To my parents, family, friends and a special...
158
1 CONTROLLED SURFACE TOPOGRAPHY OF TRANSPARENT CONDUCTING OXIDE FOR ENHANCED LIGHT EXTRACTION IN LIGHT EMITTING DEVICES By SUSHANT GUPTA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
Transcript of To my parents, family, friends and a special...
1
CONTROLLED SURFACE TOPOGRAPHY OF TRANSPARENT CONDUCTING OXIDE FOR ENHANCED LIGHT EXTRACTION IN LIGHT EMITTING DEVICES
By
SUSHANT GUPTA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2010
2
© 2010 Sushant Gupta
3
To my parents, family, friends and a special someone
4
ACKNOWLEDGMENTS
I would like to express my heartiest gratitude to Prof. Rajiv K. Singh for his support
and guidance without which none of this work would have been possible. His
encouragements and suggestions made the research more interesting and challenging.
His patience and faith made me successfully complete this endeavor even after
changing 3 projects. The dedication to his work and sleepless nights he spent for
submission of proposals, motivated me to go that extra mile. He has been a great
advisor and a mentor not just for my research at UF but also on a personal level. I
would like to acknowledge my supervisory committee: Prof. S. J. Pearton, Prof. D.
Norton, Prof. Timothy Anderson, Prof. V. Craciun and Prof. Henry Hess for serving on
my supervisory committee, for their time and suggestions.
I would like to thank, Tyler Parent, my manager at Maxim Integrated products for
his belief and confidence in me during my short stay as an intern. His candid advices
helped me shape my career and currently guiding me to choose the suitable career
path. His faith in my abilities instilled confidence in me and gave me opportunity to lead
and represent the CMP team for development of a next generation device. My sincere
gratitude for my mentor Dr. Kamal Mishra, for his guidance and training which helped
me work on various tools independently and demonstrate my skills. None of my success
at Maxim would been possible without the support of Jack Huang, Kiyoko, Sudhir
Chopra, Viral Shah, Ching Tang, and other team-members and technicians who made it
so easy to transition in to Maxim culture. Special thanks to Anuranjan, who was the
reason for getting this great internship experience. No words can be enough to express
my gratitude for him and Dhriti (his wife) for providing me with place to stay early on in
California and welcoming me in their family. Tvisha (Chiya), you might not be able to
5
read this now but your little mischief and lovely ways of making me smile made me
forget the world and made the life look like a fairy-tale. I am truly thankful to you three
for making me feel like a part of your family, without you California stay would have
been totally different and very difficult.
I am thankful to all my past and present colleagues in Dr. Singh’s group: Dr.
Karthik Ramani, Dr. Seung-Young Son, Dr. Feng Chi, Dr. Sejin Kim, Dr. Taekon Kim,
Dr. Jaesoek Lee and Myoung Hwan Oh, Jungbae Lee, Kannan Balasundar and Yu Liu.
Specials thanks to NRF staff: Bill Lewis, David Hays, Al Ogden and Dr. Brent Gila for
their guidance and help in troubleshooting my experiments. Also, I am grateful to Dr.
Mark Davidson, Jason Rowland and Chuck Rowland for their help with device
fabrication trials. I would like to thank Dr. Franky So, Dr. Do Young Kim and Wooram
Youn for their guidance and help with PL device fabrication and measurements.
I am grateful to all the people I met, all the roommates (I changed too many
apartments) who became my friends and made the stay at Gainesville a wonderful
learning experience. For making Gainesville fun and exciting, large number of my
friends played a role and I would like to thank, Aniruddh, Vibhava, Abhishek, Richa,
Karam, Arul, Parnitha Bhabhi, Anu, Ashutosh, Abu, Isha, Mamta, Ashwini and Preeti.
Special thanks to Purushottam (PK) for being sort of an advisor and a mentor after Dr.
Singh. I benefitted tremendously from his guidance and fruitful discussions on my
research work.
I can not say enough to express my gratitude and appreciation for Neetu who has
been my best friend and family away from family. Her relentless support and
encouragement pushed me to go on and overcome hurdles in my research and in a
6
way, in life. She has always been there for me and her presence made me feel that
there is always someone on whom I can always rely upon. She makes it so easy to
share even the most embarrassing moments with her. Another person who made a big
impression on me is Toral. We have been friends only for a year now but with her time
losses its count. She has been a source of support, encouragement and inspiration to
carry-on during the last few arduous semesters at UF. Although, outwardly she might be
very different person but I know from inside she is caring and loving. I cannot ever
forget our exercising and biking endeavors, HPNP discussions and of course her
delicious cooking. I am thankful to Toral, for coming in to my life and showing me a
different perspective to see life.
This acknowledgment would not be complete without thanking my parents and
family. Without my family’s (papa, mom, Anil chacha, my baba and amma’s) values and
principles, I can not imagine myself where I am right now. I am indebted by their
endless love and innumerable sacrifices which made me a successful person today. I
am thankful to my sister Surbhi for her love, patience and friendship. I express my
gratitude towards: bade and badi chachi, choti chachi, Rupali, Pragya, Sonu bhaiya,
choti boa, Ashu and all others those I have missed, for their love and support. I dedicate
this dissertation to Sangeeta didi and a special someone for their abysmal love, support,
encouragement, inspiration and sacrifices without which I would have settled for
something less and I won’t have embarked on this journey.
7
TABLE OF CONTENTS page
ACKNOWLEDGMENTS .................................................................................................. 4
LIST OF TABLES .......................................................................................................... 10
LIST OF FIGURES ........................................................................................................ 11
LIST OF ABBREVIATIONS ........................................................................................... 15
ABSTRACT ................................................................................................................... 16
CHAPTER
1 INTRODUCTION .................................................................................................... 19
Transparent Conducting Oxides ............................................................................. 19
Importance of TCOs ......................................................................................... 19 History of TCO .................................................................................................. 20
Zinc Oxide: A Promising TCO ................................................................................. 21
Properties of ZnO ............................................................................................. 22 ZnO Applications .............................................................................................. 23
TCO Roughness ..................................................................................................... 23 Light Extraction ....................................................................................................... 25
Basic Concepts ................................................................................................ 26 Light Extraction Efficiency ................................................................................ 27 Improving Light Out-coupling ............................................................................ 28
Device shaping .......................................................................................... 28 Photon recycling ........................................................................................ 28
Photon scattering ....................................................................................... 29 Summary ................................................................................................................ 30
2 LITERATURE REVIEW .......................................................................................... 36
TCO Surface Roughness ........................................................................................ 36 Light Extraction ....................................................................................................... 38
Device Shaping ................................................................................................ 39 Interface Modification Techniques .................................................................... 41
Surface roughening .................................................................................... 41 Photonic crystals ........................................................................................ 43 Patterned substrates .................................................................................. 45
External Out-couplers ....................................................................................... 47 Other Techniques ............................................................................................. 51
3 OUTLINE OF RESEARCH ..................................................................................... 62
8
4 CHEMICAL MECHANICAL POLISHING OF ZINC OXIDE ..................................... 64
Introduction ............................................................................................................. 64 Experimental Procedure ......................................................................................... 66
Results and Discussion........................................................................................... 67 Effect of Applied Down-pressure and Platen Speed ......................................... 68 Effect of Particle Size ....................................................................................... 69 Effect of Slurry pH ............................................................................................ 70 Effect of Particle Concentration ........................................................................ 71
Zinc Oxide CMP Mechanism ............................................................................ 71 Conclusion .............................................................................................................. 75
5 EFFECT OF SLURRY ADDITIVES: SURFACTANT AND SALTS .......................... 83
Introduction ............................................................................................................. 83 Experimental Procedure ......................................................................................... 84 Results & Discussion .............................................................................................. 85
Effect of Surfactant ........................................................................................... 85 Effect of Salt ..................................................................................................... 87
Conclusion .............................................................................................................. 88
6 RAY TRACING SIMULATIONS .............................................................................. 94
Introduction ............................................................................................................. 94
Device Simulation Parameters ................................................................................ 96 Results and Discussion........................................................................................... 98
Device A: Texture at Substrate/Air Interface .................................................... 98 Pyramid texture .......................................................................................... 98
Microlens texture ........................................................................................ 99 Cylindrical texture .................................................................................... 100
Device B: Texture at Substrate/TCO Interface ............................................... 101
Conclusion ............................................................................................................ 104
7 EXPERIMENTAL PROCEDURES FOR FABRICATION OF TEXTURES ............ 115
Experimental Procedure ....................................................................................... 115 Substrate Preparation .................................................................................... 115 Structure Fabrication ...................................................................................... 115
Cone/pyramid texture ............................................................................... 115 Microlens texture ...................................................................................... 117 Cylindrical texture .................................................................................... 117
Results and Discussion......................................................................................... 118
Cone/Pyramid Structure ................................................................................. 118 Microlens Structure ........................................................................................ 119 Cylindrical Structure ....................................................................................... 120
Conclusion ............................................................................................................ 121
9
8 MICRO-TEXTURED TRANSPARENT CONDUCTING OXIDE FOR ENHANCED LIGHT EXTRACTION EFFICIENCY ................................................ 128
Introduction ........................................................................................................... 128
Experimental Procedure ....................................................................................... 129 Substrate Preparation .................................................................................... 129 TCO Fabrication ............................................................................................. 130 Device Fabrication .......................................................................................... 131 Ray-tracing Simulations Parameters .............................................................. 132
Results and Discussion......................................................................................... 133 Conclusion ............................................................................................................ 136
9 CONCLUSIONS ................................................................................................... 142
Chemical Mechanical Polishing of ZnO ................................................................ 143 Light Extraction ..................................................................................................... 144 Corrugated Device ................................................................................................ 145
LIST OF REFERENCES ............................................................................................. 147
BIOGRAPHICAL SKETCH .......................................................................................... 158
10
LIST OF TABLES
Table page 1-1 Host and associated dopants in typical TCO systems ....................................... 34
1-2 Properties of ZnO ............................................................................................... 35
4-1 Fit parameters for the modified Preston equation ............................................... 82
4-2 Dissolution rates and change in RMS roughness. .............................................. 82
7-1 Maximum microlens height from square base pyramid of height H .................. 127
7-2 RIE etch recipe for silica etching ...................................................................... 127
11
LIST OF FIGURES
Figure page 1-1 Applications of transparent conducting oxides. .................................................. 31
1-2 Materials space for semimetals, good metals, transparent conductors and semiconductors based on n–µ correlations. ....................................................... 32
1-3 Schematic depicting Snell's law and total internal reflection. .............................. 33
1-4 Illustration of escape cone defined by the critical angle. ..................................... 33
1-5 Illustration depicting waveguiding of rays with emission angle greater than critical angle........................................................................................................ 33
2-1 Dark spot formation in LED active layer. ............................................................ 53
2-2 Evolution of extraction efficiency for AlGaInP (▲) and InGaN (■) LEDs. ........... 53
2-3 (a) Optical image of AlGaInP/GaP truncated inverted pyramid LED and (b) is schematic of a TIP LED device. .......................................................................... 54
2-4 SEM micrographs of an N-face GaN surface etched by a KOH-based PEC method. .............................................................................................................. 54
2-5 SEM image of Oblique angle deposited GRIN ITO layer with refractive index profile. ................................................................................................................. 55
2-6 (a) Schematic depicting fabrication of PDMS nanowires and (b) is the scanning electron micrograph of a fabricated meshed surface on PDMS film. ... 55
2-7 (a) Top lit view, (b) schematic cross-section view of the PXLED and (c) radiation patterns of the LEDs. ........................................................................... 56
2-8 (a) Schematic of a PC-OLED device with SiO2/SiNx PC layer with intensity profiles. ............................................................................................................... 56
2-9 (a) Schematic drawing of GaN based device on stripe PSS [104] and (b) cross-sectional SEM of LEPS-GaN grown on PSS. ........................................... 57
2-10 Cross-section SEM micrographs of GaN grown on V-grooved sapphire with a 6.5 μm period after 150 min growth time. ........................................................... 57
2-11 (a) Top-view SEM image of CWE-PSS and (b) is a schematic diagram of device on CWE-PSS. ......................................................................................... 58
12
2-12 (a) Schematic diagram and (b) cross-section TEM micrograph showing lateral epitaxial overgrowth of GaN layer on 6H-SiC substrate with SiO2 mask. .................................................................................................................. 58
2-13 SEM of a PDMS microlens array. ....................................................................... 59
2-14 Schematic of InGaN QWs LEDs structure utilizing SiO2/PS microspheres microlens array. .................................................................................................. 59
2-15 SEM image of (a) 1-D triangular and (b) 2-D taper-like gratings on the sapphire backplane. ........................................................................................... 60
2-16 Cross-sectional SEM image of bottom of 6H–SiC substrate with moth-eye structure fabricated using metal nanomask. ....................................................... 60
2-17 Schematic diagram of the OLED with the embedded low-index grid (LIG) in the organic layers. .............................................................................................. 61
4-1 AFM micrograph of the annealed zinc oxide film. ............................................... 76
4-2 X-ray diffraction patterns of the as-deposited and annealed ZnO films. ............. 76
4-3 Removal rate (■) with respect to the applied pressure during CMP (red line represents the linear fit). ..................................................................................... 77
4-4 Removal rate (■) versus linear velocity and RMS roughness (inset) (●) with respect to linear velocity. Red line represents the linear fit. ................................ 77
4-5 Graph representing removal rates at various particle sizes (■) whereas inset depicts the RMS surface roughness given by (●). .............................................. 78
4-6 Removal rates versus pH of the slurry. Inset represents the average RMS roughness. .......................................................................................................... 78
4-7 AFM micrographs of zinc oxide films polished with (a) pH 3.1, (b) pH 7.0 and (c) pH 10.6 slurry ................................................................................................ 79
4-8 Removal rate variation with weight percent solid loading of the polishing slurry solution. .................................................................................................... 79
4-9 Plot shows zeta potential with respect to the pH, for the silica particles (■) and the zinc oxide thin film (●). ........................................................................... 80
4-10 Optical transmission curves for the glass, unpolished and the samples polished with acidic and basic pH slurries. ......................................................... 80
4-11 (a) X-ray reflectivity curves and (b) schematic showing fitting results. ................ 81
13
5-1 Solid squares (■) show the removal rate (Å/min) variation with SDS concentration in slurry whereas RMS roughness is depicted by circles (○). ....... 89
5-2 Plot illustrating Removal rate (■) and RMS roughness (○) variation with respect to the concentration of TX-100 surfactant in the slurry........................... 89
5-3 Removal rate (■) and RMS roughness (○) with respect to the varying concentration of KCl salt in the slurry solution. ................................................... 90
5-4 AFM micrograph of the annealed zinc oxide film (scale bar is in Å). .................. 90
5-5 AFM micrographs for zinc oxide films polished with SDS slurry with SDS concentration of (a) 5 mM, (b) 10 mM, (c) 16 mM and (d) 30 mM. ..................... 91
5-6 AFM micrographs for zinc oxide films polished with Triton X-100 slurry with TX-100 concentration of (a) 0.322 g/l, (b) 0.64 g/l and (c) 1.28 g/l. .................... 92
5-7 AFM micrographs for zinc oxide films polished with (a) 0.1M, (b) 0.2 M and (c) 0.3 M KCl salt added to the polishing slurry. ................................................. 92
5-8 Optical transmission curves for samples polished using slurries with 0.1 M KCl, 5 mM SDS and 0.322 g/l TX-100. ............................................................... 93
6-1 Schematic depicting different device structures studied using ray-tracing simulations........................................................................................................ 106
6-2 Enhancement in out-coupled rays from a pyramid array textured substrate (Device A) represented by (a) and (b) whereas (c) represents the maximum simulated fractional powers obtained for the different base diameters. ............ 107
6-3 Angular distribution of the light out-coupled from device A with pyramid texture for active layer refractive index (a) nactive = 1.7 and (b) nactive = 2.5. ...... 108
6-4 (a) and (b) are enhancement in light out-coupling for a microlens array with device A configuration. The maximum simulated fractional powers at different base diameters are depicted by (c). ................................................................. 109
6-5 Angular distribution of the total light out-coupled from device A with microlens texture for (a) nactive = 1.7 and (b) nactive = 2.5. .................................................. 110
6-6 (a) and (b) depict the light out-coupled for a cylinder type texture at the substrate/air interface while (c) represents the maximum simulated fractional powers for different diameters. ......................................................................... 111
6-7 Angular spread of the light out-coupled from device A with cylindrical texture for active layer refractive index of (a) 1.7 and (b) 2.5. ...................................... 112
14
6-8 Light extraction enhancements from a microlens texture at the substrate/TCO interface (Device B) are depicted in (a) and (b) while (c) represents the maximum simulated fractional powers for different diameters. ......................... 113
6-9 Far-field angular spread of the light out-coupled from device B at different microlens heights for 6 μm base diameter and nactive of (a) 1.7 and (b) 2.5. ..... 114
6-10 Far field angular distribution of photons out-coupled from the TCO layer at different microlens heights for nactive = 1.7 and base diameter 6 μm. ................ 114
7-1 Schematic illustrating the mask designs. .......................................................... 123
7-2 Process flows for fabrication of (a) cone/pyramid and microlens and (b) cylindrical arrays. .............................................................................................. 123
7-3 SEM micrographs of cone/pyramid texture of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter. ............................................................................................ 124
7-4 2D/3D AFM topographs and line profile of cone/pyramid of (a) 6 μm and (b) 10 μm base diameter. ....................................................................................... 124
7-5 Schematic of microlens height derived from pyramid of height H. .................... 125
7-6 AFM micrographs and line profile of microlens structure after CMP of the cone/pyramid structure of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter. .. 125
7-7 SEM micrographs of microlens features with base diameter (a) 6 μm, (b) 10 μm and (c) 14 μm. ............................................................................................ 126
7-8 SEM images of fabricated cylindrical structure of 14 μm base diameter at different magnifications, (a) 2700 X, at 45º tilt and (b) 6000 X (top view). ........ 126
8-2 SEM micrograph illustrating the polished Al:ZnO on the microlens pattern of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter. ........................................... 138
8-3 Normalized photoluminescence intensity of Alq3 with Al back reflector for (a) plain microlens patterned glass, (b) as-deposited Al:ZnO and (c) polished Al:ZnO on microlens textured substrate. .......................................................... 139
8-4 Enhancement in light out-coupling in the corrugated device with (a) nactive = 1.7 and (b) nactive = 2.5. The simulated fractional powers for different diameter of the microlens at h/d = 0.5 are represented by (c). ........................................ 140
8-5 Simulated angular distribution of extracted mode for device with (a) nactive = 1.7 and (b) nactive = 2.5. ..................................................................................... 141
15
LIST OF ABBREVIATIONS
AFM Atomic force microscopy
AZO Aluminum doped zinc oxide
CMP Chemical mechanical polishing
ELOG Epitaxially laterally overgrown GaN
GaN Gallium nitride
IEP Isoelectric point
ITO Tin doped indium oxide
LED Light emitting diode/device
LEO Lateral epitaxial overgrowth
MQW Multi-quantum-well
OLED Organic light emitting diode/device
PC Photonic crystals
PL Photoluminescence
PSS Patterned sapphire substrate
RMS Root mean square
SAS Secondary alkyl sulfate
SDS Sodium dodecyl sulfate
SEM Scanning electron microscopy
TC Transparent conductors
TCO Transparent conducting oxide
TDD Threading dislocation
TFEL Thin film electroluminescent device
TX-100 Triton X 100
ZnO Zinc oxide
16
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
CONTROLLED SURFACE TOPOGRAPHY OF TRANSPARENT CONDUCTING OXIDE FOR ENHANCED LIGHT EXTRACTION IN LIGHT EMITTING DEVICES
By
Sushant Gupta
August 2010
Chair: Rajiv K. Singh Major: Materials Science and Engineering
Solid-state lighting (SSL) is an emerging technology with potential to replace the
conventional lighting. Energy efficiencies exceeding that of fluorescent and high-
intensity-discharge (HID) technologies have been demonstrated for SSL. Equipped with
high efficiencies SSL technology has potential to impact energy consumption as artificial
lighting alone amounts to more than $230 billion annually or > 8.9% of total energy
consumption. Internal quantum efficiencies nearly 100% have been reported for both
inorganic and organic LEDs. However, the external quantum efficiencies are still very
low. The major impediment for high external quantum efficiencies is the photon
extraction from various layers in the device.
The focus of this work is to enhance the light extraction efficiency in light emitting
devices. To realize this goal, surface topography of transparent conducting oxide layer
has been controlled on two scales: micro-level texturing and nano-level roughness. The
introduction of micro-scale texture at TCO/active layer interface enhances the light
output by extracting the TCO and active guided modes. By controlling the nano-scale
roughness the detrimental effects of surface roughness, large particle inclusions and
17
pinholes are eliminated. Hence, the first goal of this research is to control the surface
roughness of the TCO thin film to reduce the effect of such anomalies on the properties
of the subsequent active layer and the performance of the LED device. In order to
achieve this, a chemical mechanical polishing (CMP) process was developed for
controlling the surface roughness of zinc oxide (ZnO). ZnO was used as the TCO
material as it has been extensively researched as a replacement material for
commercially used indium tin oxide. CMP process parameters along with slurry
chemistry were systematically varied to develop a CMP process for ZnO. Based on the
findings, a CMP mechanism for ZnO polishing has been proposed. Effects of slurry
additives such as surfactants and salts were also studied to further improve the surface
quality and CMP performance. Highly smooth zinc oxide surface (with RMS roughness
between 3-6 Å) was obtained using developed CMP process for zinc oxide thin films.
The second part of this work is focused on the enhancement of extraction
efficiency by controlling the micro-topography. This involves the study of periodic micro-
scale textures (microlens, micropyramid and cylinder) for light extraction applied at
different interfaces. Ray-tracing simulations were performed to understand the ray
dynamics of these features. Two device constructions were simulated with texture at:
(A) glass/air interface and (B) glass/TCO interface (corrugated device). The impact of
feature parameters (diameter and h/d ratio or contact angle) on luminous intensity and
angular distribution of extracted photons was studied. It was found that the introduction
of these textures results in increase in the escape probability of the photon due to
change in their directionality after multiple reflections. The enhancement in light
extraction is obtained from the extraction of higher angle photons in most cases. Based
18
on the ray-tracing simulations, microlens, micropyramid and microcylinder arrays were
fabricated on the glass substrate. Photoluminescence measurements were performed
using evaporated Alq3 (as photoluminescent material) and Al (as back reflector) layers
over the substrates with controlled topography of ZnO TCO layer in corrugated device
configuration. The textured TCO device showed more than 2X improvement as
compared to control device with no structure which matches closely with the simulation
results. This work demonstrated that by controlling the topography of the TCO layer
more than 1.5 fold enhancement in extraction efficiency can be achieved. This
enhancement can be further increased by texturing the glass/air interface to reduce
substrate mode.
19
CHAPTER 1 INTRODUCTION
Transparent Conducting Oxides
Transparent conductors are the materials displaying the remarkable combination
of high electrical conductivity and optical transparency. These materials already form
the basis of many important technological applications. Transparent conducting oxides
represent an important class of materials which comprises of doped metal oxides
having both optical transparency and electric conductivity. The rapid development of
transparent conducting oxides (TCOs) has lead to a tremendous advancement and
commercialization of active and passive electronic and optoelectronic devices ranging
from transparent heating elements for windows to transparent thin film circuitry. Some of
the most common optoelectronic applications are: flat panel displays [1, 2], organic [3]
as well as inorganic [4] light emitting diodes (LEDs) , photovoltaics [5, 6], and
architectural window applications [7], low-e windows [8, 9], electro chromic devices [10],
and anti-static coatings [2, 3, 6, 9, 11-13] as shown in Figure 1-1.
Importance of TCOs
In solid-state physics, the free electron model is a simple model describing the
behavior of valence electrons in a crystal structure of a metallic solid [14]. The electron-
gas model of solids applies equally to the electrons in the conduction band of both
metals and semiconductors. As per the Boltzmann formulation, the electrical
conductivity of such a material is directly proportional to the density of free carriers (n)
and their mobility (μ). The mobility is in turn directly proportional to the free carrier
resistivity relaxation time (the time between resistive scattering events), and is inversely
proportional to the carrier effective mass. Hence, it has been demonstrated that the
20
transparent conducting oxides exhibit intermediate characteristics between the highly
conducting and reflecting elemental metals and the low carrier density, high-mobility
prototypical semiconducting materials such as Si as shown in Figure 1-2 [15].
Transparent conducting oxides belong to an important class of materials characterized
by relatively high values of both the electron density and the electron mobility sufficient
for them to exhibit substantial electrical conductivity as transparent solids especially in
form of thin films which dominate the bulk of application areas of such materials.
Hence, TCOs are conjugate property materials in which conductivity is strongly
coupled to the lossy part of the refractive index. As explained materials like metals, that
are highly conductive do not normally transmit visible light, while highly transparent
media like oxide glasses are electrically insulating in nature. Partial transparency and
good conductivity can be obtained in very thin films of metals such as aluminum or
silver but have limited scope [16, 17]. The fundamental properties governing the
structure/property relationships driving these properties of conductivity and
transparency need to be understood in order to achieve these properties. The only way
to obtain transparency along with good conductivity is to intentionally create electron
degeneracy in a wide band gap (> 3 eV) oxide by controllably introducing non-
stoichiometry or by using doping. This is achieved in oxides of various transition metals
such as tin, indium, zinc, gallium, cadmium and their multi-component alloys [2, 11].
History of TCO
TCOs have been around for a long time, the first report of a TCO film occurred
material occurred in 1907 when a sputtered thin film of cadmium metal underwent
incomplete thermal oxidation after heating in air [18]. In this case, not only the electrical
conductivity of the thin film changed with time, also the cadmium oxide was
21
representative of a free carrier like conductor in which the oxygen deficiency donate free
carriers to the defect (trap) energy levels near the conduction band of CdO. Since this
early discovery, other reports followed around 50 years ago in 1951 and 1952 [19, 20]
establishing the coexisting of optical transparency and electrical conductivity. Similar
reports followed suite for Indium oxide [21]. Since these early reports, the technological
interest in transparent conductors has grown tremendously. Newfound applications of
these films for heated windows and to diminish heat losses from sodium lamps fueled
the development of TCOs which led to concentrated research effort leading to
development of doped films of SnO2 and In2O3:Sn (indium tin oxide or ITO) with
excellent electrical and optical properties [22, 23].
Zinc Oxide: A Promising TCO
In last decade, tin doped indium oxide (ITO) has become commercially used
material for TCO application. However, due to limited nature of world indium reserves
and ubiquitous use of solid state devices in displays, lighting and solar cells; a shortage
of indium may occur in near future. In fact, a factor of ten increase in indium pricing has
been observed in recent years [5, 11]. Extensive research has been expanded to find
an alternative for ITO. For practical use as a transparent electrode, a TCO material
should exhibit a resistivity of the order of 10−3 Ω cm or less and an average
transmittance above 80% in the visible range [2, 12]. A list of researched oxides and
associated dopants with their qualitative resistivity and toxicity are listed in Table 1-1.
In addition to the low resistivity on the order of 10−4 Ω cm and transmittance
requirement, fabrication at a temperature below 200 °C along with the feasibility of
obtaining low thicknesses of approximately 15 to 100 nm are required for most of the
applications of TCO materials. Therefore, it is difficult to use practically cadmium oxide-
22
based and titanium oxide-based TCO materials because of the toxicity of cadmium and
the required high temperature heat-treatment for titania oxide-based TCOs [11].
Particularly, in titanium oxide and titanium oxide-based TCO thin films, a deposition and
heat-treatment at a high temperature above about 300 °C and/or an epitaxial growth on
a single crystal substrate are necessary in order to obtain a low resistivity [24-26]. The
impurity-doped ZnO, In2O3 and SnO2 films have been prepared on glass substrates
using various deposition methods. However, the obtained minimum resistivities of
impurity-doped SnO2 and In2O3 films have essentially remained unchanged for more
than the past twenty years whereas those of impurity-doped ZnO films are still
decreasing [12]. As a result doped zinc oxide is identified as the only promising and
practical alternative for ITO because of inexpensive and non-toxic nature of zinc oxide
[12, 27-31]. The electrical and optical properties of ZnO comparable to the ITO have
been reported in past [32-35]. However, electrical resistivity was unstable in intrinsic
ZnO at higher temperatures due to annihilation of oxygen vacancies responsible for
conductivity in intrinsic oxide [35]. Since then numerous reports on multi-component
TCs based on ZnO [36-38] or on intentionally doped ZnO [25, 39, 40] have been
published. The properties of ZnO that makes it suitable for TCO application are
delineated in next section.
Properties of ZnO
Among the vastly researched material space, zinc oxide has caught lot of attention
because of its large band gap (3.37 eV), large exciton binding energy on the order of 60
meV and high electron affinity (4.35 eV) [3, 41-43]. ZnO normally grows in the
hexagonal (wurtzite) crystal structure with a = 3.25 Å and c = 5.12 Å. ZnO exhibits a
strong room temperature, near-band-edge UV photoluminescence peak at ~3.2 eV [40].
23
The well-overlapped wave function of the conduction band of ZnO results in small
electron effective mass between 0.23–0.35me, where me is the mass of a free electron
hence high electron hall mobility of 200 cm2/Vs [40, 44]. Basic material properties of
ZnO have been tabulated in Table 1-2. Although pure zinc oxide is an insulator but
slightly reduced form of zinc oxide has oxygen vacancies which give rise to free electron
associated with metal ion. The intrinsic defect levels lie approximately 0.01–0.05 eV
below the conduction band [45]. In addition, intentional impurities such as gallium,
aluminum, indium, boron, fluorine etc have been added to obtain stable conductivity [3,
12].
ZnO Applications
In addition to ZnO used as a TCO material, it is extensively researched for various
other applications. In recent years, zinc oxide has become an important material in the
field of optoelectronics for UV and blue light emitting devices and lasers [40, 45-47],
surface acoustic wave devices [48, 49], piezoelectric transducers [50] and sensors [51,
52]. Zinc oxide exhibits a large band gap (3.37 eV) and a high exciton binding energy
(~60 meV) [42] making it suitable for UV and blue light emitting devices and lasers. The
lack of a native substrate for GaN growth has led to a search for suitable choices of
substrate. Wurtzite type single crystal zinc oxide substrates have been demonstrated as
a good template for the epitaxial growth of GaN thin films. Apart from ZnO single
crystals, zinc oxide thin films have been applied as a buffer layer for growth of GaN on
sapphire and SiC substrates [48].
TCO Roughness
Zinc oxide single crystal has been used as a substrate for epitaxial growth of GaN
thin films [53-55]. Zinc oxide thin films have been investigated as a buffer layer for
24
growth of GaN on sapphire and SiC substrates [48, 56-58]. In addition, doped-ZnO thin
films are being extensively researched for their use as transparent conducting
electrodes [40]. However, both commercially available zinc oxide single crystals and as-
grown zinc oxide thin films have a rather high surface roughness, large particle
inclusions and pinholes which hampers the growth of subsequent high quality layers
[59-62]. Such defects influence the growth of subsequent layers when zinc oxide thin
films are used for buffer layer applications or influence the interface between zinc oxide
layer and the active layer in light emitting diodes [47, 52]. In organic light emitting
diodes, these anomalies lead to color degradation and localized electrical shorts,
undermining the performance and lifetime of these devices. Further, in thin film silicon
solar cells, high roughness of the underlying ZnO layer leads to poor growth of the
subsequent silicon layers. In microcrystalline Si (µc-Si:H) solar cells, surface roughness,
orientation and grain size of the ZnO film affects the grain size and quality of the µc-Si:H
layer [62-64]. In addition to this, increased interfacial area between rough transparent
conducting oxide (TCO) and the p-layer due to TCO roughness correlates qualitatively
to the observed decrease in open circuit voltage (VOC) and fill factor for the device.
Vanecek et al., have reported that the absorption between 500 – 1100 nm wavelength
increases with increase in RMS roughness at ZnO/back reflector interface [65]. To
address these issues due to ZnO roughness, we developed a chemical mechanical
polishing process for smoothening of the zinc oxide surface.
There are very few reports published on the zinc oxide polishing. Lucca et al. first
reported the polishing of ZnO surface. In their work, the photoluminescence (PL)
response of mechanically polished and chemo-mechanically polished ZnO single crystal
25
surface was compared [66]. They showed that PL response of chemo-mechanically
polished ZnO single crystal (either O-face or Zn-face) was two orders of magnitude
higher than the mechanically polished ZnO, due to better surface roughness and
reduced sub-surface damage. However, detailed study of the chemo-mechanical
polishing process was not carried out. Recently, Lee et al. published a report on the
effect of various chemical mechanical polishing (CMP) process parameters on
morphology & optical properties of the sputtered zinc oxide thin films [59]. However, the
report does not provide a comprehensive study of the process parameters such as
slurry pH, polishing pressure and particle size and concentration. Some of the other
reported techniques for controlling surface roughness of the zinc oxide thin films involve
irradiation by gas cluster ion beam for smoothening of the surface topography [61];
controlling growth rate, thickness and doping concentration in chemical vapor deposition
grown ZnO to tune the roughness [67] and bi-layer structure of ZnO involving deposition
of second smooth ZnO layer using atomic layer deposition [68]. However, using ion
beam/electron cyclotron resonance etching to sputter material disrupts the crystal
structure of the underlying zinc oxide film [61, 69]. A CMP process however provides a
good control over the surface roughness with achievable roughnesses down to
angstrom level without disrupting the crystal structure of the ZnO film. Hence, this work
focuses on development and understanding of a CMP process for polycrystalline ZnO
thin film.
Light Extraction
The luminous intensity of any light emitting device is dependent on two factors:
internal quantum efficiency (IQE) and light coupling. The IQE is dependent on the active
material quality and properties, the charge transport and recombination in the device
26
and device fabrication. Tremendous progress has been made in improving the
properties and the quality of active semiconductor materials. Internal quantum efficiency
as high as 100% in both LED and OLED devices has been reported [70-72]. Despite of
the high IQE, the overall efficiency of the device is largely limited due to poor light out-
coupling/extraction efficiency. The primary reason for reduced light extraction is the
large disparity between the refractive indices of active layer and surrounding media.
This difference in refractive indices causes total internal reflection of majority of the
generated photons, as explained below.
Basic Concepts
According to Snell’s law, when light travels from one medium to another of
different optical densities, it changes directionality. The bending is given by,
1
2
2
1
n
n
Sin
Sin
(1-1)
where, 1 is the angle of incidence, 2 is the angle of refraction, n1 and n2 are the
refractive indices of medium 1 and 2, respectively (see Figure 1-3). When light is
traveling from denser medium to lighter medium at angle greater than what is known as
critical angle of incidence (C), the sine value of angle of refraction is greater than unity.
For angles greater than critical angle, Snell’s law cannot be used to calculate angle of
refracted light. All rays having angle of incidence greater than C are reflected back in
the medium or suffer total internal reflection. The critical angle, C is given by,
1
21
n
nSinC (1-2)
When C is viewed in three-dimensions, it results in a cone as shown in Figure 1-4,
often termed as escape cone. The emitted photons that have angle of incidence inside
27
the escape cone are extracted out of the surface whereas the rest are reflected back in
to the layer. In parallelepiped geometry since, the angle of incidence remains same, the
reflected fraction of emitted photons are wave-guided inside the layer until they are
finally absorbed (Figure 1-5).
Light Extraction Efficiency
A simple expression can be derived for the maximum emission efficiency per
escape cone (ηcoupling) based on ray optics and Snell’s law for the isotropic emission
from the active layer:
2
112
1cos1
2
1
s
accoupling n
n (1-3)
where na and ns are refractive indices of ambient and semiconductor.
Often in OLEDs, the semiconductor layer stack is reflective or cathode is
reflective. This provides the photons another chance to be out-coupled after reflection
from the back side towards the substrate/air interface. Thus reflection opens another
escape cone for photons to escape and ηcoupling can be calculated as [73],
2
2
2
111cos1
orgorg
accoupling
nnn
(1-4)
For the typical LED or OLED device using GaN and N,N′-bis-(1-naphthyl)-N,N′-
diphenyl-1,1′-biphenyl-4,4′-diamine (NPB) with index of refraction 2.4 and 1.7, the
critical angle is θc ~ 24º and 36º and the extraction efficiency would be as low as ∼ 4%
and 10% for a simple semi-infinite solid geometry without back reflector. Thus, 95 to
90% of generated light is being trapped inside the layer.
28
Improving Light Out-coupling
Numerous methods of improving ηcoupling have been reported, including the use of
device shaping [74-76], photon recycling [70, 71] and photon scattering [73, 77-87]
techniques. These techniques have been discussed briefly in this chapter and a detailed
account has been provided in chapter 2.
Device shaping
Chip-shaping for improving light extraction efficiency has been known since
1960’s. Carr and Pittman from Texas instrument first reported the hemispherical device
with 0.1 watt power output as compared to previous best of few milliwatts for a GaAs
based device [76]. Since then several other chip shapes such as a truncated cone
shape [75, 88], Weierstrass sphere [88], truncated ellipsoid [88], paraboloid [88],
truncated inverted pyramid [74], truncated hexagonal pyramid [89], rhombohedral [90],
triangular [90] and slanting wall [91] have been proposed and fabricated. Device
shaping presents numerous limitations in terms of processability and cost and has
limited applications. Transforming the device shape using packaging has been a
successful and a commercially used technique. Optical grade epoxy is used for
fabricating these hemispherical-shaped packaging. However, use of millimeter-size
packaging is not practical for various large scale, integrated devices such LED displays.
In addition, this method presents index matching issues limiting its applicability.
Photon recycling
Another way to change the directionality of the reflected photon is photon re-
absorption and re-emission [92]. The photon recycling is dependent on IQE of the
device. Using photon recycling, an IQE of ~100% and remarkable external quantum
efficiency of 72% have demonstrated [71]. Because IQE is dependent on the quality of
29
the semiconductor material, the light out-coupling hence the external quantum efficiency
(EQE) is sensitive to the parasitic losses. High EQE using photon recycling demands
high quality of material to reduce the parasitic loss mechanisms or makes it susceptible
to slight decrease in IQE. Practically, IQE is usually less than 100% and decreases with
aging. Particularly in OLEDs, active layer is considered usually absorptive (non-radiative
absorption) and hence very thin films (< 150 nm) are often used to minimize absorption
losses. Therefore, this technique is not widely applied for light extraction from
electroluminescent devices.
Photon scattering
Different methods have been adopted to scatter photons at different interfaces to
change the directionality of the photons thereby, increasing the probability of photons to
out-couple. Some of the interface modification techniques include simple surface
roughening techniques involving: natural lithography, self-assembled microbeads,
photoelectrochemical (PEC) etching, graded refractive index layers, anodic aluminum
oxide template [79, 92-96]; photonic crystals [97-103] ; patterned substrates [104-108],
metallic nanowire array [84] and even incorporation of low index grid between
ITO/organic interface [83] and 2D SiO2 nanorods between ITO/glass interface [109].
However, these methods are often accompanied by changes in the radiation pattern,
exhibit an undesirable angle dependent emission spectrum, or employ costly or
complex processing methods [85]. Application of external out-couplers provide
significant enhancement in light extraction without invoking such undesirable attributes.
External out-couplers such as hemispherical dome [76], pyramids [81, 110], cylinders
and lenses [73, 85, 111-113] have as been considered for improving the light
extraction.
30
Summary
In this chapter, a brief introduction to transparent conductors or transparent
conducting oxides, their history and various TCO materials was presented. ITO, a
commercially used TCO has Indium as its main constituent which is an expensive
material due to low availability of Indium. Zinc oxide on other hand is inexpensive and
the combination properties of ZnO, makes it a suitable candidate for replacement. Basic
properties of ZnO were presented in this chapter followed by other applications of ZnO.
It was elucidated in this chapter that the surface roughness of ZnO whether used for
TCO application or other applications can be detrimental to device performance and
reliability. Other then issue of TCO roughness, optoelectronic devices such as LED,
OLED or TFEL devices suffer from poor light out-coupling as low as < 20% despite of
nearly 100% internal efficiencies reported for both LEDs and OLEDs. Some of the basic
concepts of light trapping, methods and techniques for extract wave guided photons
were discussed in this chapter. A detailed literature review on TCO roughness issue
and adopted techniques and light extraction methods and techniques developed so far
are described in chapter 2.
31
Figure 1-1. Applications of transparent conducting oxides.
32
Figure 1-2. Materials space for semimetals, good metals, transparent conductors and semiconductors based on n–µ correlations (electron carrier density and electron mobility, respectively). Data shown are for room-temperature measurements. Constant conductivity contours are shown as straight lines ([15] - Reproduced by permission of The Royal Society of Chemistry).
33
θ1
θ2
n1
n2
θc
Total Internal
Reflection
θ1
θ2
n1
n2
θc
Total Internal
Reflection
Figure 1-3. Schematic depicting Snell's law and total internal reflection when light travels from materials with different refractive index (n1 > n2).
n2
n1θc
n2
n1θc
Figure 1-4. Illustration of escape cone defined by the critical angle when light enters from optically denser medium (n1) to lighter medium (n2).
n2
n1θ1> θc
θc
n2
n1θ1> θc
θc
Figure 1-5. Illustration depicting waveguiding of rays with emission angle greater than
critical angle.
34
Table 1-1. Host and associated dopants in typical TCO systems (adapted from ref. [11])
Binary Dopant Resistivity Toxicity
ZnO Al, Ga, B, In, Y, Sc, V, Si, Ge, Ti, Zr, Hf, F ⊚
CdO In, Sn ⊚ ××
In2O3 Sn, Ge, Mo, Ti, Zr, Hf, Nb, Ta, W, Ta ⊚ x
Ga2O3 Sn Δ
SnO2 Sb, As, Nb, Ta ○
TiO2 Nb, Ta Δ
Ternary
MgIn2O4 Δ
GaInO3, (Ga, In)2O3 Sn, Ge Δ
CdSb2O6 Y Δ ×
SrTiO3 Nb, La ×
Ternary Multi-component Resistivity Toxicity
Zn2In2O5, Zn3In2O6 ZnO–In2O3 system ○
In4Sn3O12 In2O3–SnO2 system ○
CdIn2O4 CdO–In2O3 system ○ ×
Cd2SnO4, CdSnO3 CdO–SnO2 system ○ ×
Zn2SnO4, ZnSnO3 ZnO–SnO2 system Δ
ZnO–In2O3–SnO2 system ○
CdO–In2O3–SnO2 system ○ ×
ZnO–CdO–In2O3–SnO2 system ○ ×
⊚: Very good, ○: Good, Δ: Average, ×: Bad, ××: Very bad.
35
Table 1-2. Properties of ZnO
Property Value
Lattice parameters, 300 K
a0 3.2495 Å
c0 5.2069 Å
a0/c0 1.602
Density 5.606 g/cm3
Crystal structure Wurtzite
Melting point 1975ºC
Thermal conductivity 0.6,1-1.2
Refractive index 2.008,2.029
Energy gap 3.4 eV, direct
Exciton binding energy 60 meV
36
CHAPTER 2 LITERATURE REVIEW
TCO Surface Roughness
Many techniques have been adopted to synthesize ZnO thin films mainly,
techniques such as sputtering [29, 50, 114], pulsed laser deposition [40, 47], low
pressure chemical vapor deposition [62, 115, 116] or spray pyrolysis [117-119].
Irrespective of the deposition method, deposited thin films have high surface roughness,
large particle inclusions and pinholes [60-62, 120-122]. Such defects influence the
growth of subsequent layers when TCO thin films are used in various optoelectronic
applications. Liu et al. studied the formation of dark spots due to surface roughness and
spikes in the sputtered ITO layer in an OLED device with
ITO/PEDOT:PSS/polyfluorene/Al stack [123]. The spikes in the ITO film result in local
melting of the polymer and migration of indium according to electric field to eventually
depositing on cathode surface. Indium has higher work function and continued indium
migration lead to increased heat generation and growth of dark spot in size. They
compared the failure performance of smoothened ITO surface by etching using
saturated KOH/isopropanol solution. The device fabricated on as-deposited rough ITO
shows formation of dark spot within 15 min of operation at 4 V in nitrogen glove box with
50% reduction in light output whereas chemically etched relatively smooth ITO device
gave very stable performance with no dark spot formation after 10 hours of operation
under same operating conditions. Figure 2-1 represents the SEM image of the Al
cathode surface, illustrating dark spot formation. Jung et al. reported mechanical
polishing of ITO using alumina powder of 0.05 μm organic light-emitting diodes (OLEDs)
fabricated on the polished ITO surface showed 10 times better electroluminescence and
37
current injection than the OLEDs based on as-received ITO surface [121]. The
nucleation of N,N’- diphenyl-N,N’-bis(3-methyl-phenyl)-1,1’-biphenyl-4,4’-diamine (TPD)
on ITO was found to be dependent on the features on the ITO surface. They attributed
the increased EL and current injection to reduced surface and change in surface O in
the ITO layer. Choi et al. also showed improved optical and I-V characteristics of
polished ITO layer using chemical mechanical polishing process and the MEH-
PPV/polished ITO structure [124, 125]. These reports reveal that surface roughness of
TCO can adversely affect the device performance and lifetime.
Besides application of ZnO as a TCO material, it is also known to be an effective
substrate for GaN growth [53-55] or buffer layer for GaN growth on Si, sapphire and SiC
substrates [48, 56-58, 122]. The ZnO surface morphology affects the GaN film growth
on Si substrates as studied by Xue and group [122]. It was reported that increase in
annealing temperature of ZnO buffer layer leads to reduction in film defects resulting in
smoother film up to 900ºC, beyond which dissociation of ZnO in to Zn and O2 occurs
and film quality deteriorates. The growth of GaN on rougher ZnO results in early
coalescence of GaN islands hence, increased defect density in so-deposited GaN film
evident from their SEM, AFM images and observed increased PL intensity for GaN film
grown on annealed ZnO film at 900ºC. In addition, ZnO films also find application in
optoelectronic integrated circuits as optical waveguides where surface undulations close
to dimension of wavelength lead to scattering and propagation losses [61, 120, 126].
These reports elucidate the need for polishing of zinc oxide whether ZnO is used as
TCO in organic LED or as substrate or buffer layer in GaN based devices.
38
Several techniques have been reported to for reducing the surface roughness,
large particle inclusions and pin holes in ZnO films or ZnO single crystals. Some of the
reported techniques for controlling surface roughness of the zinc oxide thin films involve
irradiation by gas cluster ion beam for smoothening of the surface topography followed
by etching of damaged layer in 0.1% HCl solution [61], controlling growth rate, thickness
and doping concentration in chemical vapor deposition grown ZnO to tune the
roughness [67], bi-layer structure of ZnO involving deposition of second smooth ZnO
layer using atomic layer deposition [68] and annealing of ZnO film at different
temperatures [122]. However, using ion beam/ electron cyclotron resonance etching to
sputter material, disrupts the crystal structure of the underlying zinc oxide film [61, 69].
Furthermore, these techniques do not provide a good control over the surface
roughness. There are very few reports published on the zinc oxide polishing. To our
knowledge, Lucca et al. first reported the polishing of ZnO surface. In their work, the
photoluminescence (PL) response of mechanically polished and chemo-mechanically
polished ZnO single crystal surface was compared [66]. They showed that PL response
of chemo-mechanically polished ZnO single crystal (either O-face or Zn-face) was two
orders of magnitude higher than the mechanically polished ZnO, due to better surface
roughness and reduced sub-surface damage. However, detailed study of the chemo-
mechanical polishing process was not carried out. Recently, Lee et al. published a
paper on the effect of various chemical mechanical polishing (CMP) process
parameters on morphology and optical properties of the sputtered zinc oxide films [59].
Light Extraction
It is well known that the large difference in refractive index between semiconductor
and air limits the escape of generated photons due to the emitting angle of large fraction
39
of these photons are larger than the critical angle (θc) defined by eq. 1-2, which get
trapped or confined by total internally reflection. For isotropic photon emission from an
active layer taking into account the internal reflection, the overall extraction efficiency (η)
is given by eq. 1-3. For the typical LED or OLED device using GaN and N,N′-bis-(1-
naphthyl)-N,N′-diphenyl-1,1′-biphenyl-4,4′-diamine (NPB) with index of refraction 2.4
and 1.7, the critical angle is θc ~ 24º and 36º and the extraction efficiency would be as
low as ∼ 4% and 10% for a simple semi-infinite solid geometry. Clearly, light extraction
efficiencies are very low and to achieve extraction efficiencies of 70% or greater for a
cost effective general purpose lighting source, simple parallelepiped geometries will not
work [127].
Much success has been achieved in improving light extraction through tailoring the
overall shape of the chip, surface texturing of the devices and other photon
manipulation techniques to enhance the light extraction. Figure 2-2 illustrates the
evolution of various light extractions methodologies in inorganic LEDs. Some of different
photon extraction techniques adopted for inorganic and organic LEDs have been
illustrated in this chapter.
Device Shaping
Chip-shaping for improving light extraction efficiency has been known since
1960’s. Carr and Pittman from Texas instrument first reported the hemispherical device
with 0.1 watt power output as compared to previous best of few milliwatts for a GaAs
based device [76]. The authors attributed the increase in output primarily to the
reduction internal reflection due to hemispherical design of the device. Franklin et al.
published a report on funnel shaped chip in a short time gap from Carr group [75]. They
40
proposed that theoretically using a truncated cone shape of cone angle 45º a device
efficiency of 18% can be achieved. Although shaping methods used were not very
efficient and practical. Other shapes such as Weierstrass sphere, truncated ellipsoid,
truncated cone and paraboloid have also been studied [88]. In 1999, Krames and group
reported a practical method for shaping of die by saw dicing using beveled saw blade
[74]. The truncated inverted pyramid (TIP) structure reduces the mean free path of the
generated photon within the device reducing internal losses (figure 2-3). Further, using
TIP geometry with 35º vertex angle and multiwell structure they demonstrated a record
peak efficiency of exceeding 100 lm/W at 100 mA dc for orange-emitting (610 nm)
device and 68 lm/W for 598 nm emitting device. Thompson et al. reported the truncated
hexagonal pyramid shaped LED based on ZnO and GaN wafer bonding for the purpose
of improving LED efficiency by fabricating highly transparent and shaped p-type
electrode [89, 128]. This ZnO wafer bonded LED provide twofold enhancement, (1) a
light extraction efficiency advantage over conventional LEDs because of the reduction
of internal light reflections and (2) reduction in internal loss mechanism because of low
absorption p-type electrode and reduction of mean photon path due to truncated
pyramid geometry. The external quantum efficiency (EQE) was 2.2 times higher at
20 mA of current injection over conventional device. There have been other geometries
and shaping strategies such as rhombohedral [90], triangular [90] chip and slanting
walls [91] which have shown enhancement in light extraction over regular parallelepiped
chip structure. However, shaping is costly and complex procedure for light extraction
and other photon randomization or redirection techniques have been demonstrated in
literature which are equally or more efficient and are easily realized experimentally.
41
Interface Modification Techniques
Surface roughening
Chaotic photon trajectories assures the maximum statistically feasible output
coupling from the LED through the escape cone [92]. Angular randomization by
photon recycling is not practical since it is vulnerable to the slightest deterioration
in material or back-reflector quality. A more practical approach is interface modification
by texturing of surface. Schnitzer et al. took advantage of random surface texturing
using natural lithography to cause chaotic photon trajectories [92]. Using self-assembled
200 nm polystyrene beads as etch mask and etching the AlGaAs surface with Cl2
assisted Xe+ ion beam, a 30% external efficiency was demonstrated. Huh et al. micro-
roughened the semiconductor surface in GaN based LED [79]. The average roughness
obtained using Pt metal cluster mask and wet etching of GaN surface was 5-6 nm. The
I-V curve of LED device showed reduced series resistance attributed to higher contact
area between GaN and light transmitting Pt-layer. The power efficiency of micro-
roughened device was 62% higher than conventional LED owing to lowered forward
voltage and enhanced light extraction. Fujii et al used hexagonal cone-shaped
roughness etched using photoelectrochemical (PEC) etching of N-face n-GaN in KOH
and in presence of Xe light source [96]. Figure 2-4 represents the SEM micrograph of
GaN surface after a 10 min etch process creating ~500 nm high hexagonal cones. The
roughened N-face n-GaN resulted in improved output power by a factor of 2.3 as
compared to non-textured device. Graded refractive index (GRIN) layers have also
been applied at air/semiconductor interface to extraction light due to their anti-reflective
properties [129]. In a GRIN layer, the refractive index is gradually varied between the
refractive index of active semiconductor and refractive index of air. Kim et al. deposited
42
ITO GRIN layer using oblique angle deposition technique on GaN surface in GaInN LED
(Figure 2-5) [77, 130]. A 24.3% enhancement in light output was observed for GRIN ITO
antireflective layer as compared to device with dense ITO layer. The mechanism for
improved light extraction is attributed to the reduction in Fresnel reflection at ITO/air
interface. Tsai et al. grew naturally textured GaN device using a growth-interruption step
and a surface treatment using biscyclopentadienyl magnesium (CP2Mg) simultaneously
to form a plurality of nuclei sites on the surface of a p-type cladding layer [131]. A rough
surface having truncated pyramids was formed after further growth of p-type contact
layer. A light output power enhancement of 66% was observed at 20 mA drive current.
Lee et al. used roughened GaN layer with ITO/Al2O3/Al stack as back reflector to extract
1.6 fold higher light as compared to non-roughened LED device [93]. They used
adhesive layer bonding technique followed by wet etching to create triangle-like
roughness. Recently, Cheng et al. created chaotic mesh-like surface roughening to
improve light extraction [95]. Using porous anodic aluminum oxide (AAO) template, they
created PDMS meshed surface which resulted in 46% enhancement in light out-
coupling from the OLED device (Figure 2-6). Kang et al. nanotextured GaN surface
using etching and ITO nanospheres [94]. They roughened the p-GaN and n-GaN
surface by an interesting wet etching technique to create ITO nanosphere by taking
advantage of faster etch rates at ITO grain boundaries and subsequent short duration
dry etch of GaN. As compared to non-textured LED, p-GaN, n-GaN and both p-GaN
and n-GaN roughened device showed 16, 12 and 33 % higher light output, respectively
without detrimental effect of dry etch process on GaN electrical properties.
43
Photonic crystals
There has been remarkable progress in photonic crystal recently. The photonic
crystals (PC) are periodic variations in the refractive index of a dielectric material in one
or more dimensions. Such variation of refractive index at wavelength of light scale give
rise to forbidden photon energy bands as well as dramatic changes to the photon
density of states [127]. Similar to electronic band gap in case of electrons, the photon
energy band allows manipulation of the photons in fundamentally new ways, including
enhancing the radiative recombination rate. It can be utilized to inhibit the emission of
the guided modes or redirect trapped light into radiated modes. In LEDs, photonic
crystals can promote increase in the radiance by increasing the directionality of the
extracted light [132]. For example, Fan and group for the first time elucidated by
simulating 2-dimensional GaAs PC slab (n = 3.5 and λ = 1.55 µm) that the extraction
efficiencies of ~80% can achieved over broader range of frequencies unlike 1-
dimensional resonant microcavities [86, 87]. They analyzed two structures one with the
dielectric slab with triangular lattice of infinite length air holes and other one with
dielectric pillars of infinite and finite length. The periodicity of these structures opened
up the gap and set the upper frequency limit. The modes between the gap are
scattering out that would have otherwise been trapped leading to average enhancement
up to ~70% and 80% in case of air holes and dielectric pillar structure, respectively.
Several reports have been published ever since on use of PCs to extract light from
various inorganic LED, OLED and thin film electroluminescent devices (TFEL) [97, 99,
100, 102, 103, 133-136]. Han-Youl et al. described both the simulation and
experimental results on effect of air-hole depth on light extraction [101]. Their findings
suggest that the light extraction decreases as the depth of air-holes decreases
44
indicating that it is desirable to transfer the PC pattern through the slab. The gain in light
extraction as large as 8-fold and 13-fold was observed for photonic crystal slab with
SiO2 and air-cladding, respectively. Also, square lattice showed better light extraction as
compared to triangular lattice due to greater density of leaky modes. Wierer et al.
demonstrated triangular lattice PC on n-GaN layer in a InGaN based MQW LED
structure [132]. A schematic of PXLED is depicted in Figure 2-7 with optical image of lit
device and angular distribution curves. Four different photonic crystal lattices were
investigated with lattice constants and hole diameters as 270, 295, 315, and 340 nm,
and 200, 220, 235, and 250 nm, respectively. The holes of 100 nm depth with 75º
sidewall angle were created using e-beam lithography and dry etching. The photonic
crystal LED (PXLEDs) displayed heavily modified far-field radiation patterns with
increased radiance up to 1.5 times. The increased out-coupling was correlated to Bragg
scattering of the guided modes out the top of the LED in the report. Lee et al. applied
the PC in an OLED device by creating 200 nm SiO2 pillars on the glass substrate
(Figure 2-8) [136]. The optimized PC pattern with 200 nm SiO2 rods embedded in SiNx
and lattice constant of 600 nm exhibited a simulated light extraction efficiency of 80%
which translated to experimentally observed 50% enhancement in light extraction over
viewing angle of 90±40º. The integrated light efficiency with triangular lattice PC
Y2O3:Eu3+ thin film device was found to be approximately 4.0 and 4.7 times higher for
the nanorods and airholes as reported by Oh et al [135]. In case of a square lattice they
integrated PL intensity 3.0 and 2.8 times that of non-PC TFEL device. They attributed
the observed enhancement in extraction efficiency to the scattering of the forward
emission excited directly by inward UV and the scattering of re-excited forward emission
45
by backscattered UV (from mercury excitation source) through the leaky and/or Bragg
scattering produced by the 2D periodic array.
Patterned substrates
Tadatomo et al. first time reported the use of patterned sapphire substrates for
growth of InGaN/GaN multi-quantum-well (MQW) UV LEDs (see Figure 2-9) [137]. They
fabricated parallel grooves with dimensions of ridges, grooves and depth of 3 μm, 3 μm
and 1.5 μm, respectively along the sapphire
0211 direction which resulted in a 24%
external quantum efficiency at emission peak of 382 nm and 20 mA operating current.
The results were approximately 5 folds improvement over the conventional LED which
they attributed to the reduction in defect density to 1.5x108 cm-2. Chang et al. claimed to
have 35% increase in the electroluminescent (EL) intensity without the shift in EL
emission peak with stripes patterned along sapphire
0011 direction [138]. Additionally,
better reliability of InGaN/GaN MQW LED on patterned sapphire substrate (PSS) was
found with maximum output at higher power and lower decay in EL intensity after 72
hours burn-in test. Hsu and group also reported similar results with 25% increase in
output power correlating to reduced threading dislocation induced non-radiative
recombination [139]. All these reports indicate that the enhancement of optical output
power is attributed to the effective suppression of leakage current using the patterned
sapphire substrates owing to reduced dislocation defects. Lee et al. argued that besides
the elimination of threading dislocations due to the lateral growth of GaN on top of stripe
patterned sapphire substrate, enhancement in light extraction could be due to extraction
of guided mode because of stripe pattern [104]. They observed an improvement of light
output efficiency of their PSS LED to be about 20% which is higher than their simulation
46
results (13%). Improvement in efficiency of their device over conventional LED was
mainly attributed to the pattern on sapphire substrate whereas the additional
improvement in light output efficiency as compared to their simulation results was
suggested to be related to the elimination of threading dislocations with better epitaxial
quality. Yamada et al. used dual approach in order to enhance the external quantum
efficiency [105]. One is to use the PSS for effective scattering of the emission light from
the active layer. Second is to use Rh as the higher reflectivity electrode. Additionally, in
order to reduce the absorption of the emission light by the electrode, they patterned the
electrode in mesh shape. Using there approach, an output power of 22 mW with 35.5%
external efficiency and 18.8 mW with 34.9% external efficiency at operating current of
20 mA at room temperature was observed for InGaN based n-UV LED of emission
wavelength at 400 nm and 426 nm, respectively. Wang et al. developed a Cantibridge
technique to reduce the threading dislocation (TDD) in the GaN film [140-142]. The
technique involves creation of V-shaped grooves on sapphire substrate by wet etching
using H2SO4 acid. Presence of V-grooves on sapphire surface resulted in selective
growth of GaN only on the (0001) mesas and no GaN growth occurs on the
k011 sidewalls of the V-grooves as shown in Figure 2-10. Recently, Lee et al. reported
using the Cantibridge technique to fabricate GaN-based light emitting diodes on
chemical wet etched PSS (CWE-PSS) with V-shaped pits on top surface (Figure 2-11)
[108]. The reduction in defect density to 3.62 x 108 cm-2 using CWE-PSS substrate
resulted in 12.5% enhancement in the internal quantum efficiency. They showed using
theoretical calculation that the V-shaped pits are stronger diffuser, increasing the
probability of trapped photons to escape and achieving a 20% enhancement in
47
estimated light extraction efficiency. Additionally, V-shaped grooved PSS demonstrated
additional 7.8% enhancement over planar sapphire due to superior guided light
extraction efficiency. A combined 63.5% light extraction efficiency was observed as
compared to 49.1% for the GaN device on flat sapphire without V-shaped pits.
GaN grown on patterned SiO2 (epitaxially laterally overgrown GaN, ELOG) has
also been developed for reducing TDD (Figure 2-12) [106, 143-145]. Mukai et al. grew
InGaN LED on ELOG and sapphire substrate. The device on ELOG substrate showed
significantly low leakage current [145]. However, the light output results illustrated no
change in output power for due to presence of SiO2 mask layer. However, Lee et al
recently used inverted hexagonal pyramid dielectric mask (IHPDM) to grown GaN
device on sapphire [106]. Compared to conventional lateral epitaxial overgrowth (LEO),
their technique provided 3-fold merit. Firstly, because large size etch pits generally
originate from the nanopipes (open-core screw dislocations), they selectively eliminate
most defective region. Secondly, the regrowth thickness of GaN layer can be controlled
less than ∼ 5 μm due to small diameter of these etch pits. Lastly, embedded inverted
pyramid structures acts as strong light scattering centers. The ray-tracing simulation for
device with IHPDM showed 56% increase in light extraction and increase in output by a
factor of 1.41 was experimentally observed. Recently, Cho et al. reported 80% increase
in light output power of InGaN LED at 20 mA using pyramidal SiO2 mask over non-LEO-
LED and 30% over standard square-slab type LEO-LED [107].
External Out-couplers
External out-couplers are any structures created on emitting side of a light emitting
device for increasing the external mode. Several structures have been considered for
48
improving the light extraction as external out-couplers such as hemispherical dome [76],
pyramids [81, 110], cylinders and lenses [73, 85, 111-113]. Spherical shapes have been
known for improving external efficiency as they subtend a large solid angle of emitted
rays resulting in extraction of light previously trapped in the substrate [73]. Madigan et
al. demonstrated the effect of spherical substrate features on external coupling
efficiency and the far-field emission pattern. The light extraction increased by factor of 3
by attaching macro-sized spherical lens using index matching gel on the backside of the
OLED. However, use of millimeter-sized lens is not practical for many applications
without index matching issues. Möller et al. suggested an alternative, simpler technique
to fabricate ordered-array of micro-sized lens without the need for alignment with
respect to LED device. A 10 um diameter microlens square array of poly-dimethyl-
siloxane (PDMS) was fabricated using mold transfer method which was attached to
emission side of glass substrate (Figure 2-13). Using this scheme, 1.5 fold increase in
light output over unlensed device was observed. Choi et al. integrated the microlens on
micro-LED substrate [111]. Microlenses were fabricated using thermal reflow method,
aligned to the emitters forming the array. Light output measured using Si detector at 5
mm distance from micro-LED was 23% higher than same for an unlensed device. The
observed enhancement was attributed to the concentrating effect of microlens. Peng et
al. studied the effect of lens diameter and height on light extraction capability of the
microlens array [112, 146]. The simulation showed that hemispherical lens results in
maximum out-coupling efficiency. With optimized microlens array, efficiency as high as
85% can be achieved whereas the experimental observed efficiency was 70% for LED
device with optical epoxy microlens array. This discrepancy was attributed to the
49
absorption loss in microlens layer media and deviation of fabricated microlenses from
spherical profile. Using similar approach but with slight modification Nakamura et al.,
fabricated pyramidal microlens array and studied the effect of using high refractive index
substrate in conjunction with microlens array [81]. Interestingly, the high refractive index
substrate did not show any enhancement in light extraction whereas pyramidal
microlens structure resulted in similar extraction efficiencies as report by Peng et al. Ee
et al. fabricated SiO2/polystyrene microlens films utilizing low cost rapid connective
deposition on InGaN based device [113]. An improvement in integrated luminescence of
270% was observed using SiO2/PS microlens arrays. A schematic of InGaN based
device with SiO2/PS microlens array is depicted in Figure 2-14. Chen et al. fabricated
the macro-pyramidal array light-enhancing layer on an OLED. The pyramidal array with
optimized spacing (15 μm) and height (150 μm) for pixel diameter of ~300 μm
demonstrated a gain factor in luminescence intensity of 2.03. Greiner analyzed the
microlens and micropyramid array as the external out-couplers in an OLED device
[147]. The simulation results elucidated that the reflectance of OLED stack or back
reflector is of paramount importance since the probability of a photon to escape in first
pace through various structures is only a fraction of total escape probability. In addition,
the aspect ratio of microlens or micropyramid structure has strong influence on the out-
coupling efficiency as escape probability of higher angle photons reduces with aspect
ratio. Compact packing of these structures results in better out-coupling efficiency
because any flat region increases the probability of TIR. Greiner findings also illustrated
that the exact geometric nature of the structure influence angular distribution much
strongly as compared to the out-coupling efficiency. Using these findings he was able to
50
demonstrate 80% extraction efficiency from substrate for microlens, micropyramid and
scattering particles with aluminum back contact OLED device. Chang et al. studied
textured encapsulating top surface (cone, hemisphere, prism, inverted hemisphere and
inverted cone) on LED [80]. The simulation results showed prism-type structure as most
efficient structure, demonstrating a 35% enhancement in light extraction efficiency with
50 μm base. The experimental results with prism-type structure concurred well with
simulation results. Bao et al. reported increase in light extraction by 18.5 and 31.9% for
imprinted 1D and 2D taper-type grating on polymer encapsulation top surface on back-
side of sapphire substrate in GaN LED (Figure 2-15) [148].
Moth-eye structure etched on InGaN based LED has been demonstrated by
Kasugai et al. for light extraction enhancement as shown in Figure 2-16 [149, 150]. GaN
based blue LED was fabricated on 6H-SiC substrate with moth-eye structure on the
bottom side using self-assembled Au nanomask and RIE of SiC. Periodic cones of
mean height 400-600 nm and periodicity 160 nm resulted in output power 3.8 times the
convention LED at 20 mA drive current. Theoretical calculations using rigorous coupled
wave analysis (RCWA) along with 3-dimensional Maxwell’s equation showed that using
periodic moth-eye structure ~80% transmittance can be achieved for broad range of
incidence angle. The calculated light extraction efficiencies for periodic moth-eye
structure and pseudo-periodic moth-eye structure were 4.27 and 4.16 times higher than
conventional LED. Hong et al. fabricated similar pseudo moth-eye structures using UV
lithography and ICP etching on p-GaN layer with 200-300 nm diameter and 80-150 nm
height [151]. They reported an enhancement in PL intensity by a gain factor 5-7 due to
scattering of photons by these moth-eye structures. Recently, Kim et al. demonstrated
51
27% enhancement in light out-coupling using deeply etched mesa holes (DEMH) at
reasonable forward voltage [78]. However, an increase in reverse currents in DEMH
LED was observed attributed to an increased surface leakage through mesa holes.
Other Techniques
Numerous other techniques for light extraction have been also been reported in
literature. Few of them have been discussed here to cover the breadth. Metallic
nanostructures have been shown to have much higher scattering cross section than the
transparent structures due to the localized electromagnetic enhancement [152]. Hsu et
al. fabricated gold nanowire array to increase light out-coupling in an OLED device [84].
A 450 nm period array doubled the extraction efficiency for an Alq3 device. Using
metallic nanowire array they demonstrated not only increased light extraction but
demonstrated that a nanowire grid can also function as narrow band color filter by
changing the periodicity of the nanowires. To improve light extraction from ZnS:Mn thin-
film phosphors, Do et al studied the effects of introducing various two-dimensional (2D)
SiO2 nanorod arrays between ITO/Glass interface [109]. They obtained a ~6.4 fold
enhancement in the CL extraction efficiency at the interface between ZnS and air.
Recently, Sun et al. demonstrated increased light extraction efficiency by incorporating
a low index grid between ITO/Organic layer interface (Figure 2-17) [83]. Low index grid
only out-couples waveguided mode not the glass mode for which they used simple
microlens array. An out-coupling of ~2.3X higher over conventional OLED was observed
which can be further improved to ~3.5 fold using even lower index grid (nLIG <1.45).
Although a significant increase in ηext was observed for the several reported
methods such as device shaping, photonic crystals, patterned substrates, surface
roughening and so on but these methods are often accompanied by changes in the
52
radiation pattern, exhibit an undesirable angle dependent emission spectrum, or employ
costly or complex processing methods [85]. Application of external out-couplers provide
significant enhancement in light extraction without invoking such undesirable attributes.
In this work, effect of feature size and aspect ratio of different pyramid, lens and
cylindrical textures have been studied using ray-tracing simulations. The simulation
results were substantiated by experimentally fabricating devices with these structures.
53
Early stage
Dark-spots
Fully developed
Dark-spots
Early stage
Dark-spots
Fully developed
Dark-spots
Figure 2-1. Dark spot formation in LED active layer (adapted with permission from ref. [123]).
Figure 2-2. Evolution of extraction efficiency for AlGaInP (▲) and InGaN (■) LEDs showing extraction efficiency as high as 80%. For the red triangles, the acronyms are: AS = Absorbing Substrate; DBR = Distributed Bragg Reflector; RS = Reflective Substrate. For the blue squares, the acronyms are: CC = Conventional Chip; FC = Flip Chip; PS = Patterned Sapphire; VTF = Vertical Thin Film; TFFC = Thin Film Flip Chip; ITO = Indium Tin Oxide (reproduced with permission from ref. [127]).
54
Figure 2-3. (a) Optical image of AlGaInP/GaP truncated inverted pyramid LED and (b) is schematic of a TIP LED device (reproduced with permission from ref. [74]).
Figure 2-4. SEM micrographs of an N-face GaN surface etched by a KOH-based PEC method where (a) is after 2-min etch and (b) after 10-min etch (reproduced with permission from ref. [96]).
(a) (b)
55
Figure 2-5. SEM image of Oblique angle deposited GRIN ITO layer with refractive index profile (reproduced with permission from ref. [77]).
Figure 2-6. (a) Schematic depicting fabrication of PDMS nanowires and (b) is the scanning electron micrograph of a fabricated meshed surface on PDMS film. The inset shows the cross section of the surface (reproduced with permission from ref. [95]).
(a) (b)
56
Figure 2-7. (a) Top lit view, (b) schematic cross-section view of the PXLED and (c) radiation patterns of the LEDs at φ=0° (top plot) and at φ = 0° and 90° (bottom plot). The upper left insets indicate the direction of the measurements with respect to the lattice (reproduced with permission from ref. [132]).
Figure 2-8. (a) Schematic of a PC-OLED device with SiO2/SiNx PC layer. Far-field intensity profiles are observed for (b) the conventional OLED and (c) the PC-OLED while (d) and (e) are Intensity profiles along the horizontal line (A–B) and diagonal line (C–D), respectively. (Dotted lines: conventional OLED, solid lines: PC-OLED) (reproduced with permission from ref. [153]).
(a) (b)
(a) (b) (c)
(d) (e)
57
Figure 2-9. (a) Schematic drawing of GaN based device on stripe PSS [104] and (b)
cross-sectional SEM of LEPS-GaN grown on PSS aligned to GaN0211
direction (reproduced with permission from ref. [137]).
Figure 2-10. Cross-section SEM micrographs of GaN grown on V-grooved sapphire with a 6.5 μm period after 150 min growth time (reproduced with permission from ref. [142]).
(a)
58
Figure 2-11. (a) Top-view SEM image of CWE-PSS and (b) is a schematic diagram of device on CWE-PSS (reproduced from ref. [108]).
Figure 2-12. (a) Schematic diagram showing lateral epitaxial overgrowth of GaN layer on 6H-SiC substrate with SiO2 mask and (b) cross-section TEM micrograph of a laterally overgrown GaN layer on a SiO2 mask (reproduced with permission from ref. [144]).
(a) (b)
(a) (b)
59
Figure 2-13. SEM of a PDMS microlens array. The inset is detailed side view of the lenses illustrating accurate replication of the mold shape (reproduced with permission from ref. [85]).
Figure 2-14. Schematic of InGaN QWs LEDs structure utilizing SiO2/PS microspheres microlens array (reproduced with permission from ref. [113]).
60
Figure 2-15. SEM image of (a) 1-D triangular and (b) 2-D taper-like gratings on the sapphire backplane. Each inset displays the corresponding image of profile (reproduced from ref. [148]).
Figure 2-16. Cross-sectional SEM image of bottom of 6H–SiC substrate with moth-eye structure fabricated using metal nanomask. Simulation results of transmittances for (a) conventional structure having flat surface (solid line), (b) pseudo-periodic moth-eye structure (dotted line) and (c) periodic moth-eye structure (dashed line) (reproduced with permission from ref. [149]).
61
Figure 2-17. Schematic diagram of the OLED with the embedded low-index grid (LIG) in the organic layers (reproduced with permission from ref. [83]).
62
CHAPTER 3 OUTLINE OF RESEARCH
This work focuses on enhancement in light extraction efficiency in light emitting
devices by controlling the surface topography of TCO thin films. Zinc oxide is a
promising material which has been extensively researched as a replacement material
for commercially used indium tin oxide TCO material. This work has been primarily
focused on ZnO as TCO material. The first objective of this research is to control the
surface roughness, large particle inclusions and pin holes on the zinc oxide thin film to
reduce the effect of such anomalies on the properties of the subsequent active layer
and the performance of the LED device. Second objective is enhancement of light
extraction efficiency by creating controlled periodic micro-topography. As part of the first
goal, a chemical mechanical polishing process for controlling the surface roughness of
ZnO was developed. A thorough study was conducted to develop an understanding of
the zinc oxide CMP process due to limited literature on the polishing of zinc oxide
material. For enhancement in light extraction, surface textures consisting of hexagonal
arrays of cones/pyramids, microlens and cylinders were fabricated. These structures
are known to improve light extraction but thorough understanding of ray dynamics
needs to be developed. The ray-tracing simulations were carried out to elucidate the ray
dynamics of these structures with respect to the effect of feature size and aspect ratio
when applied to different interfaces in a LED device structure. These two strategies
were finally combined in a thin film photoluminescent device involving Alq3 as the
photoluminescent material, to demonstrate the efficacy of these textures in improving
light extraction.
63
In this dissertation, zinc oxide chemical mechanical polishing results have been
presented in chapter 4. In this chapter, a detailed study of various CMP variables such
as pressure, platen speed, pH, slurry particle size and concentration was carried out.
Based on the results obtained, a zinc oxide CMP mechanism was proposed. The effect
of slurry additives such as surfactants and salt are illustrated in chapter 5. As part of
second objective, the ray-tracing simulations were performed on hexagonal arrays of
pyramids, microlenses and cylinders with different base diameters and aspect ratios.
The simulation results for these micro-textures when applied to different interfaces in
glass/ZnO/active layer stack with back reflector are presented in chapter 6. The chapter
7, describes the experimental procedure for fabrication of the hexagonal arrays of
cones/pyramids, microlenses and cylinders. The strategies of introducing controlled
micro-textures and polishing of zinc oxide surface to reduce surface roughness and
surface defects were combined and tested using PL measurements. The experimental
results of PL measurements with polished Al:ZnO TCO layer and micro-texture at the
glass/TCO interface are described in chapter 8. Chapter 8 also details the ray-tracing
simulation results to support the observed experimental findings. Chapter 9 summarizes
the conclusions of zinc oxide polishing results with both ray-tracing simulation and
experimental results for light extraction using various textures at different interfaces.
64
CHAPTER 4 CHEMICAL MECHANICAL POLISHING OF ZINC OXIDE
Introduction
In recent years, zinc oxide has become an important material in the field of
optoelectronics for UV and blue light emitting devices and lasers [40, 45-47],
piezoelectric transducers [50] and sensors [51, 52] due to its large band gap (3.37 eV)
and a high exciton binding energy (~60 meV) [42]. Zinc oxide single crystal has been
used as a substrate for epitaxial growth of GaN thin films [53-55]. Zinc oxide thin films
have been investigated as a buffer layer for growth of GaN on sapphire and SiC
substrates [48, 56-58]. In addition, doped-ZnO thin films are being extensively
researched for their use as transparent conducting electrodes [40]. However, both
commercially available zinc oxide single crystals and as-grown zinc oxide thin films
have a rather high surface roughness, which hampers the growth of subsequent high
quality layers. Zinc oxide thin films are largely fabricated using techniques such as
sputtering [29, 50, 114], pulsed laser deposition [40, 47], low pressure chemical vapor
deposition [62, 115, 116] or spray pyrolysis [117-119]. Irrespective of the deposition
method they have high surface roughness, large particle inclusions and pinholes [59-
62]. Such defects influence the growth of subsequent layers when zinc oxide thin films
are used for buffer layer applications or influence the interface between zinc oxide layer
and the active layer in light emitting diodes [47, 52]. In addition, in organic light emitting
diodes, these anomalies lead to color degradation and localized electrical shorts hence,
undermining the performance and lifetime of these devices. Further, in thin film silicon
solar cells, high roughness of the underlying ZnO layer leads to poor growth of the
subsequent silicon layers. In microcrystalline Si (µc-Si:H) solar cells, surface roughness,
65
orientation and grain size of the ZnO film affects the grain size and quality of the µc-Si:H
layer [62-64]. In addition to this, increased interfacial area between rough transparent
conducting oxide (TCO) and the p-layer due to TCO roughness correlates qualitatively
to the observed decrease in open circuit voltage (VOC) and fill factor for the device.
Vanecek et al., have reported that the absorption between 500 – 1100 nm wavelength
increases with increase in RMS roughness at ZnO/back reflector interface [65]. These
observations have motivated investigation of zinc oxide polishing to achieve a smooth
surface. Planarization of zinc oxide provides the opportunity for formation of controlled
surface texture and eliminates the detrimental unwanted/uncontrolled large particle
inclusions/sharp features on zinc oxide surface.
There are very few reports published on the zinc oxide polishing. To our
knowledge, Lucca et al. first reported the polishing of ZnO surface. In their work, the
photoluminescence (PL) response of mechanically polished and chemo-mechanically
polished ZnO single crystal surface was compared [66]. They showed that PL response
of chemo-mechanically polished ZnO single crystal (either O-face or Zn-face) was two
orders of magnitude higher than the mechanically polished ZnO, due to better surface
roughness and reduced sub-surface damage. However, detailed study of the chemo-
mechanical polishing process was not carried out. Recently, Lee et al. published a
paper on the effect of various chemical mechanical polishing (CMP) process
parameters on morphology & optical properties of the sputtered zinc oxide thin films
[59]. However, the report does not provide a comprehensive study of the process
parameters such as slurry pH, polishing pressure and particle size and concentration.
Some of the other reported techniques for controlling surface roughness of the zinc
66
oxide thin films involve irradiation by gas cluster ion beam for smoothening of the
surface topography [61]; controlling growth rate, thickness and doping concentration in
chemical vapor deposition grown ZnO to tune the roughness [67] and bi-layer structure
of ZnO involving deposition of second smooth ZnO layer using atomic layer deposition
[68]. However, using ion beam/ electron cyclotron resonance etching to sputter material,
disrupts the crystal structure of the underlying zinc oxide film [61, 69]. Furthermore,
these techniques do not provide a good control over the surface roughness. In this
work, a thorough study of the chemical mechanical polishing process for zinc oxide thin
film polishing was performed in order to delineate the zinc oxide polishing process and
to achieve low surface roughness ZnO thin films. We investigated the effect of various
CMP variables such as pH, applied pressure, platen speed, particle size and
concentration on the polishing of zinc oxide films. The CMP performance was evaluated
by measuring the removal rates and surface roughness obtained from thickness
measurements and atomic force microscopy (AFM) scans followed by optical and x-ray
reflectivity (XRR) measurements.
Experimental Procedure
A Struers Tegra Force 5 table-top polisher and Cabot Microelectronics D100
polishing pads were used for all the CMP experiments. The polisher variables such as
revolutions per minute (rpm) and pressure were varied from 60-120 rpm and 2.5-5 psi,
respectively, whereas slurry parameters namely, slurry pH, abrasive particle size and
slurry solid loading were varied systematically to understand the effect of each variable
on the polishing performance. Colloidal silica particles (Levosil S.p.A., Italy) with
different particle sizes (35 - 135 nm) were used throughout the study. Post-CMP
cleaning was performed using pH 10.6 (adjusted using KOH) deionized (DI) water on a
67
buffing pad for 1.5 minutes followed by ultrasonication in acetone, methanol and DI
water. For the study, silicon p-type substrates (1x1 inch2) were sputter coated with
250±20 nm thick ZnO film. The ZnO films were deposited using RF sputtering at 250 W
power, 5 mTorr working pressure with target to substrate distance of 80 mm and no
external substrate heating. All depositions were carried out at 1.5% oxygen mixed with
argon. After deposition, the samples were annealed at 150°C for 1 hour under ambient
atmosphere.
The unpolished films were characterized by x-ray diffraction (XRD) and atomic
force microscopy (AFM, Veeco Dimension 3100). The x-ray diffraction was performed
on the as-deposited and annealed samples on a Philips APD 3720 machine. The
surface morphology, before and after CMP was investigated using AFM performed in
contact mode using a silicon nitride tip. Film thicknesses were measured using
spectroscopic ellipsometry (Woollam EC110 Ellipsometer) for calculating removal rates.
X-ray reflectivity (XRR) measurements were performed on a Panalytical MRD X'Pert
System whereas optical transmission spectra of the ZnO films deposited on corning
glass were measured in the visible region using an Ocean Optics HR4000 high-
resolution spectrometer.
Results and Discussion
Figure 4-1 shows the AFM micrograph for the sputter deposited film after
annealing at 150ºC for 1 hour. The annealed zinc oxide films consist of grains varying
from 500 Å to 3000 Å in size with an average RMS roughness of 26±6 Å. Figure 4-2
shows the x-ray diffraction patterns for the as-deposited and annealed samples. The
XRD pattern of as-deposited and annealed ZnO film shows dominant (002) c-axis
orientation. The increased (002) peak intensity and reduced full width half maxima
68
(FWHM) of the annealed ZnO film are due to improvement in the film quality and
crystallinity. Annealing in ambient atmosphere resulted in a shift in (002) ZnO peak from
33.87º to 34.31º. The average crystallite size deduced from FWHM using Scherrer
equation [154, 155] increased from 154 Å to 192 Å.
Effect of Applied Down-pressure and Platen Speed
For pressure variation, samples were polished at 60 rpm, at a constant slurry flow
rate of 60 ml/min and at a basic pH (pH ~ 10.6). Down pressures above 4.5 psi (32.1
kPa) were not considered in this study since high pressure polishing is not desirable
due to higher defectivity and low yield. Figure 4-3 represents the variation of removal
rate with respect to the applied pressure during polishing. With increasing pressure,
contact area between the surface and slurry abrasives increases resulting in the
observed increase in material removal. The average RMS roughness of ZnO films was
reduced to as low as 5 Å after polishing.
Similarly, for studying the variation of removal rate with platen speed, down
pressure was fixed at 2.5 psi with a constant slurry flow rate of 60 ml/min and a
polishing time of 2 min. The variation of removal rate with linear velocity shows a typical
CMP behavior, it increases with the increase in platen speed (see figure 4-4).
Preston equation is often used to fit the removal rate variation with pressure and
platen speed. According to Preston equation:
KPVRR (4-1)
where RR is the removal rate, P is the pressure, V is the linear velocity and K is the
Preston’s constant. The linear fit of experimental data according to Preston equation is
represented by the red line in figure 4-3 and 4-4. However, careful analysis of fitting
parameters illustrate that the Preston equation is inconsistent as the removal rate
69
showed non-zero intercepts for both P→ 0 and V→ 0. In fact, a negative intercept was
observed for the case of P→ 0. To accommodate the non-zero intercepts, slightly
modified form of Preston equation as suggested by Luo et al [156] was used (eq. 4-2).
cth RVPPKRR )( (4-2)
where, Pth is a constant representing a threshold pressure and Rc is a constant removal
due to purely chemical reaction. The constants K, Pth and Rc were calculated from the
figure 4-3 and 4-4 using least square method and are enumerated in Table 4-1. The
calculated values of Preston constant, K from figure 4-3 and 4-4 were compared and
are within ± 5% error, suggesting the consistency of the equation with the experimental
data.
Effect of Particle Size
The variation of removal rate and RMS roughness (inset) with respect to varying
particle size is represented in Figure 4-5. The change in removal rate with varying
particle size is within CMP variations. However, as expected the samples polished with
smaller particle slurry show better roughness which tends to saturate for particle size
above 55 nm. This is related to the indentation depth given by the relation:
3
2
24
3
KE
Papp (4-3)
where, is the particle size, Papp is the applied load, E is the effective elastic modulus of
the two surfaces involved. As seen from the equation (4-3), smaller particles due to
shallow indentation will lead to lesser material removal and hence better surface
roughness whereas large particles will cause large indentations resulting in higher
surface roughness or defects.
70
Effect of Slurry pH
To investigate the effect of slurry pH on the material removal, pH of the slurry
solution was varied from basic to acidic pH. The zinc oxide removal rates increase
dramatically with decrease in the slurry pH from ~10.6 to ~3 as represented in Figure 4-
6. A maximum removal rate of ~670 Å/min was achieved at a pH of ~3. The 10-fold
increase in the removal rates in acidic conditions (pH ~3) as compared to the basic pH
(~10.6) illustrate that the removal rates in the ZnO polishing process are strongly
dependent on pH of the slurry.
The inset in figure 4-6 represents the RMS roughness values with respect to the
slurry pH. The RMS roughness trend demonstrates an increase in roughness with
increasing pH. However, after careful examination it was found that samples polished at
higher pH were under-polished due to lower polishing rates at those pH values. When
the samples were polished for 2 minutes with pH 10.6 slurry, the surface roughness was
comparable to that at low pH.
The AFM images of zinc oxide films polished with different pH slurries are depicted
in figure 4-7. For ease of comparison, the Z-scale of the images was adjusted. After
polishing, the zinc oxide surfaces are highly smooth and exhibit RMS roughness of 6 Å
as compared to the initial roughness of 26±6 Å. The RMS roughness values of <2 Å
were observed for the ZnO crystals polished under same conditions (results not shown),
indicating the role of grain size in the final roughness after polishing of the
polycrystalline ZnO thin films. Interestingly, the surface roughness values obtained with
acidic as well as basic pH slurry (over-polished) were very close to each other (< ±1Å
variation in RMS roughness) in spite of an order of magnitude difference in the polishing
71
rates. To explain such nature of zinc oxide CMP process, the zinc oxide polishing
mechanism has been discussed in detail in the later section.
Effect of Particle Concentration
The solid content of the slurry was varied from 2 to 10 wt% at slurry pH ~3. The
variation of removal rate with respect to solid loading is represented by the plot in figure
8 with the average RMS roughness graphed in the inset. The removal rate increases
with increasing concentration however, the RMS roughness plateaus above 5 wt%
slurry loading.
Zinc Oxide CMP Mechanism
A typical CMP process can be broken down into two components: (1) the
formation of chemically-modified surface layer and (2) mechanical removal of the
surface layer [157-159]. In case of metal oxides, the formation of chemically-modified
surface layer can be classified into two sub-processes: (1) hydration of oxide [160, 161]
and (2) bonding of abrasive particles with the material [157, 158]. The hydrolysis of
oxide is largely affected by pH of the solution and the useful pH range depends upon
the dissolution of oxide and hydroxide formed [160, 162]. For zinc oxide, it is stable
between a pH of 6 to 12, whereas dissolution of zinc oxide is relatively high in acidic
and high alkaline pH (>12) range. In alkaline pH, zinc oxide forms zinc hydroxide, which
has an inhibiting effect on the dissolution of zinc oxide. Depending upon the pH of the
solution (in alkaline range), zinc hydroxide further transforms into zincate or bizincate
ions via eq. (4-4) and (4-5). These zincate ions also passivate the zinc oxide surface
due to their low solubility in aqueous solution at pH 10 [163].
2
HZnOOHZnO (4-4)
72
2
42)(2 OHZnOHOHZnO (4-5)
OHZnHZnO2
22 (4-6)
On the other hand, dissolution of ZnO in acidic pH conditions is given by the eq.
(4-6). In acidic pH, zinc oxide dissolution is faster due to direct reaction with hydrogen
ions to give zincic ions. To delineate such chemical interactions of zinc oxide with the
slurry, the polished samples were kept in different pH slurry solutions for 5 minutes. The
polished samples with RMS roughness < 6 Å were used for this study to highlight the
change in RMS roughness due to dissolution. The dissolution rates and change in the
RMS roughness with respect to slurry pH are tabulated in Table 4-2. The dissolution
rate of 37 Å/min was observed for pH 3 slurry solution whereas the solubility of zinc
oxide was found to be negligible for other pH values tested. The change in RMS
roughness was 17 Å for pH 3 in contrast to negligible change for slurries with pH 7-11 in
5 minutes. This confirms the response of zinc oxide under various pH conditions of
slurry solution in accordance to the above discussion.
Besides dissolution, pH of the slurry also affects the mechanics of the material
removal in the CMP process [161]. An oxide surface can become charged by absorbing
either H+ or OH- ions depending on the pH of the solution. Such electrostatic charged
sites can affect the area of contact between the abrasive particles and the metal oxide
surface hence, influencing the material removal during CMP process. The number of
such charged sites and surface charge can be measured using streaming
potential/zeta-potential measurements. The zeta-potentials of the annealed zinc oxide
film and colloidal silica particles (the slurry abrasive used in this study) were measured
to understand the development of surface charge and interaction of the two surfaces
73
under various pH ranges. The zeta-potential of zinc oxide film and colloidal silica
particles with respect to the pH of media solution are plotted in figure 4-9. Isoelectric
point (IEP) of silica particles occur at pH 2.75 whereas it occurs at pH 8.5 for the zinc
oxide film. Interestingly, zinc oxide surface and silica particles are oppositely charged
between the pH region 2.75 - 8.5, implying increased interaction between the zinc oxide
surface and the silica abrasives in this range during polishing.
Based on the zinc oxide dissolution and zeta-potential measurements, CMP of the
zinc oxide can be divided in to three pH regions: (1) pH greater than ZnO IEP, (2) region
between the IEP of zinc oxide and silica and (3) pH less than silica IEP. For pH greater
than zinc oxide IEP (pH > 8.5), the zinc oxide and silica particles surface have same
charge which reduces the interaction between the two surfaces. The zinc oxide
dissolution is limited by the passivation of zinc oxide by hydroxide/zincate ions, resulting
in lower removal rates. Similarly, when solution pH < 2.75 (the IEP of silica), the two
surfaces have same charge. However, ZnO dissolution is high in such high acidic
conditions leading to higher removal rates and poor surface roughness due to dominant
dissolution process. In contrast, the region between pH 3-8, zinc oxide and silica
surfaces are oppositely charged leading to increased particle-surface interactions.
Additionally, chemical dissolution increases as pH of the solution decreases. The
combined effect of chemistry and mechanical aspect of the CMP process synergistically
mitigate the effect of dissolution on surface roughness with beneficial increase in
removal rates.
Figure 4-10 represents the optical transmission curves for plain glass substrate,
unpolished zinc oxide deposited substrate and zinc oxide substrates polished with
74
acidic (pH ~3) and basic pH (pH~10.6) slurries. The curves have been normalized with
respect to plain glass. A transmittance of 88%, 91% and 94% at the wavelength of
632.8 nm was found for the as-deposited, acidic pH slurry and basic pH slurry polished
samples, respectively. The polished samples show better transmittance which can be
attributed to the reduction in surface roughness and exclusion of peaks and large
particles. Additionally, the substrate polished using the basic pH slurry shows higher
transmission efficiency (at λ = 632.8 nm) as compared to the sample polished with the
acidic pH slurry. The sample polished with basic pH slurry shows higher transmission in
the region above ~520 nm whereas it shows lower transmission below that as
compared to both unpolished and acidic pH slurry polished samples. This behavior can
be related partially to the shifting of curves due to thickness variation between the films
and chemical interaction of the zinc oxide surface with the polishing slurry as discussed
above.
The x-ray reflectivity measurements were performed on the unpolished, acidic pH
slurry and basic pH slurry polished samples. Figure 4-11(a) depicts the measured
intensity versus angle whereas the schematic depicting the simulation results obtained
by fitting the XRR curves is shown in figure 4-11(b). As seen from the figure 4-11(a),
both acidic and basic pH slurry polished ZnO films show similar curves whereas as-
deposited film shows some dissimilarity. To model the XRR data, we have assumed a
surface layer composed of ZnO, adsorbed carbon and water. According to the fit
parameters, a surface roughness of 5-6 Å was observed for polished ZnO films (with
both acidic & basic slurries) whereas 38 Å roughness was observed for the unpolished
ZnO film. These results concur well with the RMS roughness calculated from the AFM
75
micrographs. The observed dissimilarity between polished and unpolished samples can
be attributed to the surface roughness difference and the presence of thick surface layer
for the unpolished ZnO surface.
Conclusion
Highly smooth zinc oxide surface with RMS roughness less than 6 Å was
achieved. Removal rates as high as 670 Å/min were achieved at slurry pH of ~3.
Polished zinc oxide films showed higher transmission (91% and 94%, at λ = 632.8 nm)
as compared to the unpolished surface. When various process variables were varied
typical CMP behavior was observed. Modified Preston equation was applied to explain
the removal rate variation with pressure and platen velocity. The modified Preston
equation was found to be consistent with the experimental results and accounts for the
non-zero intercepts. Low polishing rates were observed at higher pH (> 7) whereas at
low pH (~3) an order of magnitude increase in the removal rates was observed without
any detrimental effects on the surface roughness. The zinc oxide CMP process at
different slurry pH values was explained by the pH dependence of the zinc oxide
dissolution in aqueous solution and particle-surface interactions. X-ray reflectivity
simulation results concurred well with the RMS roughness values for the ZnO films
deduced from the AFM images.
76
Figure 4-1. AFM micrograph of the annealed zinc oxide film (scale bar is in Å).
26 28 30 32 34 36 38 40
As-deposited
Inte
nsity (
a.u
.)
2Theta (degrees)
ZnO (002)
Si (002)
Annealed
Figure 4-2. X-ray diffraction patterns of the as-deposited and annealed ZnO films.
1.0µm
RMS=31 Å
31.58
0.00
77
15 20 25 3040
80
120
160
200
240
Re
moval R
ate
(Å
/min
)
Pressure (kPa)
15 20 25 30
Pressure (kPa)
4
6
8
10
12
14
Roughness (
Å)
Figure 4-3. Removal rate (■) with respect to the applied pressure during CMP (red line
represents the linear fit). Inset plots RMS roughness (●) versus pressure.
30 40 50 60 70 80
60
80
100
120
Re
moval R
ate
(Å
/min
)
Linear Velocity (m/min)
30 40 50 60 70 80 4
5
6
7
8
Roughness (
Å)
Linear Velocity (m/min)
Figure 4-4. Removal rate (■) versus linear velocity and RMS roughness (inset) (●) with
respect to linear velocity. Red line represents the linear fit.
78
20 40 60 80 100 120 14020
40
60
80
100
120
Particle Size (nm)
Re
moval R
ate
(Å
/min
)
20 40 60 80 100 120 1402
3
4
5
6
Ro
ug
hn
ess (
Å)
Particle Size (nm)
Figure 4-5. Graph representing removal rates at various particle sizes (■) whereas inset
depicts the RMS surface roughness given by (●).
3 4 5 6 7 8 9 10 110
100
200
300
400
500
600
700
pH
Rem
oval R
ate
(Å
/min
)
3 4 5 6 7 8 9 10 11
6
8
10
12
14
Ro
ug
hn
ess (
Å)
pH
Figure 4-6. Removal rates versus pH of the slurry. Inset represents the average RMS roughness.
79
Figure 4-7. AFM micrographs of zinc oxide films polished with (a) pH 3.1, (b) pH 7.0 and (c) pH 10.6 slurry. Scale bar is in Å
2 4 6 8 10400
500
600
700
800
900
1000
1100
Concentration (wt%)
Re
moval R
ate
(Å
/min
)
2 4 6 8 10 3
4
5
6
7
8 R
oug
hn
ess (
Å)
Concentration (wt%)
Figure 4-8. Removal rate variation with weight percent solid loading of the polishing slurry solution. Measured RMS roughness is shown in the inset.
1.0µm
13
.77
0.0
0
(b) RMS=6Å
1.0µm
13
.06
0.0
0
(c) RMS=5Å
1.0µm
13
.12
0.0
0
(a) RMS=6Å
80
1 2 3 4 5 6 7 8 9 10
-50
-40
-30
-20
-10
0
10
Zeta
Pote
ntial (m
V)
pH
Silica Particles
ZnO
ZnO
ZnO
ZnO
Figure 4-9. Plot shows zeta potential with respect to the pH, for the silica particles (■) and the zinc oxide thin film (●).
400 500 600 700 8000.0
0.5
1.0
No
rmaliz
ed Inte
nsity (
a.u
.)
Wavelength (nm)
Glass
Non-CMP
Acidic pH
Basic pH
Figure 4-10. Optical transmission curves for the glass, unpolished and the samples
polished with acidic and basic pH slurries.
81
0.46 0.92 1.38 1.84 2.30 2.76 3.22 3.68 4.141
10
100
1k
10k
100k
1M
10MIn
tensity (
counts
/s)
Angle ()
Non-CMP
Acidic pH
Basic pH
Acidic pH
Basic pH
Non-CMP
(a)
Substrate
ZnO
(2697Å, 38Å)
Surface layer
(~52Å)
Substrate
ZnO
(1516Å, ~5Å)
Surface layer
(~7Å)
Substrate
ZnO
(1775Å, ~5Å)
Surface layer
(~4Å)
Non-CMP Acidic pH Basic pH
(b)
Substrate
ZnO
(2697Å, 38Å)
Surface layer
(~52Å)
Substrate
ZnO
(1516Å, ~5Å)
Surface layer
(~7Å)
Substrate
ZnO
(1775Å, ~5Å)
Surface layer
(~4Å)
Non-CMP Acidic pH Basic pH
Substrate
ZnO
(2697Å, 38Å)
Surface layer
(~52Å)
Substrate
ZnO
(2697Å, 38Å)
Surface layer
(~52Å)
Substrate
ZnO
(1516Å, ~5Å)
Surface layer
(~7Å)
Substrate
ZnO
(1516Å, ~5Å)
Surface layer
(~7Å)
Substrate
ZnO
(1775Å, ~5Å)
Surface layer
(~4Å)
Substrate
ZnO
(1775Å, ~5Å)
Surface layer
(~4Å)
Non-CMP Acidic pH Basic pH
(b)
Figure 4-11. (a) X-ray reflectivity curves and (b) schematic showing fitting results
(values indicate film thickness and roughness, respectively).
82
Table 4-1. Fit parameters for the modified Preston equation
Plot Rc (Å/min) Pth K(x10-4 Pa-1)
RR versus pressure 9 1.235x104 2.69
RR versus velocity 2.78
Table 4-2. Dissolution rates and change in roughness for the samples soaked in different pH slurries.
pH Dissolution Rate
(Å/min) ∆Roughness
(Å)
10.5 <1 ~1
9 <1 ~1
7 ~2 3
3 37 17
83
CHAPTER 5 EFFECT OF SLURRY ADDITIVES: SURFACTANT AND SALTS
Introduction
Surfactants are surface active agents that can modify the surface properties of
liquids or solids. In the field of integrated circuit fabrication, adsorption of surfactants at
the solid-liquid interface has been mainly applied to the stabilization of particulate
dispersions in chemical mechanical planarization [164, 165] and wet cleaning [166-168].
Surfactants can selectively adsorb at any interface. Spontaneous adsorption of
surfactants on surface can cause charge reversal. Presence of surfactants is also
known to result in lubrication between the two interacting surfaces [165]. This concept
has been applied to CMP, not only for particulate stabilization but for the maximization
of the topographical selectivity to increase planarization efficiency [169]. Surfactant
layers can also adsorb on a metal or a dielectric surface and act as inhibiting layers
against unfavorable chemical etching or corrosion during CMP [164, 170].
The addition of salt into the solution increases the ionic strength of the solution.
The increase in ionic strength of the slurry solution results in screening of the charges
on various surfaces, lowering the repulsion between polishing pad, particles or wafer
surface depending upon the pH of the slurry [171]. The charge screening enables a
closer contact between the pad and the wafer surface; and between the particles and
the wafer surface. This phenomenon is known to increase the frictional forces between
the pad and the wafer surface [161, 171]. Additionally, it has been suggested that some
transient agglomerates may form during polishing in the presence of salt which can
further increase the frictional forces between different interacting surfaces in the CMP
process [161, 171, 172].
84
In this chapter, the effect of slurry additives such as surfactants and salt has been
studied to increase the polishing efficiency of zinc oxide. The concentrations of
surfactants (anionic and non-ionic) and potassium chloride (KCl) salt were varied to
understand the effect of these additives on chemical mechanical polishing (CMP) of zinc
oxide films. The CMP performance was evaluated by measuring the removal rates and
surface roughness obtained from thickness measurements and AFM scans. Finally,
optical measurements were performed to evaluate the optical properties of polished zinc
oxide surface.
Experimental Procedure
CMP experiments were carried out on a Struers Tegra Force 5 table-top polisher
with a Cabot microelectronics D100 polishing pad. The effect of concentration of
surfactants namely, sodium dodecyl sulfate (SDS) and Triton X 100 (TX-100); and
potassium chloride (KCl) salt on the surface roughness of zinc oxide was studied. The
concentrations of additives were varied systematically to understand the dependence of
polishing performance on the concentration. The polishing was carried out at 2.5 psi
pressure with platen speed of 60 rpm under acidic pH (~3) conditions. Colloidal silica
(Silbond Corp.) with mean particle size of 75 nm was used for all the experiments.
Silicon p-type substrates (1x1 inch2) were sputter deposited with 250±20 nm thick ZnO
film. The ZnO oxide films were deposited using RF sputtering at 250 W power, 5 mTorr
ambient pressure with target to substrate distance of 80 mm and no external substrate
heating. All films were deposited at 1.5% oxygen mixed with Argon and were
subsequently annealed at 150°C for 1 hour in ambient atmosphere.
The CMP performance was evaluated by measuring the RMS roughness and
removal rate. The surface morphology, before and after CMP was investigated using
85
atomic force microscope (AFM, Veeco Dimension 3100). Initial and final thicknesses
were measured using spectroscopic ellipsometer (Woollam EC110 Ellipsometer). The
optical transmission spectrum was measured on samples deposited on glass using an
Ocean optics spectrophotometer (HR4000).
Results & Discussion
It is known that surfactants in a polsihing slurry can effect CMP process in two
ways, (1) modifying particle-particle and (2) particle-surface interactions [165]. By tuning
the concentration of surfactant, optimal CMP performance in terms of removal rates and
surface roughness can be achieved. In addition, ionic strength of the slurry solution also
affects the mechanical aspect of CMP process by charge screening between various
surfaces. To elucidate the effect of slurry additives such as surfactants and salts, ionic
and non-ionic surfactants and KCl salt concentrations were varied systematically. The
CMP variables such as pressure and platen speed, were fixed at 2.5 psi, 60 rpm
whereas slurry pH, particle size and concentration were ~3, 75 nm and 5 wt %
respectively as per optimized process parameters described in chapter 4.
Effect of Surfactant
Sodium dodecyl sulfate (SDS) surfactant was used for the polishing experiments.
Figure 5-1 represents the variation of removal rate and RMS roughness (inset) with
SDS concentration. A reduction of greater than 18 % in removal rates was observed as
compared to the reference slurry. Additionally, the zinc oxide removal rate decreases
with increasing concentration of the SDS surfactant whereas RMS roughness shows
slight increase. This can be attributed to the surfactant-zinc oxide surface interactions.
At slurry pH ~3, the ZnO surface is positively charged and silica particles are negatively
charged according the zeta potential curves (chapter 4, Figure 4-9). SDS being an
86
anionic surfactant attaches to the positively charged zinc oxide surface, thereby
reducing the charged sites on the surface. This charge screening effect contributes to
the reduced interaction between the zinc oxide surface and the silica particles as
compared to the case with no surfactant. The reduction in mechanical removal can be
correlated to the observed increase in RMS roughness as chemical dissolution of zinc
oxide slightly dominates.
The concentration of Triton X 100 surfactant in the slurry was varied to study the
effect of non-ionic surfactant on the polishing of zinc oxide. Figure 5-2 represents the
variation of removal rate with respect to the concentration (in g/l) of TX-100. The
removal rates with triton X-100 surfactant were found to be lower than the rates
observed for slurry without any surfactant. Addition of non-ionic surfactant is known to
have lubricating effect between the surfaces. Also, non-ionic surfactant forms a
surfactant layer on substrate, silica particle and polishing pad surface which leads to the
charge screening effect causing observed decrease in removal rates as compared to
the slurry without the TX-100 surfactant. However, removal rate shows slight increase
with increasing concentration of TX-100 while RMS roughness illustrates a decreasing
trend with respect to increasing surfactant concentration. This corresponds to the fact
that as surfactant concentration increases the screening (reducing electrostatic
repulsion) between silica particles and CMP pad (negatively charged in pH range 4-9
[161]) thereby, the number of particles between pad and zinc oxide substrate increases.
Such increase in the number of particles between pad and substrate leads to higher
friction due to larger contact area hence higher observed polishing rates. However, due
87
to the distribution of down force (reduction of force per particle), surface roughness
decreases with increasing concentration of surfactant [161].
Effect of Salt
Salts increase the removal rates in a typical CMP process by increasing the ionic
strength of slurry solution resulting in screening of charge. Such charge screening leads
to the reduction in repulsive forces between various surfaces as depicted by the
increase in removal rate with salt concentration in Figure 5-3. The removal rates for
concentrations greater than 0.2 M were observed to be even more than that for
reference slurry (without salt, 670 Å/min). The increase in the surface roughness of the
polished zinc oxide surface with the addition of salt can also be related to the increased
frictional forces. Additionally, it is suggested that some transient agglomerates may form
during polishing in presence of salts in the slurry due to local variations in the particle
concentration. Such increased mechanical removal and formation of transient
agglomerates cause the observed increase in roughness as depicted by Figure 5-3
(inset).
AFM micrograph for the sputter deposited film after annealing at 150ºC for 1 hour
is represented by Figure 5-4. The AFM image elucidates that the zinc oxide films
consist of grains varying from 500 Å to 3000 Å in size with an average RMS roughness
of 26±6 Å. The AFM micrographs of zinc oxide films polished with different
concentrations of SDS, TX-100 and KCl in the slurry are represented in Figure 5-5, 5-6
and 5-7, respectively. The zinc oxide surfaces after polishing were highly smooth and
exhibit RMS roughness between 3-6 Å as compared to the initial roughness of 26±6 Å.
RMS roughness between 3-6 Å after polishing irrespective of pH or slurry additives
88
indicates that the minimum roughness is limited by the grain structure of the
polycrystalline zinc oxide films.
Figure 5-8 is the optical transmission plots for plain glass substrate, as-deposited
zinc oxide on glass substrate and zinc oxide on glass substrates polished with different
slurry compositions. The curves have been normalized with respect to the plain glass
substrate. The average transmittance varies between 92-93% for the as-deposited zinc
oxide film and polished zinc oxide with slurry consisting of 0.1 M KCl, 5 mM SDS and
0.322 g/l TX-100. The observed difference in the spectrum of various films polished with
different additives is due to variation in the final thicknesses of these films.
Conclusion
Effect of anionic and non-ionic surfactants on the zinc oxide CMP was studied.
Anionic surfactant (SDS in this case) was found to lower the polishing rates and
increase the RMS roughness with increasing concentration. Addition of 5 mM SDS was
found to give the best surface finish with RMS roughness ~3 Å. However, reverse
trends were obtained with Triton X-100 (non-ionic) surfactant due to reduced repulsive
forces between slurry particles and CMP pad. The removal rates increased to as high
as 900 Å/min with addition of KCl salt at the cost of surface roughness. The RMS
roughness of polycrystalline film of ZnO was found to limit at or greater than 3 Å for any
slurry composition. This behavior is related to the film morphology dependence (grain
size) of CMP process for zinc oxide. Optical transmission of zinc oxide films deposited
on glass substrate exhibited an average transmission between 92-93% for various
slurry compositions.
89
5 10 15 20 25 30200
250
300
350
400
450
500
550
600
Rate
RMS Roughness (Center)
Conc. (mM)
Re
moval R
ate
(Å
/min
)
2
3
4
5
6
7
RM
S R
oughness (Å
)
Figure 5-1. Solid squares (■) show the removal rate (Å/min) variation with SDS concentration in slurry whereas RMS roughness is depicted by circles (○).
0.2 0.4 0.6 0.8 1.0 1.2 1.4500
550
600
650
700
Rate
RMS Roughness (Center)
Conc. (g/l)
Rem
oval R
ate
(Å
/min
)
2
3
4
5
6
RM
S R
oughness (Å
)
Figure 5-2. Plot illustrating Removal rate (■) and RMS roughness (○) variation with respect to the concentration of TX-100 surfactant in the slurry.
90
0.0 0.1 0.2 0.3 0.4
200
400
600
800
1000
1200
Rate
RMS Roughness (Center)
Salt Conc. (M)
Re
moval R
ate
(Å
/min
)
2
3
4
5
6
7
8
9
RM
S R
oughness (Å
)
Figure 5-3. Removal rate (■) and RMS roughness (○) with respect to the varying concentration of KCl salt in the slurry solution.
Figure 5-4. AFM micrograph of the annealed zinc oxide film (scale bar is in Å).
1.0µm
RMS=31 Å
31.58
0.00
91
1.0µm
(c)
6.7
0
0.0
0
6.7
0.0
1.0µm
(d)
6.8
6
0.0
0
6.9
0.0
1.0µm
(a)
6.8
6
0.0
0
6.8
0.0
1.0µm
(b)
6.8
6
0.0
0
7.0
0.0
RMS=3Å RMS=5Å
RMS=4Å RMS=6Å
1.0µm
(c)
6.7
0
0.0
0
6.7
0.0
1.0µm
(c)
6.7
0
0.0
0
6.7
0.0
6.7
0
0.0
0
6.7
0.0
1.0µm
(d)
6.8
6
0.0
0
6.9
0.0
1.0µm
(d)
6.8
6
0.0
0
6.9
0.0
6.8
6
0.0
0
6.9
0.0
1.0µm
(a)
6.8
6
0.0
0
6.8
0.0
1.0µm
(a)
6.8
6
0.0
0
6.8
0.0
6.8
6
0.0
0
6.8
0.0
1.0µm
(b)
6.8
6
0.0
0
7.0
0.0
1.0µm
(b)
6.8
6
0.0
0
7.0
0.0
6.8
6
0.0
0
7.0
0.0
RMS=3Å RMS=5Å
RMS=4Å RMS=6Å
Figure 5-5. AFM micrographs for zinc oxide films polished with SDS slurry with SDS concentration of (a) 5 mM, (b) 10 mM, (c) 16 mM and (d) 30 mM (scale bar is in Å).
92
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
1.0µm
(c)6
.86
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
10.0
0.0
RMS=5Å RMS=4Å RMS=4Å
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
6.8
6
0.0
0
12.0
0.0
1.0µm
(c)6
.86
0.0
0
11.0
0.0
1.0µm
(c)6
.86
0.0
0
11.0
0.0
6.8
6
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
10.0
0.0
1.0µm
(b)
6.8
6
0.0
0
10.0
0.0
6.8
6
0.0
0
10.0
0.0
RMS=5Å RMS=4Å RMS=4Å
Figure 5-6. AFM micrographs for zinc oxide films polished with Triton X-100 slurry with TX-100 concentration of (a) 0.322 g/l, (b) 0.64 g/l and (c) 1.28 g/l (scale bar is in Å).
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
1.0µm
(c)
6.8
6
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
11.0
0.0
RMS=5Å RMS=5Å RMS=6Å
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
1.0µm
(c)
6.8
6
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
11.0
0.0
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
1.0µm
(a)
6.8
6
0.0
0
12.0
0.0
6.8
6
0.0
0
12.0
0.0
1.0µm
(c)
6.8
6
0.0
0
11.0
0.0
1.0µm
(c)
6.8
6
0.0
0
11.0
0.0
6.8
6
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
11.0
0.0
1.0µm
(b)
6.8
6
0.0
0
11.0
0.0
6.8
6
0.0
0
11.0
0.0
RMS=5Å RMS=5Å RMS=6Å
Figure 5-7. AFM micrographs for zinc oxide films polished with (a) 0.1M, (b) 0.2 M and (c) 0.3 M KCl salt added to the polishing slurry (scale bar is in Å).
93
400 500 600 700 8000.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Glass
Non-CMP
0.1M KCl
5mM SDS
0.322g/l TX-100
No
rmaliz
ed Inte
nsity (
a.u
.)
Wavelength (nm)
Figure 5-8. Optical transmission curves for samples polished using slurries with 0.1 M KCl, 5 mM SDS and 0.322 g/l TX-100.
94
CHAPTER 6 RAY TRACING SIMULATIONS
Introduction
Light extraction efficiency is primary reason for low overall efficiency of light
emitting devices. Although a significant increase in ηext was observed for several
reported methods such as device shaping, photonic crystals, patterned substrates,
surface roughening etc. but these methods are often accompanied by changes in the
radiation pattern, exhibit an undesirable angle dependent emission spectrum, or employ
costly or complex processing methods [85]. Application of external out-couplers provide
significant enhancement in light extraction without invoking such undesirable attributes.
Several structures as external out-couplers have been considered for improving the light
extraction such as hemispherical dome [76], pyramids [81, 110], cylinders and lenses
[73, 85, 111-113].
Spherical shapes have been known for improving external efficiency as they
subtend a large solid angle of emitted rays resulting in extraction of light previously
trapped in the substrate [73]. Madigan et al. demonstrated the effect of spherical
features on external coupling efficiency and the far-field emission pattern. The light
extraction increased by factor of 3 by attaching macro-sized spherical lens using index
matching gel on the backside of the OLED. However, use of millimeter-sized lens is not
practical for many applications due to alignment and index matching issues. Möller et al.
suggested an alternative, simpler technique to fabricate ordered-array of micro-sized
lens without the need for alignment with respect to LED device. A 10 μm diameter
microlens square array of poly-dimethyl-siloxane (PDMS) was fabricated using mold
transfer method which was attached to emission side of the glass substrate. This
95
architecture resulted in 1.5 fold increase in the light output over unlensed device.
Nakamura et al., fabricated pyramidal microlens array and studied the effect of using
high refractive index substrate in conjunction with microlens array [81]. Interestingly, the
high refractive index substrate did not show any enhancement in light extraction
whereas pyramidal microlens structure resulted in extraction efficiency enhancement by
a factor of 1.5. Greiner analyzed the microlens and micropyramid array as the external
out-couplers in an OLED device [147]. The simulation results elucidated that the
reflectance of OLED stack or back reflector is of paramount importance since the
probability of a photon to escape in first pace through various structures is only a
fraction of total escape probability. In addition, the aspect ratio of microlens or
micropyramid structure has strong influence on the out-coupling efficiency as escape
probability of higher angle photons reduces with aspect ratio. Compact packing of these
structures results in better out-coupling efficiency because any flat region increases the
probability of TIR. Greiner findings also illustrated that the exact geometric nature of the
structure influence angular distribution much strongly as compared to the out-coupling
efficiency. Using these findings he was able to demonstrate 80% extraction efficiency
from substrate for microlens, micropyramid and scattering particles with aluminum back
contact OLED device.
In this chapter, Monte-Carlo ray-tracing simulations in 3D have been performed to
understand the behavior of pyramid, microlens and cylinder arrays as external out-
couplers. A comparative study on effect of feature size and aspect ratio of pyramid, lens
and cylindrical features on output intensity and angular distribution of out-coupled
photons has been conducted using ray-tracing simulations. For the first time, corrugated
96
device structures with micron scale textures applied to substrate/TCO interface were
analyzed to simulate the controlled micro-scale topography on TCO surface. The
enhancement in light output and angular spread of out-coupled intensity from
corrugated structures were studied for microlens structure with different feature sizes
and h/d ratios. Commercially available LightTools (v6.3.0) ray-tracing simulation
software from Optical Research Associates was used for all the simulations. The details
of simulation parameters and obtained results are discussed below.
Device Simulation Parameters
Textured substrates with pyramid, microlens and cylinder arrays were simulated.
Figure 6-1 is the schematic of the device simulated with the glass as substrate, a TCO
layer and a light emitting layer. The light emitting layer of thickness, t = 1 μm and
refractive index, nactive = 2.5, 1.7 over ZnO layer as the TCO with thickness, t = 0.2 μm,
nZnO = 1.9 and optical density (O.D.) = 0.09691 was placed on a glass substrate (t = 0.5
mm, n = 1.46). The two refractive indices of active layer were considered for simulation
to understand the effect of refractive index of the active layer on the ray-dynamics. The
nactive of 1.7 represents typical refractive index of organic materials whereas nactive = 2.5
is representative of inorganic materials. The device size was 1x1 mm2 with 100%
reflecting sides. This structure represents the simplified structure of an OLED/TFEL type
of device. The values of the material refractive index and thicknesses may change in
actual device but the ray dynamics will hold true for classical ray optics regime. Two
kinds of structures were simulated (A) texture on the top-side of glass substrate which
results in enhancement in extracted mode and (B) texture on the bottom-side (TCO and
active layer side) of glass substrate for extraction of TCO/active layer mode (Figure 6-
1). For device structure A, pyramid, microlens and cylinder arrays with different base
97
diameters and aspect ratios were simulated whereas for device structure B, only
microlens was simulated. The cylinder arrays and pyramid structure were not
considered for the simulation in device structure B configuration due to deposition of
conformal film on straight wall structures is difficult and pyramid structure presents
sharp edges which can serve as site for probable shorts. The film thickness of active
layer was changed to 0.15 μm for device B simulations to match closely with the device
fabricated with Tris(8-hydroxyquinolinato)aluminium (Alq3) organic active layer with
nactive = 1.7. For comparison, the nactive value equal to 2.5 was also simulated with same
thickness of 0.15 μm. The rear surface of active layer was made reflective to replicate
the metal back contact present in a typical OLED device. The absorption of 20% was
deliberately incorporated in the properties of reflecting rear surface to account for
absorption losses in the active layer and metal back contact. The emission of the active
layer was assumed to be isotropic with 550 nm used as the emission wavelength for all
simulations. The ray-tracing simulations were performed using 100,000 rays for
measuring out-coupling using large plane detectors near front and rear side of the
device. The angular distribution of the out-coupled rays was measured using a far-field
detector centered to the device with 1 million rays used for the simulation. Both planar
and far-field detector measured total fractional power of the out-coupled rays. The
aspect ratio of pyramid structures was varied by changing the contact angle from 0-85º
whereas for microlens and cylinders, height to diameter (h/d) ratio was varied from 0 to
0.5 (0.5 being hemisphere) and 0 to 1, respectively. The base diameters (1, 3, 6, 10, 14
and 26 μm) were considered for Device A all the structures with individual texture
elements arranged in hexagonal pattern with 3 μm spacing. The side of the pyramid
98
base was calculated as 2r , considering an inscribed square in the circular base. For
Device B, microlens of diameter 6, 10 and 14 μm were simulated.
Results and Discussion
Device A: Texture at Substrate/Air Interface
Pyramid texture
Figure 6-2 (a) & (b) represents the light out-coupled from a pyramid array for
device configuration A normalized to a planar device for the active layer with refractive
index of 1.7 and 2.5, respectively. It can be seen from the figures that irrespective of the
refractive index of active layer, the enhancement in light out-coupling increases with
contact angle for a given base diameter. Extraction of substrate mode is higher for the
device with larger base diameter. Interestingly, the enhancement shows slight decrease
for structures having contact angle greater than 55º. This loss in light extraction at
higher contact angles is because of re-entry of the extracted rays from a cone into the
adjacent cone. Such re-entrant rays finally get absorbed in the different layers of the
device leading to loss in luminance intensity. Figure 6-2(c) plots the maximum obtained
enhancement with respect to the diameter of pyramids for active layer with different
refractive indices. Lower refractive index active layer demonstrates higher enhancement
due to absence of active layer mode as the refractive index of active layer is less than
the TCO layer.
Figure 6-3 (a) and (b) represent the angular distribution curves for different
diameter structures at most efficient contact angle for nactive = 1.7 and 2.5, respectively.
The angular distribution curves illustrate a deviation from Lambertian distribution profile
of a planar device. A superposition of peak for angles greater than critical angle (~43º)
99
can be seen from the curves. The observed increase in intensity at higher angle is a
result of extraction of emitted photons by the pyramidal structures at angles greater than
critical angle. The intensity in forward direction within escape cone is expected to be
reduced due to introduction of pyramid structure. However, it remains same or
demonstrates slight increase than the planar device as the reflected photons incident in
forward direction on the surface are recovered after multiple reflections.
Microlens texture
Microlens array texture demonstrates light extraction as high as ~1.87X for nactive =
1.7 and ~1.74X for nactive = 2.5. The light out-coupling efficiency increases both with h/d
ratio and base diameter of the lens as depicted in Figure 6-4 (a) and (b). As observed in
case of pyramid, the trend with respect to aspect ratio and diameter remain unaffected
by the refractive index of the active layer. The lower refractive index active layer
demonstrates higher enhancement values (see Figure 6-4 (c)) because of nactive being
lower than nZnO which results in elimination of active layer mode. A strong dependence
of light out-coupling on diameter and h/d ratio of the lens for feature size ≤ 14 μm is
clearly evident from the figure. The hemispherical lens (h/d = 0.5) demonstrates
maximum efficiency in light extraction for a given diameter. A microlens structure
reduces the extraction of photons emitted at normal incidence due to curvature of the
feature. However, the reduction in the light extraction of normal incident photons is more
than compensated by the enhancement from the extraction of higher angle photons. In
addition, even these reflected photons are directed back towards substrate by the
mirrored back side with changed directionality due to non-parallelepiped geometry of
the substrate. The multiple reflections and changed directionality upon reflection
improves the probability of such photons to be extracted. Figure 6-5 represents the
100
angular distribution of extracted mode photons for the most efficient h/d ratio structures
for each diameter. The angular spread of photons elucidates the extraction of higher
angle photons although the enhancement is observed for both forward and at higher
angles. The observed improvement of the extracted light intensity is related to the
simultaneous increase in photon extraction due to back reflector and extraction of
higher angle photons coupled with the focusing effect of lens structure.
Cylindrical texture
The simulation of cylindrical structures divulges a different behavior as compared
to the pyramid and microlens structures. The reason for such disparity can be explained
by the difference in geometry of these structures. The cylindrical structures are straight
wall features with flat top unlike inclined or spherical profile of the pyramid and
microlens, respectively with no flat region. In case of cylinders, the introduction of
vertical wall leads to improved light out-coupling however, efficacy of the cylindrical
structure is limited to low values of h/d ratio (< ~0.4) as evident from Figure 6-6. The
flattening observed for enhancement curves for the cylindrical features at low h/d can be
explained by the losses due to re-entry of extracted photons from one cylinder into
adjacent cylinder. Further, the cylindrical arrays with diameter greater than 10 μm lead
to no significant increase in light out-coupling as compared to one with 10 μm diameter.
This behavior is a result of steep increase in the area of top surface of cylinder as
compared to surface area of the side walls which undermines the enhancement in light
extraction. The simulation results show enhancements as high as ~1.68 and 1.57 folds
for nactive = 1.7 and 2.5, respectively for base diameter of 14 μm.
The angular spread of extracted photons is depicted in Figure 6-7. The light
extraction in direction normal to the surface show no significant improvement,
101
corroborating the reason of reduced efficacy of cylindrical structure with increase in
diameter of the features. The formation of butterfly wing like features between 45-90º
illustrates the redirection of higher angle photons into escape cone. However, the
increase is less drastic and the intensity distribution is less spread as observed for
pyramid and microlens arrays. The observed low improvement in light extraction in
cylindrical structure as compared to the pyramid and microlens can be due to
parallelepiped geometry of the features which limits the ability of texture to randomize
the directionality of reflected rays. Hence, the probability of photon extraction is not
significantly enhanced even after multiple reflections from back-side reflector.
Device B: Texture at Substrate/TCO Interface
Only 4% and 10% of generated photons are extracted finally from the device with
nactive = 2.5 and 1.7, respectively. Simple calculations show that the active layer mode
account for trapping of ~ 82% of generated photons whereas TCO layer mode accounts
for 8% of the 18% entering TCO layer for nactive = 2.5, nTCO = 1.9 and nsubstrate = 1.46
system. Similarly, in case of nactive = 1.7, since active layer mode is absent, the
combined TCO/active layer mode amounts for trapping of roughly 76% of generated
photons in TCO layer without taking into account the high absorption losses in organic
layer. These calculations illustrate a clear need to extract the TCO and active layer
modes which account for significant amount of generated photons. Few reports have
been published on attempts to reduce TCO/active layer modes because of the
increased probability of electrical shorts due to intentionally introduced roughness at
active layer interface [82-84, 109, 123]. This restricts the feature dimensions much less
than film thicknesses. In this section, an attempt has been made to understand the ray
dynamics when microlens textures with dimensions much greater than the emission
102
wavelength are applied on the substrate at active layer side. In order to understand the
ray dynamics in such system, simulations were carried-out with varying diameter and
h/d ratios of microlens texture applied on the active layer side. Due to the spherical
profile of a lens structure, the shape is ideal from fabrication viewpoint and has less
probability of possible shorts due to sharp features or edges. Further, the dimensions
considered for these structures are in micron range, making microlens structures locally
flat compared to the scale of active layer thicknesses. It should be noted here that as
the film thicknesses (TCO and active layer) are much less than the height of microlens
features, a corrugated geometry of the device is obtained assuming conformal nature of
the deposited films.
The simulation results on the enhancement in light extraction are presented by
Figure 6-8 (a) and (b) for the microlens texture at substrate/TCO interface (device B
configuration) whereas (c) presents the maximum fractional powers obtained for
different diameters and the two refractive indices of active layer considered here. In
order to understand the effect of feature dimensions alone on the light extraction
efficiency, the simulations were carried out with fixed number of rays (100,000 rays)
eliminating the effect of increased surface area. The enhancement demonstrates a
steep increase with increasing h/d ratio with maximum enhancement of 1.45 and 1.55
for nactive = 1.7 and 2.5, respectively. A significant portion of generated photons are
waveguided within active layer due to high refractive index of the layer as compared to
TCO (nZnO = 1.9). This active layer mode is absent in case of nactive = 1.7, as the
refractive index is less than nZnO. The application of texture on the active layer side
distorts the parallelepiped geometry of active layer leading to extraction of the active
103
layer mode. Additional photons extracted results in higher observed relative
enhancement in higher refractive index material as compared to lower index material.
However, the absolute extracted power of the photons in nactive = 2.5 are still less as
compared to the nactive = 1.7 (see Figure 6-8 (c)). Additionally, although smaller diameter
structure shows better enhancement but the difference in enhancement values is very
small. This small difference in enhancement can be attributed to the fact that as feature
size increases, the radius of curvature of the feature reduces resulting in reduced
efficacy of the texture.
Figure 6-9 depicts the angular spread of photons for nactive = 1.7 and 2.5 with 6 μm
microlens base diameter at different microlens heights. The angular spread illustrates
Lambertian spread as the measured output is extracted mode from glass substrate
without any texturing. However, intensity increases with height (or h/d ratio) of the
microlens feature as observed from the enhancement curves for a given diameter. The
increase in intensity in all direction is attributed to the reduction in TCO/active layer
modes and extraction of higher angle photons leading to higher fraction of generated
photons entering substrate. Simulation results also illustrate a reduction in
enhancement after h/d of 0.3 for nactive = 1.7 and 0.4 for nactive = 1.7 as observed from
both enhancement and angular distribution plots. The reason for this decrease can be
understood from the far-field angular distribution of photons represented by Figure 6-10.
In the figure, substrate has been immersed in a medium of refractive index same as that
of substrate. This leads to measurement of only the photon out-coupled from the TCO
layer. With special emphasis on the forward direction in the figure, it can be inferred that
the enhancement is coming from increased extraction in forward direction. However, as
104
h/d (or height) reaches close to hemispherical profile the distribution becomes well
spread and more off-normal photons are extracted as compared to forward direction
photons. The increase in forward for h/d less than certain value and off-forward for h/d
greater than that value can be explained by considering the fact that a microlens in the
device B configuration is also acting like a concave mirror. Because there is no texture
at glass/air interface the out-coupling of photons from this interface is still restricted by
the escape cone (θc ~ 43) and has Lambertian distribution. Hence, the extracted
angular spread is a superposition of Lambertian distribution of glass/air interface over
the non-Lambertian spread of photons entering the substrate thereby preventing the
higher angle photons to escape reducing overall efficiency for high aspect ratio
structures.
Conclusion
Light extraction efficiencies of various external light out-couplers such as pyramid,
microlens and cylinder arrays were analyzed. The use of such external light out-
couplers in conjunction with back-side mirror results in enhancement of light due to
extraction of higher angle photons and change in directionality of the reflected photons
outside the escape cone. In case of pyramids, the intensity of extracted light increased
with contact angle less than ~55º after which the structure demonstrates slight decrease
due to re-entry of the extracted photons back into device. The ray simulations
elucidated similar behavior of the microlens arrays. Microlens arrays demonstrate
higher light extraction efficiency and circular symmetry in angular distribution of
extracted photons. In contrast, efficacy of the cylinder arrays is limited to the lower h/d
ratios and the light extraction efficiency becomes independent of increase in feature
diameter after 10 μm. In conclusion, microlens and pyramid arrays were found to be
105
more efficient structures as compared to cylindrical arrays. The trends remain
independent of refractive index of the active layer as expected because these features
only affect the substrate mode.
Ray dynamics slightly changes when the microlens structure is applied at the
substrate/TCO interface in the device B architecture. Application of microlens texture at
this interface changes the geometry of the TCO and active layer to a corrugated
geometry. The deviation from parallelepiped geometry leads to increased extraction of
TCO/active mode in case with nactive = 1.7 and higher extraction of both TCO and active
layer mode in case of nactive = 2.5. The enhancement increases steeply up to 0.3 and
0.4 h/d ratio for nactive = 1.7 and nactive = 2.5, respectively. Thereafter, decrease in
extraction efficiency was observed in simulated results. The increase in extraction of
normal angle photons from TCO layer up to 0.3 and 0.4 h/d ratio for nactive = 1.7 and
nactive = 2.5, respectively was found to be responsible for highest enhancement at these
h/d ratios when fraction of photons exiting TCO layer were measured alone. However,
at higher h/d ratios the extraction of higher angle photons increased but due to
Lambertian profile of extracted light at glass/air interface, the combined effect of the
TCO extracted and glass extracted mode profiles resulted in the observed decrease in
the overall enhancement.
106
Glass Glass Glass
Glass
Device A
Device B
- Glass
- ZnO
- Active layer
Glass Glass GlassGlassGlassGlass GlassGlassGlass GlassGlass
GlassGlassGlass
Device A
Device B
- Glass
- ZnO
- Active layer
Figure 6-1. Schematic depicting different device structures studied using ray-tracing simulations.
107
0 10 20 30 40 50 60 70 80 901.0
1.2
1.4
1.6
1.8E
nhancem
ent
Contact Angle
1 um 3 um 6 um
10 um 14 um 26 umn
active = 1.7
(a)
0 10 20 30 40 50 60 70 80 90
1.0
1.2
1.4
1.6
1.8(b)
nactive
= 2.5 1 um 3 um 6 um
10 um 14 um 26 um
Enhancem
ent
Contact Angle
0 5 10 15 20 25 30
0.1
0.2
0.3
Fra
ctional P
ow
er
Radius (m)
nactive
= 1.7
nactive
= 2.5
(c)
Figure 6-2. Enhancement in out-coupled rays from a pyramid array textured substrate
(Device A) for (a) nactive = 1.7 and (b) nactive = 2.5 whereas (c) represents the maximum simulated fractional powers obtained for the different base diameters of pyramid arrays and refractive indices of active layer.
108
0
45
90
135
180
Planar 1 um 3 um 6 um
10 um 14 um 26 um
(a) nactive
= 1.7
0
45
90
135
180
nactive
= 2.5
Planar 1 um 3 um 6 um
10 um 14 um 26 um
(b)
Figure 6-3. Angular distribution of the light out-coupled from device A with pyramid texture for active layer refractive index (a) nactive = 1.7 and (b) nactive = 2.5.
109
0.0 0.1 0.2 0.3 0.4 0.51.0
1.2
1.4
1.6
1.8
2.0E
nh
an
ce
me
nt
h/d
1 um 3 um 6 um
10 um 14 um 26 umn
active = 1.7
(a)
0.0 0.1 0.2 0.3 0.4 0.5
1.0
1.2
1.4
1.6
1.8n
active = 2.5
Enhancem
ent
h/d
1 um 3 um 6 um
10 um 14 um 26 um
(b)
0 5 10 15 20 25 30
0.1
0.2
0.3
Fra
ctional P
ow
er
Radius (m)
nactive
= 1.7
nactive
= 2.5
(c)
Figure 6-4. Enhancement in light out-coupling for a microlens array with device A configuration represented for (a) nactive = 1.7, (b) nactive = 2.5. The maximum simulated fractional powers at different base diameters for the two refractive indices of active layer are depicted by (c).
110
0
45
90
135
180
Planar 1 um 3 um 6 um
10 um 14 um 26 um
(a) nactive
= 1.7
0
45
90
135
180
Planar 1 um 3 um 6 um
10 um 14 um 26 um
nactive
= 2.5(b)
Figure 6-5. Angular distribution of the total light out-coupled from device A with microlens texture for (a) nactive = 1.7 and (b) nactive = 2.5.
111
0.0 0.2 0.4 0.6 0.8 1.01.0
1.2
1.4
1.6
1.8 nactive
= 1.7E
nhancem
ent
h/d
1 um 3 um 6 um
10 um 14 um 26 um
(a)
0.0 0.2 0.4 0.6 0.8 1.0
1.0
1.2
1.4
1.6
(b)
Enhancem
ent
h/d
1 um 3 um 6 um
10 um 14 um 26 um
nactive
= 2.5
0 5 10 15 20 25 30
0.1
0.2
0.3
Fra
ctional P
ow
er
Radius (m)
nactive
= 1.7
nactive
= 2.5
(c)
Figure 6-6. (a) and (b) depict the light out-coupled normalized to a planar device for a cylinder type texture at the substrate/air interface (Device A) for nactive = 1.7 and 2.5, respectively while (c) represents the maximum simulated fractional powers for different diameters and active layer refractive indices.
112
0
45
90
135
180
Planar 1 um 3 um 6 um
10 um 14 um 26 um
nactive
= 1.7(a)
0
45
90
135
180
Planar 1 um 3 um 6 um
10 um 14 um 26 um
(b) nactive
= 2.5
Figure 6-7. Angular spread of the light out-coupled from device A with cylindrical texture for active layer refractive index of (a) 1.7 and (b) 2.5.
113
0.0 0.1 0.2 0.3 0.4 0.51.0
1.2
1.4
1.6E
nhancem
ent
h/d
6 um 10 um 14 um(a)
nactive
= 1.7
0.0 0.1 0.2 0.3 0.4 0.51.0
1.2
1.4
1.6
Enhancem
ent
h/d
6 um 10 um 14 um(b)
nactive
= 2.5
0 5 10 150.0
0.1
0.2
0.3
Fra
ctional P
ow
er
Radius (m)
nactive
= 1.7
nactive
= 2.5
(c)
Figure 6-8. Light extraction enhancement from a microlens texture at the substrate/TCO
interface (Device B) normalized to the planar device is represented by (a) for nactive = 1.7 and (b) for nactive = 2.5 while (c) represents the maximum simulated fractional powers for different diameters and refractive indices of active layer.
114
0
45
90
135
180
Planar 0.6 um 1.2 um
1.8 um 2.4 um 3 um
nactive
= 1.7(a)
0
45
90
135
180
Planar 0.6 um 1.2 um
1.8 um 2.4 um 3 um
nactive
= 2.5(b)
Figure 6-9. Far-field angular spread of the light out-coupled from device B at different
microlens heights for 6 μm base diameter and nactive of (a) 1.7 and (b) 2.5.
0
45
90
135
180
Planar 0.6 um 1.2 um
1.8 um 2.4 um 3 um
Figure 6-10. Far field angular distribution of photons out-coupled from the TCO layer at different microlens heights for nactive = 1.7 and base diameter 6 μm.
115
CHAPTER 7 EXPERIMENTAL PROCEDURES FOR FABRICATION OF TEXTURES
Experimental Procedure
Substrate Preparation
Devices were fabricated on 1x1 inch2 glass substrates (Corning 2497). As-
received glass substrates were cleaned in acetone, methanol, deionized water in a
sonicator. For wet etching, not only the etch selectivity between etch mask material and
the etched material but the interface between the etch mask and the etched material is
also important. To remove organic residue, piranha clean with 2:1 sulfuric acid and
hydrogen peroxide was carried out immediately before deposition of etch mask layer.
Samples were kept in piranha solution for 20-30 min and were rinsed thoroughly with DI
water and blown dried with nitrogen. The substrates were placed in nitrogen filled oven
at 120ºC for 15 min to remove the adsorbed water from the surface.
Structure Fabrication
Three different types of features (pyramid/cones, cylinder and microlens) were
fabricated on the substrate with base diameters of 6, 10 and 14 μm. Figure 7-1 shows
the mask design along with the dimensions and lattice spacing. Figure 7-2 is the
process flow schematic for different features fabricated. The description of process
flows has been divided in to three sub-sections based on structure type as described
below.
Cone/pyramid texture
For any etching process the etch selectivity between etch mask material and the
etched material is an important parameter to be considered when selecting a material
for etch mask. Amorphous silicon (a-Si) demonstrates infinite etch selectivity with
116
respect to glass in buffered oxide etchant (BOE) commonly used for glass etching. For
this reason, a-Si was chosen as a etch mask for creating cone/pyramid structures. The
cleaned substrates were deposited with amorphous silicon after 30 s Ar plasma pre-
clean in the Kurt J Lesker CMS-18 Multi-Source RF sputter tool. A ~3000 Å thick a-Si
film was deposited at room temperature at the power setting of 200 W and 5 mTorr
chamber pressure using argon as the sputter gas. Silicon target was pre-sputtered for 5
min to remove any oxide or impurity layer on the target surface. The substrates with
sputtered a-Si films were annealed at 400ºC for 6.5 hours in nitrogen ambient at 100
mTorr pressure. The annealed substrates were primed with HMDS vapors in a YES
oven. Immediately after priming, the substrates were patterned using standard UV
photolithography technique. A ~1.2 μm thick S1813 photoresist was spun on the
substrates followed by exposing to 365 nm Hg i-line UV radiation using a Karl Suss
MA6 Aligner. The photoresist was soft-baked at 105ºC for 30 minutes prior to UV
exposure and was post-baked at 120ºC for about 30 minutes. Standard AZ300 MIF
developing solution was used for 1 min to develop the pattern. Patterned photoresist
(PR) coated substrates were descummed in a reactive ion etcher (RIE/ICP, Unaxis
SLR) under oxygen plasma for 20 s. The RIE conditions used for descumming process
were 100/600 W RF/ICP power with 20 sccm O2 flow at chamber pressure of 5 mTorr.
To transfer pattern from PR to the a-Si layer, standard SF6 chemistry was used in a STS
deep reactive ion etcher (DRIE) with 800 W RF power and 12 W DC platen power.
Cone/pyramid structures on glass substrates were etched using a 6:1 buffered oxide
etchant (BOE) solution for 6 and 10 μm size structures whereas for 14 μm diameter
structure 6:1 BOE solution was diluted by adding 15 vol % 30-44% hydrochloric acid.
117
End-point for BOE etch was determined by first visual sign of Si mask curling, when
substrate was retracted from the BOE solution. To ensure complete removal of a-Si
mask, DRIE step was repeated before final clean with acetone, methanol and DI water.
Microlens texture
Cone/pyramid structures fabricated according to the method described above
were used to create microlens profile using chemical mechanical fabrication (CMF)
technique. Substrate with cone structures fabricated as above were polished using
standard silica slurry on a soft politex pad. Details of the CMP process are described in
ref [173]. Post-polish cleaning was carried out by buffing with DI water followed by
acetone, methanol and DI water rinse in an ultrasonicator.
Cylindrical texture
Dry etching process was used to attain high aspect ratio vertical wall cylindrical
features. Glass substrates were coated with low frequency PECVD silica deposited at
60 W (RF) and 300ºC by flowing 7.2 sccm SiH4 mixed with 1279 sccm N2O and 352
sccm N2. For etching PECVD silica, chromium (Cr) metal was used as the etch mask.
The S1813 PR was spin-coated at 5000 rpm for a nominal thickness of ~1.2 μm. To
pattern Cr mask layer, lift-off process was used by image reversing the positive S1813
photoresist after soft bake and exposing the PR layer to UV radiation in presence of a
photomask. For image reversal, PR was exposed to ammonia (NH3) gas in a YES
vacuum oven at 90ºC temperature. The NH3 reacts with acid in the exposed resist
rendering it insoluble in the developer. The proceeding flood exposure causes acid to
form in the previously unexposed areas allowing them to be removed in development
step, leaving behind negative image of the first exposure. A 3500 Å of Cr metal was
118
deposited using DC sputter after the lithography process. Metal deposited substrates
were sonicated in acetone to remove PR for Cr lift-off.
Dry etching of PECVD silica was performed in an Unaxis reactive ion etcher with
inductively coupled plasma module using CHF3 recipe. A 600 W ICP power with 30 W
RF power was applied with 25 sccm of CHF3 and 3 sccm of O2 gas under 5mTorr
pressure. An etch selectivity of 33:1 was obtained between PECVD silica and Cr metal
mask for the given optimized recipe. After etching of the cylindrical structures of desired
heights controlled by changing the etch time, the Cr metal mask was removed using
Transgene chromium etchant. Substrates were finally cleaned with acetone, methanol
and DI water and blow dried using N2 gas.
The topography of the fabricated structures was characterized using atomic force
microscope (AFM, Veeco Dimension 3100) and field emission scanning electron
microscope (JEOL SEM 6335F).
Results and Discussion
Cone/Pyramid Structure
Various masking strategies have been published on micropatterning of glass.
Metal masks such as Cr/Au [174-176] or Cr/Cu [177, 178] have been reported. LPCVD
poly-Si [177], PECVD a-Si [179] or even bulk Si [180] are also commonly used etch
masks. Amorphous Si is resistant to hydrofluoric (HF) acid and offers an infinite etch
selectivity with glass (SiO2) etching in HF [178, 179]. However, the a-Si mask/glass
interface and internal stresses in a-Si layer are two very important parameters that
determine the performance of the silicon masking layer. Poor adhesion between the two
surfaces and internal stress in masking layer can lead to premature peeling of mask in
the etchant solution. To create a good interface between silicon mask and glass, after
119
standard acetone, methanol and DI water rinse in an ultrasonicator, the glass substrates
were further stripped of any organic residue using piranha solution (a mixture of 2:1
sulfuric acid and hydrogen peroxide) for 20-30 min and rinsed in DI water thoroughly.
Substrates were subjected to 400ºC for 6.5 hours to reduce the internal stress in the
deposited silicon layer on glass.
Figure 7-3 represents the SEM micrographs at 45º tilt angle for cone/pyramid
structures fabricated. SEM images reveal that due to isotropic nature of wet etching
process, the cones/pyramids have concave slant walls. The low magnification SEM
images (not shown here) illustrate highly smooth etched glass surface with very low
density of wet etching related defects such as notching, pinholes etc. These
observations in conjunction with observed high aspect ratio demonstrate the good
interface between Si mask and glass and optimum internal stress in a-Si layer required
to withstand long etch times. AFM was performed to measure the height of fabricated
structures. Figure 7-4 (a) and (b) are the AFM topographs of cones/pyramid structures
with base diameter of 6 and 10 μm. As maximum h/d ratio achievable is 1:2 due to
isotropic nature of wet etching process. The obtained h/d ratio in this work was ~1:2.6
for both 6 and 10 μm diameter structures. The glass etch rates of ~ 0.6 μm/min were
observed for 6 and 10 μm whereas for 14μm, high etch rates ~ 4.5 μm/min were
obtained. The addition of HCl in the BOE solution dissolves the insoluble oxides of alkali
metals present in glass acting as micro-masks. This helps in accelerating the glass
etching and obtaining smooth surface after the etch process.
Microlens Structure
Microlens structures were created by polishing the cone/pyramid structures. The
details of the process are describe in ref [173]. Because the microlens feature is created
120
from a cone/pyramid structure, it limits the maximum achievable aspect ratio according
to the Figure 7-5. The maximum height of microlens is calculated according to equation
7-1 and is tabulated in Table 7-1.
xHxH
xh 22 (7-1)
where, h represents the height of the segment (microlens) of a sphere, H is the height of
square base pyramid assuming the pyramid has vertex angle of 45º and square base
with side x (x = r√2, r is radius of pattern). The AFM and SEM micrographs of fabricated
structures are given in Figure 7-6 and Figure 7-7. Figure 7-6 also shows a line profile
along the center of lens feature extracted from AFM data. The rounding of surface after
polishing can be observed from the curves. It should be noted that the polishing time to
reach the maximum achievable lens height varies with the base diameter of pyramid
structure. The variability of smoothening rates was observed to be high which is
attributed to the variation in local pressure with contact area of the feature (higher
pressures for smaller features). For this reason a good control over polishing is difficult.
In addition to polishing variability, the concave geometry of fabricated cone/pyramid
structure also leads to deviation of obtained microlens h/d ratios (approximately 0.21,
0.17 and 0.21 for 6, 10, and 14 μm) from the calculated height values. SEM images in
Figure 7-7 are obtained at 45º tilt angle. The images illustrate rounding of the pyramid
tip confirming the AFM results.
Cylindrical Structure
Cylindrical structures were created using dry etch process as creation of vertical
wall features necessitates anisotropic etching. The Corning 2947 glass slide used in this
study is a soda-lime glass which has various metal oxides other than silica such as
121
Na2O, MgO, CaO and Al2O3. These metals do not form gaseous products in a dry etch
process which makes the dry etching of glass to be physical in nature. The etch
selectivity between masking layer and the glass is severely impeded by the physical
nature of glass etching. To avoid such complications, 10 μm thick PECVD silica layer
was deposited on the glass substrate. Chromium metal used as etch mask for the dry
etching of PECVD silica presented an etch selectivity (oxide:Cr) of ~33:1 using CHF3
plasma chemistry at 30/600 RF/ICP powers. The etch selectivities and etch rates for
silica obtained by variation of RIE process variables have been tabulated in Table 7-2.
Figure 7-8 represents the SEM images of fabricated cylinder array of 14 μm base
diameter. The Figure 7-8 (a) reveals the straight wall structure of the cylinder with height
of cylinder approximately 5.38 μm. The top view of the cylindrical pillar is presented in
Figure 7-8 (b). The figure illustrates a very close match between the mask and
fabricated pillar diameter (see annotation). However, the side walls are highly rough
which is due to striations in the photoresist layer and formation of fluorinated amorphous
carbon residue - a side product of CHF3 etch.
Conclusion
Cone/pyramid arrays with aspect ratio as high as 1:2.6 were fabricated on the
Corning 2947 glass using sputtered a-Si mask and 6:1 buffered HF solution. The etch
rates of 0.6 μm/min and 4.5 μm/min were observed for the Corning 2947 soda-lime
glass when etched with 6:1 BOE and diluted BOE with15 vol % 30-44% HCl,
respectively. Pre-clean of glass substrate was found to be of utmost importance in the
wet etching process for better adhesion between glass and a-Si mask. The release of
internal stress in the masking layer after optimum post-annealing reduces pre-mature
failing of the etch mask. Smoothening process to create spherical profile from pyramid
122
structures was tested for higher aspect ratio structures using CMF process. As
expected features with smaller diameters present poor process control as compared
features with larger diameters. Microlens arrays with aspect ratio as high as 1:6 were
fabricated using the CMF process. Cylindrical structures with aspect ratio up to 1.7:1
were fabricated using CHF3 dry etch recipe and Cr metal mask.
123
a
r
r = 3, 5 μm and a = 3 μm
r = 7 μm and a = 4 μm
a
r
r = 3, 5 μm and a = 3 μm
r = 7 μm and a = 4 μm
Figure 7-1. Schematic illustrating the mask designs.
Glass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
(a) (b)
Glass
Glass
Glass
Glass
Glass
Glass
GlassGlass
GlassGlass
GlassGlass
GlassGlass
Glass
Glass
Glass
Glass
Glass
Glass
Glass
GlassGlass
GlassGlass
GlassGlass
GlassGlass
GlassGlass
GlassGlass
GlassGlass
(a) (b)
Figure 7-2. Process flows for fabrication of (a) cone/pyramid and microlens and (b) cylindrical arrays.
124
Figure 7-3. SEM micrographs of cone/pyramid texture of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter (at tilt angle of 45º).
4.0
7 µ
m
0.0
0 µ
m
4.0
μ
m
(b)
2.5
8 µ
m
0.0
0 µ
m
2.6
μ
m4
.0
7 µ
m
0.0
0 µ
m
4.0
μ
m
(b)
4.0
7 µ
m
0.0
0 µ
m
4.0
μ
m4.07 µ
m
0.0
0 µ
m
4.0
μ
m
(b)
2.5
8 µ
m
0.0
0 µ
m
2.6
μ
m2.58 µ
m
0.0
0 µ
m
2.6
μ
m
Figure 7-4. 2D/3D AFM topographs and line profile of cone/pyramid of (a) 6 μm and (b) 10 μm base diameter.
(a) (b) (c)
125
xh
R
H
Side view
2x
r
Top view
xh
R
H
Side view
xh
R
H
xh
R
H
Side view
2x
r
Top view
2x
r
Top view Figure 7-5. Schematic of microlens height derived from pyramid of height H.
10µm
10
µm
10µm
10µ
m
1.38
0.00
(a)
(b)
4.47
252015105
5
4
3
2
1
X[µm]
Hei
ght[
µm
]
987654321
5
4
3
2
1
X[µm]
Heig
ht[
µm
]
10987654321
5
4
3
2
1
X[µm]
Hei
gh
t[µ
m]
(c)
1.95
0.00
10µm
10
µm
10µm
10µ
m
1.38
0.00
(a)
(b)
4.47
252015105
5
4
3
2
1
X[µm]
Hei
ght[
µm
]
987654321
5
4
3
2
1
X[µm]
Heig
ht[
µm
]
10987654321
5
4
3
2
1
X[µm]
Hei
gh
t[µ
m]
(c)
1.95
0.00
Figure 7-6. AFM micrographs and line profile of microlens structure after CMP of the
cone/pyramid structure of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter.
126
Figure 7-7. SEM micrographs of microlens features with base diameter (a) 6 μm, (b) 10 μm and (c) 14 μm (at 45º tilt).
Figure 7-8. SEM images of fabricated cylindrical structure of 14 μm base diameter at different magnifications, (a) 2700 X, at 45º tilt and (b) 6000 X (top view).
(a) (b)
(a) (b) (c) (a)
127
Table 7-1. Maximum microlens height from square base pyramid of height H
2r (μm) H (μm) 2x (μm) h (μm)
6 2.12 4.24 1.00
10 3.53 7.07 1.67
14 4.95 9.9 2.34
Table 7-2. RIE etch recipe for silica etching
RF1 (W)
ICP Power
(W)
Gas mixture
Gas flow (sccm)
Chamber (mTorr)
Oxide removal (nm/min)
Cr removal (nm/min)
Selectivity (Oxide:Cr)
50 600 SF6/Ar 20/5 5 115.5 7.6 15
50 600 CHF3/O2 25/3 5 139 4.8 29
50 600 CHF3/O2 25/10 5 109.5 15.7 7
30 600 CHF3/O2 25/3 5 112 3.4 33
128
CHAPTER 8 MICRO-TEXTURED TRANSPARENT CONDUCTING OXIDE FOR ENHANCED LIGHT
EXTRACTION EFFICIENCY
Introduction
Only 4% and 10% of the generated photons are extracted finally from the device
with nactive = 2.5 (inorganic active layer materials) and 1.7 (organic active layer
materials), respectively. Simple calculations show that the active layer mode accounts
for trapping of ~ 82% of generated photons for nactive = 2.5 and nTCO = 1.9 system
whereas TCO layer mode again accounts for 8% of the 18% photons entered TCO layer
in same system with nsubstrate = 1.46. In case of OLEDs, the refractive index of the active
layer is roughly 1.7 which is less than or close to that of TCO layer. This eliminates the
guided mode in the active layer. The TCO or combined TCO/active layer mode still
amounts for trapping of roughly 76% generated photons in TCO layer. These
calculations illustrate a clear need to extract the TCO and active layer modes which
account for significant amount of generated photons. Few reports have been published
on attempts of reducing TCO/active layer modes because of the increased probability of
electrical shorts due to intentionally introduced roughness at active layer interface [82-
84, 109, 123]. In order to avoid shorts due to light extraction structures, the feature
dimensions are restricted to much less than film thicknesses. At these scales
(wavelength or sub-wavelength range), the enhancement in light extraction is often
accompanied by undesirable change in the radiation pattern, exhibit angle dependent
emission, or employ costly or complex processing methods [83-85].
In this chapter, an attempt has been made to understand the ray dynamics when
microlens textures with dimensions much greater than the emission wavelength are
applied on the substrate at active layer side. In order to understand the ray dynamics in
129
such system, ray-tracing simulations have been carried-out with varying diameter and
h/d ratios of the microlens texture applied on the active layer side. Due to spherical
profile of a lens structure, the shape is ideal from fabrication viewpoint and has less
probability of possible shorts due to sharp features or edges. Further, the dimensions
considered for these structures are in micron range, making microlens structures locally
flat compared to the scale of active layer thicknesses. In addition, results on chemical
mechanical polishing of TCO layer in order to reduce the nano-scale roughness have
been presented in this chapter. Alq3 layer with aluminum back metal were evaporated
on the fabricated microlens textured substrates to experimentally evaluate the efficacy
of microlens structure and corroborate the simulation results.
Experimental Procedure
Substrate Preparation
Devices were fabricated on 1x1 inch2 glass substrates (Corning 2497). As-
received glass substrates were cleaned in acetone, methanol, deionized water in a
sonicator and 30 min in piranha solution followed by deionized water rinse and nitrogen
blow drying. For creating textures, micro-pyramid patterns were fabricated on the
substrates using standard lithography with S1813 photoresist to define pattern on the Si
etch mask. The cleaned substrates were first deposited with amorphous silicon after 30
s Ar plasma pre-clean in the Kurt J Lesker CMS-18 Multi-Source RF sputter tool. A
~3000 Å thick a-Si film was deposited at room temperature at the power setting of 200
W and 5 mTorr chamber pressure using argon as the sputter gas. Silicon target was
pre-sputtered for 5 min to remove any oxide or impurity layer on the target surface. The
substrates with a-Si were annealed at 400ºC for 6.5 hours. The annealed substrates
were primed with HMDS vapors in a YES oven. Immediately after priming, the
130
substrates were patterned with ~1.2 μm thick S1813 photoresist spun on the substrate.
The photoresist was exposed to 365 nm Hg i-line UV radiation using a Karl Suss MA6
Aligner to create the pattern. The photoresist was soft-baked at 112ºC for 2 minutes
prior to UV exposure and was post-baked at 125ºC for 2 minutes on CEE hotplate.
Standard AZ300 MIF developing solution was used for 1 min to develop the pattern.
Patterned photoresist (PR) coated substrates were descummed in a reactive ion etcher
(RIE/ICP, Unaxis SLR) under oxygen plasma for 20 s. The RIE conditions used for
descumming process were 100/600 W RF/ICP power with 20 sccm O2 flow at chamber
pressure of 5 mTorr. To transfer pattern from PR to the a-Si layer, SF6 chemistry was
used in a STS deep reactive ion etcher (DRIE) with 800 W RF power and 12 W DC
platen power. Pyramid structures on glass substrates were made by etching glass with
a 6:1 buffered oxide etchant (BOE) solution or 15% HCl/BOE solution. The patterned
substrate with micro-pyramidal array were polished using standard silica slurry on a soft
politex pad. Details of the CMP process are described elsewhere [173]. Post-polish
cleaning was carried out by buffing with DI water followed by acetone, methanol and DI
water rinse in an ultrasonicator. Immediately before deposition of transparent
conducting oxide layer of Al doped zinc oxide, a 30 minute piranha clean with 2:1
sulfuric acid and hydrogen peroxide was carried out followed by rinse with DI water and
nitrogen blow drying. The Al doped zinc oxide was deposited on the cleaned substrate
using a RF sputter tool, details of which are described in next section.
TCO Fabrication
Al doped ZnO (AZO) films were deposited using Kurt J Lesker CMS-18 Multi-
Source RF sputter tool. For the study, all substrates were coated with 1500 Å thick film
of Al:ZnO on the patterned side of the substrate. The RF power used was 150 W
131
whereas working pressure of 10 mTorr with target to substrate distance 60 mm and
substrate temperature of 255ºC was found to be optimum with reasonably good
conductivity of the AZO film with high optical transmission. The Al doped ZnO
depositions were carried out in pure argon atmosphere. The as-deposited AZO films
were polished using silica slurry with 0.5 wt% secondary alkyl sulfate (SAS) surfactant
at pH ~3. The CMP down pressure was kept constant at ~3.5 psi with linear velocity
reduced to ~4.8 m/min and slurry flow rate fixed to 70 ml/min. Post-CMP cleaning was
performed using pH 10.6 (adjusted using KOH) deionized (DI) water on a buffing pad for
1.5 minutes followed by ultrasonication in acetone, methanol and DI water.
Resistivity of the deposited AZO thin films was characterized using Pro4 four
probe resistivity system (Lucas Signatone Corp.) with Keithley 2601 source meter.
Optical transmission was measured using Ocean optics HR4000 spectrophotometer.
Surface morphology of the deposited films was investigated by Field Emission Scanning
Electron Microscope (FESEM, JEOL JSM6330F).
Device Fabrication
A 1500 Å thick tris(8-hydroxyquinolinato)aluminium (Alq3) layer was evaporated on
to the substrates on the patterned side after sonicating samples in acetone and
isopropanol (IPA) for 15 minutes each. To form the backside metal reflector, 600 Å
aluminum metal was deposited on top of Alq3 layer. The Alq3 and Al layers were
deposited through a shadow mask to define the active layer area over patterned portion
of the substrate. Three different sets of devices were fabricated viz., Alq3 and Al layer
deposited on (a) patterned glass, (b) patterned glass with AZO layer and (c) patterned
glass with polished AZO layer with unpatterned glass used as a reference with similar
layer stack.
132
Photoluminescence measurements were carried out in Perkin Elmer LS-55
fluorescence spectrometer using a front surface solid sample measurement assembly.
The measurements were made in reflective mode due to geometry of the sample
assembly and presence of Al back reflector.
Ray-tracing Simulations Parameters
Textured substrates with microlens arrays were simulated for base diameter 6, 10
and 14 μm for h/d ratio varied between 0.1-0.5. The device architecture simulated was
kept close to the experimentally fabricated device with 0.15 μm thick active light emitting
layer (Alq3) having refractive index, nactive = 1.7 over the TCO layer. The ZnO as the
TCO layer with thickness, t = 0.2 μm, nZnO = 1.9 and optical density (O.D.) = 0.09691
was placed on a glass substrate (t ~ 8.4 μm, n = 1.46) to simulate the complete stack.
Due to the software limitations, instead of having array of texture elements arranged in
hexagonal pattern, the simulations were simplified by considering only one texture
element with square field area equal to that of the hexagonal unit cell area in actual
texture. All vertical sides of each layer were made 100% reflecting whereas bottom
surface of Alq3 layer was made 80% reflecting to account for absorption losses in the
Alq3 layer and metal back contact. The emission of the active layer was assumed to be
isotropic with 550 nm emission wavelength for all the simulations. Since, the software
strictly operates in classical ray regime; the simulation results are independent of the
wavelength of emission light. The ray-tracing simulations were performed using 100,000
rays for measuring the out-coupled photons using large plane detectors near front and
rear side of the device. The angular distribution of the out-coupled rays was measured
using a far-field detector centered to the device with 1 million rays used for the
133
simulation. Both planar and far-field detector measured total fractional power of the out-
coupled rays.
Results and Discussion
The electrical and optical properties of deposited Al doped zinc oxide films were
characterized on a plain glass substrate without patterns. Four probe measurements
were made to characterize the electrical resistivity of the deposited films. The as-
deposited films had average sheet resistance of 5 kΩ/sq. Figure 8-1 is the optical
transmission plot for plain glass reference and as-deposited AZO film. The film shows
transmission as high as ~93% for a film thickness of ~1450 Å as compared to the plain
glass substrate.
Figure 8-2 depicts SEM micrographs of the fabricated microlens after polishing of
Al:ZnO film. The SEM images illustrate that only microlens tops are polished whereas
on rest of the area, zinc oxide grains are still discernable, illustrating no polishing. Due
to close packing of the features and high aspect ratio, the field area is relatively
inaccessible to the pad asperities to cause any polishing. Further, the feature height,
diameter and pad mechanical properties affect the ability to polish the surface with
closely packed features. This is evident from the extent of polished area along the slope
of the microlens, 6 μm has more polished area followed by 10 and 14 μm base diameter
microlens. The SEM micrographs also elucidate the conformal nature of deposited AZO
films. It should be noted here that because the film thicknesses (TCO and active layer)
are much less than the height, the diameter and periodicity of the texture. The deposited
subsequent films are conformal which results in corrugated geometry of the device.
Photoluminescence measurements were performed to investigate the
enhancement in light out-coupling efficiency of the microlens texture. Figure 8-3 plots
134
relative PL intensity with respect to the unpatterned reference for 6, 10 and 14 μm
diameter microlens texture on (a) patterned glass substrate, (b) patterned glass
substrate with AZO film and (c) patterned glass substrate with polished AZO film. The
obtained broad PL peak centered at 520 nm matches closely with the literature reported
peak position for Alq3 emission [181]. The UV excitation wavelength used was 325 nm.
Figure 8-3 (a) elucidates the enhancement in PL output obtained using the structures of
different diameters as compared to the reference non-patterned glass substrate. The
results demonstrate increase in enhancement with increasing diameter of the microlens
with maximum enhancement of ~ 4 fold for 14 μm microlens texture. Figure 8-3 (b) and
(c) are relative PL intensities of the substrates with AZO and polished AZO. The relative
PL intensities demonstrate a maximum enhancement of 2.1 and 2.6X for as-deposited
Al:ZnO and polished Al:ZnO samples, respectively. The PL spectra for substrate with Al
doped zinc oxide layer show broadening of Alq3 peak and additional shoulder peaks
between 450-460 and 480-490 nm wavelength. The broadening and appearance of the
peak shoulders are related to the weak deep level PL emission peaks of AZO film which
occur in the range of 510-600 nm [182, 183]. Further, it should be noted here that the
measured PL intensity is the cumulative of enhancement due to micro-texture and the
increased surface area measured at an angle of 30º with respect to surface normal of
the substrate.
Ray-tracing simulations were performed to understand the origin of improvement
in the light extraction due to corrugated device structure. Figure 8-4 (a) plots the
simulated enhancement in light extraction efficiency using microlens arrays of different
diameters. The plotted curves take in to account the fractional increase in the surface
135
area relative to planar device to obtain the total combined enhancement. For
comparison, simulation results for nactive = 2.5 have also been presented here (see
Figure 8-4 (b)). Figure 8-4 (c) depicts the fractional power output for nactive = 1.7 and 2.5
at h/d ratio of 0.5. The ray-tracing simulation results presented in chapter 6 show that
the smaller diameter is slightly efficient and a maxima is observed in enhancement
versus h/d plot (see Figure 6-8). However, when the increase in surface area is taken in
to account, the increase in number of generated photons compensates for the decrease
in the extraction efficiency of structure with larger diameter and greater h/d ratio. The
simulations results for high refractive index active layer (nactive = 2.5) demonstrate more
enhancement as more light is trapped in the active layer with high refractive index which
is extracted from the active and TCO layer by the microlens texture.
The far-field angular spreads of the extracted photons (depicted by Figure 8-5)
reiterate the similar trend observed with total out-coupled light when they are multiplied
by the fractional increase in area. The angular distribution curves show increase in
received power at all angles proportionally. However, there is preferential enhancement
in forward direction as it was observed in chapter 6 (Figure 6-10). The introduction of
microlens features at glass/TCO interface actually leads to preferential enhancement in
forward direction of the extracted photons from the TCO layer. This enhancement leads
to observed improvement in out-coupled photons. The resultant angular spread of the
finally extracted photons from substrate is superposition of the non-Lambertian profile of
glass/TCO and Lambertian spread of non-textured glass/air interface. The angular
distributions of extracted photons from simulated stack with nactive = 2.5 demonstrate
similar results but the spread is expectedly compact as compared to that for nactive = 1.7.
136
The reason for this behavior is high refractive index of the active layer which opens
another guided mode in active layer with escape cone angle ~ 50º at TCO/active layer
interface.
Based on the height calculated from AFM scans on the textured surface (see
Figure 7-5), the fabricated features have h/d ratio approximately 0.21, 0.17 and 0.21 for
6, 10 and 14 μm, respectively. The enhancement values obtained from the ray-tracing
simulations for h/d ratio of 0.2 and measured at detection angle of 30º are close to the
observed experimental results. The slight deviation from the simulation results can
originate from the approximation of absorption within the Alq3 and the reflectivity of the
Al layer in the simulation model or from interference effect between broad closely
occurring emission peaks of Alq3 and AZO around 520 nm and the thickness variation in
the film thicknesses which can alter the measured absolute intensity. In addition, the
simulation software used for this study simulates ray-tracing only in classical ray-optics
regime. However, the film thicknesses of TCO, active layer and metal contact are much
smaller than emission wavelength which invokes additional quantum mechanical
phenomena. It has been shown in literature that changes in refractive index at sub-
wavelength scales demonstrate micro-cavity effects which are known to cause spatial
and spectral changes in the emitted light [184].
Conclusion
The microlens structures with h/d ratio ~ 0.2 were fabricated with 1500 Å thick Al
doped zinc oxide thin film. The SEM micrographs elucidate that the polishing of AZO
film was limited to tip of the features due to compactness and relatively high aspect ratio
of the features. The enhancement in light extraction efficiency was experimentally
determined by measuring the PL intensity of evaporated Alq3 layer on top of the AZO
137
layer. The experimentally observed extraction efficiency from the PL results show
maximum enhancement as high as 2.6 folds for the polished Al doped zinc oxide film.
The experimental values are close to the simulated ray-tracing results. The results
presented here illustrate that the introduction of structures at glass/TCO interface leads
to more extraction of photons from active layer and TCO layer. The cumulative effect of
improvement in extraction efficiency and increased surface area results in the observed
enhancement of more than 1.5X. The angular spread obtained from simulation follows
the Lambertian profile of a rectilinear geometry providing further room for enhancement
of the extraction efficiency by placing textures on the glass/air interface.
138
400 500 600 700 8000.0
0.2
0.4
0.6
0.8
1.0
No
rmaliz
ed Inte
nsity (
a.u
.)
Wavelength (nm)
Glass
Al:ZnO
Figure 8-1. Optical transmission of plain glass and Al doped ZnO film.
Figure 8-2. SEM micrograph illustrating the polished Al:ZnO on the microlens pattern of (a) 6 μm, (b) 10 μm and (c) 14 μm base diameter.
(a) (b) (c) (a)
139
450 500 550 6000.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
No
rmaliz
ed Inte
nsity
Wavelength (nm)
Reference
6 um
10 um
14 um
14 m
10 m
6 m
Microlens - ex
= 325 nm
Reference
(a)
450 500 550 6000.0
0.5
1.0
1.5
2.0
2.5
No
rmaliz
ed Inte
nsity
Wavelength (nm)
Reference
6 um
10 um
14 um
(b) Microlens with Al:ZnO - ex
= 325 nm
Reference
6 m
10 m
14 m
450 500 550 6000.0
0.5
1.0
1.5
2.0
2.5
3.0
Norm
aliz
ed Inte
nsity
Wavelength (nm)
Reference
6 um
10 um
14 um
Reference
6 m
10 m
14 m
Microlens with polished Al:ZnO - ex
= 325 nm(c)
Figure 8-3.Normalized photoluminescence intensity of Alq3 with Al back reflector for (a) plain microlens patterned glass, (b) as-deposited Al:ZnO and (c) polished Al:ZnO on microlens textured substrate.
140
0.0 0.1 0.2 0.3 0.4 0.51.0
1.2
1.4
1.6
1.8
2.0
2.2E
nhancem
ent
h/d
6 um 10 um 14 umnactive
= 1.7
(a)
0.0 0.1 0.2 0.3 0.4 0.51.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
Enhancem
ent
h/d
6 um 10 um 14 umnactive
= 2.5
(b)
0 5 10 15
0.1
0.2
0.3
0.4(c)
Fra
ctional P
ow
er
Radius (m)
nactive
= 1.7
nactive
= 2.5
Figure 8-4.Enhancement in light out-coupling in the corrugated device with (a) nactive = 1.7 and (b) nactive = 2.5. The simulated fractional powers for different diameter of the microlens and active layer refractive indices at h/d = 0.5 are represented by (c).
141
0
45
90
135
180
Planar 0.6 um 1.2 um
1.8 um 2.4 um 3 um
nactive
= 1.7(a)
0
45
90
135
180
Planar 0.6 um 1.2 um
1.8 um 2.4 um 3 um
nactive
= 2.5(b)
Figure 8-5. Simulated angular distribution of the extracted mode for device with (a) nactive = 1.7 and (b) nactive = 2.5.
142
CHAPTER 9 CONCLUSIONS
The increasing energy needs of the world have led to a quest for alternative
energy efficient lighting sources. This has led to tremendous research efforts in SSL
devices such as thin film EL devices, OLEDs and LEDs. One of the main issues limiting
the performance of all these devices is the light extraction efficiency limit of currently
used architectures. Most of these devices are now achieving close to 100% internal
quantum efficiencies, however, the external quantum efficiencies are usually limited to
20% or lower. Hence, there is an urgent need to push the limit of extraction efficiency
higher in order to tap all the photons generated in the device. Hence, this work has
focused on enhancing the light extraction efficiency in light emitting devices.
In order to enhance the extraction efficiency, surface topography of transparent
conducting oxide layer has been controlled on two scales: micro-level texturing and
nano-level roughness. The introduction of micro-scale texture at TCO/active layer
interface enhances the light output by extracting the TCO and active guided modes. By
controlling the nano-scale roughness the detrimental effects of surface roughness, large
particle inclusions and pinholes are eliminated.
For controlling the surface roughness of ZnO (the TCO considered in this work), a
CMP process was developed, the results and findings of which are summarized below.
For enhancing the light extraction efficiency, periodic micro-scale textures (microlens,
micropyramid and cylinder) applied at different interfaces were investigated using ray-
tracing simulations. To corroborate the ray-simulation results, photoluminescence study
was conducted on the fabricated substrate with periodic micro-scale arrays. The
143
findings of these ray tracing simulations and photoluminescence study confirmed the
light extraction enhancement with these structures as summarized below.
Chemical Mechanical Polishing of ZnO
As part of the first goal of this research a zinc oxide polishing process was
developed in order to control the surface roughness of the TCO layer. Highly smooth
zinc oxide surface with RMS roughnesses between 4-6 Å were obtained with developed
CMP process. Removal rates as high as 670 Å/min were achieved at slurry pH of ~3.
Acidic and basic pH slurry polished zinc oxide films showed higher transmission (91%
and 94%, at λ = 632.8 nm) as compared to the unpolished surface. Modified Preston
equation was applied to explain the removal rate variation with pressure and platen
velocity. The modified Preston equation was found to be consistent with the
experimental results and accounts for the non-zero intercepts. Low polishing rates were
observed at higher pH (> 7) whereas at low pH (~3) an order of magnitude increase in
the removal rates was observed without any detrimental effects on the surface
roughness. The zinc oxide CMP process at different slurry pH values was explained by
the pH dependence of the zinc oxide dissolution in aqueous solution and particle-
surface interactions. X-ray reflectivity simulation results concurred well with the RMS
roughness values for the ZnO films deduced from the AFM images.
Effect of anionic and non-ionic surfactant on the zinc oxide CMP was also
investigated. Anionic surfactant, SDS in this case was found to lower the polishing rates
and increase the RMS roughness with increasing concentration. Addition of 5 mM SDS
was found to give the best surface finish with RMS roughness ~3 Å. However, reverse
trends were obtained with Triton X-100 (non-ionic) surfactant due to reduced repulsive
forces between slurry particles and CMP pad because of the screening of surface
144
charges by the surfactant molecules adsorbed on the pad and silica particles. The
removal rates increased to as high as 900 Å/min with addition of KCl salt but at the cost
of surface roughness. The lowest RMS roughness of ZnO polycrystalline films after
polishing was found to limit at or greater than 3 Å. This is related to the film morphology
(grain size) dependence of the CMP process for zinc oxide and the changes in surface-
particle and particle-pad interactions by addition of different slurry additives. Optical
transmission of zinc oxide films deposited on glass substrate exhibited an average
transmission between 92-93% for various slurry compositions. The effects of various
CMP process variables and slurry chemistry and additives were delineated in this work
which resulted in understanding of CMP process for zinc oxide and achieving highly
smooth zinc oxide surface after chemical mechanical polishing.
Light Extraction
Light extraction efficiencies of various external light out-couplers such as pyramid,
microlens and cylinder arrays were analyzed. The use of such external light out-
couplers in conjunction with back-side mirror results in enhancement of light due to
extraction of higher angle photons and change in directionality of reflected photons
outside the escape cone. In case of pyramids, the intensity of extracted light increased
with contact angle less than ~55º after which the structure demonstrates slight decrease
due to re-entry of the extracted photons back into device. The ray simulations
elucidated similar behavior of microlens array. Microlens arrays demonstrate higher light
extraction efficiency and circular symmetry in angular distribution of the extracted
photons. In contrast, efficacy of the cylinder arrays is limited to the lower h/d ratios and
light extraction efficiency becomes independent of increase in the feature diameter after
10 μm. Microlens and pyramid arrays were found to be more efficient structures as
145
compared cylindrical arrays. The trends remain independent of refractive index of the
active layer as expected because these features only affect the substrate mode. In
addition the enhancement in light out-coupling originates from the extraction of photons
outside the escape cone by changing their directionality upon multiple reflections.
Ray dynamics slightly change when the microlens structure is applied at the
substrate/TCO interface. Application of microlens texture at this interface changes the
geometry of TCO and active layer to a corrugated geometry. The deviation from
parallelepiped geometry leads to increased extraction of TCO/active mode in case with
nactive = 1.7 and higher extraction of both TCO and active layer mode in case of nactive =
2.5. The enhancement increases steeply up to 0.3 and 0.4 h/d ratio for nactive = 1.7 and
nactive = 2.5, respectively. Thereafter, decrease in extraction efficiency was observed in
simulated results. The increase in extraction of normal angle photons from TCO layer
up to 0.3 and 0.4 h/d ratio for nactive = 1.7 and nactive = 2.5, respectively was found to be
responsible for highest enhancement at these h/d ratios when fraction of photons exiting
TCO layer were measured alone. However, at higher h/d ratios the extraction of higher
angle photons increased but due to Lambertian profile of extracted light at glass/air
interface, the combined effect of TCO extracted and glass extracted mode profiles
resulted in the observed decrease in the overall enhancement.
Corrugated Device
For the first time, micro-scale textures applied to glass/TCO interface to form
corrugated device structures were analyzed and demonstrated experimentally. The
microlens structures with h/d ratio ~ 0.2 were fabricated with 1500 Å thick Al doped zinc
oxide thin film. The polishing of the AZO films was limited to tip of the features due to
compactness and relatively high aspect ratio of the features. The enhancement in light
146
extraction efficiency was experimentally determined by measuring the PL intensity of
evaporated Alq3 layer on top of the AZO layer. The experimentally observed extraction
efficiency from the PL results show maximum enhancement as high as 2.6 folds for
polished Al doped zinc oxide film. The experimental values were close to the simulated
ray-tracing results. The results presented here illustrate that the introduction of
structures at glass/TCO interface leads to more extraction of photons from active layer
and TCO layer. The cumulative effect of extraction efficiency improvement and
increased surface area leads to observed enhancement of more than 1.5X. The angular
spread obtained from simulation follows the Lambertian profile of a rectilinear geometry
providing further room for enhancement of extraction efficiency by applying textures at
glass/TCO interface to reduce substrate mode.
147
LIST OF REFERENCES
[1] G. Chae, Sung, Jpn. J. Appl. Phys. 2001, 40, 1282.
[2] K. L. Chopra, S. Major, D. K. Pandya, Thin Solid Films 1983, 102, 1.
[3] G. J. Exarhos, X.-D. Zhou, Thin Solid Films 2007, 515, 7025.
[4] H. Ohta, K.-i. Kawamura, M. Orita, M. Hirano, N. Sarukura, H. Hosono, Appl. Phys. Lett. 2000, 77, 475.
[5] W. Beyer, J. Hüpkes, H. Stiebig, Thin Solid Films 2007, 516, 147.
[6] C. G. Granqvist, Sol. Energy Mater. Sol. Cells 2007, 91, 1529.
[7] S. K. Deb, S.-H. Lee, C. Edwin Tracy, J. Roland Pitts, B. A. Gregg, H. M. Branz, Electrochim. Acta 2001, 46, 2125.
[8] B. Karlsson, E. Valkonen, T. Karlsson, C. G. Ribbing, Thin Solid Films 1981, 86, 91.
[9] J. D. Perkins, J. A. del Cueto, J. L. Alleman, C. Warmsingh, B. M. Keyes, L. M. Gedvilas, P. A. Parilla, B. To, D. W. Readey, D. S. Ginley, Thin Solid Films 2002, 411, 152.
[10] J. Nagai, Solid State Ionics 1990, 40-41, 383.
[11] T. Minami, Thin Solid Films 2008, 516, 5828.
[12] T. Minami, Semicond. Sci. Technol. 2005, 20, S44.
[13] R. G. Gordon, MRS Bull. 2000, 25, 52.
[14] S. Elliott, The Physics and Chemistry of Solids, John Wiley & Sons, Inc., Chichester 1998.
[15] P. P. Edwards, A. Porch, M. O. Jones, D. V. Morgan, R. M. Perks, Dalton Trans. 2004, 2995.
[16] P. Melpignano, C. Cioarec, R. Clergereaux, N. Gherardi, C. Villeneuve, L. Datas, Org. Electron. 2010, 11, 1111.
[17] S. Y. Ryu, J. H. Noh, B. H. Hwang, C. S. Kim, S. J. Jo, J. T. Kim, H. S. Hwang, H. K. Baik, H. S. Jeong, C. H. Lee, S. Y. Song, S. H. Choi, S. Y. Park, Appl. Phys. Lett. 2008, 92, 023306.
[18] K. Bädeker, Ann. Phys. 1907, 327, 749.
[19] L. Hiesinger, H. König, in Festschrift W.C. Heraeus, 1951, 376.
148
[20] G. Helwig, Zeitschrift für Physik A Hadrons and Nuclei 1952, 132, 621.
[21] A. Thelen, H. König, Naturwissenschaften 1956, 43, 297.
[22] R. Groth, E. Kauer, Philips Tech. Rev. 26, 1965.
[23] R. Groth, physica status solidi (b) 1966, 14, 69.
[24] Y. Furubayashi, T. Hitosugi, Y. Yamamoto, K. Inaba, G. Kinoda, Y. Hirose, T. Shimada, T. Hasegawa, Appl. Phys. Lett. 2005, 86, 252101.
[25] T. Minami, Thin Solid Films 2008, 516, 1321.
[26] T. Tomio, H. Miki, H. Tabata, T. Kawai, S. Kawai, J. Appl. Phys. 1994, 76, 5886.
[27] J. Mass, P. Bhattacharya, R. S. Katiyar, "Mater. Sci. Eng., B" 2003, 103, 15.
[28] X. Jiang, F. L. Wong, M. K. Fung, S. T. Lee, Appl. Phys. Lett. 2003, 83, 1877.
[29] H. Nanto, T. Minami, S. Shooji, S. Takata, J. Appl. Phys. 1984, 55, 1034.
[30] H. Kim, A. Piqué, J. S. Horwitz, H. Murata, Z. H. Kafafi, C. M. Gilmore, D. B. Chrisey, Thin Solid Films 2000, 377-378, 802.
[31] S.-Y. Kuo, W.-C. Chen, F.-I. Lai, C.-P. Cheng, H.-C. Kuo, S.-C. Wang, W.-F. Hsieh, J. Cryst. Growth 2006, 287, 78.
[32] P. S. Nayar, A. Catalano, Appl. Phys. Lett. 1981, 39, 105.
[33] J. B. Webb, D. F. Williams, M. Buchanan, Appl. Phys. Lett. 1981, 39, 640.
[34] T. Minami, H. Nanto, S. Takata, Appl. Phys. Lett. 1982, 41, 958.
[35] T. Minami, H. Nanto, S. Takata, Jpn. J. Appl. Phys. 1984, 23, L280
[36] T. Minami, "Transparent and conductive multicomponent oxide films prepared by magnetron sputtering", Baltimore, Maryland (USA), 1999.
[37] C.-Y. Lo, C.-L. Hsu, Q.-X. Yu, H.-Y. Lee, C.-T. Lee, J. Appl. Phys. 2002, 92, 274.
[38] C.-T. Lee, Q.-X. Yu, B.-T. Tang, H.-Y. Lee, F.-T. Hwang, Appl. Phys. Lett. 2001, 78, 3412.
[39] M. Tadatsugu, Semicond. Sci. Technol. 2005, 20, S35.
149
[40] D. P. Norton, Y. W. Heo, M. P. Ivill, K. Ip, S. J. Pearton, M. F. Chisholm, T. Steiner, Mater. Today 2004, 7, 40.
[41] C. Klingshirn, physica status solidi (b) 1975, 71, 547.
[42] D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen, T. Goto, Appl. Phys. Lett. 1997, 70, 2232.
[43] V. Srikant, D. R. Clarke, J. Appl. Phys. 1998, 83, 5447.
[44] T. Kamiya, M. Kawasaki, MRS Bull. 2008, 33.
[45] S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo, T. Steiner, Prog. Mater Sci. 2005, 50, 340.
[46] Y. Dai, Y. Zhang, Q. K. Li, C. W. Nan, Chem. Phys. Lett. 2002, 358, 86.
[47] S. Choopun, R. D. Vispute, W. Noch, A. Balsamo, R. P. Sharma, T. Venkatesan, A. Iliadis, D. C. Look, Appl. Phys. Lett. 1999, 75, 3949.
[48] C. R. Gorla, N. W. Emanetoglu, S. Liang, W. E. Mayo, Y. Lu, M. Wraback, H. Shen, J. Appl. Phys. 1999, 85, 2595.
[49] Y. Yoshino, T. Makino, Y. Katayama, T. Hata, Vacuum 2000, 59, 538.
[50] W. S. Shi, O. Agyeman, C. N. Xu, J. Appl. Phys. 2002, 91, 5644.
[51] M. Izaki, T. Omi, Appl. Phys. Lett. 1996, 68, 2440.
[52] A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. Sumiya, K. Ohtani, S. F. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koinuma, M. Kawasaki, Nat. Mater. 2005, 4, 46.
[53] F. Hamdani, A. Botchkarev, W. Kim, H. Morkoc, M. Yeadon, J. M. Gibson, S. C. Y. Tsen, D. J. Smith, D. C. Reynolds, D. C. Look, K. Evans, C. W. Litton, W. C. Mitchel, P. Hemenger, Appl. Phys. Lett. 1997, 70, 467.
[54] K.-P. Hsueh, C.-T. Pan, C.-T. Li, H.-C. Lin, Y.-M. Hsin, J.-I. Chyi, J. Electrochem. Soc. 2010, 157, H381.
[55] F. Hamdani, M. Yeadon, D. J. Smith, H. Tang, W. Kim, A. Salvador, A. E. Botchkarev, J. M. Gibson, A. Y. Polyakov, M. Skowronski, H. Morkoc, J. Appl. Phys. 1998, 83, 983.
[56] R. D. Vispute, V. Talyansky, Z. Trajanovic, S. Choopun, M. Downes, R. P. Sharma, T. Venkatesan, M. C. Woods, R. T. Lareau, K. A. Jones, A. A. Iliadis, Appl. Phys. Lett. 1997, 70, 2735.
[57] X. W. Sun, R. F. Xiao, H. S. Kwok, J. Appl. Phys. 1998, 84, 5776.
150
[58] G. Popovici, W. Kim, A. Botchkarev, H. Tang, H. Morkoc, J. Solomon, Appl. Phys. Lett. 1997, 71, 3385.
[59] W.-S. Lee, G.-W. Choi, Y.-J. Seo, Microelectron. J. 2009, 40, 302.
[60] J. Hüpkes, J. Müller, B. Rech, in Transparent Conductive Zinc Oxide, 2008, 359.
[61] H. Chen, S. W. Liu, X. M. Wang, M. N. Iliev, C. L. Chen, X. K. Yu, J. R. Liu, K. Ma, W. K. Chu, "Nucl. Instrum. Methods Phys. Res., Sect. B" 2005, 241, 635.
[62] S. Faÿ, A. Shah, in Transparent Conductive Zinc Oxide, 2008, 235.
[63] J. Bailat, E. Vallat-Sauvain, L. Feitknecht, C. Droz, A. Shah, J. Non-Cryst. Solids 2002, 299-302, 1219.
[64] Y. Nasuno, M. Kondo, A. Matsuda, Sol. Energy Mater. Sol. Cells 2002, 74, 497.
[65] J. Springer, A. Poruba, L. Mullerova, M. Vanecek, O. Kluth, B. Rech, J. Appl. Phys. 2004, 95, 1427.
[66] D. A. Lucca, D. W. Hamby, M. J. Klopfstein, G. Cantwell, C. J. Wetteland, J. R. Tesmer, M. Nastasi, CIRP Annals - Manufacturing Technology 2001, 50, 400.
[67] W. W. Wenas, K. Dairiki, A. Yamada, M. Konagai, K. Takahashi, J. H. Jang, S. Lim, "High efficiency a-Si solar cells with ZnO films", presented at Photovoltaic Energy Conversion, 1994., Conference Record of the Twenty Fourth. IEEE Photovoltaic Specialists Conference - 1994, 1994 IEEE First World Conference on, 1994.
[68] B. Sang, K. Dairiki, A. Yamada, M. Konagai, Jpn. J. Appl. Phys. 1999, 38, 4983.
[69] M. Kadota, M. Minakata, T. Kasanami, Jpn. J. Appl. Phys. 1992, 31S1, 224.
[70] R. J. Nelson, R. G. Sobers, J. Appl. Phys. 1978, 49, 6103.
[71] I. Schnitzer, E. Yablonovitch, C. Caneau, T. J. Gmitter, Appl. Phys. Lett. 1993, 62, 131.
[72] C. Adachi, M. A. Baldo, M. E. Thompson, S. R. Forrest, J. Appl. Phys. 2001, 90, 5048.
[73] C. F. Madigan, M. H. Lu, J. C. Sturm, Appl. Phys. Lett. 2000, 76, 1650.
151
[74] M. R. Krames, M. Ochiai-Holcomb, G. E. Hofler, C. Carter-Coman, E. I. Chen, I. H. Tan, P. Grillot, N. F. Gardner, H. C. Chui, J. W. Huang, S. A. Stockman, F. A. Kish, M. G. Craford, T. S. Tan, C. P. Kocot, M. Hueschen, J. Posselt, B. Loh, G. Sasser, D. Collins, Appl. Phys. Lett. 1999, 75, 2365.
[75] A. R. Franklin, R. Newman, J. Appl. Phys. 1964, 35, 1153.
[76] W. N. Carr, G. E. Pittman, Appl. Phys. Lett. 1963, 3, 173.
[77] J. K. Kim, S. Chhajed, M. F. Schubert, E. F. Schubert, A. J. Fischer, M. H. Crawford, J. Cho, H. Kim, C. Sone, Adv. Mater. 2008, 20, 801.
[78] H. Kim, K.-S. Ahn, Electrochem. Solid-State Lett. 2010, 13, H131.
[79] C. Huh, K.-S. Lee, E.-J. Kang, S.-J. Park, J. Appl. Phys. 2003, 93, 9383.
[80] M.-W. Chang, H. S. Won, H. Kim, J. M. Lee, Jpn. J. Appl. Phys. 2007, 46, 6673
[81] T. Nakamura, N. Tsutsumi, N. Juni, H. Fujii, J. Appl. Phys. 2005, 97, 054505.
[82] A. O. Altun, S. Jeon, J. Shim, J.-H. Jeong, D.-G. Choi, K.-D. Kim, J.-H. Choi, S.-W. Lee, E.-S. Lee, H.-D. Park, J. R. Youn, J.-J. Kim, Y.-H. Lee, J.-W. Kang, Org. Electron. 2010, 11, 711.
[83] Y. Sun, S. R. Forrest, Nat. Photon. 2008, 2, 483.
[84] S.-Y. Hsu, M.-C. Lee, K.-L. Lee, P.-K. Wei, Appl. Phys. Lett. 2008, 92, 013303.
[85] S. Moller, S. R. Forrest, J. Appl. Phys. 2002, 91, 3324.
[86] E. F. Schubert, N. E. J. Hunt, M. Micovic, R. J. Malik, D. L. Sivco, A. Y. Cho, G. J. Zydzik, Science 1994, 265, 943.
[87] S. Fan, P. R. Villeneuve, J. D. Joannopoulos, E. F. Schubert, Phys. Rev. Lett. 1997, 78, 3294.
[88] W. N. Carr, Infrared Physics 1966, 6, 1.
[89] D. B. Thompson, A. Murai, M. Iza, S. Brinkley, S. P. DenBaars, U. K. Mishra, S. Nakamura, Jpn. J. Appl. Phys. 2008, 47, 3447
[90] S. J. Lee, S. W. Song, "Efficiency improvement in light-emitting diodes based on geometrically deformed chips", San Jose, CA, USA, 1999.
[91] C. H. Chiu, H. C. Kuo, C. F. Lai, H. W. Huang, W. C. Wang, T. C. Lu, S. C. Wang, C. H. Lin, J. Electrochem. Soc. 2007, 154, H944.
152
[92] I. Schnitzer, E. Yablonovitch, C. Caneau, T. J. Gmitter, A. Scherer, Appl. Phys. Lett. 1993, 63, 2174.
[93] Y. J. Lee, H. C. Kuo, S. C. Wang, T. C. Hsu, M. H. Hsieh, M. J. Jou, B. J. Lee, Photonics Technology Letters, IEEE 2005, 17, 2289.
[94] J. H. Kang, H. G. Kim, H. K. Kim, H. Y. Kim, J. H. Ryu, P. Uthirakumar, N. Han, C.-H. Hong, Jpn. J. Appl. Phys. 2009, 48, 102104.
[95] Y.-H. Cheng, J.-L. Wu, C.-H. Cheng, K.-C. Syao, M.-C. M. Lee, Appl. Phys. Lett. 2007, 90, 091102.
[96] T. Fujii, Y. Gao, R. Sharma, E. L. Hu, S. P. DenBaars, S. Nakamura, Appl. Phys. Lett. 2004, 84, 855.
[97] M. Fujita, T. Ueno, T. Asano, S. Noda, H. Ohhata, T. Tsuji, H. Nakada, N. Shimoji, Electron. Lett 2003, 39, 1750.
[98] C. Wiesmann, K. Bergenek, N. Linder, U. T. Schwarz, Laser & Photonics Review 2009, 3, 262.
[99] K. Kim, J. Choi, S. C. Jeon, J. S. Kim, H. M. Lee, Appl. Phys. Lett. 2007, 90, 181115.
[100] K.-Y. Ko, Y. K. Lee, Y. R. Do, Y.-D. Huh, J. Appl. Phys. 2007, 102, 013509.
[101] R. Han-Youl, H. Jeong-Ki, L. Yong-Jae, L. Yong-Hee, Selected Topics in Quantum Electronics, IEEE Journal of 2002, 8, 231.
[102] T. Baba, K. Inoshita, H. Tanaka, J. Yonekura, M. Ariga, A. Matsutani, T. Miyamoto, F. Koyama, K. Iga, Lightwave Technology, Journal of 1999, 17, 2113.
[103] M. Boroditsky, R. Vrijen, T. F. Krauss, R. Coccioli, R. Bhat, E. Yablonovitch, Lightwave Technology, Journal of 1999, 17, 2096.
[104] Y. J. Lee, T. C. Hsu, H. C. Kuo, S. C. Wang, Y. L. Yang, S. N. Yen, Y. T. Chu, Y. J. Shen, M. H. Hsieh, M. J. Jou, B. J. Lee, Materials Science and Engineering: B 2005, 122, 184.
[105] M. Yamada, T. Mitani, Y. Narukawa, S. Shioji, I. Niki, S. Sonobe, K. Deguchi, M. Sano, T. Mukai, Jpn. J. Appl. Phys. 2002, 41, L1431.
[106] J. W. Lee, C. Sone, Y. Park, S. N. Lee, J. H. Ryou, R. D. Dupuis, C. H. Hong, H. Kim, Appl. Phys. Lett. 2009, 95, 011108.
153
[107] C.-Y. Cho, J.-B. Lee, S.-J. Lee, S.-H. Han, T.-Y. Park, J. Won Kim, Y. C. Kim, S.-J. Park, Opt. Express 2010, 18, 1462.
[108] Y.-J. Lee, H.-C. Kuo, T.-C. Lu, S.-C. Wang, K. W. Ng, K. M. Lau, Z.-P. Yang, A. S. P. Chang, S.-Y. Lin, Lightwave Technology, Journal of 2008, 26, 1455.
[109] Y. R. Do, Y.-C. Kim, D.-H. Park, Y.-D. Huh, J. Electrochem. Soc. 2005, 152, H48.
[110] M.-L. Chen, A.-C. Wei, H.-P. D. Shieh, Jpn. J. Appl. Phys. 2007, 46, 1521
[111] H. W. Choi, C. Liu, E. Gu, G. McConnell, J. M. Girkin, I. M. Watson, M. D. Dawson, Appl. Phys. Lett. 2004, 84, 2253.
[112] P. Huajun, H. Yeuk Lung, Y. Xing-Jie, W. Man, K. Hoi-Sing, Display Technology, Journal of 2005, 1, 278.
[113] Y.-K. Ee, R. A. Arif, N. Tansu, P. Kumnorkaew, J. F. Gilchrist, Appl. Phys. Lett. 2007, 91, 221107.
[114] K. B. Sundaram, A. Khan, Thin Solid Films 1997, 295, 91.
[115] S. F. E. Vallat-Sauvain, S. Dubail, J. Meier, J. Bailat, U. Kroll, and A. Shah "Improved Interface Between Front TCO and Microcrystalline Silicon p-i-n Solar Cells", presented at MRS Spring Meeting, San francisco, 2001.
[116] S. Faÿ, L. Feitknecht, R. Schlüchter, U. Kroll, E. Vallat-Sauvain, A. Shah, Sol. Energy Mater. Sol. Cells 2006, 90, 2960.
[117] J. Ebothe, A. El Hichou, P. Vautrot, M. Addou, J. Appl. Phys. 2003, 93, 632.
[118] P. Nunes, B. Fernandes, E. Fortunato, P. Vilarinho, R. Martins, Thin Solid Films 1999, 337, 176.
[119] J. Wienke, A. S. Booij, Thin Solid Films 2008, 516, 4508.
[120] A. Ashida, H. Ohta, T. Nagata, Y. Nakano, N. Fujimura, T. Ito, J. Appl. Phys. 2004, 95, 1676.
[121] S. Jung, N. G. Park, M. Y. Kwak, B. O. Kim, K. H. Choi, Y. J. Cho, Y. K. Kim, Y. S. Kim, Opt. Mater. 2003, 21, 241.
[122] S. Xue, X. Zhang, R. Huang, H. Zhuang, Appl. Phys. A 2009, 94, 287.
[123] G. Liu, J. B. Kerr, S. Johnson, Synth. Met. 2004, 144, 6.
[124] G. W. Choi, N. H. Kim, Y. J. Seo, W. S. Lee, Electron. Lett 2006, 42, 489.
154
[125] Y.-J. Seo, G.-W. Choi, W.-S. Lee, Microelectron. Eng. 2008, 85, 1776.
[126] D. J. Channin, J. M. Hammer, M. T. Duffy, Appl. Opt. 1975, 14, 930.
[127] J. M. Phillips, M. E. Coltrin, M. H. Crawford, A. J. Fischer, M. R. Krames, R. Mueller-Mach, G. O. Mueller, Y. Ohno, L. E. S. Rohwer, J. A. Simmons, J. Y. Tsao, Laser & Photonics Review 2007, 1, 307.
[128] A. Murai, D. B. Thompson, H. Masui, N. Fellows, U. K. Mishra, S. Nakamura, S. P. DenBaars, Appl. Phys. Lett. 2006, 89, 171116.
[129] W. H. Southwell, Opt. Lett. 1983, 8, 584.
[130] B.-J. Kim, J. Bang, S. H. Kim, J. Kim, J. Electrochem. Soc. 2010, 157, H449.
[131] C. M. Tsai, J. K. Sheu, W. C. Lai, Y. P. Hsu, P. T. Wang, C. T. Kuo, C. W. Kuo, S. J. Chang, Y. K. Su, Electron Device Letters, IEEE 2005, 26, 464.
[132] J. J. Wierer, M. R. Krames, J. E. Epler, N. F. Gardner, M. G. Craford, J. R. Wendt, J. A. Simmons, M. M. Sigalas, Appl. Phys. Lett. 2004, 84, 3885.
[133] Y. R. Do, Y. C. Kim, Y.-W. Song, C.-O. Cho, H. Jeon, Y.-J. Lee, S.-H. Kim, Y.-H. Lee, Adv. Mater. 2003, 15, 1214.
[134] J. R. Oh, Y. K. Lee, H. K. Park, Y. R. Do, J. Appl. Phys. 2009, 105, 043103.
[135] J. R. Oh, H. K. Park, Y. R. Do, Electrochem. Solid-State Lett. 2009, 12, J58.
[136] K. Ishihara, M. Fujita, I. Matsubara, T. Asano, S. Noda, H. Ohata, A. Hirasawa, H. Nakada, N. Shimoji, Appl. Phys. Lett. 2007, 90, 111114.
[137] K. Tadatomo, H. Okagawa, Y. Ohuchi, T. Tsunekawa, Y. Imada, M. Kato, T. Taguchi, Jpn. J. Appl. Phys. 2001, 40, L583.
[138] S. J. Chang, Y. C. Lin, Y. K. Su, C. S. Chang, T. C. Wen, S. C. Shei, J. C. Ke, C. W. Kuo, S. C. Chen, C. H. Liu, Solid-State Electron. 2003, 47, 1539.
[139] Y. P. Hsu, S. J. Chang, Y. K. Su, J. K. Sheu, C. T. Lee, T. C. Wen, L. W. Wu, C. H. Kuo, C. S. Chang, S. C. Shei, J. Cryst. Growth 2004, 261, 466.
[140] J. Wang, L. W. Guo, H. Q. Jia, Z. G. Xing, Y. Wang, H. Chen, J. M. Zhou, J. Vac. Sci. Technol., B 2005, 23, 2476.
[141] J. Wang, L. W. Guo, H. Q. Jia, Z. G. Xing, Y. Wang, H. Chen, J. M. Zhou, Jpn. J. Appl. Phys. 2005, 44, L982
155
[142] J. Wang, L. W. Guo, H. Q. Jia, Z. G. Xing, Y. Wang, H. Chen, J. M. Zhou, Thin Solid Films 2006, 515, 1727.
[143] O.-H. Nam, M. D. Bremser, B. L. Ward, R. J. Nemanich, R. F. Davis, Jpn. J. Appl. Phys. 1997, 36, L532
[144] O.-H. Nam, M. D. Bremser, T. S. Zheleva, R. F. Davis, Appl. Phys. Lett. 1997, 71, 2638.
[145] T. Mukai, K. Takekawa, S. Nakamura, Jpn. J. Appl. Phys. 1998, 37, L839
[146] H. J. Peng, Y. L. Ho, C. F. Qiu, M. Wong, H. S. Kwok, SID Symposium Digest of Technical Papers 2004, 35, 158.
[147] H. Greiner, Jpn. J. Appl. Phys. 2007, 46, 4125
[148] K. Bao, X. N. Kang, B. Zhang, T. Dai, C. Xiong, H. Ji, G. Y. Zhang, Y. Chen, Photonics Technology Letters, IEEE 2007, 19, 1840.
[149] H. Kasugai, Y. Miyake, A. Honshio, S. Mishima, T. Kawashima, K. Iida, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki, H. Kinoshita, H. Shiomi, Jpn. J. Appl. Phys. 2005, 44, 7414
[150] M. Iwaya, H. Kasugai, T. Kawashima, K. Iida, A. Honshio, Y. Miyake, S. Kamiyama, H. Amano, I. Akasaki, Thin Solid Films 2006, 515, 768.
[151] E.-J. Hong, K.-J. Byeon, H. Park, J. Hwang, H. Lee, K. Choi, G. Y. Jung, Materials Science and Engineering: B 2009, 163, 170.
[152] J. P. Kottmann, O. J. F. Martin, D. R. Smith, S. Schultz, Chem. Phys. Lett. 2001, 341, 1.
[153] Y.-J. Lee, S.-H. Kim, J. Huh, G.-H. Kim, Y.-H. Lee, S.-H. Cho, Y.-C. Kim, Y. R. Do, Appl. Phys. Lett. 2003, 82, 3779.
[154] P. Scherrer, Göttinger Nachrichten, Vol. 2, 1918.
[155] R. Zsigmondy, Kolloidchemie 1920.
[156] Q. Luo, S. Ramarajan, S. V. Babu, Thin Solid Films 1998, 335, 160.
[157] L. M. Cook, J. Non-Cryst. Solids 1990, 120, 152.
[158] T. Hoshino, Y. Kurata, Y. Terasaki, K. Susa, J. Non-Cryst. Solids 2001, 283, 129.
[159] J. T. Abiade, S. Yeruva, W. Choi, B. M. Moudgil, D. Kumar, R. K. Singh, J. Electrochem. Soc. 2006, 153, G1001.
156
[160] A. Degen, M. Kosec, J. Eur. Ceram. Soc. 2000, 20, 667.
[161] U. Mahajan, M. Bielmann, R. K. Singh, Electrochem. Solid-State Lett. 1999, 2, 80.
[162] X. G. Zhang, Corrosion and electrochemistry of zinc, Plenum Press, New York 1996.
[163] M. Pourbaix, Atlas of electrochemical equilibria in aqueous solutions, Pergamon Press, Oxford; New York 1966.
[164] Youngki Hong, Udaya B. Patri, Suresh Ramakrishnan, S. V. Babu, "Novel Use of Surfactants in Copper Chemical Mechanical Polishing (CMP)", presented at Mater. Res. Soc. Symp. Proc., 2005.
[165] G. Bahar Basim, B. M. Moudgil, Kona 2003, 21, 178.
[166] M. Itano, T. Kezuka, M. Ishii, T. Unemoto, M. Kubo, T. Ohmi, J. Electrochem. Soc. 1995, 142, 971.
[167] R. Vos, M. Lux, K. Xu, W. Fyen, C. Kenens, T. Conard, P. Mertens, M. Heyns, Z. Hatcher, M. Hoffman, J. Electrochem. Soc. 2001, 148, G683.
[168] D. Ng, S. Kundu, M. Kulkarni, H. Liang, J. Electrochem. Soc. 2008, 155, H68.
[169] H.-G. Kang, T. Katoh, M.-Y. Lee, H.-S. Park, U. Paik, J.-G. Park, Jpn. J. Appl. Phys. 2004, 43, L1060
[170] Y. Hong, D. Roy, S. V. Babu, Electrochem. Solid-State Lett. 2005, 8, G297.
[171] G. B. Basim, B. M. Moudgil, J. Colloid Interface Sci. 2002, 256, 137.
[172] U. Mahajan, M. Bielmann, R. K. Singh, Electrochem. Solid-State Lett. 1999, 2, 46.
[173] P. Kumar, S.-Y. Son, J. Lee, F.-C. Chang, A. Khanna, A. C. Arjunan, R. K. Singh, Electrochem. Solid-State Lett. 2009, 12, H271.
[174] C. Iliescu, J. Jing, F. E. H. Tay, J. Miao, T. Sun, Surf. Coat. Technol. 2005, 198, 314.
[175] C. Iliescu, B. Chen, J. Miao, "Sens. Actuators, A" 2008, 143, 154.
[176] S. Shoji, H. Kikuchi, H. Torigoe, "Sens. Actuators, A" 1998, 64, 95.
[177] D. C. S. Bien, P. V. Rainey, S. J. N. Mitchell, H. S. Gamble, J. Micromech. Microeng. 2003, 13, S34.
157
[178] C. Iliescu, F. E. H. Tay, J. Miao, "Sens. Actuators, A" 2007, 133, 395.
[179] C. Iliescu, J. Miao, F. E. H. Tay, Surf. Coat. Technol. 2005, 192, 43.
[180] C. Thierry, et al., J. Micromech. Microeng. 1998, 8, 84.
[181] D. Z. Garbuzov, V. Bulovic, P. E. Burrows, S. R. Forrest, Chem. Phys. Lett. 1996, 249, 433.
[182] M. Wang, K. E. Lee, S. H. Hahn, E. J. Kim, S. Kim, J. S. Chung, E. W. Shin, C. Park, Mater. Lett. 2007, 61, 1118.
[183] K. Yim, C. Lee, Cryst. Res. Technol. 2006, 41, 1198.
[184] J. Lee, N. Chopra, F. So, Appl. Phys. Lett. 2008, 92, 033303.
158
BIOGRAPHICAL SKETCH
Sushant Gupta was born in 1984, in Delhi, India. He completed his high school in
2001 from Delhi. He graduated from the Delhi College of Engineering, Delhi, India in
2005 with a bachelor’s degree in polymer science and chemical technology. He interned
at two prestigious national laboratories in India, Bhabha Atomic Research Centre,
Mumbai and National Physical Laboratory, Delhi. The research experience at these
national laboratories motivated him to further his career by pursuing graduate studies.
He received his Master’s degree in material science and engineering from University of
Florida in 2008 while continuing with his Doctorate at same place.
During Sushant’s graduate studies, he worked on several projects during his
doctorate studies at UF, including synthesis of polymeric thin films for Fuel cell
applications, fabrication of PTFE thin films using pulse electron deposition technique
and organic LED material thin film fabrication using pulse laser deposition. In 2008, he
interned at Maxim Integrated products, San Jose, CA where he worked in chemical
mechanical polishing (CMP) area. His skills and in-depth knowledge in material science
and CMP were greatly appreciated and he was given the task to manage the project
independently. His recommendations and suggestions on the projects were well
received. The internship opportunity for him was a great learning experience both
professionally and personally.
His doctoral research was in the area of light out-coupling from solid state lighting
devices. He developed a CMP process for zinc oxide polishing and fabricated textured
ZnO films for improving the light extraction efficiency of light emitting devices. He hopes
to join a semiconductor manufacturing company to demonstrate his acquired skills and
knowledge. He feels learning is a continuous process and the journey has just begun.
