in the strain-relief ring region. Recently, E.Y.
Xie et,al. [86] reported a similar discovery by CL imaging but our work provides a direct
link of strain level to the locations based on Raman spectroscopy. As the Raman
technique is high spatially resolved with 100X objective, the widths of the strain-relief
ring regions can be figured out in micro-LED-pillars with diameter as small as 1 µm (see
Figure 35). Although the width of the strain-relief ring region decreases as the micro-
LED-pillar is further scaled down, the proportion of this area, on the contrary, increases
continuously as the pillar diameter reduces, from 31% at 15 µm to 92% at 1 µm. These
results suggested that for the smaller micro-LED-pillars, especially D < 3 µm, the strain
within the majority of the volume is significantly reduced.
Figure 36. The width of the strain-relief ring region derived from Raman line-mapping as a function of pillar diameter.
67
Meanwhile, it is necessary to consider the strain in center of micro-LED-pillars
when scaling down the size of micro-LED-pillars before we can get a complete picture of
the strain distribution in micro-LED-pillars. By Lorentz fitting of E2 phonon peak, the
changes of the peak wavenumber and FWHM with the pillar diameter were plotted in
Figure 36 (a) and (b), respectively. The Raman peak position of E2 phonon from bulk area
was labeled in Figure 36 (a) as well. Considering that the maximum strain relief at the
etching edge was observed to be -0.4 GPa, corresponding to 568.3 cm-1, the shift of
strain within the ring region, or the difference between the maximum strain relief at the
etching edge to the uniform strain relief in the center of the pillar, experiences a
decrease from 0.6 GPa at D = 15 µm to 0.05 GPa at D = 1 µm. Thus the experimental
results suggested that the strain variation inside the small pillars with D < 3 µm was
reduced to a relatively small value. So this further leads us to the conclusion that a
uniform strain relief across the entire pillar is achieved for D < 3 µm.
68
The reduction of FWHM along with shrinking pillar size supports the argument
that a reduction of strain variation detected on small micro-LED-pillars. S. H. Margueron
et,al. reported an increase of FWHM with reduced pillar diameter and they attributed
(a)
(b)
Figure 37 . (a). E2 phonon Raman peak positions at the center of each micro-pillar, compared with that of the bulk area. (b). FWHM of fitted E2 Raman peak as a function of pillar diameter.
69
the cause to be the inhomogeneous strain field caused by defects [59]. Therefore, our
experimental results suggested that the process developed in this work would introduce
no significant defects and etching damage. It is also worth mentioning that the phonon
FWHM was further reduced after RTA. The only exception for micro-LED-pillar with D = 1
µm to be an incomparable case is due to the E2 phonon splitting in Figure 31. So we can
conclude that RTA is proved to be a measure to improve the uniformity of strain relief in
micro-LED-pillars.
The strain relief is analyzed in detail with regarding to the size of the micro-LED-
pillars. Based on the discussion above, It was noted that the strain relief region become
dominant only when D was further reduced to 3 μm or less. Raman spectra from micro-
LED-pillars with different sizes further confirmed our conclusion (see Figure 37).
Figure 38. Stack of Raman spectra from micro-LED-pillars with different sizes.
70
The last parameter in this study for designing strain-relieved micro-LED-pillar is
the micro-machining depths or the micro-LED-pillar heights (L). Micro-LED-pillars with a
fixed diameter of 1 µm were fabricated. Raman spectra were analyzed from micro-LED-
pillars with heights ranging from 460 nm to 3.01 µm (see Figure 38). At L = 462 nm, the
depth reached the top of active region, while at L = 710 nm the depth reached the
bottom of MQW region. Within L = 998 nm to 2.5 µm, the n-GaN region is exposed.
Finally, the whole pillar consists of the entire epitaxy, exposing the undoped-GaN grain
layer at L = 3.01 µm. Maximum strain relief was found in micro-LED-pillar with depth of
3 µm.
Figure 39. Raman spectra from bulk LED epitaxy and micro-LED-pillars with different micro-machining depth.
71
3.4 Summary
In this chapter, the strain relief and its distribution across micro-structures were
investigated by Micro-Raman and PL measurements. A significant wavenumbers down-
shift for E2 and A1 peak were observed in the Raman spectra of as-prepared micro-LED-
pillars. With a blue shift of PL emission, the strain relief was identified in micro-LED-
pillars. In as-prepared pillars, a sharp transition from strained to relaxed epitaxy region
was discernible based on the observed two competing E2 phonon peaks in Raman
spectra. RTA process was proved to promote the distribution of strain release, leading
to a uniform strain relief. Benefiting from Raman line-scan, the transition of strain relief
was found along the narrow circumference of the pillars. Reducing the dimension of
micro-pillar enabled not only a higher level of strain relief but also a higher volume ratio
of strain distributing ring, resulted in a uniform strain relief across the entire pillar. Our
work supports that micro-Raman serves as a powerful tool for strain analysis. Besides,
the design of strain relief micro-LED-pillars was investigated targeting at achieving high
brightness LEDs. The results of this study suggested that thermal annealed micro-LED-
pillars with diameter < 3 µm and depth (height) > 3 µm are promising strain relief
building blocks for further development of micro-LED array devices.
72
4 Chapter Four: Building up of micro-LED-pillar arrays
4.1 Introduction
In Chapter 2, the design of strain relief micro-LED-pillar for high efficiency low
droop light emitter applications was discussed. To build a practical device, we propose
the approach by arranging micro-LED-pillar in an array configuration. Through CMOS-
compatible passivation and planarization processes, micro-LED-pillar array forms the
building block for large effective-area, high brightness emitter.
The fabrication of nano-rod array [104, 105] and nano-strip array [106, 107] has
been reported recently. The fabrication was usually done by Ni self-assembly
lithography [108] and anodic alumina oxide (AAO) templates [109]. However, we
suggested that there was no significant benefit going to nano-sized structures as no
further strain relief could be achieved but an inhomogeneous size and height
distribution (see Figure 40) lead to difficulty in interconnection and device fabrication.
Moreover, an increasing surface-to-volume ratio in nano-LED-pillars was expected to
reduce electrical and optical properties in fabricated devices due to the growing impact
of sidewall damage. Thus we tend to favor micro-LED-pillars instead of nano-structures
in this research.
Two configurations of micro-LED-pillar array were fabricated and evaluated in this
chapter. The feasibility of achieving high brightness LED is discussed based on micro-
LED-pillar array design.
73
4.2 Experiments
In our study, micro-LED-pillar arrays were fabricated by direct FIB milling. The
design of micro-LED-pillar array can be found in Figure 41. The fabricated pillar arrays
with the pillar size of 2 µm (array (a), etching depth L = 2.8 µm) and 1 µm (array (b),
etching depth L = 1.1 µm) were observed by SEM in Figure 42. The effective area ratio,
defined as the area of micro-LED-pillars over the entire area is calculated to be 0.313 in
this design. A low ratio is designed to grantees each pillar to be independent with a
relative large gap between micro-LED-pillars.
Figure 40. (a), (b) The SEM images of nanopillars fabricated by self-assembly Ni dots lithography. (c) cross-sectional SEM and (d)cross-sectional TEM images of nanopillars prepared by AAO template etching.
74
Figure 41. Design of micro-LED-pillar array for FIB milling.
(a)
(b)
Figure 42. SEM images of FIB fabricated micro-LED-pillar arrays with D = 2 µm (a) and 1 µm (b).
75
4.3 Results and discussion
PL spectra obtained from the 405 nm laser excitation is plotted in Figure 43. The
effective PL intensity was calculated based on the effective area ratio in our array design.
A 62 meV and 47 meV blue shift of PL emission peak was identified from micro-LED-
pillar array (a) and (b), respectively. Thus the strain relief in micro-LED-pillar arrays was
evidenced. Raman spectra collected from the two samples clarified this argument as
well (see Figure 44), where a downshift of GaN E2 phonon peak can be observed. The
FWHM in both cases showed no significant changes, leading to the conclusion that dry
etching damage was not introduced to the InGaN/GaN MQWs active layer during the
fabrication process. Besides, a higher intensity from the micro-LED-pillar array (a)
compared with that from the bulk-area revealed improved photon emission efficiency. It
should be noted that no further treatment, such as RTA, was followed in this
Figure 43. PL of micro-LED-pillar arrays compared with the bulk area.
76
experiments to reduce the sidewall damage. So a further improvement of efficiency
characteristics is expected after proper post-patterning treatment.
We presented here that the strain relief structures could be produced in a large
scale by configuring micro-LED-pillars in an array form. Although the electrical
performance of the device has not been measured yet, the improvement of electrical
properties with reduction of efficiency droop can be expected based on the findings
detailed in Chapter 2 and 3.
Figure 44. Raman spectra of micro-LED-pillar arrays compared with the bulk area.
77
5 Conclusions and future work
In summary, we confirmed the improved electrical and optical characteristics,
with reduced efficiency droop in InGaN micro-LED-pillars when these devices were
scaled down in size. We demonstrated that strain relief indeed contributed to further
improvement in EQE characteristics in small InGaN µLED-pillars (D < 50 µm), apart from
the uniform current spreading effect. The strain relief and its distribution across micro-
structures were investigated by Micro-Raman and PL measurements. A significant
wavenumbers down-shift for E2 and A1 peak were observed in the Raman spectra of as-
prepared micro-pillars. With a blue shift of PL emission, the strain relief was identified in
micro-pillars. In as-prepared pillars, a sharp transition from strained to relaxed epitaxy
region was discernible based on the observed two competing E2 phonon peaks in Raman
spectra. RTA process was proved to promote the distribution of strain release, leading
to a uniform strain relief. Using the line-scan feature in Raman spectroscopy, the
transition of strain relief was found along the narrow circumference of the pillars.
Reducing the dimension of micro-pillar enabled not only a higher level of strain relief
but also a higher volume ratio of strain distributing circumference, resulting in a uniform
strain relief across the entire pillar. Our work demonstrates that micro-Raman serves as
a powerful tool for strain analysis.
The design of strain relief micro-LED-pillar was investigated in terms of height and
size. Our experimental study suggested that micro-pillars with diameter < 3 µm and
depth (height) > 3 µm are promising candidates for GaN based high-brightness LEDs. We
demonstrated that micro-LED-pillar in an array configuration could act as the building
78
block for large effective-area, high brightness emitter, which can be realized using
CMOS-compatible passivation and planarization processes.
The significance and major finding of this work are as follows,
Top-down approach of high quality InGaN/GaN MQW micro-LED-pillars was
demonstrated to be feasible.
A reduction of efficiency droop effect was observed in micro-LED-pillars, which
was attributed to both current spreading effect and strain relief effect.
A peak-splitting of E2 optical phonon peak is found in as-fabricated micro-LED-
pillars, indicating a sharp transition from strain relief to strained region.
For micro-LED-pillars with D > 5 μm, a significant “strain-relax ring” is found at
the vicinity of pillars with a width of 1-2 μm.
Thermal processing using RTA spread out the strain relaxation and restored
potential processing damage, which was proved to be a necessary step in future
device fabrication.
A minimum stress of -0.4GPa is observed in micro-LED-pillar with D < 3 μm.
The strain relief micro-LED-pillar could be arranged in an array for building large-
effective area, high brightness LEDs with improved EQE.
In this thesis, the feasibility of building high efficiency LEDs based on micro-LED-
pillar array by top-down approach is discussed. The design of micro-LED-pillar with D < 3
79
µm and L > 3 µm is favored for high efficiency, low droop devices. Further research and
exploration are required in order to manufacture practical large effective area LEDs.
First of all, in our demonstrated micro-LED-pillar array, FIB technique was applied.
However, for cost-effectiveness and mass production, CMOS based dry etching process
for micro-LED-pillar array formation has to be developed. To minimize the etching
damage on the device, further study on the selection of hard mask material,
optimization of dry etching condition are required. Furthermore, our study suggested
that post-patterning RTA could bring a uniform strain relief as well as a recovery of
sidewall damage. Therefore, the RTA process is another area to be further optimized.
Finally, the sidewall passivation is also a vital process to be investigated for a reliable
device. Aiming at reducing the dangling bonds and surface states, those processes
involve evaporating parylene, spin-coated bisbenzocyclobutene (BCB), SU-8 or PDMS
and the etch-back process. The success of the process integration will allow various high
brightness LEDs (blue, green, white) to be thoroughly investigated and eventually mass
produced based on the micro-structure method proposed in this thesis.
80
REFERENCES
[1] U. S. D. O. Energy, "Solid-State Lighting," Report on, January, 2013 2013.
[2] S. N. Laboratories. (2013). Solid-State Lighting Science EFRC. Available: http://energy.sandia.gov/?page_id=452
[3] J. Y. Tsao, M. E. Coltrin, M. H. Crawford, and J. A. Simmons, "Solid-State Lighting: An Integrated Human Factors, Technology, and Economic Perspective," Proceedings of the IEEE, vol. 98, pp. 1162-1179, Jul 2010.
[4] J. M. Phillips, M. E. Coltrin, M. H. Crawford, A. J. Fischer, M. R. Krames, R. Mueller-Mach, et al., "Research challenges to ultra-efficient inorganic solid-state lighting," Laser & Photonics Reviews, vol. 1, pp. 307-333, Dec 2007.
[5] S. Nakamura, T. Mukai, and M. Senoh, "Candela-Class High-Brightness InGaN/AlGaN Double-Heterostructure Blue-Light-Emitting Diodes," Applied Physics Letters, vol. 64, pp. 1687-1689, Mar 28 1994.
[6] Y. Uchida, T. Setomoto, T. Taguchi, Y. Nakagawa, and K. Miyazaki, "Characteristics of high-efficient InGaN-based white LED lighting," in Display Technologies Iii. vol. 4079, I. W. Wu and H. Uchiike, Eds., ed, 2000, pp. 120-126.
[7] G. Z. i. Anibal de Almeida, Michel Quicheron, Paolo Bertoldi "Accelerating the deployment of Solid State Lighting (SSL) in Europe," Publications Office of the European Union, Luxembourg JRC 71058, 2013.
[8] E. F. Schubert, Light-Emitting Diodes: Cambridge University Press, 2006.
[9] T. Gessmann and E. F. Schubert, "High-efficiency AlGaInP light-emitting diodes for solid-state lighting applications," Journal of Applied Physics, vol. 95, pp. 2203-2216, Mar 1 2004.
[10] D. H. Newman, GaAs optoelectronic devices: Wiley, 1985.
[11] K. Eberl, M. O. Lipinski, Y. M. Manz, W. Winter, N. Y. Jin-Phillipp, and O. G. Schmidt, "Self-assembling quantum dots for optoelectronic devices on Si and GaAs," Physica E, vol. 9, pp. 164-174, Jan 2001.
[12] P. Lian, T. Yin, Z. T. Xu, H. D. Zhao, D. S. Zou, G. Gao, et al., "High quality carbon doped GaAs/AlGaAs material growth by MOCVD and its application for optoelectronic devices," Semiconductor Lasers Iii, vol. 3547, pp. 278-284, 1998.
[13] Z. Q. Zhou, Y. W. Lin, Z. C. Niu, and C. Y. Li, "MBE growth of high quality GaAs heterostructures for optoelectronic devices," Semiconductor Lasers Ii, vol. 2886, pp. 313-318, 1996.
[14] S. Pimputkar, J. S. Speck, S. P. DenBaars, and S. Nakamura, "Prospects for LED lighting," Nature Photonics, vol. 3, pp. 179-181, Apr 2009.
81
[15] J. Y. Kim, Y. Tak, J. Kim, H. G. Hong, S. Chae, J. W. Lee, et al., "Highly efficient InGaN/GaN blue LED on 8-inch Si (111) substrate," Gallium Nitride Materials and Devices Vii, vol. 8262, 2012.
[16] Y. K. Su, "Nitride-Based LEDs and Superluminescent LEDs," in Comprehensive Semiconductor Science and Technology, B. Editors-in-Chief: Pallab, F. Roberto, and K. Hiroshi, Eds., ed Amsterdam: Elsevier, 2011, pp. 28-100.
[17] C. H. Liao, C. Y. Chen, H. S. Chen, K. Y. Chen, W. L. Chung, W. M. Chang, et al., "Emission Efficiency Dependence on the p-GaN Thickness in a High-Indium InGaN/GaN Quantum-Well Light-Emitting Diode," IEEE Photonics Technology Letters, vol. 23, pp. 1757-1759, Dec 1 2011.
[18] W. C. Huang, C. F. Lin, T. H. Hsieh, S. H. Chen, M. S. Lin, K. T. Chen, et al., "InGaN light emitting diodes with a laser-treated tapered GaN structure," Optics Express, vol. 19, pp. A1126-A1134, Sep 12 2011.
[19] P. T. Torma, O. Svensk, M. Ali, S. Suihkonen, M. Sopanen, M. A. Odnoblyudov, et al., "Effect of InGaN underneath layer on MOVPE-grown InGaN/GaN blue LEDs," Journal of Crystal Growth, vol. 310, pp. 5162-5165, Nov 15 2008.
[20] Y. J. Lee, T. C. Lu, H. C. Kuo, and S. C. Wang, "High brightness GaN-based light-emitting diodes," Journal of Display Technology, vol. 3, pp. 118-125, Jun 2007.
[21] S. C. Wei, Y. K. Su, S. J. Chang, S. M. Chen, and W. L. Li, "Nitride-based MQW LEDs with multiple GaN-SiN nucleation layers," IEEE Transactions on Electron Devices, vol. 52, pp. 1104-1109, Jun 2005.
[22] A. G. Bhuiyan, A. Hashimoto, and A. Yamamoto, "Indium nitride (InN): A review on growth, characterization, and properties," Journal of Applied Physics, vol. 94, pp. 2779-2808, Sep 2003.
[23] I. NSM. Semiconductors on NSM. Available: http://www.ioffe.ru/SVA/NSM/Semicond/
[24] J. Karpinski, J. Jun, and S. Porowski, "Equilibrium Pressure of N-2 over Gan and High-Pressure Solution Growth of Gan," Journal of Crystal Growth, vol. 66, pp. 1-10, 1984.
[25] T. Paskova, D. A. Hanser, and K. R. Evans, "GaN Substrates for III-Nitride Devices," Proceedings of the IEEE, vol. 98, pp. 1324-1338, Jul 2010.
[26] A. Honshio, T. Kitano, M. Imura, Y. Miyake, H. Kasugai, K. Iida, et al., "Group III nitrides grown on 4H-SiC (3038) substrate by metal-organic vapor phase epitaxy," GaN, AIN, InN and Their Alloys, vol. 831, pp. 685-690, 2005.
[27] K. Jeganathan, M. Shimuzu, H. Okumura, F. Hirose, and S. Nishizawa, "Initial stage of GaN nucleation on root 3 x root 3 R30 degrees-Ga reconstructed 4H-SiC(0001)(Si) by molecular-beam epitaxy," Surface Science, vol. 527, pp. L197-L202, Mar 10 2003.
82
[28] K. Jeganathan, M. Shimuzu, T. Ide, and H. Okumura, "Control of GaN surface morphologies grown on 6H-SiC (0001) using plasma-assisted molecular beam epitaxy," Journal of Crystal Growth, vol. 244, pp. 33-38, Sep 2002.
[29] J. Y. Kim, Y. Tak, H. G. Hong, S. Chae, J. W. Lee, H. Choi, et al., "Highly efficient InGaN/GaN blue LEDs on large diameter Si (111) substrates comparable to those on sapphire," Eleventh International Conference on Solid State Lighting, vol. 8123, 2011.
[30] X. H. Huang, J. P. Liu, Y. M. Fan, J. J. Kong, H. Yang, and H. B. Wang, "Improving InGaN-LED performance by optimizing the patterned sapphire substrate shape," Chinese Physics B, vol. 21, Mar 2012.
[31] H. Y. Ryu, J. I. Shim, C. H. Kim, J. H. Choi, H. M. Jung, M. S. Noh, et al., "Efficiency and Electron Leakage Characteristics in GaN-Based Light-Emitting Diodes Without AlGaN Electron-Blocking-Layer Structures," Ieee Photonics Technology Letters, vol. 23, pp. 1866-1868, Dec 15 2011.
[32] S. A. Kukushkin, A. V. Osipov, V. N. Bessolov, B. K. Medvedev, V. K. Nevolin, and K. A. Tcarik, "Substrates for epitaxy of gallium nitride: New materials and techniques," Reviews on Advanced Materials Science, vol. 17, pp. 1-32, Mar 2008.
[33] A. Nishikawa, T. Kawasaki, N. Furukawa, Y. Terai, and Y. Fujiwara, "Electroluminescence properties of Eu-doped GaN-based red light-emitting diode by OMVPE," Physica Status Solidi a-Applications and Materials Science, vol. 207, pp. 1397-1399, Jun 2010.
[34] Y. K. Ee, J. M. Biser, W. J. Cao, H. M. Chan, R. P. Vinci, and N. Tansu, "Metalorganic Vapor Phase Epitaxy of III-Nitride Light-Emitting Diodes on Nanopatterned AGOG Sapphire Substrate by Abbreviated Growth Mode," IEEE Journal of Selected Topics in Quantum Electronics, vol. 15, pp. 1066-1072, Jul-Aug 2009.
[35] T. Honda, T. Kobayashi, S. Egawa, M. Sawada, K. Sugimoto, and T. Baba, "Integrated light-emitting diodes grown by MOVPE for flat panel displays," Journal of Crystal Growth, vol. 298, pp. 736-739, Jan 2007.
[36] A. Dadgar, C. Hums, A. Diez, J. Blasing, and A. Krost, "Growth of blue GaN LED structures on 150-mm Si(111)," Journal of Crystal Growth, vol. 297, pp. 279-282, Dec 29 2006.
[37] Y. P. Hsu, S. J. Chang, Y. K. Su, S. C. Chen, J. M. Tsai, W. C. Lai, et al., "InGaN-GaN MQW LEDs with Si treatment," Ieee Photonics Technology Letters, vol. 17, pp. 1620-1622, Aug 2005.
[38] P. De Mierry, J. M. Bethoux, H. P. D. Screnk, M. Vaille, E. Feltin, B. Beaumont, et al., "Vertical cavity InGaN LEDs grown by MOVPE," Physica Status Solidi a-Applied Research, vol. 192, pp. 335-340, Aug 16 2002.
[39] A. H. Blake, D. Caselli, C. Durot, J. Mueller, E. Parra, J. Gilgen, et al., "InGaN/GaN multiple-quantum-well light-emitting diodes grown on Si(111) substrates with ZrB2(0001) buffer layers," Journal of Applied Physics, vol. 111, Feb 1 2012.
83
[40] M. Wakui, H. Sameshima, F. R. Hu, and K. Hane, "Fabrication of GaN light emitting diode membrane on Si substrate for MEMS applications," Microsystem Technologies-Micro-and Nanosystems-Information Storage and Processing Systems, vol. 17, pp. 109-114, Jan 2011.
[41] M. Zhang, P. Bhattacharya, W. Guo, and A. Banerjee, "Mg doping of GaN grown by plasma-assisted molecular beam epitaxy under nitrogen-rich conditions," Applied Physics Letters, vol. 96, Mar 29 2010.
[42] Y. Cordier, J. C. Moreno, N. Baron, E. Frayssinet, J. M. Chauveau, M. Nemoz, et al., "Growth of GaN based structures on Si(110) by molecular beam epitaxy," Journal of Crystal Growth, vol. 312, pp. 2683-2688, Sep 15 2010.
[43] Y. S. Park, C. M. Park, J. W. Lee, H. Y. Cho, T. W. Kang, K. H. Yoo, et al., "Electrical transport properties of a nanorod GaN p-n homojunction grown by molecular-beam epitaxy," Journal of Applied Physics, vol. 103, Mar 15 2008.
[44] A. Dussaigne, J. Brault, B. Damilano, and J. Massies, "Monolithic white light emitting diodes with a broad emission spectrum," Physica Status Solidi C - Current Topics in Solid State Physics, Vol 4, No 1, vol. 4, pp. 57-60, 2007.
[45] I. A. Buyanova, M. Izadifard, L. Storasta, W. M. Chen, J. Kim, F. Ren, et al., "Optical and electrical characterization of (Ga,Mn)N/InGaN multiquantum well light-emitting diodes," Journal of Electronic Materials, vol. 33, pp. 467-471, May 2004.
[46] M. A. L. Johnson, Z. H. Yu, J. D. Brown, F. A. Koeck, N. A. El-Masry, H. S. Kong, et al., "A critical comparison between MOVPE and MBE growth of III-V nitride semiconductor materials for opto-electronic device applications," MRS Internet Journal of Nitride Semiconductor Research, vol. 4, pp. art. no.-G5.10, 1999.
[47] M. H. Crawford, "LEDs for Solid-State Lighting: Performance Challenges and Recent Advances," IEEE Journal of Selected Topics in Quantum Electronics, vol. 15, pp. 1028-1040, Jul-Aug 2009.
[48] S. Q. Li, L. Wang, Y. J. Han, Y. Luo, H. Q. Deng, J. S. Qiu, et al., "A new growth method of roughed p-GaN in GaN-based light emitting diodes," Acta Physica Sinica, vol. 60, Sep 2011.
[49] M. K. Kwon, J. Y. Kim, I. K. Park, K. S. Kim, G. Y. Jung, S. J. Park, et al., "Enhanced emission efficiency of GaN/InGaN multiple quantum well light-emitting diode with an embedded photonic crystal," Applied Physics Letters, vol. 92, Jun 23 2008.
[50] Y. Jiang, Y. Luo, L. Wang, H. T. Li, G. Y. Xi, W. Zhao, et al., "Influence of pillar- and hole- patterned sapphire substrates on MOVPE grown GaN bulk and LED structures," Acta Physica Sinica, vol. 58, pp. 3468-3473, May 2009.
[51] S. M. Pan, R. C. Tu, Y. M. Fan, R. C. Yeh, and J. T. Hsu, "Improvement of InGaN-GaN light-emitting diodes with surface-textured indium-tin-oxide transparent ohmic contacts," IEEE Photonics Technology Letters, vol. 15, pp. 649-651, May 2003.
84
[52] D. W. Merfeld, X. A. Cao, S. F. Leboeuf, S. D. Arthur, J. W. Kretchmer, and M. P. D'Evelyn, "Influence of GaN material characteristics on device performance for blue and ultraviolet light-emitting diodes," Journal of Electronic Materials, vol. 33, pp. 1401-1405, Nov 2004.
[53] S. F. Chichibu, A. Uedono, T. Onuma, B. A. Haskell, A. Chakraborty, T. Koyama, et al., "Origin of defect-insensitive emission probability in In-containing (Al, In, Ga) N alloy semiconductors," Nature Materials, vol. 5, pp. 810-816, Oct 2006.
[54] J. Chen, G. H. Fan, W. Pang, S. W. Zheng, and Y. Y. Zhang, "Improvement of Efficiency Droop in Blue InGaN Light-Emitting Diodes With p-InGaN/GaN Superlattice Last Quantum Barrier," Ieee Photonics Technology Letters, vol. 24, pp. 2218-2220, Dec 15 2012.
[55] J. Iveland, L. Martinelli, J. Peretti, J. S. Speck, and C. Weisbuch, "Direct Measurement of Auger Electrons Emitted from a Semiconductor Light-Emitting Diode under Electrical Injection: Identification of the Dominant Mechanism for Efficiency Droop," Physical Review Letters, vol. 110, p. 177406, 04/25/ 2013.
[56] Y. L. Li, Y. R. Huang, and Y. H. Lai, "Investigation of Efficiency Droop Behaviors of InGaN/GaN Multiple-Quantum-Well LEDs With Various Well Thicknesses," IEEE Journal of Selected Topics in Quantum Electronics, vol. 15, pp. 1128-1131, Jul-Aug 2009.
[57] L. Zhu, P. T. Lai, and H. W. Choi, "An emissive GaN micro-LED array for visible-light multi-channel communication," in Photonics Global Conference (PGC), 2010, 2010, pp. 1-3.
[58] Z. Shuailong, J. J. D. McKendry, G. Zheng, B. R. Rae, S. Watson, X. Enyuan, et al., "Directly color-tunable smart display based on a CMOS-controlled micro-LED array," in Photonics Conference (IPC), 2012 IEEE, 2012, pp. 435-436.
[59] S. H. Margueron, P. Bourson, S. Gautier, A. Soltani, D. Troadec, J. C. De Jaeger, et al., "Residual stress relaxation in GaN/sapphire circular pillars measured by Raman scattering spectroscopy," Journal of Crystal Growth, vol. 310, pp. 5321-5326, Dec 1 2008.
[60] N. Thillosen, K. Sebald, H. Hardtdegen, R. Meijers, R. Calarco, S. Montanari, et al., "The state of strain in single GaN nanocolumns as derived from micro-photoluminescence measurements," Nano Letters, vol. 6, pp. 704-708, Apr 2006.
[61] T. Martensson, P. Carlberg, M. Borgstrom, L. Montelius, W. Seifert, and L. Samuelson, "Nanowire arrays defined by nanoimprint lithography," Nano Letters, vol. 4, pp. 699-702, Apr 2004.
[62] S. Bhunia, T. Kawamura, Y. Watanabe, S. Fujikawa, and K. Tokushima, "Metalorganic vapor-phase epitaxial growth and characterization of vertical InP nanowires," Applied Physics Letters, vol. 83, pp. 3371-3373, Oct 20 2003.
85
[63] H. S. Chen, C. F. Lu, D. M. Yeh, C. F. Huang, J. J. Huang, and C. C. Yang, "Orange-red light-emitting diodes based on a prestrained InGaN-GaN quantum-well epitaxy structure," IEEE Photonics Technology Letters, vol. 18, pp. 2269-2271, Nov-Dec 2006.
[64] T. Egawa, B. Zhang, N. Nishikawa, H. Ishikawa, T. Jimbo, and M. Umeno, "InGaN multiple-quantum-well green light-emitting diodes on Si grown by metalorganic chemical vapor deposition," Journal of Applied Physics, vol. 91, pp. 528-530, Jan 1 2002.
[65] J. K. Sheu, G. C. Chi, Y. K. Su, C. C. Liu, C. M. Chang, W. C. Hung, et al., "Luminescence of an InGaN/GaN multiple quantum well light-emitting diode," Solid-State Electronics, vol. 44, pp. 1055-1058, Jun 2000.
[66] Y. C. Lin, S. J. Chang, Y. K. Su, T. Y. Tsai, C. S. Chang, S. C. Shei, et al., "InGaN/GaN light emitting diodes with Ni/Au, Ni/ITO and ITO p-type contacts," Solid-State Electronics, vol. 47, pp. 849-853, May 2003.
[67] S. N. Mohammad, "Contact mechanisms and design principles for alloyed ohmic contacts to n-GaN," Journal of Applied Physics, vol. 95, pp. 7940-7953, Jun 15 2004.
[68] D. F. Wang, S. W. Feng, C. Lu, A. Motayed, M. Jah, S. N. Mohammad, et al., "Low-resistance Ti/Al/Ti/Au multilayer ohmic contact to n-GaN," Journal of Applied Physics, vol. 89, pp. 6214-6217, Jun 1 2001.
[69] J. K. Sheu, Y. K. Su, G. C. Chi, M. J. Jou, C. C. Liu, C. M. Chang, et al., "Investigation of the mechanism for Ti/Al ohmic contact on etched n-GaN surfaces," Journal of Vacuum Science & Technology B, vol. 18, pp. 729-732, Mar-Apr 2000.
[70] H. Kim, J. Cho, J. W. Lee, S. Yoon, H. Kim, C. Sone, et al., "Measurements of current spreading length and design of GaN-based light emitting diodes," Applied Physics Letters, vol. 90, Feb 5 2007.
[71] X. Guo and E. F. Schubert, "Current crowding and optical saturation effects in GaInN/GaN light-emitting diodes grown on insulating substrates," Applied Physics Letters, vol. 78, pp. 3337-3339, May 21 2001.
[72] V. K. Malyutenko, S. S. Bolgov, and A. D. Podoltsev, "Current crowding effect on the ideality factor and efficiency droop in blue lateral InGaN/GaN light emitting diodes," Applied Physics Letters, vol. 97, Dec 20 2010.
[73] X. H. Wang, "Optical studies of focused ion beam fabricated GaN microstructures andnanostructures," Master of Philosophy Department of Electrical and Electronic Engineering, The University of Hong Kong Hong Kong 2011.
[74] T. J. Yu, X. N. Kang, Y. B. Pan, Z. X. Qin, Z. Z. Chen, Z. J. Yang, et al., "Strain effect on photoluminescences from InGaN MQWs with different barriers grown by MOCVD," Chinese Physics Letters, vol. 24, pp. 1365-1367, May 2007.
[75] J. H. Son and J. L. Lee, "Strain engineering for the solution of efficiency droop in InGaN/GaN light-emitting diodes," Optics Express, vol. 18, pp. 5466-5471, Mar 15 2010.
86
[76] A. G. Kontos, Y. S. Raptis, N. T. Pelekanos, A. Georgakilas, E. Bellet-Amalric, and D. Jalabert, "Micro-Raman characterization of InxGa1-xN/GaN/Al2O3 heterostructures," Physical Review B, vol. 72, Oct 2005.
[77] I. Gorczyca, N. E. Christensen, E. L. P. Y. Blanca, and C. O. Rodriguez, "Optical Phonon Modes in Gan and Aln," Physical Review B, vol. 51, pp. 11936-11939, May 1 1995.
[78] A. Hushur, M. H. Manghnani, and J. Narayan, "Raman studies of GaN/sapphire thin film heterostructures," Journal of Applied Physics, vol. 106, p. 054317, 2009.
[79] T. Sugiura, E. H. Kim, Y. Honda, H. Takagi, T. Tsukamoto, H. Andoh, et al., "Raman Spectroscopic Study Of Residual Strain In (1-101) GaN And (0001) GaN Layers Grown On Si Substrate," Physics of Semiconductors: 30th International Conference on the Physics of Semiconductors, vol. 1399, 2011.
[80] V. Laneuville, F. Demangeot, R. Pechou, P. Salles, A. Ponchet, G. Jacopin, et al., "Double strain state in a single GaN/AlN nanowire: Probing the core-shell effect by ultraviolet resonant Raman scattering," Physical Review B, vol. 83, Mar 10 2011.
[81] J. Wagner, A. Ramakrishnan, H. Obloh, and M. Maier, "Effect of strain and associated piezoelectric fields in InGaN GaN quantum wells probed by resonant Raman scattering," Applied Physics Letters, vol. 74, pp. 3863-3865, Jun 21 1999.
[82] I. DeWolf, "Micro-Raman spectroscopy to study local mechanical stress in silicon integrated circuits," Semiconductor Science and Technology, vol. 11, pp. 139-154, Feb 1996.
[83] S. F. Chichibu, A. C. Abare, M. S. Minsky, S. Keller, S. B. Fleischer, J. E. Bowers, et al., "Effective band gap inhomogeneity and piezoelectric field in InGaN/GaN multiquantum well structures," Applied Physics Letters, vol. 73, pp. 2006-2008, Oct 5 1998.
[84] K. P. O'Donnell, M. A. D. Maur, A. Di Carlo, K. Lorenz, and S. Consortium, "It's not easy being green: Strategies for all-nitrides, all-colour solid state lighting," Physica Status Solidi-Rapid Research Letters, vol. 6, pp. 49-52, Feb 2012.
[85] P. Yu, C. H. Chiu, Y.-R. Wu, H. H. Yen, J. R. Chen, C. C. Kao, et al., "Strain relaxation induced microphotoluminescence characteristics of a single InGaN-based nanopillar fabricated by focused ion beam milling," Applied Physics Letters, vol. 93, pp. 081110-3, 08/25/ 2008.
[86] E. Y. Xie, Z. Z. Chen, P. R. Edwards, Z. Gong, N. Y. Liu, Y. B. Tao, et al., "Strain relaxation in InGaN/GaN micro-pillars evidenced by high resolution cathodoluminescence hyperspectral imaging," Journal of Applied Physics, vol. 112, Jul 1 2012.
[87] H. J. Chang, Y. P. Hsieh, T. T. Chen, Y. F. Chen, C. T. Liang, T. Y. Lin, et al., "Strong luminescence from strain relaxed InGaN/GaN nanotips for highly efficient light emitters," Optics Express, vol. 15, pp. 9357-9365, Jul 23 2007.
87
[88] T. Y. Tang, C. H. Lin, Y. S. Chen, W. Y. Shiao, W. M. Chang, C. H. Liao, et al., "Nitride Nanocolumns for the Development of Light-Emitting Diode," IEEE Transactions on Electron Devices, vol. 57, pp. 71-78, Jan 2010.
[89] F. Sacconi, M. A. D. Maur, and A. Di Carlo, "Optoelectronic Properties of Nanocolumn InGaN/GaN LEDs," IEEE Transactions on Electron Devices, vol. 59, pp. 2979-2987, Nov 2012.
[90] J. M. Lee, C. Huh, D. J. Kim, and S. J. Park, "Dry-etch damage and its recovery in InGaN/GaN multi-quantum-well light-emitting diodes," Semiconductor Science and Technology, vol. 18, pp. 530-534, Jun 2003.
[91] N. Yao, Focused Ion Beam Systems: Basics and Applications: Cambridge University Press, 2011.
[92] C. S. Kim, S. H. Ahn, and D. Y. Jang, "Review: Developments in micro/nanoscale fabrication by focused ion beams," Vacuum, vol. 86, pp. 1014-1035, Feb 29 2012.
[93] M. K. Miller, K. F. Russell, K. Thompson, R. Alvis, and D. J. Larson, "Review of atom probe FIB-Based specimen preparation methods," Microscopy and Microanalysis, vol. 13, pp. 428-436, Dec 2007.
[94] J. M. Poate and K. Saadatmand, "Ion beam technologies in the semiconductor world (plenary)," Review of Scientific Instruments, vol. 73, pp. 868-872, Feb 2002.
[95] S. Reyntjens and R. Puers, "A review of focused ion beam applications in microsystem technology," Journal of Micromechanics and Microengineering, vol. 11, pp. 287-300, Jul 2001.
[96] J. Martinez, F. Barbagini, A. Bengoechea-Encabo, S. Albert, M. A. S. Garcia, and E. Calleja, "Fabrication of GaN nanorods by focused ion beam," Microelectronic Engineering, vol. 98, pp. 250-253, Oct 2012.
[97] S. Dhara, C. Y. Lu, C. T. Wu, C. W. Hsu, W. S. Tu, K. H. Chen, et al., "Focused Ion Beam Induced Nanojunction and Defect Doping as a Building Block for Nanoscale Electronics in GaN Nanowires," Journal of Physical Chemistry C, vol. 114, pp. 15260-15265, Sep 16 2010.
[98] S. Dhara, "Formation, dynamics, and characterization of nanostructures by ion beam irradiation," Critical Reviews in Solid State and Materials Sciences, vol. 32, pp. 1-50, 2007.
[99] C. Y. Nam, J. Y. Kim, and J. E. Fischer, "Focused-ion-beam platinum nanopatterning for GaN nanowires: Ohmic contacts and patterned growth," Applied Physics Letters, vol. 86, May 9 2005.
[100] H. A. R. El-Ella, D. P. Collins, M. J. Kappers, R. A. Taylor, and R. A. Oliver, "Optical cavity efficacy and lasing of focused ion beam milled GaN/InGaN micropillars," Journal of Applied Physics, vol. 112, Dec 1 2012.
88
[101] C. H. Liang, L. C. Chen, J. S. Hwang, K. H. Chen, Y. T. Hung, and Y. F. Chen, "Selective-area growth of indium nitride nanowires on gold-patterned Si(100) substrates," Applied Physics Letters, vol. 81, pp. 22-24, Jul 1 2002.
[102] S. B. Dun, Y. Jiang, J. Q. Li, Y. L. Fang, J. Y. Yin, B. Liu, et al., "Micro-Raman spectroscopy observation of field-induced strain relaxation in AlGaN/GaN heterojunction field-effect transistors," Physica Status Solidi a-Applications and Materials Science, vol. 209, pp. 1174-1178, Jun 2012.
[103] P. Yu, M.-A. Tsai, C.-H. Chiu, H.-c. Kuo, and Y.-R. Wu, "Strain relaxation characteristics of a single InGaN-based nanopillar fabricated by focused ion beam milling," Optoelectronic Materials and Devices III , Hangzhou, China, 2008, pp. 71351Y-6.
[104] C. C. S. Chan, P. A. Shields, M. J. Holmes, Y. D. Zhuang, X. Wang, B. P. L. Reid, et al., "Optical studies of quantum dot-like emission from localisation centres in InGaN/GaN nanorod array LEDs," Physica Status Solidi C: Current Topics in Solid State Physics, Vol 9, No 3-4, vol. 9, pp. 635-638, 2012.
[105] H. W. Lin, Y. J. Lu, H. Y. Chen, H. M. Lee, and S. Gwo, "InGaN/GaN nanorod array white light-emitting diode," Applied Physics Letters, vol. 97, Aug 16 2010.
[106] S. Keller, N. A. Fichtenbaum, C. Schaake, C. J. Neufeld, A. David, E. Matioli, et al., "Optical properties of GaN nanopillar and nanostripe arrays with embedded InGaN/GaN multi quantum wells," Physica Status Solidi B-Basic Solid State Physics, vol. 244, pp. 1797-1801, Jun 2007.
[107] S. Keller, C. Schaake, N. A. Fichtenbaum, C. J. Neufeld, Y. Wu, K. McGroddy, et al., "Optical and structural properties of GaN nanopillar and nanostripe arrays with embedded InGaN/GaN multi-quantum wells," Journal of Applied Physics, vol. 100, Sep 1 2006.
[108] J. H. Zhu, S. M. Zhang, X. Sun, D. G. Zhao, J. J. Zhu, Z. S. Liu, et al., "Fabrication and optical characterization of GaN-based nanopillar light emitting diodes," Chinese Physics Letters, vol. 25, pp. 3485-3488, Sep 2008.
[109] Y. D. Wang, S. J. Chua, S. Tripathy, M. S. Sander, P. Chen, and C. G. Fonstad, "High optical quality GaN nanopillar arrays," Applied Physics Letters, vol. 86, Feb 14 2005.
89
APPENDICES
Appendix A. Design of mask for large-area LEDs and micro-LED-pillars
fabrication
In this study, two masks are designed for MESA definition and N contact formation.
The masks polarities are digitized dark/brightfield and digitized clear/darkfield,
respectively. Those were chosen due the fact that MESA was defined by direct dry
etching while N contact was achieved by lift-off process. The design of masks in each die
can be found in Figure 45 and Figure 46.
Figure 45. Mask design for MESA of large area LEDs and micro-LED-pillars.
90
Figure 46. Mask design for N contact of large area LEDs and micro-LED-pillars.
91
Appendix B. Fabrication process flow of InGaN/GaN MQW LEDs
Step number and name
Sub steps Equipment and bench
Process recipes Note
1 Wafer
cleaning Wet bench
a) HCl solution, 1 minute
b) IPA solution, ultrasonic, 3 minutes
c) Acetone, ultrasonic, 3 minutes
2 P-contact formation
Ni deposition DC sputter
DC power: 100W Height: 160H Pressure: 10 mTorr Ar flow: 20 sccm Time: 30 seconds
ITO deposition
RF sputter
RF power: 50W Height: 180H Pressure: 2 mTorr Ar flow: 20 sccm Time: 45 minutes
Annealing RTP 600 ℃ , atmosphere, 60 seconds
3 MESA
formation
Pre cleaning Wet bench
a) IPA solution, ultrasonic, 2 minutes
b) Acetone, ultrasonic, 2 minutes
Sample preparation*
JST Resist Spin/Bake
a) Spin-coat AZ 5214 on 4’’ Si wafer
b) Put GaN LED wafer on the Si wafer
c) Bake at 120 ℃ for 5 minutes
*stick sample on Si
PR spin-coating
JST Resist Spin/Bake
PR: AZ 5214 Step 1: 3 seconds,
speed:800 rpm, Ramp: 1000rpm/s.
92
Step 2: 3 seconds, speed:1500 rpm, Ramp: 1000rpm/s.
Step 3: 30 seconds, speed:1500 rpm, Ramp: 1500rpm/s.
Soft-bake: 120 ℃ for 120 seconds
UV-exposure EVG 6200
UV contact aligner
Dose: 90 mJ/cm2 Separation: 30 µm.
Development Wet bench
Developer: MIF 726 Time: 30 seconds Post-bake *: 100 ℃ for 15 seconds
*Water removal
ICP-RIE etching*
Oxford Plasma 100
Temperature: 30 ℃ Gas flow: Ar = 5 sccm, Cl2 = 15 sccm, BCl3 = 25 sccm. Power: RF = 100W, ICP = 500W Pressure = 10 mTorr
*Helium back
cooling is applied
PR removal Wet bench Acetone, 5 minutes
4 N-contact
formation*
PR removal Wet bench Acetone, 5 minutes * omitted
Sample preparation
JST Resist Spin/Bake
Stick sample on 4’’ Si wafer
PR spin-coating
JST Resist Spin/Bake
Spin 1.6um AZ5214, Soft bake: 100 ℃ , 1 minute
UV-exposure EVG 6200
UV contact aligner
Dose: 110 mJ/cm2 Separation: 30 µm.
Reversal bake
JST Resist Spin/Bake
120 ℃ , 2 minutes
93
Flood-exposure
EVG 6200 UV contact
aligner
Dose: 200 mJ/cm2 Separation: 30 µm.
Development Wet bench Developer: MIF 726 Time: 60 seconds
Metal deposition
DC sputter Sputter Ti/Al/Ti/Au
Lift-off Wet bench Acetone, 5 minutes
94
Appendix C. Programs for I-V and L-I measurements
Interface for I-V measurements:
1. Select the voltage source range and compliance value. 3. Type in the range of injection and
delay time if applicable.
3.
5. 6.
2. File Output
95
Labview code for I-V measurements (selected):
96
Interface for L-I measurements:
1. Select the current source range and power meter channel.
3. Type in the range of injection.
4. 5.
2. File Output
6.
97
Labview code for I-V measurements (selected):