Analysis of Light Extraction Efficiency Enhancement for Deep

ANALYSIS OF LIGHT EXTRACTION EFFICIENCY

ENHANCEMENT FOR DEEP ULATRAVIOLET AND VISIBLE

LIGHT-EMITTING DIODES WITH III-NITRIDE MICRO-DOMES

by

PENG ZHAO

Submitted in partial fulfillment of requirements

For the degree of Master of Science

Thesis Adviser: Professor Hongping Zhao

Electrical Engineering and Computer Science Department

CASE WESTERN RESERVE UNIVERSITY

January 2013

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Peng Zhao

candidate for the Master of Science degree *.

(signed) Hongping Zhao. (chair of the committee)

Christian A. Zorman

Francis Merat

(date) 08/24/2012 .

*We also certify that written approval has been obtained for any

proprietary material contained therein.

Acknowledgement

I would like to thank my advisor Dr. Hongping Zhao, for her dedicate help on my

research project and graduate education training. Without her assistance and training, I

cannot have publications in this field and have the opportunity to attend and present at

top-leading academic conferences.

Then, I would like to thank Dr. Zorman for his constructive advice for my thesis

work as well as his advanced vision on my career plan. I would also like to thank Dr.

Frank Merat for being on my thesis committee member and providing insightful

comments.

Many thanks to members within this Micro/nano field, especially Christopher

Robert and Andrew Barnes, for giving me motivation and consistent support.

Finally, I would like to thank my family for their unlimited support and

encouragement.

1

Table of Content

List of Figures…………………………………………………………………………….5

Abstract………………………………………………………………………………….11

Chapter 1: Introduction………………………………………………………………..12

1.1 Lighting Energy Consumption and Solid State Lighting for Energy Saving…...12

1.2 III-nitride LEDs Introduction…………………………………………………...17

1.3 Current Challenges to Pursue High Efficiency III-nitride LEDs………………..18

1.3.1 Internal Quantum Efficiency Limitation…………………………………...18

1.3.2 Light Extraction Efficiency Limitation…………………………………….20

1.4 Recent Approaches to Enhance Light Extraction Efficiency for III-nitride

LEDs…………………………………………………………………………….21

1.5 Thesis Organization…………………………………………………………….24

Chapter 2: Finite Difference Time Domain Method for Light Extraction Efficiency

Calculation of III-nitride LEDs……………………………………….…..26

2.1 Finite Difference Time Domain Method (FDTD)……………………………...26

2.1.1 Introduction………………………………………………………………..26

2.1.2 Three Dimensional FDTD method and Yee’s Mesh……………………...27

2

2.2 Light Extraction Efficiency Calculation based on 3D-FDTD Method…………30

2.2.1 Light Extraction Efficiency Calculation Method………………………….30

2.2.2 Far Field Radiation Pattern Calculation Method………………………….32

Chapter 3: Analysis of Light Extraction Efficiency Enhancement for Deep

Ultraviolet AlGaN Quantum Wells LEDs with III-nitride Micro-

domes……………………………………………………………………….34

3.1 Concepts of III-nitride Micro-domes…………………………………………...34

3.2 Introduction of Deep Ultraviolet AlGaN Quantum Wells LEDs………………38

3.2.1 Deep UV AlGaN QWs LED Configuration and Applications……………38

3.2.2 Challenges in Achieving High Efficiency Deep UV AlGaN QWs LED….41

3.3 Polarization Analysis of Spontaneous Emission from AlGaN Quantum Wells..43

3.4 Calculation of Light Extraction Efficiency by 3-D FDTD Method…………….44

3.5 Light Extraction Efficiency Optimization of III-nitride Micro-domes…………46

3.5.1 Effect of Micro-dome (Micro-hemisphere) Size………………………….47

3.5.2 Effect of p-type Layer Thickness………………………………………….50

3.5.3 Effect of Micro-dome Size and Height (Micro-hemiellipsoid)…………...53

3.6 Far Field Radiation Pattern Calculation………………………………………..57

3.7 Light Extraction Efficiency Enhancement from Other Approaches…………...60

3

3.7.1 Surface Roughness………………………………………………………..60

3.7.2 SiO2 microlens array………………………………………………………63

3.8 Summary of Light Extraction Efficiency Enhancement for Deep UV AlGaN QWs

LEDs……………………………………………………………………………..65

Chapter 4: Analysis of Light Extraction Efficiency Enhancement for Thin-Film-

Flip-Chip (TFFC) InGaN Quantum Wells Light-Emitting Diodes with

GaN Micro-Domes…………………………………………………………67

4.1 Introduction of InGaN Quantum Wells LEDs…………………………………...67

4.1.1 Structure of III-nitride InGaN QWs LED…………………………………67

4.1.2 Efficiency Challenges in III-nitride InGaN QWs LED…………………...69

4.2 Polarization Analysis of Spontaneous Emission from InGaN Quantum Well…..70

4.3 Thin-Film-Flip-Chip Technology………………………………………………..71

4.4 Light Extraction Efficiency Calculation of TFFC InGaN QWs LED by 3-D FDTD

Method…………………………………………………………………………..72

4.5 Effect of P-GaN Layer Thickness on Light Extraction Efficiency for Conventional

TFFC InGaN LEDs………………………………………………………………74

4.5.1 Theoretical Prediction of Extracted Interference Pattern…………………..75

4.5.2 3D-FDTD Calculation of Extracted Interference Pattern………………….77

4.6 Effect of Micro-dome Size on Light Extraction Efficiency for InGaN LEDs…...80

4

4.6.1 Conventional Package of InGaN LED with Micro-dome (Micro-

hemisphere)…………………………………………………………………80

4.6.2 TFFC InGaN LED with Micro-dome (Micro-hemisphere)……………….82

4.6.3 TFFC InGaN LED with Micro-dome (Micro-hemiellipsoid)……………..84

4.7 P-GaN Layer Thickness Dependence of Light Extraction Efficiency for TFFC

InGaN QWs LEDs with GaN Micro-domes……………………………………..86

4.8 Summary of Light Extraction Efficiency Enhancement for TFFC InGaN QWs

LEDs…………………………………………………………………..…………88

Chapter 5: Conclusion and Future Work……………………………………………..89

5.1 Conclusions……………………………………………………………………...89

5.2 Future Work……………………………………………………………………..90

References……………………………………………………………………………….91

5

List of Figures

Figure 1-1. U.S. Annual Energy Outlook 2012 from Energy Information Administration

(EIA) - electricity (a) delivered energy and (b) related losses in sectors of residential,

commercial, industrial and transportation.

Figure 1-2. Ratio of electricity consumption used for lighting in total energy for sectors

of residential and commercial.

Figure 1-3. Annual electricity consumption of lighting technologies and the electricity

savings resulting from the increased use of LEDs in general illumination applications,

disaggregated by sectors.

Figure 1-4. Energy bandgap as a function of lattice constant for both zinc-blende III-

phosphide and wurtzite III-nitride semiconductor alloy.

Figure 1-5. External quantum efficiency for visible spectrum LEDs. V() represents the

luminous eye response curve from CIE (International Commission on Illumination).

Figure 1-6. Light trapping in GaN due to total internal reflection with critical angle .

Figure 1-7. Existing approaches to enhance the light extraction efficiency by (a) surface

roughness; (b) photonic crystals; (c) sapphire lens and (d) SiO2 / polystyrene microlens

array.

Figure 1-8. Absorption spectra from 1. Polystyrene 2. Ethylbenzeze.

Figure 2-1. Yee’s mesh cell: Maxwell’s equations are solved discretely in unit of Yee’s

mesh cell.

Figure 2-2. Power collecting setting for light extraction efficiency calculation: (a)

extracted power detection plane; (b) power box surrounding around the dipole source.

6

Figure 2-3. Far field projection spherical coordinates system: (a) 3D far field hemisphere

projection surface at 1 meter away; (b) 3D spherical coordinates system and

corresponding Cartesian coordinates system.

Figure 3-1. The schematics of (a) conventional flat emission surface LEDs with a narrow

photon escape cone, (b) III-nitride micro-domes LEDs with increased effective escape

cone.

Figure 3-2. 3D schematics for the formation process of III-nitride micro-domes: (a)

original sample with flat surface; (b) coated with microspheres monolayer; (c) the

intermediate state of RIE pattern transferring; (d) formation of III-nitride micro-domes.

Figure 3-3. The schematic of III-nitride micro-domes LED with different diameter (D) vs.

height (h) aspect ratios (a) D/2>h; (b) D/2=h (c) D/2<h.

Figure 3-4. Ultraviolet (UV) radiation spectrum: UV Vacuum (100nm-200nm), UV C

(200nm-280nm), UV B (280nm-320nm) and UV A (320nm-400nm).

Figure 3-5. Ultraviolet (UV) applications chart in spectrum: UV A, UV B and UV C.

Figure 3-6. Configuration schematic of deep UV AlGaN QW LED grown on sapphire

substrate.

Figure 3-7. The state of the art of III-nitride UV LEDs external quantum efficiency.

Figure 3-8. Spontaneous emission spectra (TE and TM modes) for AlxGa1-xN QWs LEDs

with x=0.58, 0.62, 0.66, and 0.7. TM spontaneous emission component becomes

dominant with x>0.66.

Figure 3-9. 2D Schematics of the computational domain in 3D FDTD analysis of deep

UV AlGaN QWs LEDs with (a) flat surface; and (b) AlGaN micro-domes on p-type

emission surface for enhancing light extraction efficiency.

7

Figure 3-10. Source positioning dependence analysis of light extraction efficiency for

Deep UV AlGaN QW LEDs with III-nitride micro-dome (Micro-hemisphere) at diameter

sizes: D=0, D=100nm, D=200nm, D=300nm, D=400nm and D=500nm. The other

conditions are peak=250nm, FWHM=10nm and Ptype = 300nm.

Figure 3-11. Light extraction efficiency enhancement of the TM polarized spontaneous

emission component for AlGaN QWs LEDs with AlGaN micro-hemispheres emitting at

250nm as a function of the micro-hemisphere diameter. The P-type layer thickness is 300nm,

and the light extraction efficiency of conventional deep UV LEDs with flat surface is

normalized to 1.


Deep UV AlGaN QW LEDs with III-nitride micro-dome (Micro-hemisphere) with

various p-type layer thickness: Ptype=300, Ptype=350nm, Ptype=400nm and Ptype=700nm.

The other conditions are peak=250nm, FWHM=10nm and D = 500nm. Blue dash line

represents the conventional deep UV AlGaN QWs LED with flat emission and p-type

layer thickness 300nm.

Figure 3-13. Light extraction efficiency of the TM polarized spontaneous emission

component for AlGaN QWs LEDs with AlGaN micro-domes (micro-hemispheres) emitting

at 250nm as a function of the p-type layer thickness. The micro-hemisphere diameter D is

500nm. The light extraction efficiency of conventional deep UV LEDs with flat surface is

plotted as a comparison.


Deep UV AlGaN QW LEDs with III-nitride micro-dome (Micro-hemiellipsoid) at

D=200nm/Ptype=300nm with micro-dome height h ranging from h=0 to h=275nm. Note

8

that h=0 and h=100nm represent deep UV LED with flat emission surface and micro-

hemisphere structure on p-type layer.






Figure 3-16. The total light extraction enhancement ratio as a function of micro-dome

height h for the diameter cases at D=100nm, D=125nm, D=200nm, D=300nm and

D=500nm. Yellow block dot on each curve indicates the light extraction efficiency with

micro-hemisphere structure (h=D/2). The P-type layer thickness is 300nm, and the light

extraction efficiency of conventional deep UV LEDs with flat surface is normalized to 1.

The optimized enhancement ratio of 2.4, 7.2, 7.3, 6.8 and 6.1 are obtained from micro-

dome at the diameter D=100nm, D=125nm, D=200nm, D=300nm and D=500nm,

respectively.

Figure 3-17. Far field emission pattern of the TM polarized spontaneous emission

component for AlGaN QWs LEDs with (a)flat surface; and (b)micro-hemispheres

(D=200nm); (c)micro-hemispheres (D=500nm); (d) micro-domes (D=200nm, h=175nm).

Figure 3-17. 3D far field emission pattern of the TM polarized spontaneous emission



Figure 3-19. The light extraction efficiency enhancement for deep UV AlGaN QWs LED

with roughened surface: (a) the schematic of deep UV LED with roughened surface

9

(RMS=40nm); (b) light extraction efficiency enhancement ratio as a function of

roughened surface RMS value from 0 to 40nm.

Figure 3-20. Far field radiation pattern from deep UV AlGaN QWs LED with roughened

surface (RMS=35nm) in forms of (a) 2D angular polar plot and (b)3D far field pattern.

Figure 3-21. Wavelength dependent refractive index for (a) AlN and (b) SiO2.


emission component for AlGaN QWs LEDs with SiO2 microspheres emitting at 250nm

as a function of the SiO2 microspheres diameter. The optimized light extraction efficiency

enhancement ratio is 1.6x by SiO2 microspheres with diameter 190nm.

Figure 4-1. Schematic of the configuration of III-nitride InGaN QWs LED device.

Figure 4-2. Spontaneous emission spectra (TE component and TM component (x50)) for

3-nm InxGa1-xN QWs LEDs with x= 0.1, 0.2, 0.25 and 0.3. TE spontaneous emission

component is dominant in InGaN QWs LED emitting in visible spectrum region.

Figure 4-3. 2D Schematics of III-nitride LED based on Thin-Film-Flip-Chip technology

package design (a) flip chip technology package design; and (b) Thin-Film-Flip-Chip

technology package design.

Figure 4-4. 2D schematics of the computational domain of the thin film flip-chip (TFFC)

InGaN QWs LEDs with (a) flat surface and (b) GaN micro-domes.

Figure 4-5. Schematic of TFFC InGaN QWs LED with interference between emission of

QWs and from bottom mirror.

Figure 4-6. Light extraction efficiency for the conventional TFFC InGaN QWs LEDs at

wavelength (a) peak=460nm and (b) peak=550nm with flat surface as a function of the p-

10

GaN layer thickness. N-GaN thickness is 2.5m. Dash line and solid dots represent the

theoretical fitting curve and the FDTD calculation results, respectively.

Figure 4-7. Light extraction efficiency analysis for conventional package design InGaN

QWs LED with p-GaN micro-domes: (a) source positioning dependence analysis along

micro-dome diameter; (b) total light extraction enhancement ratio.

Figure 4-8. Light extraction efficiency enhancement of TE polarized spontaneous

emission component for TFFC InGaN QWs LEDs with optimized p-GaN thickness

(195nm for peak=460nm, 230nm for peak=550nm as a function of GaN micro-

hemisphere diameter (D)).

Figure 4-9. (a) Geometric structure of the general micro-dome structure on n-GaN

emission surface of TFFC InGaN QWs LED with diameter D and height h; (b) light

extraction efficiency at peak=460nm from InGaN QWs TFFC LED with micro-dome

structures as a function of micro-dome height h for diameters D=500nm, D=1000nm and

D=1500nm.

Figure 4-10 Light extraction efficiency for the conventional TFFC InGaN QWs LEDs

(peak=460nm) with flat surface and with GaN micro-domes as a function of the p-GaN

layer thickness. Dash lines and solid dots represent the theoretical fitting curves and the

FDTD calculation results, respectively.

11

Analysis of Light Extraction Efficiency Enhancement for Deep Ultraviolet and Visible

Light-Emitting Diodes with III-Nitride Micro-Domes

Abstract

by

PENG ZHAO

III-nitride (In, Al, Ga-N) semiconductors are considered as wide-bandgap materials that

have promising applications in the next generation of lighting technology such as

ultraviolet (UV) / visible light-emitting diodes and laser diodes. The limitation occurred

in light extraction efficiency of III-nitride LEDs is attributed to the large refractive index

difference between III-nitride semiconductor and free space, which leads to significant

total internal reflection at the semiconductor and air interface. This thesis presents the

design of III-nitride micro-dome surface structures at the emission surface of LEDs to

enhance the extraction efficiency. The extraction efficiency analysis and micro-dome

structure designs were conducted for both deep-UV AlGaN quantum wells (QWs) LEDs

and thin-film flip-chip (TFFC) InGaN QWs visible LEDs by using three dimensional

finite difference time domain (3D-FDTD) method. The analysis results shows significant

enhancement of light extraction efficiency from both deep-UV LEDs and TFFC visible

LEDs with III-nitride micro-domes.

12

Chapter 1: Introduction

1.1 Lighting Energy Consumption and Solid State Lighting for Energy

Saving

Based on current archaeological record, the first lighting technology started 70000

years ago, which attempted to fill some non-flammable containers with natural

combustible materials such as wood or grass. As time passed, the first generation lamp

was developed by controlling rate of burning for whatever fuel-based combustible

materials such as natural oil and wax. The design for the controlling of burning fuel was

basically realized by applying wicks, chimneys or other similar device to make the

burning steady and continuous [1, 2].

The first electric lights were developed and commercialized by Sir Joseph Swan in

England and Thomas Edison in the United States in 1870 [1], since then lighting

technology powered by electricity becomes dominant in human society. The conversion

efficiency from electricity to light power turns into a significant issue in light

illumination technology.

The percentage of electricity power used in lighting was investigated by most recent

annual reported data from U.S. Energy Information Administration (EIA) - Annual

Energy Outlook 2012 (AEO 2012), which provides the basis for examination and

discussion of energy market trends and serves as a database for electricity usage

estimation [3]. Reference 3 is the recent published data provided by AEO2012 early

release overview, which shows the energy consumption record from 2009 to the

13

estimation till 2035 by four different sectors - residential, commercial, industrial and

transportation.

Figure 1-1. U.S. Annual Energy Outlook 2012 from Energy Information Administration

(EIA) - electricity delivered energy (a) and related losses (b) in sectors of residential,

commercial, industrial and transportation.

0

5

10

15

20

25

30

35

2005 2010 2015 2020 2025 2030 2035 2040

Ele

ctr

icit

y d

eli

ve

red

en

erg

y

by s

ecto

rs

(qu

ad

rill

ion

Btu

/ y

ea

r)

Reference years

Transportation

Industrial

Residential

Commercial

(a)

0

2

4

6

8

10

12

14

2005 2010 2015 2020 2025 2030 2035 2040

Ele

ctr

icit

y r

ela

ted

lo

sses

by s

ecto

rs

(qu

ad

rill

ion

Btu

/ y

ear)

Reference years

Transportation

Industrial

Residential

Commercial

(b)

14

By summarizing the energy in the form of electricity, the delivered energy and the

related energy losses for each sectors are plotted in figure 1-1 (a) and figure 1-1(b)

respectively in the unit of quadrillion Btu (1.05505585 × 1018

joules). In the sectors of

residential, commercial and industrial, the related losses are comparable with the total

delivered electricity energy, but the losses in transportation sector are trivial, which

means energy efficiency in residential, commercial and industrial applications needs to be

improved.

Based on the most recent data record, the ratio of lighting electricity used in

residential and commercial sectors was also studied from ‘EIA’s AEO2012 sectors key

indicator and consumption’, since industrial and transportation report do not include

lighting applications. [3] Based on approximation from EIA lighting energy consumption,

the ratio of light electricity of the total energy consumption for residential and

commercial application are calculated and plotted in the year range from 2009 to 2035 in

figure 1-2. The results indicate that in 2010 the electricity used for lighting by residential

and commercial sectors takes more than 10% of total U.S. electricity consumption.

15

Figure 1-2. Ratio of electricity consumption used for lighting in total energy for sectors

of residential and commercial.

The U.S. Department of Energy provides a comprehensive strategy to explain the

forecast for the ratio of lighting electricity consumption in total U.S. electricity

consumption - energy savings from potential application of solid state lighting. In figure

1-2, the electricity consumption using for lighting will significantly decrease in future

decades. This econometric model for light market relies on the estimation from projected

solid state lighting efficiency development, cost and reliability life time, which leads to

the reasonable forecast by U.S. Department of Energy and Next Generation Lighting

Industry Alliance [4-6].

As the next generation of illumination technology, light-emitting diodes, a type of

solid-state lighting offer the lighting market an advanced light source, which has the

0

2

4

6

8

10

12

14

2005 2010 2015 2020 2025 2030 2035 2040

Commercial

Residential

Ra

tio

of

lig

hti

ng

en

erg

y (

%)

Reference years

16

advantages of better light quality, high efficiency and longer operating life, as compared

with the conventional white light sources such as incandescent, halogen, fluorescent and

high intensity discharge [7-10]. The energy saving report especially benefits from LEDs

white sources is shown in figure 1-3, which indicates that due to market penetration of

LED, the electricity consumption for lighting in 2030 could potentially decrease ~46%

percent, relative to the baseline scenario [4].

Figure 1-3. Annual electricity consumption of lighting technologies and the electricity

savings resulting from the increased use of LEDs in general illumination applications,

disaggregated by sectors [4].

17

1.2 III-nitride LEDs Introduction

The main materials for solid-state lighting are III-V group semiconductors, which

are mainly based on III-phosphide and III-nitride materials. Figure 1-4 shows the energy

bandgap as a function of lattice constant for both zinc-blende III-phosphide and wurtzite

III-nitride semiconductor alloy. In the III-phosphide material system, (Alx, Ga1-x)1-yInyP is

lattice-matched to GaAs substrate which is able to grown thick bulk material by

straightforward epitaxial process. The direct bandgap available for AlGaInP is from

1.9eV (6500nm) to 2.2eV (560nm). Due to low the electron confinement potential inside

of AlInP, the carrier confinement becomes a challenge which leads to the consequence

that the emission wavelength from AlInGaP-based LED cannot be shorter than 580nm

[11-13].

Figure 1-4. Energy bandgap as a function of lattice constant for both zinc-blende III-

phosphide and wurtzite III-nitride semiconductor alloy [12].

18

The energy bandgap information for III-nitride wurtzite materials are also shown in

figure 1-4, in which InxGa1-xN is employed for visible LED for illumination application.

The tunable bandgap is related to Indium content fraction. By varying Indium content

ratio x from 0 (GaN) to 1 (InN), the bandgap of InxGa1-xN alloy shows tuning from 3.4eV

(365nm, UVA) to 0.7eV (1800nm, infrared), which exhibits much larger emission

wavelength range compared to III-phosphide materials [11-13].

1.3 Current Challenges to Pursue High Efficiency III-nitride LEDs

1.3.1 Internal Quantum Efficiency Limitation

Internal quantum efficiency (IQE) is defined as the proportion of all the electron-

hole recombination in III-nitride LED active region that are radiative and producing

photons. In the group of III-nitride wurtzite semiconductors, spontaneous polarization

exists in the material, which leads to significant reduction of the radiative recombination

of electrons and holes. Besides, the mismatch between GaN and InxGa1-xN will also lead

to a piezoelectric polarization. With higher Indium content in InxGa1-xN alloy, the

polarization effect in the active region of LEDs increases, which causes the spatial

separation and decreasing of electron and hole wavefunctions overlap, and in turn results

in low radiative efficiency [13]. Figure 1-5 shows the external quantum efficiency for

visible spectrum InxGa1-xN LEDs, which could be defined as the product of internal

quantum efficiency, electrical current injection efficiency and light extraction efficiency.

From the data shown in figure 1-5, the blue LEDs have higher external quantum

efficiency than green LEDs, which is due to the high Indium content in the active region

of green LEDs significantly suppresses the radiative efficiency. Recently, some

19

approaches based on novel QWs designs have been proposed to engineer the band

lineups of the InGaN QWs for enhancing the electron-hole wave function overlap,

including the staggered InGaN QW [14-19], type-II InGaN-GaNAs QW [20-22], strain-

compensated InGaN-AlGaN QW [23-25], InGaN-delta-AlGaN QW [26, 27], InGaN-

delta-InN QW [28, 29], and triangular InGaN QW [30]. In addition, significant efforts

have been focused on nonpolar / semipolar InGaN QWs growths for removing / reducing

polarization field in the QW [31-33].

Another factor that leads to low radiative efficiency for III-nitride LEDs is attribute

to very high dislocation density which acts as nonradiative recombination center [12].

The high dislocation density is caused by the lattice mismatch between underlying GaN

layer and sapphire substrate.

Figure 1-5. External quantum efficiency for visible spectrum LEDs. V() represents the

luminous eye response curve from CIE (International Commission on Illumination) [12].

20

1.3.2 Light Extraction Efficiency Limitation

Even if the internal quantum efficiency is optimized, the external quantum

efficiency of III-nitride LEDs still largely depends on the light extraction efficiency. The

low light extraction efficiency for conventional III-nitride LEDs with flat emission

surface is subjected to the total internal reflection due to the large refractive index

difference between III-nitride materials (~2.5) and free space (~1) interface, thus a large

amount of the light generated inside the LED active region will be trapped within the

semiconductors.

The theory analysis for total internal reflection could be predicted by Snell’s Law.

As shown in figure 1-6, the refractive index of GaN in visible spectrum is approximately

2.5. Parameter represents the critical angle for total internal reflection.

Figure 1-6. Light trapping in GaN due to total internal reflection with critical angle

The critical angle for total internal reflection could be calculated by equation 1-1

below:

1-1

AIR

p-GaN c

Escape cone

21

At the interface between GaN and free space in visible spectrum, the critical angle

for total internal reflection was calculated as 23.58º. Based on the calculated critical angle

in equation 1-1, the light extraction efficiency for conventional III-nitride LED with flat

surface is estimated by taking the ratio of emission area and total spherical area from

point source. The spherical emission area ( ) is calculated by the equation 1-2, as

a function of critical angle for total internal reflections.

∫

1-2

Assuming the total spherical area from point source could be expressed as , the

estimated light extraction efficiency is calculated by equation 1-3 as followed.

1-3

For GaN with refractive index 2.5 in visible spectrum region, the light extraction

estimated from total internal reflection analysis was calculated as ~4% of the total power

generated in the active region of LED. Thus, most of the light emitted from the active

region is trapped inside the semiconductor. Therefore, the enhancement of the light

extraction efficiency for III-nitride LEDs is extremely important to achieve high external

quantum efficiency.

1.4 Recent Approaches to Enhance Light Extraction Efficiency for III-

nitride LEDs

Many approaches have been proposed to enhance the light extraction efficiency for

III-nitride LEDs. One approach is to create random roughened emission surface at the top

22

layer, which is able to scatter more light from the interface of III-nitride semiconductor

and free space. The GaN surface roughness has been realized by photochemical etching

(PEC) technology [34, 35]. Figure 1-7 (a) shows the surface morphology after roughness

process, which exhibits non-uniform and uncontrollable surface feature shapes and sizes.

Thus, the variation of light extraction efficiency would be a serious issue from surface

roughness approach.

Figure 1-7. Existing approaches to enhance the light extraction efficiency by (a) surface

roughness; (b) photonic crystals; (c) sapphire lens and (d) SiO2 / polystyrene microlens

array [34-41].

The Fabrication of photonic crystal on the emission surface of III-nitride LED is

another commonly used approach to enhance the light extraction efficiency, which has

photonic crystals

sapphire lens

surface roughness

SiO2/polystyrene microlens

(a) (b)

(c) (d)

23

the advantages of large enhancement and high uniformity, as shown in figure 1-7 (b) The

mechanism for the enhancement stems from the photonic bandgap and scattering effects.

However, the feature size of photonic crystals is in the range of 300 to 500nm, which

requires high cost e-beam or holographic lithography process [36]. Another approach is

sapphire lens structure fabricated on substrate side, as shown in figure 1-7 (c). Sapphire

lens were also reported to contribute large light extraction efficiency enhancement, but

the fabrication process is complex [37].

Recently, Ee and co-workers proposed to enhance light extraction efficiency for

InGaN QWs LEDs by self-assembled SiO2 / polystyrene microlens array as shown in

figure 1-7 (d), which has the advantages of low cost, large production scale and high

uniformity [38-41]. But this approach leads to the limitation on the device operation

temperature, since the melting point for polystyrene material is ~240°C. Although the

absorbance of light by polystyrene in the visible portion of the electromagnetic spectrum

is very low, in the ultraviolet range the polystyrene tends to absorb light and the material

itself will have degradation issue, as shown in figure 1-7. There is a new ultraviolet

absorption band exhibiting at ~290nm for polystyrene as compared with Ethylbenzeze,

which is due to the associative interaction of pendant phenyl groups. [42] Thus, this

approach is not suitable to be applied for ultraviolet LEDs.

24

Figure 1-8. Absorption spectra from 1. Polystyrene 2. Ethylbenzeze [42].

1.5 Thesis Organization

This thesis is composed of 5 chapters. The introduction presents the motivation of

energy saving by solid state lighting technology, the challenges to pursue high efficiency

III-nitride LEDs as well as the existing approaches to enhance the light extraction

efficiency. In chapter 2, the finite difference time domain method was introduced for III-

nitride LED light extraction efficiency calculation. Chapter 3 will focus on analysis of

light extraction efficiency enhancement for deep ultraviolet AlGaN quantum wells LEDs

with III-nitride micro-domes by 3D-FDTD method. Chapter 4 will discuss the light

extraction efficiency enhancement for Thin-Film-Flip-Chip (TFFC) InGaN quantum

25

wells LEDs with GaN micro-domes. Lastly, Chapter 5 will summarize the work in this

thesis and suggest possible future works.

26

Chapter 2: Finite Difference Time Domain Method for Light Extraction

Efficiency Calculation of III-nitride LEDs

2.1 Finite Difference Time Domain Method (FDTD)

2.1.1 Introduction

The conventional ray tracing calculation method is widely used for designing optics

in the feature size much larger than the simulated wavelength of optical rays. However,

as the simulated objects dimensions are in the range of micron or submicron, the

traditional method of tracing of optical rays is not accurate, since in micro- and nano-

scale, the optical components will interact with the light by scattering, absorption and

polarization effects, in which the optical rays should be treated as electromagnetic waves.

The finite difference time domain (FDTD) method is a state of the art method to

calculate Maxwell’s equations with specific boundary conditions in complex geometries.

The differential form of Maxwell’s equations is solved by FDTD method to obtain

rigorous and accurate solutions for the electromagnetic wave propagation, instead of

approximate solutions like traditional ray tracing method. Since the FDTD algorithms

solve the Maxwell’s equations in time-domain, the calculation could cover full frequency

range by Fourier transformation. The Maxwell’s curl equations in differential forms

could be expressed in equations 2-1 and 2-2 as followed [43]:

221

121

Et

H

EHt

E

27

where E

and H

are electric and magnetic fields. , and represent the medium

permittivity, permeability and conductivity, respectively.

The permittivity of the material could be defined as:

32),(20)(0)( nr

where )( r is the complex relative dielectric constant which is related to refractive

index )(n as follows:

42)(2)( nr

The conductivity of the material is defined by the equation 2-5:

520)(5.0)( n

where , 0 and represent the material absorption coefficient, permittivity in vacuum

and the light angular frequency, respectively. In this work, Commercial simulation

software ‘FDTD Solutions 7.5’ from ‘Lumerical Solutions, Inc’ was employed to

calculate the light extraction efficiency for III-nitride LEDs.

2.1.2 Three Dimensional FDTD method and Yee’s Mesh

In three dimensions, the Maxwell’s curl equations could be developed into the

time-differentiated, spatial components in Cartesian coordinate system, which have six

28

electromagnetic field components in each respective axis: E zE yE x ,, and

H zH yH x ,, .The corresponding set of equations are derived as follows in equations 2-6:

fE zx

H y

y

H x

t

E z

eE yx

H z

z

H x

t

E y

dE xy

H z

z

H y

t

E x

cx

E y

y

E x

t

H z

bx

E z

z

E x

t

H y

az

E y

y

E z

t

H x

621

621

621

621

621

621

In the three spatial directions, all field components could be solved based on the set

of differential form of Maxwell’s equations above. For example, electric field component

in x direction could be solved by magnetic field components in y and z direction as well

as the previous value of itself.

In order to solve the time dependent Maxwell’s equations, FDTD algorithm

requires to calculate the electromagnetic field in grid steps in both spatial and time

domain, which should be allowed to be small enough than certain fractions of wavelength.

In this condition, the FDTD simulation is hard to be applied for optical components with

large dimensions, since the large computational work from spatial grids is restricted by

practical limitations of computational resources. In order to apply the Maxwell’s

29

equations by numerical method so as to realize FDTD calculation, Yee’s mesh was used

to enable analyzing the electromagnetic fields discretely in both space and time [44].

Figure 2-1 shows the unit cell of Yee’s mesh grid, which involves positioning of the E

and H

in the cubic-cell space lattice with offset 1n and2

1n (n is integer),

respectively. In three dimensions, the Maxwell’s curl equations could be developed into

the time-differentiated, spatial components in Cartesian coordinate system, which have

six electromagnetic field components in each respective axis as shown in figure 2-1:

E zE yE x ,, and H zH yH x ,, . All field components could be solved based on the set of

differential form of Maxwell’s equations. The spatial size of mesh is defined by x , y

and z , which directly related to the simulation time and memory requirements in

practical FDTD calculation.

Based on this model, the Maxwell’s equations were solved discretely in time, where

the time step t here is defined by the mesh size through the speed of light. When the

limit of mesh size goes to zero, the technique should be supposed to exactly represent the

accurate calculation of Maxwell’s equations. This model has been proved to be able to

simulate the typical geometrical structure of LEDs, integrated optics, planar waveguides

and devices, photonic crystal based devices and optical fibers.

In this calculation, the loss from material absorption and wavelength dependence of

the refractive index n() were taken into account [45]. The absorption in one unit Yee’s

mesh cell could be calculated as below:

30

)(722

1

2

1

2

/2

2/1

2/11

)(722

1

2

1

2

/2

2/1

2/11

)(722

1

2

1

2

/2

2/1

2/11

cHnxH

nzt

stE

nzt

tE

ny

bHnzH

nyt

stE

nyt

tE

nx

aHnxH

nyt

stE

nxt

tE

nz

where and are the permittivity and conductivity of the material, respectively.

Fig. 2-1. Yee’s mesh cell: Maxwell’s equations are solved discretely in unit of Yee’s

mesh cell.

2.2 Light Extraction Efficiency Calculation based on 3D-FDTD Method

2.2.1 Light Extraction Efficiency Calculation Method

The light extraction efficiency could be defined by taking the ratio of total extracted

light power to the total power emitted from active region of LEDs. To collect the escaped

Hx

Hz

Hy

Ex

Ex

Ey

Ey

Ey

Ez

Ez

Ez

Y

X

Z

31

light from emission surface, a far field power detection plane was set /n away from

emission surface as shown in figure 2-2 (a), where n is the wavelength dependent

refractive index. In this simulation work, the extracted power from LED emission

surface can be obtained by integrating the Poynting vectors over far field projection

surface.

The total power emitted from active region were calculated by surrounding the

dipole with a box of power monitors to record the net outflowing power through

integrating Poynting vectors in the near field of dipole source, as shown in figure 2-2 (b).

Figure 2-2. Power collecting setting for light extraction efficiency calculation: (a)

extracted power detection plane; (b) power box surrounding around the dipole source.

The power integral equations are shown below by equation 2-8:

822

1 SdPreal

AIR

Detection Plane

p-GaN

n-GaN1

(a)

PML

Perfect

Mirror

InGaN/GaN

MQW

Reflective layer(b)

Surrounding power box

32

where is the calculated energy, P is the Poynting vector depending on light angular

frequency, and dS is the surface normal.

In both far field and dipole source power box, Poynting vectors could be calculated

from electric field component E

based on the plane wave approximation, as follows:

922

0

0

EnP

2.2.2 Far Field Radiation Pattern Calculation Method

The far field radiation pattern for LEDs is always an important issue in practical

application. At far field, the LED could be studied as an equivalent point light source,

which begins at a distance much larger than the dimensions of light source itself. Thus,

the far field intensity is not related to the distance from source, but depends on the radiant

angle.

In this study, all the far field data were projected onto a hemisphere upon the

emission surface with a radius mR 1 . In spherical coordinates, the power integral

calculated in (2-8) could be specified as follows:

102

,

sin2,,2

1

ddRP

33

where and correspond the angle variables in spherical coordinates system, as shown

in figure 2-3 below.

Figure 2-3. Far field projection spherical coordinates system: (a) 3D far field hemisphere

projection surface at 1 meter away; (b) 3D spherical coordinates system and

corresponding Cartesian coordinates system.

(a) (b)

34

Chapter 3: Analysis of Light Extraction Efficiency Enhancement for Deep

Ultraviolet AlGaN Quantum Wells LEDs with III-nitride Micro-domes

In this chapter, the III-nitride micro-domes were proposed to be applied to III-nitride

LEDs on the emission surface to enhance light extraction efficiency. The analysis of light

extraction efficiency enhancement for AlGaN quantum wells (QWs) based deep

ultraviolet (UV) light-emitting diodes (LEDs) with III-nitride micro-dome structures

(mciro-hemisphere and mciro-hemiellipsoid) on the p-type layer were studied and

compared to that of the conventional deep UV LEDs with flat surface. First, the

transverse electric (TE) and transverse magnetic (TM) components of the spontaneous

emission of AlGaN QWs with AlN barriers were calculated by using self-consistent 6-

band k∙p method. The light extraction efficiency of the AlGaN/AlN QWs deep UV LEDs

emitting at 250 nm with AlGaN micro-domes was calculated by three dimensional finite

difference time domain (3D-FDTD) method. The optimization based on the effects of the

III-nitride micro-dome diameter and height as well as the p-type layer thickness on the

light extraction efficiency was studied systematically. To confirm the superiority of III-

nitride micro-domes on the light extraction efficiency enhancement of deep UV AlGaN

QWs LED, other proposed techniques such as surface roughness and SiO2 microlens

array were also calculated as comparison.

3.1 Concepts of III-nitride Micro-dome Structures

For III-nitride LEDs with p-GaN layer emission surface, the large refractive index

difference GaN (n~2.5) and free space (n~1.0) leads to significant total internal reflection.

35

The critical angle for light escape cone from conventional LED with flat p-GaN emission

surface was calculated as 23.58° in chapter 1 section 1.3.2, which indicates that the

dominant optical modes for light generated from LED active region are trapped inside of

semiconductor. Figure 3-1 (a) shows the schematic of conventional III-nitride LED with

flat emission surface with a narrow photon escape cone, due to significant total internal

reflection.

Figure 3-1. The schematics of (a) conventional flat surface LEDs with a narrow photon

escape cone; (b) III-nitride micro-domes LEDs with increased effective escape cone.

AIR

n-type

p-typeActive

Region

1

(a)

Escape cone

AIR

n-type

p-type

(b)

Increased effective escape cone

Active

Region

36

In order to suppress the effect from total internal reflection and increase the effective

photon escape cone, III-nitride micro-dome structure was designed on the top emission

surface by self-assembled approaches. The GaN micro-domes could be formed by

reactive ion etching (RIE) of the GaN layer with a self-assembled dielectric microspheres

monolayer as mask. The 3D schematics for the formation of III-nitride micro-dome

structures are shown in figure 3-2.

Figure 3-2. 3D schematics for the formation process of III-nitride micro-domes: (a)

original sample with flat surface; (b) coated with microspheres monolayer; (c) the

intermediate state of RIE pattern transferring; (d) formation of III-nitride micro-domes.

The III-nitride micro-domes are hexagonally close-packed and shown as periodical

distributed micro-hemisphere or micro-hemiellipsoid structures on p-type emission

(a) (b)

(c) (d)

37

surface. Due to the dome-like morphology on the top of p-GaN layer, photons generated

from QWs active region of III-nitride LED are allowed to uniformly scattered out from

the interface of GaN and free space, thus resulting in a wider effective escape cone

compared to the conventional III-nitride LED with flat emission surface. Figure 3-1 (b)

shows the III-nitride LED with micro-domes array on p-type emission surface. With the

increased effective photon escape cone, the light extraction efficiency could be enhanced

uniformly by the photon scattering from hemi-spherical micro-dome structures.

The shape profile of III-nitride micro-dome structures could be externalized by

elliptic function in terms of micro-dome diameter (D) and micro-dome height (h), as

shown in figure 3-3. The optimization of micro-dome structures for maximizing light

extraction efficiency of III-nitride LED will involve the tuning of micro-dome diameter

(D), micro-dome height (h) and p-type layer thickness (Ptype). Figure 3-3 (a), (b) and (c)

show the schematic of III-nitride micro-domes LED with different diameter (D) vs.

height (h) aspect ratios as D/2>h, D/2=h and D/2<h respectively. Note that figure 3-2 (b)

represents the case of micro-hemisphere structure. We have already experimentally

demonstrated that the feature shape of III-nitride individual micro-dome structure could

be precisely controlled in practical fabrication.

38

n-type

p-type

D

(c)

Ptype

Figure 3-3. The schematic of III-nitride micro-domes LED with different diameter (D) vs.

height (h) aspect ratios (a) D/2>h; (b) D/2=h (c) D/2<h.

3.2 Introduction of Deep Ultraviolet AlGaN Quantum Wells LEDs

3.2.1 Deep UV AlGaN QWs LED Configuration and Application

Ultraviolet (UV) light is from the solar spectrum region between the wavelength

range from 100nm to 400nm [46-48]. As shown in figure 3-4, the wavelength region of

UV light is longer than X-ray but short than visible light. Within the UV spectrum, the

electromagnetic spectrum of UV light could be subdivided into four regimes: UV vacuum,

UV C, UV B and UV A, from shorter wavelength to longer wavelength. Deep UV

spectrum region refers to UV C band with wavelength range from 200nm to 280nm.

D

n-type

p-typeh

(a)

n-type

p-type

D

(b)

39

Figure 3-4. Ultraviolet (UV) radiation spectrum: UV Vacuum (100nm-200nm), UV C

(200nm-280nm), UV B (280nm-320nm) and UV A (320nm-400nm).

Figure 3-5. Ultraviolet (UV) applications chart in spectrum: UV A, UV B and UV C.

10-1 1 10 103 104 105 106 107

Visible Light

X-ray-ray Infrared Microwaves

102

UVWavelength (nm)

UV Vacuum UV C UV B UV A

100 200 280 320 400

Mercury Detectors

Disinfection

Decontamination of surfaces/water

Optical sensors

Photolithography

Drug detection

Protein analysis

Medical imaging cells

Label tracking

Bacterial identification

Specimen staining

Clean room inspection

Examination of fine art

UV curing

Gradient sampling

Solar experimentation

Gel Electrophoresis

UV-C

(200~280nm)

UV-A

(320~400nm)

UV-B

(280~320nm)

Solid state lighting

Light therapy

40

The applications from ultraviolet light mainly stem from the high energy nature of

itself. On the basis of three spectrum regions: UV C, UV B and UV A, figure 3-4 shows

the typical ultraviolet (UV) applications chart with overlapping of spectrum. The major

applications focus on solid state light, disinfection, detecting and etc.

Ultraviolet (UV) light-emitting diodes (LEDs) based on wide band gap AlGaN

quantum wells (QWs) with AlN barriers have a wide range of applications in water/air

purification, white light illumination, spectrometry and medical phototherapy [49-52].

Different from the InGaN QWs based LEDs [53-55]. The typical configuration for deep

UV layer structure is schematically shown in figure 3-6 [50], which have been widely

adopted by existing research [56-58].

Figure 3-6. Configuration schematic of deep UV AlGaN QW LED grown on sapphire

substrate [50].

In practical fabrication process, the deep UV epitaxial layers were deposited on

sapphire substrate by metal-organic chemical vapor deposition (MOCVD) technique. The

41

first deposition process involves the AlN buffer layer directly grown on sapphire

substrate, which aims to decrease the screw dislocation density [58]. The AlN/AlxGa1-xN

super lattice structures were subsequently grown above AlN buffer layer, which allows

the growing of crack free thick n- AlxGa1-xN cladding layer. On the top of cladding layer

AlGaN multiple quantum wells active region was deposited, which was capped by p-type

AlGaN electron block layer and p-type GaN contact layer successively.

3.2.2 Challenges in Achieving High Efficiency Deep UV AlGaN QWs LED

It is still challenging to pursue high performance UV LEDs with high internal

quantum efficiency due to the challenges to grow high Al-content AlGaN QWs active

region and the challenges to grow high material quality, efficient conducting and

transparent p-type layer. The low external quantum efficiency of the UV and deep UV

III-nitride LEDs is also significantly attributed to the low light extraction efficiency, due

to 1) total internal reflection from the high refractive index of the nitride semiconductors

(n~2.4) in contrast to that of the free space (n=1), and more importantly, 2) domination of

the transverse magnetic (TM) component of the spontaneous emission from the high Al-

content AlGaN QWs active region, where TM polarization is polarized along the

direction normal to the surface, leading to extremely low total light extraction efficiency.

Figure 3-7 shows the state of the art of the external quantum efficiency (EQE) for III-

nitride UV LEDs. Compared with visible III-nitride LEDs, the EQE for UV LEDs is

fairly low especially at deep UV spectrum region around 250nm (EQE≤1%) [59-61].

42

Figure 3-7. The state of the art of III-nitride UV LEDs external quantum efficiency [59-

61].

In this chapter, we propose the computational analysis of light extraction efficiency

enhancement for AlGaN/AlN QWs based deep UV LEDs emitting at 250nm by forming

the III-nitride micro-domes on the p-type layer. Studies show that significant light

extraction efficiency enhancement is achievable by optimizing micro-dome size and

height, compared to conventional deep UV AlGaN QWs LED with flat p-type surface.

Since recent approaches for enhancing the light extraction efficiency of III-nitride LEDs

mainly focus on the visible InGaN LEDs by using surface roughening, photonic crystals,

SiO2/polystyrene colloidal microspheres, and graded refractive index materials, which are

introduced in Chapter 1, section 1.4. Very few studies focus on the light extraction

efficiency for UV and deep UV III-nitride LEDs.

43

3.3 Polarization Analysis of Spontaneous Emission from AlGaN

Quantum Well

In this section, the polarization of AlGaN QW spontaneous emission from deep UV

LED was studied by calculating the transverse Electric (TE) and transverse

Magnetic(TM) components ratio. In AlGaN QWs based UV LEDs, the heavy hole (HH),

light hole (LH) and crystal-field split-off hole (CH) energy bands in the valence band

cross over between HH/LH and CH bands [62]. For low Al-content AlxGa1-xN QWs

(x<0.66), the dominant transition is between the conduction and HH/LH bands, that is

transverse electric (TE) polarized spontaneous emission component. For high Al-content

AlxGa1-xN QWs (x>0.66), the dominant transition is between the conduction band and

CH band, which is TM polarized spontaneous emission component. To illustrate this,

figure 3-8 plots the spontaneous emission spectra (Rsp) for 3nm AlxGa1-xN QWs with

AlN barriers with x=0.58, 0.62, 0.66 and 0.7, respectively. The calculations of the band

structure and wave functions for AlGaN QWs were carried out by using self-consistent 6-

band k∙p method for wurtzite semiconductors, taking into account the valence band

mixing, strain effect, polarization fields, and carrier screening effect [63, 64]. The band

parameters for the III-nitride alloys utilized in our calculations were obtained from

references 65 and 66. The spontaneous emission spectra were calculated at carrier density

n=1x1019

cm-3

. From figure 3-8, for the AlGaN QWs with x<0.66 (>250nm), the

spontaneous emission is dominant with the TE polarized component. As the Al-content

increases, the TM polarized component becomes the domination of the total spontaneous

emission spectra Rsp. Therefore, in AlGaN QWs based deep UV LEDs, it is important to

44

design device structures to enhance the light extraction efficiency for the TM polarized

spontaneous emission component.

Figure 3-8. Spontaneous emission spectra (TE and TM modes) for AlxGa1-xN QWs LEDs

with x=0.58, 0.62, 0.66, and 0.7. TM spontaneous emission component becomes

dominant with x>0.66.

3.4 Calculation of Light Extraction Efficiency by 3-D FDTD Method

In this work, we propose to computational analysis of light extraction efficiency

enhancement for AlGaN/AlN QWs based deep UV LEDs emitting at 250nm by forming

micro-dome structures on the p-type layer, which was compared with the conventional

AlGaN/AlN QWs based deep UV LEDs with flat emission surface. In this study, the

LED light extraction efficiency calculation was performed by using three-dimensional

0

50

100

150

200

250

300

350

400

450

230 240 250 260 270

3 nm AlxGa1-xN/AlN QW

x=0.66

Sp

on

tan

eo

us

Em

iss

ion

(a

. u

.)

Wavelength (nm)

x=0.7

x=0.62x=0.58

TE

TM

45

finite difference time domain (FDTD) method. Note that the feature size of III-nitride

micro-domes set in simulation work is in the range of submicron to micron, which is

comparable to the emission wavelength from the AlGaN QWs active region, so the

traditional calculation of light extraction efficiency based on ray tracing is not as

accurate. The 3-D FDTD simulation takes into account the frequency dependence of the

refractive index of the III-nitride alloys [45]. The light extraction efficiency is defined as

the ratio of total extracted light power to the total power emitted from AlGaN QWs. In

this simulation, the extracted power from LED surface can be obtained by integrating the

Poynting vectors over far field projection surface, and the total power emitted from

AlGaN/AlN QWs based deep UV LEDs were calculated by Poynting vectors integrated

surrounding the near field of dipole source. As shown in figure 3-9, a single dipole source

model with TM polarization was placed in the AlGaN QWs active region. The boundary

condition for the simulation area is perfectly matched layer (PML) boundaries which

absorb electromagnetic energy incident upon them. The near field detection plane is set

as /n away from the top surface of p-type emission surface, where is the peak

emission wavelength in vacuum from the QWs and n represents the refractive index of

the media. In AlGaN QWs deep UV LEDs, n = 2.39643 was chosen as the refractive

index at 250nm wavelength in AlN medium. The mesh step is set as less than /10n, and

the average grid points are estimated around 500000 in the computational domain, which

is supposed to generate reasonable accuracy in light extraction efficiency calculation. The

light extraction efficiency calculations were performed for both conventional AlGaN

QWs deep UV LEDs with flat surface and the deep UV LEDs with III-nitride micro-

domes on top of the device. Studies show that the light extraction efficiency of a single

46

dipole source has strong dependence on the position of the single dipole source relative to

the micro-dome structures. In this study, we took into account the position dependence of

the light extraction efficiency of the dipole sources and obtained the average value of

light extraction efficiency of the deep UV AlGaN QWs LEDs with III-nitride micro-

domes on p-type emission surface.

Figure 3-9. 2D Schematics of the computational domain in 3D FDTD analysis of deep

UV AlGaN QWs LEDs with (a) flat surface; and (b) AlGaN micro-domes on p-type

emission surface for enhancing light extraction efficiency.

3.5 Light Extraction Efficiency Optimization of III-nitride Micro-domes

To study the effect of the III-nitride micro-dome (hemisphere and hemiellipsoid)

diameter D and height h, as well as p-type layer thickness Ptype on the light extraction

efficiency for the TM polarized spontaneous emission component, the source positioning

dependence analysis of light extraction efficiency was performed for deep UV AlGaN

AIR

Detection Plane

PMLn-type

p-typeAlGaN/AlN

MQW

1

(a)

AIR

n-type

p-type

(b)

Detection Plane

47

QW LEDs with III-nitride Micro-dome (Micro-hemisphere) emitting at peak=250nm

with full width half maximum (FWHM) 10nm.

3.5.1 Effect of Micro-dome (Micro-hemisphere) Size

As shown schematically in Figure 3-3, two parameters: micro-dome diameter (D)

and p-type layer thickness (Ptype), were used for the extraction efficiency calculation for

deep UV AlGaN QWs LED with III-nitride micro-domes (micro-hemisphere) on the p-

type surface. Figure 3-10 plots the light extraction efficiency of the TM dipole sources

(=250nm) located along the diameter of the micro-dome with various diameters

(D=100nm up to 500nm). The top p-type layer thickness is 300nm. Studies show that the

light extraction efficiency strongly depends on the position of the TM dipole source along

the diameter of the micro-domes with D>300nm - the extraction efficiency reaches

maximum for the dipole source locating around the center region under the micro-domes,

and decreases along the radius of the micro-domes. For D<300nm, the light extraction

efficiency of the TM polarized Rsp is relatively constant for the dipoles locating along the

diameter of the microspheres, which is probably due to the fact that smaller size of micro-

domes induce more flat surface morphology of top surface so that the position

dependence of TM dipole source is not as obvious as larger size of micro-domes.

48


Deep UV AlGaN QW LEDs with III-nitride micro-dome (Micro-hemisphere) at diameter

sizes: D=0, D=100nm, D=200nm, D=300nm, D=400nm and D=500nm. The other

conditions are peak=250nm, FWHM=10nm and Ptype = 300nm.

The light extraction efficiency of the AlGaN QWs LEDs (=250nm) after

considering source position dependence for different diameter of micro-dome

(hemisphere) were calculated. Figure 3-11 plots the ratio of the light extraction efficiency

enhancement of the AlGaN QWs LEDs (peak=250nm, FWHM=10nm) with III-nitride

micro-domes (hemispheres) as a function of the micro-dome (hemisphere) diameter (D).

The top p-type layer thickness is constant of 300nm. Note that the extraction efficiency

enhancement for D=0 represents the case for the conventional LEDs with flat surface,

which is normalized to 1. As the micro-hemisphere diameter D increases, the light

extraction efficiency enhancement ratio will increase. The enhancement ratio increases

0

2

4

6

8

10

12

-150 -100 -50 0 50 100 150

500nm

400nm

300nm

200nm

100nm

flatlL

igh

t E

xtr

acti

on

Eff

icie

ncy (

a.u

.)

TM Dipole Source PositionD/2 0D/3 D/6 D/2D/3D/6

D=500nm

D=400nm

D=300nm

D=200nm

D=100nm

D= 0flat surface

P type thickness = 300nm

TM Mode

= 250nm

49

significantly from 1 (D=0) to 5.7 (D=200nm). As the micro-hemisphere diameter D

increases from D=200nm to D=600nm, the enhancement ratio increases slightly. From

the device fabrication point of view, it is favorable to form relatively small diameter of

the mciro-domes (D<200nm) due to 1) the limited thickness of the top p-type layer (~

200-300nm) and 2) the potential effect of the micro-hemispheres fabrication on the

AlGaN QWs active region if the radius of the dielectric microspheres is close to the p-

type layer thickness.


emission component for AlGaN QWs LEDs with AlGaN micro-hemispheres emitting at

250nm as a function of the micro-hemisphere diameter. The P-type layer thickness is 300nm,

and the light extraction efficiency of conventional deep UV LEDs with flat surface is

normalized to 1.

0

1

2

3

4

5

6

7

0 100 200 300 400 500 600Micro-hemisphere Diameter D (nm)

To

tal L

igh

t E

xtr

ac

tio

n E

ffic

ien

cy

En

ha

nc

em

en

t R

ati

o

P-type thickness = 300nm TM Mode

= 250nm

conventional LEDs with flat surface

50

3.5.2 Effect of p-type Layer Thickness

The source positioning dependence for the light extraction efficiency of the TM

dipole sources located along the diameter of the micro-domes with various p-type layer

thicknesses was also investigated. Figure 3-12 shows the light extraction efficiency as a

function of TM dipole source position along the micro-dome diameter. The p-type layer

thickness ranges from Ptype=300nm to Ptype=700nm. The diameter of micro-dome is

chosen as D=500nm, because figure 3-11 shows that 500nm microsphere has enough

large light extraction efficiency enhancement without the potential fabrication effect on

the AlGaN QWs active region. Note that in the conventional LEDs calculation with flat

surface, the p-type layer thickness was chosen as Ptype = 300nm. The studies show that

the light extraction efficiency keeps having dependence on the position of the TM dipole

source along the diameter of the micro-domes with Ptype 400nm, and in this regime, the

extraction efficiency reaches maximum for the dipole source locating around the center

region under the micro-domes, and decreases along the radius of the microspheres. For

Ptype > 400nm, the extraction efficiency shows less dependence of the dipoles locating

along the diameter of the micro-domes. As comparison, the conventional deep UV

AlGaN QWs LED with flat emission was calculated with p-type layer thickness 300nm,

shown as blue dash line in figure 3-12.

51


Deep UV AlGaN QW LEDs with III-nitride micro-dome (Micro-hemisphere) with

various p-type layer thickness: Ptype=300, Ptype=350nm, Ptype=400nm and Ptype=700nm.

The other conditions are peak=250nm, FWHM=10nm and D = 500nm. Blue dash line

represents the conventional deep UV AlGaN QWs LED with flat emission and p-type

layer thickness 300nm.

The total the light extraction efficiency of the TM polarized spontaneous emission

component for the deep UV AlGaN QWs LEDs (peak=250nm) with various p-type layer

thickness was studied based on the integration of source positioning dependence, and the

results are shown in figure 3-13. The light extraction efficiency of the TM polarized

component for the conventional LEDs with p-type thickness of 300nm is plotted as

reference, shown as blue dash line in figure 3-13. The light extraction efficiencies of the

TM polarized component for the LEDs with AlGaN micro-dome (micro-hemisphere)

0

2

4

6

8

10

12

-150 -100 -50 0 50 100 150

300nm

350nm

400nm

700nm

CV_300

D = 500nm Ptype=300nm

Ptype=350nm

Ptype=400nm

Ptype=700nm

Ptype=300nm

flat surfaceL

igh

t E

xtr

ac

tio

n E

ffic

ien

cy (

a.u

.)

D/2 0D/3 D/6 D/2D/3D/6TM Dipole Source Position

TM Mode

= 250nm

52

(D=500nm) were calculated for various p-type layer thickness from 300nm up to 700nm.

Large enhancement of the light extraction efficiency was observed for different p-type

layer thickness. The enhancement factor ranges between 5.8-6.2 times for the deep UV

LEDs with III-nitride micro-domes (D=500nm) as compared to that of the conventional

LEDs with flat surface. The LEDs with p-type layer thickness of 300nm shows the

largest enhancement of 6.2 times. Considering the growth challenges of high quality p-

type layer, relative thin p-type layer thickness ranges between 200-300nm is preferable

for the epitaxy of the LED device.

Figure 3-13. Light extraction efficiency of the TM polarized spontaneous emission

component for AlGaN QWs LEDs with AlGaN micro-domes (micro-hemispheres) emitting

at 250nm as a function of the p-type layer thickness. The micro-hemisphere diameter D is

500nm. The light extraction efficiency of conventional deep UV LEDs with flat surface is

plotted as a comparison.

0

0.5

1

1.5

2

2.5

3

3.5

4

0 200 400 600 800 1000

LEDs with micro-hemispheres (D =500nm)

conventional LEDs with flat surface (D=0)

P-type Layer Thickness (nm)To

tal L

igh

t E

xtr

acti

on

Eff

icie

ncy (

a.u

.)

TM Mode = 250nm

53

3.5.3 Effect of Micro-dome Size and Height (Micro-hemiellipsoid)

The light extraction efficiency from deep UV AlGaN QWs LED with micro-domes

(micro-hemiellipsoid) was studied by tuning the micro-dome height h (h≠D/2). The

results show that optimized micro-dome structure for the maximum light extraction

efficiency is not necessary occurred from the micro-domes with h=D/2 (micro-

hemisphere). The geometric structure of the general micro-dome structure on p-type layer

is shown in figure 3-3 (a) and (c), where h≠D/2. The light extraction efficiency of the TM

dipole sources located along the diameter of the micro-domes with different micro-dome

height was also investigated.






0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

-150 -100 -50 0 50 100 150

h=0

h=25nm

h=50nm

h=75nm

h=100nm

h=125nm

h=150nm

h=175nm

h=200nm

h=225nm

h=250nm

h=275nm

D/6D/3D/2 D/6 D/3 D/2

D=200nm

TM mode

= 250nm

Lig

ht

ex

tra

cti

on

eff

icie

nc

y (

a.u

.)

0

TM dipole source position

flat surface

hemisphere

54

Figure 3-14 shows the light extraction efficiency as a function of TM dipole source

position along the micro-dome with diameter D=200nm with micro-dome height ranging

from h=0 (flat surface) to h=275nm. The study indicates that the position dependence of

TM source along diameter is obvious when h>125nm which means the micro-dome with

h>D/2 (micro-hemiellipsoid) have potential to further optimize the light extraction

efficiency from deep UV AlGaN QWs LED.






Similarly, the source position dependence for micro-domes with diameter 500nm

was also investigated and the results are shown in figure 3-15 for height increasing from

h=0 (flat surface) to h=250nm (micro-hemisphere). The results show the TM source

0

2

4

6

8

10

12

-150 -100 -50 0 50 100 150

h=0

h=25nm

h=50nm

h=75nm

h=100nm

h=125nm

h=150nm

h=175nm

h=200nm

h=225nm

h=250nm

D/6D/3D/2 D/6 D/3 D/2

D=500nm

TM mode

= 250nm

TM dipole source position

Lig

ht

extr

acti

on

eff

icie

ncy (

a.u

.)

flat surface

hemisphere

55

position dependence for light extraction efficiency becomes more obvious with the

increasing of micro-dome height - the extraction efficiency reaches maximum for when

dipole source locates around the center region, and also decreases along the radius of the

micro-domes, and the total extraction efficiency increases with the increasing of height

from h=0 (flat surface) to h=250nm (micro-hemisphere). Due to the limitation of the total

p-type layer thickness (Ptype=300nm), the structure only calculate till h=D/2. With thicker

p-type layer, the extraction efficiency for micro-dome at D=500nm could be further

optimized by calculation the micro-dome structure with h>D/2.

Taking into account the TM source position dependence, the light extraction

efficiency enhancement ratio was calculated in figure 3-16 for micro-dome diameter sizes

D=100nm, D=125nm, D=200nm, D=300nm and D=500nm with different micro-dome

height ranging from h=0 (flat surface) to h=250nm. The cases for h>250nm were not

calculated further due to the limitation of p-type layer thickness (Ptype=300nm). For

practical LED device fabrication, less potential effect on AlGaN QWs active region if the

III-nitride micro-domes are distant from the QWs. Note that the yellow block dot on each

curve indicates the light extraction efficiency with micro-hemisphere structure (h=D/2).

The light extraction efficiency at h=0 represents the case for the conventional LED with

flat surface. From figure 3-16, the optimized micro-dome structure for the highest light

extraction efficiency shows when h>D/2 for the diameter cases at D=100nm, D=125nm

and D=200nm, while for the diameter cases at D=300nm and D=500nm, the light

extraction efficiency increases as the increasing of micro-dome height, which is due to

the limitation of p-type layer thickness. For deep UV AlGaN QWs LED with III-nitride

micro-domes emitting at 250nm, the light extraction efficiency could be optimized by

56

tuning both the diameter D and the height h of the micro-domes. The optimized

enhancement ratio of 2.4, 7.2, 7.3, 6.8 and 6.1 could be achieved at peak=250nm for the

diameter at D=100nm, D=125nm, D=200nm, D=300nm and D=500nm, respectively.

Figure 3-16. The total light extraction enhancement ratio as a function of micro-dome

height h for the diameter cases at D=100nm, D=125nm, D=200nm, D=300nm and

D=500nm. Yellow block dot on each curve indicates the light extraction efficiency with

micro-hemisphere structure (h=D/2). The P-type layer thickness is 300nm, and the light

extraction efficiency of conventional deep UV LEDs with flat surface is normalized to 1.

The optimized enhancement ratio of 2.4, 7.2, 7.3, 6.8 and 6.1 are obtained from micro-

dome at the diameter D=100nm, D=125nm, D=200nm, D=300nm and D=500nm,

respectively.

0

1

2

3

4

5

6

7

8

0 50 100 150 200 250 300

D=100nm

D=300nm

D=500nm

D=125nm

D=200nm

Lig

ht

ex

trac

tio

n e

ffic

ien

cy

en

ha

nc

em

en

t ra

tio

Micro-dome height h (nm)

hemisphere

TM mode

= 250nm

57

3.6 Far Field Radiation Pattern Calculation

Figure 3-17 shows the far field TM polarized emission pattern in 2D angular plot

(with 90º as the normal to the LED emission surface) for AlGaN QWs deep UV LEDs

with flat surface, with micro-hemispheres (D=500nm and D=200nm, Ptype =300nm) and

with micro-domes (D=200nm, h=175nm, Ptype =300nm). The far field emission pattern

indicates that the LED structure with III-nitride micro-hemispheres has significant

enhancement of the TM emission component for a wide range of angles, especially in the

directions normal to the LED device surface. After tuning the height of the micro-

hemisphere to change to micro-dome structures, the light extraction efficiency could be

further enhanced.

Figure 3-17. Far field emission pattern of the TM polarized spontaneous emission



0

30

60

90

120

150

180

(a)

(b)

(c)

(d)

(a)

X3

3X Conventional LEDs with flat surface

(b) LEDs with micro-hemispheres (D=200nm; Ptype=300nm)

(c) LEDs with micro-hemispheres (D=500nm; Ptype=300nm)

(d) LEDs with micro-domes (D=200nm; Ptype=300nm; h=175nm)

58

The increase in far field radiance of the deep UV LEDs with III-nitride micro-

domes can be attributed to the enhanced scattering of photons and enlargement of the

photon escape cone from the hemispherical shaped micro-domes. In order to further show

the light extraction efficiency enhancement from far field pattern, figure 3-18 shows the

3D far field pattern corresponding to each curve in figure 3-17, which also indicates the

large significant enhancement from III-nitride micro-domes especially in the directions

normal to the emission surface.

(a)

conventional deep UV LEDs

deep UV LEDs with nitride micro-hemispheres

D=500nm

(b)

59

Figure 3-18. 3D far field emission pattern of the TM polarized spontaneous emission



deep UV LEDs with nitride micro-hemispheres

D=200nm

(c)

deep UV LEDs with nitride micro-domes

D=200nm; h=175nm

(d)

60

3.7 Light Extraction Efficiency Enhancement from Other Approaches

To enhance the light extraction efficiency for deep UV AlGaN QWs LED, other

popular approaches such as surface roughness and SiO2 microlens array were calculated

by 3D-FDTD method. The enhancement ratio by surface roughness and SiO2 microlens

array were obtained from rough optimization for each technique, which are compared to

the light extraction efficiency enhancement from III-nitride micro-dome structures. The

comparison confirms the superiority of III-nitride micro-domes for the light extraction

efficiency enhancement of deep UV AlGaN QWs LED with TM polarized spontaneous

emission in terms of high uniformity and large enhancement ratio.

3.7.1 Surface Roughness

The surface roughness on p-type emission layer was defined by correlated length

and root mean square (RMS) which are statistical values to quantitatively the magnitude

of a varying level. In this study, the correlated length was set constantly typical value

30nm and the RMS value increase from 0 (flat emission surface) to 40nm. For deep UV

LED with p-type layer thickness 300nm, the roughness at RMS=40nm will potentially

effect the AlGaN QWs active region, as shown in figure 3-19 (a). Above surface

roughness with RMS=40nm, the surface dent feature will reach AlGaN QWs active

region. Figure 3-19 (b) shows the light extraction efficiency enhancement ratio as a

function of roughened surface RMS value from 0 to 40nm. The results indicate that the

light extraction efficiency for deep UV AlGaN QWs LED with roughened surface will

increase with the level of roughness. The highest extraction efficiency enhancement ratio

61

(5.3) occurs at the roughness of RMS=40nm, which is not as beneficial as the

enhancement from III-nitride micro-domes on p-type emission surface.

Figure 3-19. The light extraction efficiency enhancement for deep UV AlGaN QWs LED

with roughened surface: (a) the schematic of deep UV LED with roughened surface

(RMS=40nm); (b) light extraction efficiency enhancement ratio as a function of

roughened surface RMS value from 0 to 40nm.

Mean line

Quantum well

active region

N-AlN

P-AlN

RMS = 40nm

(a)

P-type

N-type

0

1

2

3

4

5

6

0 10 20 30 40 50

Lig

ht

extr

acti

on

eff

icie

ncy

en

ha

nc

em

en

t ra

tio

Roughened surface RMS (nm)

TM mode

= 250nm

conventional LED with flat surface

(b)

62

Figure 3-20. Far field radiation pattern from deep UV AlGaN QWs LED with roughened

surface (RMS=35nm) in forms of (a) 2D angular polar plot and (b)3D far field pattern.

Another disadvantage of using surface roughness to enhance the light extraction

efficiency of deep UV AlGaN QWs LED is non-uniformly distributed far field radiation

pattern, which is attributed to the randomness of roughening technique. Figure 3-20

shows the far field radiation pattern from deep UV AlGaN QWs LED with roughened

surface (RMS=35nm) in forms of 2D angular polar plot in figure 3-20 (a) and 3D far field

0

30

60

90

120

150

180

(a)

(b)

(a)

X3

3X Conventional LEDs with flat surface(b) LEDs with roughened surface (RMS=35nm)

(a)

(b)

63

pattern in figure 3-20 (b). The results show evident non-uniformity from the light

extraction far field pattern.

3.7.2 SiO2 microlens array

The light extraction efficiency of deep UV LED can also be enhanced by simply

deposit SiO2 microsphere monolayer on p-type emission surface, which provides the

grading refractive index between III-nitride material and free space. Figure 3-21 plots the

wavelength dependent refractive index n for both AlN and SiO2. At deep UV emitting

wavelength 250nm, the refractive index for AlN and SiO2 are ~2.4 and ~1.5, respectively.

The SiO2 microsphere monolayer act as a grading refractive index layer (n=~1.5) between

III-nitride material (n=~2.4) and free space (n=1). In the meantime, the spherical

emission morphology effectively scatters out the light from deep UV AlGaN QWs LED

to enhance the extraction efficiency.

2.10

2.15

2.20

2.25

2.30

2.35

2.40

250 300 350 400 450 500 550 600

Wavelength (nm)

Re

fra

cti

ve

In

de

x n

AlN

(a)

64

Figure 3-21. Wavelength dependent refractive index for (a) AlN and (b) SiO2

The size effect of SiO2 microsphere diameter on light extraction efficiency

enhancement was also investigated for deep UV AlGaN QWs LED to achieve the

optimized feature size at TM polarized emitting wavelength at 250nm. Figure 3-22 shows

the light extraction efficiency enhancement ratio for deep UV AlGaN QWs LED with

SiO2 microspheres in various diameter sizes. The enhancement was normalized to

conventional LED with flat surface. The overall enhancement ratios for different size of

SiO2 microspheres are fairly small compared to deep UV LEDs with III-nitride micro-

domes, and the optimized light extraction efficiency enhancement (1.60X) occurs when

D=190nm.

1.45

1.46

1.47

1.48

1.49

1.50

1.51

250 300 350 400 450 500 550 600

Wavelength (nm)

Re

fra

cti

ve

In

de

x n

SiO2

(b)

65


emission component for AlGaN QWs LEDs with SiO2 microspheres emitting at 250nm

as a function of the SiO2 microspheres diameter. The optimized light extraction efficiency

enhancement ratio is 1.6x by SiO2 microspheres with diameter 190nm.

3.8 Summary of Light Extraction Efficiency Enhancement for Deep UV

AlGaN QWs LEDs

In summary, 3-D FDTD simulations were performed to calculate the light extraction

efficiency for both conventional UV LEDs with flat surface and the LEDs with III-nitride

micro-domes. The study focuses on the enhancement of light extraction efficiency from

TM-polarized spontaneous emission, which was proved to be the dominant component of

dipole source in deep UV LEDs. With p-type layer thickness of 300nm and III-nitride

micro-dome diameter of 200nm and height of 175nm, the deep UV LED shows 7.3 times

To

tal

lig

ht

extr

ac

tio

n e

ffic

ien

cy

en

ha

nc

em

en

t ra

tio

0.8

0.9

1.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

0 100 200 300 400 500

Microsphere diameter (nm)

TM mode

= 250nm

LEDs with SiO2 microlens

66

enhancement of the TM polarized spontaneous emission as compared to that of the

conventional LEDs. The far field pattern shows significant enhancement of TM-polarized

extraction efficiency for a wide range of polar angles. The design of the LEDs with III-

nitride micro-domes has great potential to significantly enhance the total light extraction

efficiency of the deep UV LEDs, which will lead to enhancement of the total external

quantum efficiency of the deep UV LEDs.

67

Chapter 4: Analysis of Light Extraction Efficiency Enhancement for Thin-

Film-Flip-Chip (TFFC) InGaN Quantum Wells Light-Emitting Diodes with

GaN Micro-Domes

In this chapter, the analysis of the light extraction efficiency enhancement for thin-

film flip-chip (TFFC) InGaN quantum wells (QWs) light-emitting diodes (LEDs) with

GaN micro-domes on n-GaN layer was studied. The light extraction efficiency of TFFC

InGaN QWs LEDs with GaN micro-domes were calculated and compared to that of the

conventional TFFC InGaN QWs LEDs with flat surface. The three dimensional finite

difference time domain (3D-FDTD) method was used to calculate the light extraction

efficiency for the InGaN QWs LEDs emitting at 460nm and 550 nm, respectively. The

effects of the GaN micro-dome feature size and the p-GaN layer thickness on the light

extraction efficiency were studied systematically. Studies indicate that the p-GaN layer

thickness is critical for optimizing the TFFC LED light extraction efficiency. Significant

enhancement of the light extraction efficiency is achievable from TFFC InGaN QWs

LEDs with optimized GaN micro-dome diameter D and height h.

4.1 Introduction of InGaN Quantum Wells LEDs

4.1.1 Structure of III-nitride InGaN QWs LED Device

As the promising candidate for the next generation lighting technology, visible

emission light-emitting diodes (LEDs) play an important role in solid state lighting [7-10].

68

InGaN quantum wells (QWs) are employed as active region for LEDs emitting in the

near ultraviolet, blue and green spectral region.

Figure 4-1 shows the schematic of the typical configuration of III-nitride InGaN

QWs LED device. The LED chip structure also grown on sapphire substrate by MOCVD

technology mentioned. First, a buffer GaN layer was grown at low temperature, followed

by a thick n-doped GaN layer. The InGaN multiple quantum wells with GaN barriers

were grown above n-GaN layer and capped with thinner p-type doped GaN layer. In

order to expose deposit n-electrode metal on n-GaN, reactive ion etching was normally

employed on LED chip to selectively etch to form mesa area and expose n-GaN.

Additionally, p-electrode metal and transparent metal were deposited on p-GaN layer and

mesa, respectively. Nickle, Titanium and Gold are typical metal materials used for

contacting layer.

Figure 4-1. Schematic of the configuration of III-nitride InGaN QWs LED device.

69

4.1.2 Efficiency Challenges in III-nitride InGaN QWs LED

However, major challenges still exist for high performance InGaN QWs LEDs,

including 1) the existence of the spontaneous and piezoelectric polarizations in the III-

nitride semiconductor materials [67] leads to the charge separation in the QWs, resulting

in low radiative recombination rate of electrons and holes, and 2) low light extraction

efficiency in InGaN QWs LEDs due to the total internal reflection at the GaN/air

interface. Recently, approaches based on novel QWs designs have been proposed to

engineer the band lineups of the InGaN QWs for enhancing the electron-hole

wavefunction overlap, including the staggered InGaN QW, type-II InGaN-GaNAs QW,

strain-compensated InGaN-AlGaN QW, InGaN-delta-AlGaN QW, InGaN-delta-InN

QW, and triangular InGaN QW. In addition, significant efforts have been focused on

nonpolar / semipolar InGaN QWs growths for removing / reducing polarization field in

the QW, as mentioned in chapter 1.

The limitation occurred in light extraction efficiency of III-nitride LEDs is attributed

to the large refractive index difference between III-nitride semiconductor (n~2.4) and free

space (n~1), which leads to severe total internal reflection at the semiconductor and air

interface. Most of the generated photons in the active region are trapped inside the LED

device and finally absorbed by the material. In order to achieve high performance InGaN

QWs LEDs with high total external quantum efficiency, the enhancement of the light

extraction efficiency is extremely important.

The approaches that have been proposed for enhancing the light extraction

efficiency of III-nitride LEDs were introduced in Chapter 1, section 1.3, including

surface roughness, photonic crystal, patterned sapphire substrate, nanopyramids, graded

70

refractive index material and SiO2/polystyrene microlens arrays. Potential issues such as

non-uniformity, high cost, limited efficiency enhancement, material degradation and

reliability are still required to be addressed in these approaches.

4.2 Polarization Analysis of Spontaneous Emission from InGaN

Quantum Well

In order to study the transverse electric (TE) and transverse magnetic (TM)

components of the InGaN QWs spontaneous emission (Rsp), the calculations of the band

structure and wavefunctions for InGaN QWs were carried out by using self-consistent 6-

band k∙p method for wurtzite semiconductors, taking into account the valence band

mixing, strain effect[63, 64], polarization fields, and carrier screening effect. The band

parameters for the III-nitride alloys utilized in our calculations were obtained from

references 65 and 66.

Figure 4-2 plots the TE component and TM component (x50) of the spontaneous

emission spectra for 3-nm InxGa1-xN QWs with In-contents of x=0.1, 0.2, 0.25 and 0.3,

respectively. From figure 4-2, the TE spontaneous emission component dominates the

total Rsp in the visible wavelength regime. In this calculation, the TE polarized dipole

source is used for the light extraction efficiency calculation.

71

Figure 4-2. Spontaneous emission spectra (TE component and TM component (x50)) for

3-nm InxGa1-xN QWs LEDs with x= 0.1, 0.2, 0.25 and 0.3. TE spontaneous emission

component is dominant in InGaN QWs LED emitting in visible spectrum region.

4.3 Thin-Film-Flip-Chip Technology

Thin-Film-Flip-Chip package design has been widely used in the current commercial

LEDs, which possesses high light extraction efficiency as compared to that of the

conventional LED package [68-70]. Thin film LEDs technology was first proposed by

OSRAM in 2003, which could be realized by removing sapphire substrate by laser lift-off

technique [71], as shown in figure 4-3 (a). Since the thermal and electrical conductivity

of original sapphire substrate are poor, by separating sapphire substrate and sub-mounting

thin-film LED chip on other desire materials, the performance of III-nitride LED could be

significantly improved. Flip-chip LEDs are achieved by submounting the p-GaN on a

high reflectance metallic mirror to form the vertical LED configuration as shown in

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

300 350 400 450 500 550 600

Sp

on

tan

eo

us E

mis

sio

n (

a.u

.)

Wavelength (nm)

3nm InxGa1-xN / GaN QW

TE

TM

(x50)

x=0.1

x=0.2

x=0.3x=0.25

72

figure 4-3 (b), which allows the photons to emit from the n-GaN layer side, and allows

flexible surface texturing and patterning process on relative thick n-GaN to enhance the

extraction efficiency without the potential effect on InGaN QWs active region. The

vertical-injection configuration also leads to better current injection efficiency, longer

LED reliability and more efficient thermal dissipation process due to shortened thermal

path. Also, less photon will be obscured by contact metal layers and bonding wires. The

Thin-Film-Flip-Chip LEDs combine these two techniques together.

Figure 4-3. 2D Schematics of III-nitride LED based on Thin-Film-Flip-Chip technology

package design (a) flip chip technology package design; and (b) Thin-Film-Flip-Chip

technology package design.

4.4 Light Extraction Efficiency Calculation of TFFC InGaN QWs LED

by 3-D FDTD Method

3-D FDTD method was applied to calculate the light extraction efficiency of the

TFFC InGaN QWs LEDs with both flat surface and GaN micro-domes. Figure 4-4 shows

Thin film

technology

Sapphire

(a) (b)

73

the schematics for the computational domain for both conventional TFFC InGaN QWs

LEDs with flat surface (figure 4-4 (a)) and with GaN micro-domes (figure 4-4 (b)). Note

that the feature size of GaN micro-domes is in the range of submicron to micron, which is

comparable to the wavelength emitted from InGaN QWs active region. As compared to

the traditional calculation approach based on ray tracing, the FDTD method is more

accurate to solve the differential forms of Maxwell’s equations with specific boundary

conditions in such complex geometries.

Here, the lateral dimension of the computational domain is set as 10m which is

much smaller than that of the real LEDs (~300-1000m). The boundary conditions for

the four lateral boundaries as shown in figure 4-4 are set as perfect mirror for

representing the limited lateral dimension as infinite [72, 73]. The reflections from the

boundary perfect mirror takes into account the light extraction beyond the computational

domain. With the perfect mirror lateral boundary conditions and the reflective layer at the

bottom, photons emitted directly from InGaN QWs active region and the light

propagating after reflections can only be extracted out from the top n-GaN surface. The

boundary condition for top simulation area is set as perfectly matched layer (PML)

boundary condition, which absorbs electromagnetic energy incident upon it.

In the simulation, a single polarized dipole source was placed in the InGaN QWs

active region, and the detection plane is set as /n() away from the emission surface of

n-GaN, where represents the peak emission wavelength from the InGaN QWs and n()

is the wavelength dependent refractive index of the media. The maximum mesh step is

set as /10*n(), and the average grid points are estimated around 1800000 in the

74

computational domain, which generates good accuracy in the light extraction efficiency

calculation.

The light extraction efficiency is defined as the ratio of total extracted light power

to the total power emitted from InGaN QWs. In this simulation, the extracted power from

TFFC LED surface can be also obtained by integrating the Poynting vectors over far field

projection surface, and the total power emitted from QWs were calculated by Poynting

vectors integrated surrounding the near field of dipole source.

Figure 4-4. 2D schematics of the computational domain of the Thin Film Flip-Chip

(TFFC) InGaN QWs LEDs with (a) flat surface and (b) GaN micro-domes.

4.5 Effect of P-GaN Layer Thickness on Light Extraction Efficiency for

Conventional TFFC InGaN LEDs

Note that in the TFFC LEDs, the QWs active region was placed close enough to the

reflective metallic mirror (on the order of 150-400 nm). The light emitted from QWs will

interfere with the reflected waves, and the coupled interference patterns in the escape

AIR

Detection Plane

p-GaN

n-GaN1

(a)

PML

Perfect

Mirror

InGaN/GaN

MQW

Reflective layer

InGaN/GaN

MQW

AIR

Detection Plane

n-GaN

PML

p-GaN

(b)Perfect

Mirror

Reflective layer

75

cone will lead to significant changes in light extraction efficiency from conventional flip

chip LEDs [74-76]. In the FDTD calculations for conventional TFFC InGaN LEDs with

flat surface, the distance between InGaN QWs active region and reflective layer could be

modified by varying the p-GaN layer thickness, which is critical for optimizing the LED

light extraction efficiency.

4.5.1 Theoretical Prediction of Extracted Interference Pattern

Theoretical analysis on the optical cavity effects provides a reasonable model to

analyze the dependence trend by calculating the monochromatic transmittance curve, in

which the optical cavity is based on planar micro-cavity structure with a high-reflectance

bottom mirror [77].

Figure 4-5. Schematic of TFFC InGaN QWs LED with interference between emission of

QWs and from bottom mirror.

p-GaN

n-GaN

InGaN/GaN

MQW

Reflective layer

t

Interference

(wavelengh )Incident light

reflected light

76

As shown in figure 4-5, if a monochromatic oscillating dipole source with vacuum

wavelength is located at a distance t above the bottom mirror, the interference patterns

of the electric field upon dipole source and mirror could be expressed as follows [78]:

)(2cos221)(2

0

2)(21)(

2

0

2 rrEe

irEE (4-1)

where )(0E is the electric field without mirror underneath, r is the amplitude

reflectivity of bottom mirror and )( is the phase shift due to optical path difference

which is related to the source-mirror distance t, the incident angle (from the normal)

and associated wave vector

)(2 nk . Note that half-cavity approximation is used

here to assume the media thickness above the mirror is infinite. Also, the lifetime,

polarization, and orientational effects of the dipole source are not taken into account for

simplicity so as to get the upward-radiated electric far field in equation (4-1). Since phase

shift upon the reflections off the mirror is neglected here, the total phase shift from

optical path difference could be calculated as follows:

cos22 kt (4-2)

Another approximation here is made to assimilate normal incident light ( 0 )

and incident light within critical angle ( c ). For an dipole source located near

bottom mirror, the phase changing difference for rays propagating between normal

incidence and critical angle incident direction is trivial, hence cos could assume to be

1 when c . Constructive interference coupling happens when mkt 22 with

77

integer m for +r and half-integer m for –r. Under perfect mirror approximation ( 1r )

in equation (4-1), two constructive interference with +r and –r will bring the maximum

( )(2

04)(max EE ) and minimum ( 0)(min E ) electric field, respectively. Then

the source-mirror distance t corresponding to constructive interference between emitted

light and reflected light could be achieved as follows:

)(2)(

n

mt (4-3)

which indicates the periodicity of t is )(2

n.

4.5.2 3D FDTD Calculation of Extracted Interference Pattern

The dependence of the light extraction efficiency of TE polarized spontaneous

emission component for the TFFC InGaN QWs LEDs on p-GaN layer thickness was

calculated with two wavelengths of peak=460nm and peak=550nm, as shown in figure 4-

6(a) and figure 4-6(b) respectively. The solid dots represent FDTD simulated results, and

the dash line is the theoretical fitting curve, which was obtained from equations (4-1) and

(4-3) regarding oscillation periodicity and amplitude. The results indicate theoretical

analysis for interference pattern discussed in section 4.5.1 gives reasonable predictions

for the FDTD calculation results.

78

Figure 4-6. Light extraction efficiency for the conventional TFFC InGaN QWs LEDs at

wavelength (a) peak=460nm and (b) peak=550nm with flat surface as a function of the p-

GaN layer thickness. N-GaN thickness is 2.5m. Dash line and solid dots represent the

theoretical fitting curve and the FDTD calculation results, respectively.

0

0.05

0.1

0.15

0.2

0.25

0 50 100 150 200 250 300 350 400

Ex

tra

cti

on

Eff

icie

nc

y

p-GaN Thickness (nm)

TE mode

= 460nm

(a)

0

0.05

0.1

0.15

0.2

0.25

0 50 100 150 200 250 300 350 400

Extr

acti

on

Eff

icie

nc

y


TE mode

= 550nm

(b)

79

In figure 4-6(a), the p-GaN layer thickness (t) periodicity calculated from equation

(4-3) at peak=460nm is 92.9 nm and the peak light extraction efficiency amplitude is

0.198, which provides a good fitting to the solid dots obtained from FDTD calculation. In

figure 4-6(b), the p-GaN layer thickness periodicity of t and light extraction efficiency

amplitude corresponding to peak=550nm are 113.3 nm and 0.204 respectively, which also

shows good agreement with the solid dots.

From figure 4-6, the strong effect of the p-GaN layer thickness on the light

extraction efficiency indicates the importance of optimizing the p-GaN thickness for the

TFFC InGaN QWs LEDs. The peak light extraction efficiency and oscillation periodicity

are determined by the emission wavelength and the material. The typical p-GaN layer

thickness in InGaN QWs LEDs is around 200 nm. From figure 4-6, the optimized p-GaN

layer thickness of 195 nm (peak=460nm) and 230nm (peak=550nm) were obtained for

maximum light extraction of the conventional InGaN QWs TFFC LEDs with flat surface.

4.6 Effect of Micro-dome Size on Light Extraction Efficiency for InGaN

LEDs

In this work, the effects of the micro-dome diameter (D) and height (h) on the light

extraction efficiency of the TE polarized spontaneous emission component were studied

for both conventional package design of InGaN QWs LED as well as TFFC InGaN QWs

LED with GaN micro-domes on emission surface. The light extraction efficiency was

calculated based on source positioning dependence relative to individual micro-dome

structure.

80

4.6.1 Conventional Package of InGaN LED with Micro-dome (Micro-hemisphere)

The light extraction efficiency from conventional package design of InGaN LED

with p-GaN micro-domes were calculated at first. In this study, since there is no bottom

mirror coated below the InGaN QWs active region, the photon recycling due to mirror

reflections was not taken into account in the calculation. Figure 4-7 (a) plots the light

extraction efficiency of the TE dipole sources (=460nm) located along the diameter of

the micro-dome with various diameters (D=250nm up to 1000nm). The top p-type layer

thickness is 300nm. Studies show that the light extraction efficiency strongly depends on

the position of the TE dipole source along the diameter of the micro-domes with all

diameter cases - the extraction efficiency reaches maximum for the dipole source locating

around the center region under the micro-domes, and decreases along the radius of the

micro-domes. The light extraction efficiency from conventional InGaN QWs LED with

flat emission surface was calculated as comparison, which is shown as blue dash line in

figure 4-7 (a). Note that the when the TE polarized dipole source located around the edge

or void region of GaN micro-domes, the light extraction efficiency will be suppressed

below the flat emission surface LED level (blue dash line), which can significantly

restrain the light extraction efficiency enhancement from p-GaN micro-domes.

81

Figure 4-7. Light extraction efficiency analysis for conventional package design InGaN

QWs LED with p-GaN micro-domes: (a) source positioning dependence analysis along

micro-dome diameter; (b) total light extraction enhancement ratio.

0

5

10

15

20

25

-100 -80 -60 -40 -20 0 20 40 60 80 100

x 0

.01

Flat suface

D=500nm

D=1000nm

D=750nm

D=800nm

D=350nm

D=250nm

D/10 2D/5 D/2

TE mode

= 460nm

Lig

ht

extr

acti

on

eff

icie

ncy (

a.u

.)

0

TE dipole source position

3D/10D/5D/10D/53D/102D/5D/2

(a)

0.96

0.98

1.00

1.02

1.04

1.06

1.08

1.10

0 200 400 600 800 1000

Lig

ht

extr

ac

tio

n e

ffic

ien

cy

en

han

cem

en

t ra

tio

Microsphere diameter (nm)

TE mode

= 460nm

(b)

82

Taking into account the source positioning dependence, the overall light extraction

efficiency enhancement ratio was calculated in figure 4-7 (b), where the light extraction

efficiency from conventional InGaN QWs LED with flat emission surface was

normalized to 1. The results show trivial light extraction efficiency enhancement from p-

GaN micro-domes. At small size of micro-domes (D=250 and D=350), the light

extraction efficiency are not enhanced. The trivial enhancement is probably due to the

significant suppression around the edge and void region of GaN micro-domes.

4.6.2 TFFC InGaN LED with Micro-dome (Micro-hemisphere)

Effect of micro-dome (micro-hemisphere) size on light extraction efficiency for was

studied TFFC package design of InGaN QWs LEDs. Figure 4-8 plots the light extraction

efficiency of the TFFC InGaN QWs LEDs with GaN micro-domes (h=D/2) as a function

of the micro-dome diameter (D) at peak=460nm and peak =550nm. The optimized p-GaN

layer thickness of 195 nm (peak=460nm) and 230nm (peak=550nm) were used in the

calculation. The top n-GaN layer thickness is set as 2.5m. Note that the extraction

efficiency at D=0 represents the case for conventional LEDs with flat surface. From

figure 4-8, the LEDs with GaN micro-domes show significant enhancement of the light

extraction efficiency at different micro-dome diameter. For emission wavelength of

peak=460nm, the light extraction efficiency of LEDs with GaN micro-domes increases

significantly from 0.209 (D=0) to 0.476 (D=500nm) and forms a peak at D=500nm. Then

the light extraction efficiency increases slightly and saturates (0.53) when D > 1m. The

light extraction efficiency of LEDs with GaN micro-domes for emission wavelength

83

peak=550nm is also calculated for comparison. The result shows similar trend: it forms a

peak with extraction efficiency of 0.477 at D=600nm and saturates at 0.53 when D >

1.25m.

Fig. 4-8. Light extraction efficiency enhancement of TE polarized spontaneous emission

component for TFFC InGaN QWs LEDs with optimized p-GaN thickness (195nm for

peak=460nm, 230nm for peak=550nm as a function of GaN micro-hemisphere diameter

(D).

Note that the light extraction efficiency of TFFC InGaN LEDs with GaN micro-

domes as a function of the micro-dome diameter shows similar trend for emission

wavelength of 460 nm and 550 nm. There exists peak extraction efficiency before the

light extraction efficiency saturates at larger micro-dome diameter. The diameter

corresponding to the peak light extraction efficiency shifts from D=500 nm for

0

0.1

0.2

0.3

0.4

0.5

0.6

0 500 1000 1500 2000 2500

Ex

tra

cti

on

Eff

icie

nc

y

Micro-hemisphere Diameter D (nm)

Conventional LED with Flat Surface (D=0)

TE mode GaN (peak=460nm )

GaN (peak=550nm )

84

peak=460nm to D=600 nm for peak=550nm, which indicates its wavelength dependence

characteristics. The maximum light extraction efficiency enhancement of 2.6 times

(peak=460nm) and 2.8 times (peak=550nm) are obtained from LEDs with GaN micro-

domes when D~2m.

For practical LED device fabrication, smaller size micro-domes are preferable due

to 1) the short etching time required to form GaN micro-domes, and 2) less potential

effect on InGaN QWs active region if the GaN micro-domes are distant from the QWs.

Thus, the suitable micro-domes (h=D/2) for peak=460nm and peak=550nm are D=1m

(2.53 times) and D=1.25m (2.74 times), respectively.

4.6.3 TFFC InGaN LED with Micro-dome (Micro-hemiellipsoid)

The light extraction efficiency of TFFC InGaN QWs LED with micro-domes was

studied by tuning the micro-dome height h (h≠D/2). The studies indicate the optimized

micro-dome structure for highest light extraction efficiency is not necessary from the

micro-domes with h=D/2 (micro-hemisphere). Figure 4-9(a) shows the geometric

structure of the general micro-dome structure on n-GaN layer, where h≠D/2. Figure 4-

9(b) plots the light extraction efficiency of TFFC InGaN QWs LED with micro-dome

structure as a function of the micro-dome height h for emission wavelength at

peak=460nm.

85

Fig. 4-9. (a) Geometric structure of the general micro-dome structure on n-GaN emission

surface of TFFC InGaN QWs LED with diameter D and height h; (b) light extraction

efficiency at peak=460nm from InGaN QWs TFFC LED with micro-dome structures as a

function of micro-dome height h for diameters D=500nm, D=1000nm and D=1500nm.

Three micro-dome diameter sizes D=500nm, D=1000nm and D=1500nm were

studied and plotted in figure 4-9(b). Note that the yellow triangular dots on each curve

indicate the light extraction efficiency with micro-hemisphere structure (h=D/2). The

light extraction efficiency at h=0 represents the case for the conventional LED with flat

InGaN

MQW

p-GaN

n-GaN

Dh

(a)

0

0.1

0.2

0.3

0.4

0.5

0.6

0 250 500 750 1000 1250 1500 1750 2000

D=500nm

D=1000nm

D=1500nm

TE mode

peak=460nm

Micro-dome height h (nm)

To

tal li

gh

t e

xtr

acti

on

eff

icie

nc

y

Conventional LED

Micro-hemisphere

(b)

86

surface. From figure 4-9(b), the optimized micro-dome structure for highest light

extraction efficiency occurs at h>D/2. For TFFC InGaN QWs LEDs with GaN micro-

domes emitting in the visible wavelength region, the light extraction efficiency could be

optimized by tuning both the diameter and the height of the micro-domes. The

enhancement of 2.71 times was achieved at peak=460nm when D=1m and h=1.7m. In

practical devices, the aspect ratio of the GaN micro-domes is not required to be

accurately controlled to form exact micro-hemisphere shape, which provides tolerance

for device fabrication.

4.7 P-GaN Layer Thickness Dependence of Light Extraction Efficiency

for TFFC InGaN QWs LEDs with GaN Micro-domes

The light extraction efficiency dependence on the p-GaN layer thickness for the

TFFC InGaN QWs LEDs was calculated and plotted, as shown in figure 4-10. The

periodic oscillation of the light extraction efficiency as a function of the p-GaN layer

thickness for the conventional TFFC InGaN QWs LEDs with flat surface was also plotted

for comparison. The results show good agreement between the theoretical fitting curves

(dash lines) and the FDTD simulation results (triangular dots) for the TFFC InGaN LEDs

with GaN micro-domes.

Note that the LED with GaN micro-domes show significant light extraction

efficiency enhancement at different p-GaN layer thickness as compared to that of the

conventional LED. The light extraction efficiency for the LED with micro-domes also

oscillates as a function of the p-GaN layer thickness, but with smaller amplitude of

oscillation as compared to that of the conventional LEDs. This indicates the TFFC LED

87

light extraction efficiency has weaker dependence on the p-GaN layer thickness by

employing the GaN micro-domes.

For practical conventional TFFC InGaN LED epitaxy, the p-GaN layer thickness is

required to be precisely controlled in order to obtain the maximum light extraction

efficiency. By employing GaN micro-domes, it provides more tolerance for the p-GaN

layer growth and contributes to much higher light extraction efficiency in a wider range

of p-GaN layer thickness.

Fig. 4-10 Light extraction efficiency for the conventional TFFC InGaN QWs LEDs

(peak=460nm) with flat surface and with GaN micro-domes as a function of the p-GaN

layer thickness. Dash lines and solid dots represent the theoretical fitting curves and the

FDTD calculation results, respectively.

0

0.1

0.2

0.3

0.4

0.5

0.6

0 50 100 150 200 250 300 350

TE mode

= 460nm

Extr

acti

on

Eff

icie

ncy


Conventional LED with Flat Surface

GaN Micro-hemisphere

88

4.8 Summary of Light Extraction Efficiency Enhancement for TFFC

InGaN QWs LEDs

The light extraction efficiency for TFFC InGaN QWs LEDs was studied by using 3-

D FDTD method. TFFC InGaN QWs LEDs with GaN micro-domes on top of n-GaN

layer show significant enhancement of light extraction efficiency. The optimized light

extraction efficiency enhancement of 2.5-2.7 times (peak=460nm) and 2.7~2.8 times

(peak=550nm) for LEDs with GaN micro-domes were achieved with micro-dome size of

D~1m and h~0.5-1.7 m (peak=460nm) and D~1.25m, h~0.625m (peak=550nm).

The design of the LEDs with GaN micro-domes has great potential to significantly

enhance the total light extraction efficiency of the TFFC InGaN QWs LEDs and allows

more tolerance in p-GaN layer growth thickness, which in turn leads to enhancement of

the total external quantum efficiency of InGaN QWs LEDs in a wider range of p-GaN

layer thickness.

89

Chapter 5: Conclusions and Future Work

5.1 Conclusions

In this work, the light extraction efficiency issue for III-nitride LEDs was

investigated for solid state lighting. III-nitride micro-dome structures were designed on

LEDs emission surface to enhance the light extraction efficiency for III-nitride LEDs.

Three dimensional finite difference time domain (3D-FDTD) method analysis of

light extraction efficiency enhancement for AlGaN quantum wells (QWs) based deep

ultraviolet (UV) light-emitting diodes (LEDs) with III-nitride micro-hemisphere and

micro-dome structures on the p-type layer are studied first, which was compared to that

of the conventional deep UV LEDs with flat surface. The transverse electric (TE) and

transverse magnetic (TM) components of the spontaneous emission of AlGaN QWs with

AlN barriers were calculated by using self-consistent 6-band k∙p method, which shows

the TM component overtakes the TE component and becomes the dominant contribution

of the spontaneous emission when the Al-content of the AlGaN QWs is larger than 0.66.

The light extraction efficiency of the AlGaN/AlN QWs deep UV LEDs emitting at 250

nm with AlGaN micro-domes as well as of the conventional LEDs with flat surface was

calculated. The effects of the III-nitride micro-dome diameter and height as well as the p-

type layer thickness on the light extraction efficiency were studied systematically. The

studies indicate optimized light extraction efficiency enhancement (>7.3 times) of the

dominant TM polarized spontaneous emission for deep-UV LEDs with III-nitride micro-

domes.

90

The enhancement of light extraction efficiency for thin-film flip-chip (TFFC)

InGaN quantum wells (QWs) light-emitting diodes (LEDs) with GaN micro-domes on n-

GaN layer was also studied by 3D-FDTD method for wavelength emitting at 460nm and

550nm. The effects of the GaN micro-dome feature size and the p-GaN layer thickness

on the light extraction efficiency were studied and the optimized the TFFC LED light

extraction efficiency could be achieved with p-GaN layer thickness at 190nm (460nm)

and 230nm (550nm). After studying the effects from GaN micro-dome size and height,

significant enhancement of the light extraction efficiency (2.5-2.7 times for peak=460nm

and 2.7-2.8 times for peak=550nm) is achievable from TFFC InGaN QWs LEDs with

optimized GaN micro-dome diameter and height.

5.2 Future Work

The future work will mainly focus on the experimental light extraction efficiency

enhancement characterization for both deep UV AlGaN QWs LEDs and visible InGaN

QWs LEDs with III-nitride micro-domes on emission surface. III-nitride micro-domes

with different aspect ratio will be fabricated on LEDs sample and photoluminescence

measurement will be applied to show the enhancement of extraction efficiency, as

compare to conventional LEDs with flat emission surface. Also, further optimization

work of III-nitride micro-domes could be performed by 3D-FDTD calculation to enhance

light extraction efficiency by analyzing nonclose-packed distributed micro-domes with

various micro-dome center to center distances.

91

References

[1] A Brief Outline of the History of Stage Lighting: http://www3.northern.edu/

[2] Bel Geddes, Norman. Miracle in the Evening. Garden City, NY: Double Day and Co.,

Inc. 1960.

[3] AEO2012 Early Release Overview

http://www.eia.gov/forecasts/aeo/er/tables_ref.cfm

[4] Navigant Consulting, “Energy Savings Potential of Solid-State Lighting in General

Illumination Applications”, U.S. Department of Energy, 2012.

[5] Navigant Consulting, “Life-Cycle Impact Analysis: LED Manufacturing and

Performance”, U.S. Department of Energy, Solid State Lighting Program, 2012.

[6] Navigant Consulting, “Energy Savings Estimates of Light Emitting Diodes in Niche

Lighting Applications” report prepared for US Department of Energy, 2011.

[7] S. Nakamura, M. Senoh, N. Iwasa, and S. Nagahama, “High-brightness InGaN blue,

green and yellow light-emitting diodes with quantum well structures,” Jpn. J. Appl.

Phys., 34, 797–799, 1995.

[8] M. Krames, O. Shchekin, R. Mueller-Mach, G. Mueller, L. Zhou, G. Harbers, and M.

Craford, “Status and future of high-power light-emitting diodes for solid-state lighting,”

IEEE J. Display Technol., 3, 160–175, 2007.

[9] M. H. Crawford, “LEDs for Solid-State Lighting: Performance Challenges and Recent

Advances,” IEEE J. Sel. Top. Quantum Electron., 15, 1028-1040, 2009.

[10] N. Tansu, H. Zhao, G. Liu, X. H. Li, J. Zhang, H. Tong, and Y. K. Ee,

“Breakthrough in Photonics 2009: III-Photonics”, IEEE Photonics Journal, 2, 236-243,

2010.

http://www.eia.gov/forecasts/aeo/er/tables_ref.cfm

92

[11] F. A. Kish, and R. M. Fletcher, “AlInGaP Light-Emitting Diodes”, Semiconductors

and Semimetals, 48, 149, 1997.

[12] M. R. Krames, O. B. Shchekin, R. M. Mueller-Mach, G. O. Mueller, L. Zhou, G.

Harbers, M.G. Craford, “Status and Future of High-Power Light-Emitting Diodes for

Solid-State Lighting” J. Disp. Tech., 3, 160, 2007.

[13] F. Bernardini, V. Fiorentini, and D. Vanderbilt, “Spontaneous polarization and

piezoelectric constants of III-V nitrides”, Phys. Rev. B, 56, R10024, 1997.

[14] H. Zhao, G. Liu, J. Zhang, J. D. Poplawsky, V. Dierolf, and N. Tansu, “Approaches

for high internal quantum efficiency green InGaN light-emitting diodes with large

overlap quantum wells,” Optics Express, 19, 991-1007, 2011.

[15] R. A. Arif, Y. K. Ee, and N. Tansu, “Polarization engineering via staggered InGaN

quantum wells for radiative efficiency enhancement of light emitting diodes,” Appl.

Phys. Lett., 91, 091110, 2007.

[16] H. Zhao, R. A. Arif, N. Tansu, “Design Analysis of Staggered InGaN Quantum

Wells Light-Emitting Diodes at 500–540 nm,” IEEE J. Sel. Top. Quantum Electron., 15,

1104-1114, 2009.

[17] H. Zhao, G. S. Huang, G. Liu, X. H. Li, J. D. Poplawsky, S. Tafon Penn, V. Dierolf,

and N. Tansu, “Growths of staggered InGaN quantum wells light-emitting diodes

emitting at 520–525 nm employing graded growth-temperature profile,” Appl. Phys.

Lett., 95, 061104, 2009.

[18] S. H. Park, D. Ahn, and J. W. Kim, “High-efficiency staggered 530 nm

InGaN/InGaN/GaN quantum-well light-emitting diodes,” Appl. Phys. Lett., 94, 041109,

2009.

93

[19] H. Zhao and N. Tansu, “Optical gain characteristics of staggered InGaN quantum

wells lasers,” J. Appl. Phys., 107, 113110, 2010.

[20] R. A. Arif, H. Zhao, and N. Tansu, “Type-II InGaN-GaNAs quantum wells for lasers

applications,” Appl. Phys. Lett., 92, 011104, 2008.

[21] H. Zhao, R. A. Arif, and N. Tansu, “Self-consistent gain analysis of type-II ‘W’

InGaN–GaNAs quantum well lasers,” J. Appl. Phys., 104, 043104, 2008.

[22] S. H. Park, Y. T. Lee, and J. Park, "Optical properties of type-II InGaN/GaAsN/GaN

quantum wells," Optical and Quantum Electronics, 41, 779-785, 2009.

[23] H. Zhao, R. A. Arif, Y. K. Ee, and N. Tansu, “Self-Consistent Analysis of Strain-

Compensated InGaN–AlGaN Quantum Wells for Lasers and Light-Emitting Diodes,”

IEEE J. Quantum Electron., 45, 66-78, 2009.

[24] C. L. Tsai, G. C. Fan, and Y. S. Lee, "Effects of strain-compensated AlGaN/InGaN

superlattice barriers on the optical properties of InGaN light-emitting diodes," Appl.

Phys. A, 104, 319-323, 2011.

[25] S. H. Park, Y. T. Moon, J. S. Lee, H. K. Kwon, J. S. Park, and D. Ahn,

"Spontaneous emission rate of green strain-compensated InGaN/InGaN LEDs using

InGaN substrate," Phys. Stat. Sol. A, 208, 195-198, 2011.

[26] J. Park and Y. Kawakami, “Photoluminescence property of InGaN single quantum

well with embedded AlGaN δ layer,” Appl. Phys. Lett., 88, 202107, 2006.

[27] S. H. Park, J. Park, E. Yoon, “Optical gain in InGaN/GaN quantum well structures

with embedded AlGaN δ layer”, Appl. Phys. Lett., 90, 023508, 2007.

[28] H. Zhao, G. Liu, and N. Tansu, “Analysis of InGaN-delta-InN quantum wells for

light-emitting diodes”, Appl. Phys. Lett., 97, 131114, 2010.

94

[29] Y. Li, B. Liu, R. Zhang, Z. Xie, Y. Zheng, Physica E, “Investigation of optical

properties of InGaN-InN-InGaN/GaN quantum-well in the green spectral regime”, 44,

821-825, 2012.

[30] R. J. Choi, Y. B. Hahn, H. W. Shim, M. S. Han, E. K. Suh, and H. J. Lee, “Efficient

blue light-emitting diodes with InGaN/GaN triangular shaped multiple quantum wells,”

Appl. Phys. Lett., 82, 2764-2766, 2003.

[31] S. H. Park, D. Ahn, S. L. Chuang, “"Electronic and optical properties of a- and m-

plane Wurtzite InGaN/GaN quantum wells," IEEE J. Quantum Electron., 43, 1175-1182,

2007.

[32] A. Chakraborty, B. A. Haskell, S. Keller, J. S. Speck, S. P. Denbaars, S. Nakamura

and U. K. Mishra, “Demonstration of Nonpolar m-Plane InGaN/GaN Light-Emitting

Diodes on Free-Standing m-Plane GaN Substrates”, Jpn. J. Appl. Phys.,44, 173-175,

2005.

[33] H. Masui, S. Nakamura, S. P. DenBaars, and U. K. Mishra, “Nonpolar and

Semipolar III-Nitride Light-Emitting Diodes: Achievements and Challenges”, IEEE

Transactions on Electron Devices, 57, 88-100, 2010.

[34]. C. Huh, K. S. Lee, E. J. Kang, and S. J. Park, J. Appl. Phys., “Improved light-output

and electrical performance of InGaN-based, light-emitting diode by microroughening of

the p-GaN surface”, 93, pp. 9383–9385, 2003.

[35]. T. Fujii, Y. Gao, R. Sharma, E. L. Hu, S. P. DenBaars, and S. Nakamuraa, “Increase

in the extraction efficiency of GaN-based light-emitting diodes via surface roughening “,

Appl. Phys. Lett., 84, 855–857, 2003.

95

[36]. J. J. Wierer, A. David, and M. M. Megens, “III-nitride photonic-crystal light-

emitting diodes with high extraction efficiency”, Nat. Photon., 3(3), pp. 163-169, 2009).

[37]. H.W. Choi, C. Liu, E. Gu, G. McConnell, J. M. Girkin, I.Watson, and M. Dawson,

“GaN micro-light-emitting diode arrays with monolithically integrated sapphire

microlenses “, Appl. Phys. Lett., 84, 2253, 2004.

[38]. Y. K. Ee, R. A. Arif, N. Tansu, P. Kumnorkaew, J. F. Gilchrist, “Enhancement of

light extraction efficiency of InGaN quantum wells light emitting diodes using

SiO2/polystyrene microlens arrays “, Appl. Phys, Lett., 91, 221107, 2007.

[39]. Y. K. Ee, P. Kumnorkaew, R. Arif, H. Tong, H. Zhao, J. Gilchrist and N. Tansu,

“Optimization of Light Extraction Efficiency of III-Nitride LEDs With Self-Assembled

Colloidal-Based Microlenses”, IEEE J. Quantum. Electronics, 15, 1218, 2009.

[40]. Y. K. Ee, P. Kumnorkaew, R. A. Arif, H. Tong, J. F. Gilchrist, and N. Tansu, “Light

extraction efficiency enhancement of InGaN quantum wells light-emitting diodes with

polydimethylsiloxane concave microstructures”, Optics Express, 17 (16), 13747, 2009.

[41]. X. H. Li, R. B. Song, Y. K. Ee, P. Kumnorkaew, J. F. Gilchrist, and N. Tansu,

“Light Extraction Efficiency and Radiation Patterns of III-Nitride Light-Emitting Diodes

With Colloidal Microlens Arrays With Various Aspect Ratios”, IEEE Photonics Journal,

3 (3), 489, 2011.

[42]. T. Li, C. Zhou and M. Jiang, “UV absorption spectra of polystyrene”, polymer

Bulletin, 25, 211-216, 1991.

[43] A. Taflove and S. C. Hagness, Computational Electrodynamics: The Finite-

Difference Time-Domain Method. Norwood, MA: Artech House, pp. 51–80. 1995.

96

[44] K. Yee, “Numerical Solution of Initial Boundary Value Problem Involving

Maxwell’s Equations in Isotropic Media,” IEEE Trans. Antenna Propagat., 14, pp. 302-

307, 1966.

[45] M. Bass, Handbook of Optics, (Optical Society of America, vol. 2: Devices,

Measurements, and Properties, 1994.

[46] Collins, D.G., and W.G. Blattner, M.B. Wells, H.G. Horak, “Backward Monte Carlo

Calculations of Polarization Characteristics of the Radiation Emerging from Spherical

Shell Atmospheres”, Applied Optics, 11, 2684-2696, 1972.

[47] Bird, R.E. and R.L. Hulstrom, L.J. Lewis, “Terrestrial Solar Spectral Data Sets”,

Solar Energy, 30, 563, 1983.

[48] Gueymard, C, “Parameterized transmittance model for direct beam and circumsolar

spectral irradiance”, Solar Energy, 71, 325-346, 2011.

[49] H. Hirayama, T. Yatabe, N. Noguchi, T. Ohashi and N. Kamata, “231–261 nm

AlGaN deep-ultraviolet light-emitting diodes fabricated on AlN multilayer buffers grown

by ammonia pulse-flow method on sapphire”, Appl. Phys. Lett. 91, 071901, 2007.

[50] M. Asif Khan, “AlGaN multiple quantum well based deep UV LEDs and their

applications”, Phys. Stat. Sol. (a). 203, 1764, 2006.

[51] A. A. Allerman, M. H. Crawford, A. J. Fischer, K. H. A. Bogart, S. R. Lee, D. M.

Follstaedt, P. P. Provencio, D. D. Koleske, “Growth and design of deep-UV (240-290

nm) light emitting diodes using AlGaN alloys”, J. Cryst. Growth, 272, 227, 2004.

[52] H. Hirayama, S. Fujikawa, N. Noguchi, J. Norimatsu, T. Takano, K. Tsubaki, N.

Kamata, “222–282 nm AlGaN and InAlGaN-based deep-UV LEDs fabricated on high-

quality AlN on sapphire”, Phys. Status Solidi A 206, 1176, 2009.

97

[53] Q. Dai, M. F. Schubert, M. H. Kim, E. F. Schubert, D. D. Koleske, M. H. Crawford,

S. R. Lee, A. J. Fischer, G. Thaler, M. A. Banas, “Internal quantum efficiency and

nonradiative recombination coefficient of GaInN/GaN multiple quantum wells with

different dislocation densities”, Appl. Phys. Lett., 94, 111109, 2009.

[54] J.-H. Ryou, W. Lee, J. Limb, D. Yoo, J. P. Liu, R. D. Dupuis, Z. H. Wu, A. M.

Fischer, F. A. Ponce, “Control of quantum-confined Stark effect in InGaN/GaN multiple

quantum well active region by p-type layer for III-nitride-based visible light emitting

diodes”, Appl. Phys. Lett. 92, 10113, 2008.

[55] T. Koyama, T. Onuma, H. Masui, A. Chakraborty, B. A. Haskell, S. Keller, U. K.

Mishra, J. S. Speck, S. Nakamura, S. P. DenBaars, T. Sota, S. F. Chichibu, “Prospective

emission efficiency and in-plane light polarization of nonpolar m-plane InxGa1−xN/GaN

blue light emitting diodes fabricated on freestanding GaN substrates”, Appl. Phys. Lett.

89, 091906, 2006.

[56] V. Adivarahan, J. P. Zhang, A. Chitnis, S. Wu, J. Sun, R. Pachipulusu, M. Shatalov,

and M. Asif Khan, “Sub-Milliwatt Power III-N Light Emitting Diodes at 285 nm”, Jpn. J.

Appl. Phys. 41, L435, 2002.

[57] J. P. Zhang, A. Chitnis, V. Adivarahan, S. Wu, V. Mandavilli, R. Pachipulusu, M.

Shatalov, G. Simin, J. W. Yang, and M. Asif Khan, “Milliwatt power deep ultraviolet

light-emitting diodes over sapphire with emission at 278 nm”, Appl. Phys. Lett. 81, 4910,

2002.

[58] A. J. Fischer, A. A. Allerman, M. H. Crawford, K. H. A. Bogart, S. R. Lee, R. J.

Kaplar, W. W. Chow, S. R. Kurtz, K. W. Fullmer, and J. J. Figiel, “Room-temperature

98

direct current operation of 290 nm light-emitting diodes with milliwatt power levels”,

Appl. Phys. Lett. 84, 3394, 2004.

[59] T. Nishida, N. Kobayashi and T. Ban, “GaN-free transparent ultraviolet light-

emitting diodes “, Appl. Phys. Lett, vol. 82, pp. 1533851, 2003.

[60] 1. M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, A. Knauer, H.

Rodriguez, S. Einfeldt, Z. Yang, N. Johnson and M. Weyers, “Advances in group III-

nitride-based deep UV light-emitting diode technology”, Semicond, Sci, Technol, vol. 26,

pp. 014036, 2011.

[61] Y. Taniyasu, M. Kasu and T. Makimoto, “An aluminium nitride light-emitting diode

with a wavelength of 210 nanometres”, Nature, vol. 441, pp. 325-328, 2006.

[62] J. Zhang, H. Zhao, and N. Tansu, “Effect of crystal-field split-off hole and heavy-

hole bands crossover on gain characteristics of high Al-content AlGaN quantum well

lasers”, Appl. Phys. Lett. 97, 111105, 2010.

[63] S. L. Chuang, “Optical gain of strained wurtzite GaN quantum-well lasers”, IEEE J.

Quantum Electron, 32, 1791, 1996.

[64] H. Zhao, R. A. Arif, Y. K. Ee, and N. Tansu, “Self-consistent analysis of strain-

compensated InGaN-AlGaN quantum wells for lasers and light emitting diodes”, IEEE J.

Quantum Electron., 45, 66, 2009.

[65] I. Vurgaftman and J. R. Meyer, in Nitride Semiconductor Devices, edited by J.

Piprek, (Wiley, New York, 2007), Chap. 2.

[66] I. Vurgaftman and J. R. Meyer, “Band parameters for nitrogen-containing

semiconductors”, J. Appl. Phys. 94, 3675, 2003

99

[67] I. H. Brown, P. Blood, P. M. Smowton, J. D. Thomson, S. M. Olaizola, A. Mark

Fox, P. J. Parbrook, and W. W. Chow, “Time Evolution of the Screening of Piezoelectric

Fields in InGaN Quantum Wells,” IEEE J. Quantum Electron., 42, 1202-1208, 2006.

[68] O. B. Shchekin, J. E. Epler, T. A. Trottier, T. Margalith, D. A. Steigerwald, M. O.

Holcomb, P. S. Martin, and M. R. Krames, “High performance thin-film flip-chip InGaN-

GaN light-emitting diodes,” Appl. Phys. Lett., 89, 071109, 2006.

[69] H. Kim, K. K. Kim, K. K. Choi, H. Kim, J. O. Song, J. Cho, K. H. Baik, C. Sone,

and Y. Park, “Design of high-efficiency GaN-based light emitting diodes with vertical

injection geometry,” Appl. Phys. Lett., 91, 023510, 2007.

[70] S. Chang, W. Chen, S. Shei, C. Kuo, T. Ko, C. Shen, J. Tsai, W. Lai, J.Sheu and A.

Lin, “High-Brightness InGaN-GaN Power Flip-Chip LEDs,” J. lightwave tech, 27, 1985-

1989, 2009.

[71] C. Chu, F. Lai, J. Chu, C. Yu, C. Lin, H. Kuo and S. Wang, “Study of GaN light-

emitting diodes fabricated by laser lift-off technique,” Appl. Phys. Lett., 95, 3916-3922,

2004.

[72] H. K. Cho, J. Jang, J. H. Choi, J. Choi, J. Kim, J. S. Lee, B. Lee, Y. H. Choe, K. D.

Lee, S. H. Kim, K. Lee, S. K. Kim, and Y. H. Lee, “Light extraction enhancement from

nano-imprinted photonic crystal GaN-based blue light-emitting diodes,” Opt. Express,

14, 8655–8660, 2006.

[73] D. H. Long, I. K. Hwang, and S. W. Ryu, “Design optimization of photonic crystal

structure for improved light extraction of GaN LED,” IEEE. J. Sel. Topics Quantum

Electron., 15, 1257–1263, 2009.

100

[74] Y. S. Choi, M. Iza, E. Matioli, G. Koblmüller, J. S. Speck, C. Weisbuch, and E. L.

Hu, “2.5microcavity InGaN light-emitting diodes fabricated by a selective dry-etch

thinning process,” Appl. Phys. Lett., 92, 061120, 2007.

[75] Y. C. Shen, J. J. Wierer, M. R. Krames, M. J. Ludowise, M. S. Misra, F. Ahmed, A.

Y. Kim, G. O. Mueller, J. C. Bhat, S. A. Stockman, and P. S. Martin, “Optical cavity

effects in InGaN/GaN quantum-well heterostructure flip-chip light-emitting diodes,”

Appl. Phys. Lett., 82, 2221–2223, 2003.

[76] D. G. Deppe and C. Lei, “Spontaneous emission and optical gain in a Fabry-perot

microcavity,” Appl. Phys. Lett., 60, 527–529, 1992.

[77] H. Benisty, H. De Neve, and C. Weisbuch, “Impact of planar microcavity effects on

light extraction—Part I: Basic concepts and analytical trends,” IEEE J. Quantum

Electron., 34, 1612-1631, 1998.

[78] W. Lukosz, “Theory of optical-environment-dependent spontaneous emission rates

for emitters in thin layers,” Phys. Rev. B, 22, 3030–3038, 1980.

Documents

Analysis of Light Extraction Efficiency Enhancement for Deep