design of high performance organic light emitting diodes · organic light emitting diodes (oleds)...

Design of High Performance Organic Light Emitting

Diodes

by

Zhibin Wang

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Graduate Department of Materials Science and Engineering University of Toronto

© Copyright by Zhibin Wang 2012

ii

Design of High Performance Organic Light Emitting Diodes

Zhibin Wang

Doctor of Philosophy

Materials Science and Engineering

University of Toronto

2012

Abstract

Organic light emitting diodes (OLEDs) are being commercialized in display applications,

and will be potentially in lighting applications in the near future. This thesis is about the design

of high performance OLEDs, which includes both the electrical and optical design of OLEDs. In

particular, the following work is included in this thesis: i) Energy level alignment and charge

injection at metal/organic interfaces have been systematically studied. ii) Transition metal oxide

anodes have been developed to inject sufficient holes into the OLEDs due to their high work

function. The oxide anodes have also been used to systematically study the transport properties

in organic semiconductors. iii) Highly simplified OLED devices with unprecedentedly high

efficiency have been realized using both fluorescent and phosphorescent emitters. The high

performance was enabled by using a high work function metal oxide anode and a hole transport

material with very a deep highest occupied molecular orbital (HOMO). iv) An optical model has

been developed to describe the optical electric field across the OLED device. By using the

model, a high performance flexible OLED using metal anode was designed and realized.

iii

Table of Contents

TABLE OF CONTENTS ............................................................................................................................................ III

LIST OF TABLES ..................................................................................................................................................... V

LIST OF FIGURES.................................................................................................................................................. VI

ABBREVIATIONS AND SYMBOLS ........................................................................................................................... XI

CHAPTER 1 INTRODUCTION ............................................................................................................................ 1

1.1. BRIEF REVIEW ON OLED DEVELOPMENTS .................................................................................................................... 1

1.2. MOTIVATIONS ........................................................................................................................................................ 6

1.3. OUTLINE ............................................................................................................................................................... 7

CHAPTER 2 EXPERIMENTAL METHODS ............................................................................................................ 8

2.1. DEVICE DESIGNS ..................................................................................................................................................... 8

2.1.1. Single carrier devices .................................................................................................................................. 8

2.1.2. OLED devices ............................................................................................................................................ 10

2.2. DEVICE FABRICATION ............................................................................................................................................. 10

2.3. DEVICE CHARACTERIZATION .................................................................................................................................... 16

CHAPTER 3 ENERGY LEVEL ALIGNMENT AT METAL/ORGANIC INTERFACES .................................................... 18

3.1. INTRODUCTION .................................................................................................................................................... 18

3.2. RESULTS AND DISCUSSION ...................................................................................................................................... 20

3.3. SUMMARY .......................................................................................................................................................... 26

CHAPTER 4 ANALYSIS OF CHARGE INJECTION CHARACTERISTICS AT ELECTRODE-ORGANIC INTERFACES........ 27

4.1. INTRODUCTION .................................................................................................................................................... 27

4.2. THEORY .............................................................................................................................................................. 29

4.2.1. Space charge limited current ................................................................................................................... 29

4.2.2. Injection limited current ........................................................................................................................... 30

4.2.3. In between SCLC and ILC (quasi-Ohmic) ................................................................................................... 31

4.3. RESULTS AND DISCUSSION ...................................................................................................................................... 34

4.3.1. IV characteristics ...................................................................................................................................... 34

4.3.2. Fitting IV characteristics and transport parameters ................................................................................ 36

iv

4.3.3. Built-in potential and device thickness dependence ................................................................................ 38

4.3.4. Criterion for SCLC, quasi-Ohmic and ILC ................................................................................................... 40

4.4 SUMMARY ........................................................................................................................................................... 45

CHAPTER 5 ORGANIC/ORGANIC INTERFACE DESIGNS OF OLEDS ................................................................... 47

5.1. INTRODUCTION .................................................................................................................................................... 47

5.2. CBP INTERLAYER TO REDUCE EXCITON QUENCHING ...................................................................................................... 48

5.3. DEEP HOMO HTL: ENABLE SIMPLE STRUCTURE WITH HIGH EFFICIENCY .......................................................................... 54

5.4. DEEP HOMO HTL FOR PHOSPHORESCENT OLEDS .................................................................................................... 60

5.4. SUMMARY .......................................................................................................................................................... 67

CHAPTER 6 OPTICAL DESIGNS OF OLEDS ....................................................................................................... 68

6.1. INTRODUCTION .................................................................................................................................................... 68

6.2. OPTICAL MODEL OF OLEDS .................................................................................................................................... 69

6.2.1. Theory ...................................................................................................................................................... 69

6.2.2. Evaluation of the model ........................................................................................................................... 78

6.3. FLEXIBLE OLEDS .................................................................................................................................................. 79

6.4. SUMMARY .......................................................................................................................................................... 85

CHAPTER 7 SUMMARY AND FUTURE WORK .................................................................................................. 86

7.1. SUMMARY .......................................................................................................................................................... 86

7.2. FUTURE WORK ..................................................................................................................................................... 87

REFERENCES ................................................................................................................................................. 90

APPENDIX A. ERROR ANALYSIS ............................................................................................................... 100

APPENDIX B. LIST OF PUBLICATIONS RELATED TO THIS THESIS ................................................... 104

v

List of Tables

Table 1-1 Phosphorescent OLEDs performance from UDC (adapted from Ref. 15) .................................. 4

Table 2-1 Molecular structures of the organic molecules used in this study.............................................. 13

Table 5-1 Hole injection barrier heights (Bp ) and interfacial dipoles ( ) at different organic/organic

interfaces extracted from the UPS spectra. ........................................................................................ 52

vi

List of Figures

Figure 1.1 Schematic structure of the first efficient OLED as well as the molecular structures. Adapted

from Ref. 2. .......................................................................................................................................... 1

Figure 1.2 Schematic energy diagram of an OLED with an organic heterostructure. ................................. 2

Figure 1.3 Schematic energy diagram of single (fluorescent) and triplet (phosphorescent) emission. ........ 3

Figure 1.4 Photograph of (left) the Samsung Galaxy S cell phone using 4 inch OLED display and (right)

the LG 55 inch OLED television prototype. ........................................................................................ 4

Figure 2.1 Energy diagram of a single carrier hole-only device with a structure of metal/organic/metal. .. 8

Figure 2.2 Schematic device structure of a single carrier device. ................................................................ 9

Figure 2.3 Schematic device structure of a typical multi-layer OLED. ....................................................... 9

Figure 2.4 Layout of patterned ITO used in this study. ............................................................................. 11

Figure 2.5 Schematic structure of the devices (single carrier devices or OLED). ..................................... 11

Figure 2.6 Picture of Kurt J. Lesker LUMINOS® cluster tool. ................................................................. 12

Figure 2.7 Home-made closed-loop He3 low-temperature cryostat. .......................................................... 17

Figure 3.1 Energy diagrams of (a) metal and inorganic semiconductors and (b) metal and organic

semiconductors before making contact. ............................................................................................. 19

Figure 3.2 Current density with Au and Au/C60 anodes at room temperature as a function of electric field

for single carrier devices. The average electric field (F) is taken as F = V/d, where V is the applied

voltage and d is the device thickness. ................................................................................................ 21

Figure 3.3 He Iα (hν = 21.22 eV) spectra of (a) secondary electron cut-off (SEC) of sputter cleaned Au

and Ag with and without 3 nm of C60, and (b) SEC and valence band of Au, Au/α-NPD (3 nm) and

Au/C60 (3 nm)/α-NPD (3 nm); SEC for Au/C60 (dashed line) is also shown for reference. Interfacial

dipoles ( ) and hole injection barrier heights ( Bp ) are as indicated. ............................................. 22

vii

Figure 3.4 Effective work function ,m eff as a function of the pristine anode metal work function m .

The data shown with open points is obtained from Ref. 47,48, while the data shown with solid

points is from this work. The inset is the schematic energy level diagram of the band alignment at

Metal/C60 interface. The Fermi level of the metal is pinned by the adsorbed C60 molecules. ........... 23

Figure 3.5 Schematic energy-level diagram of the band alignment at (a) Au/organic, (b) Au/C60/organic,

(c) Ag/organic and (d) Ag/C60/organic interfaces. The symbol of d indicates the thickness of the C60

layer.................................................................................................................................................... 24

Figure 4.1 Energy Current density (J) as a function of average electric field (F = V/d) for single carrier

hole-only devices with different metal oxide anodes at room temperature (297 K). The structure of

the devices is anode/α-NPD (~ 500 nm)/Au. The solid line is the calculated SCLC from Eq. (4.3) for

α-NPD using the field dependent mobility measured by the TOF technique and reported previously

in Ref. 79. Notice that the different oxides fall into three different groups in terms of the

current-voltage (s) characteristics. ..................................................................................................... 35

Figure 4.2 Current density (J) as a function of average electric field (F = V/d) for ITO/α-NPD,

V2O5/α-NPD and Ni2O3/α-NPD plotted as ln(J/F2 ) vs. F

1/2. The organic layer thickness (d) in all

cases is ~ 500 nm. The linear fits are used to extract the apparent mobility from the IV

characteristics using Eq. (4.3). However, since a good fit is achieved for all three anodes, despite the

significant difference in injection properties (see Fig. 4.1), the goodness of fit cannot be used to

distinguish an Ohmic contact. What is more, the extracted mobility values are significantly different

for each anode, and deviates significantly from the value measured by time of flight (TOF). This

serves as an example that transport parameters cannot be extracted from IV characteristics, without

verifying that a true Ohmic contact has been made using another technique. ................................... 37

Figure 4.3 Current density (J) as a function of average electric field (F = V/d) for Ni2O3/α-NPD/Ag with

different organic layer thicknesses (d) of α-NPD, of which (a) is before the subtraction of built-in

potential (Vbi) and (b) is after subtraction of an estimated ~ 0.9 eV built-in potential. The inset of (a)

is the current density as a function of voltage in the case of (a). Ag was chosen as cathode so as to

increase the built-in potential. ............................................................................................................ 39

Figure 4.4 Current density (J) as a function of voltage (V) for different injection barrier heights ( Bp ).

The solid symbols correspond to the time-domain simulation results. The solid line is the calculated

SCLC by Eq. (4.3) using the field dependent mobility measured by the TOF technique and reported

previously in Ref. 79. The dashed line is the injection limited current (ILC) calculated using Eq.

viii

(4.4). The current density at 10 V (i.e., F = V/d = 0.2 MV/cm) for group A oxides, group B oxides

and ITO (see Fig. 4.1) is also shown as solid star symbols for comparison. ..................................... 41

Figure 4.5 The calculated electric field at the charger-injecting contact as a function of barrier height

( Bp ) for α-NPD with an organic layer thickness (d) of 100 nm and 1000 nm respectively. The

average electric field (F = V/d) for each case is 0.5 MV/cm. Notice that the electric field at the

interface converges to the average value for increasing barrier height, and tends to zero for

decreasing barrier height; the region in between defines a quasi-Ohmic contact. ............................. 43

Figure 4.6 The calculated injection “phase” diagram for α-NPD indicating the boundaries of the

quasi-Ohmic regime (i.e., the criterion for ILC and SCLC) as a function of the injection barrier

height ( ) and the organic layer thickness (d). The first region (left side) defines the criterion for

an Ohmic contact (SCLC), the second region (middle) is for a quasi-Ohmic contact and the third

region (right side) is for an injection limited contact (ILC). The boundaries for these regimes are

dependent on applied bias. The solid symbols correspond to an average electric field of 0.5 MV/cm

while the open symbols to an electric field of 0.1 MV/cm. The results for the group A and group B

oxides as well as ITO are also shown for comparison and are obtained from the experimental results

shown in Fig. 4.1................................................................................................................................ 44

Figure 5.1 Efficiency and (b) IV characteristics of the OLED devices with the following structures: (I)

α-NPD/Alq3 (standard reference); (II) α-NPD/CBP (3nm)/Alq3 and (III) α-NPD/TPBi (3nm)/Alq3.

........................................................................................................................................................... 49

Figure 5.2 Normalized EL spectra of the OLED devices with different thickness (0, 3, 10 nm) interlayer

of CBP and TPBi................................................................................................................................ 50

Figure 5.3 He Iα (hν = 21.22 eV) valence band spectra for: (a) CuPc/α-NPD/Alq3; (b) α-NPD/CBP/Alq3;

and (c) α-NPD/TPBi/Alq3. In (b) and (c) the CuPc/α-NPD interface is not shown for clarity since it

is identical to (a). ............................................................................................................................... 51

Figure 5.4 Schematic energy diagram for the device structure: (a) CuPc/α-NPD/Alq3; (b) CuPc / α-NPD /

CBP / Alq3; and (c) CuPc/α-NPD/TPBi/Alq3. The LUMOs are estimated from cyclic voltammetry

measurements.92

................................................................................................................................. 53

Figure 5.5 Device structure of (a) standard reference device, and (b) device with non-blocking exciton

formation zone. .................................................................................................................................. 54

Bp

ix

Figure 5.6 (a) Luminance-Voltage and Current-Voltage characteristics and (b) efficiency of the OLED

devices with the following structures: ITO/CuPc/α-NPD (square); ITO/WO3/CBP (circle); ITO/CBP

(triangle). ........................................................................................................................................... 55

Figure 5.7 He Iα (hν = 21.22 eV) valence band spectra of ITO and ITO/WO3 with a 3 nm thick layer of

CBP showing (a) the secondary electron cut-off, (b) the valence band, and (c) the HOMO of CBP. 57

Figure 5.8 (a) Current efficiency and (b) power efficiency of the OLED devices with the following

structures: ITO/MoO3, V2O5, WO3 (1 nm)/CBP(50 nm); ITO/MoO3, V2O5, WO3 (1 nm) /α-NPD(50

nm) and ITO/CuPc (25 nm)/α-NPD(45 nm). ..................................................................................... 58

Figure 5.9 Electroluminescence (EL) spectra of the devices shown in Fig. 5.9. ....................................... 59

Figure 5.10 Schematic energy level diagram for the device structure: (a) CuPc/α-NPD/Alq3; (b)

CBP/Alq3. The energy offsets were obtained from UPS measurements. ........................................... 59

Figure 5.11 Schematic device structures and energy-level diagrams of the devices in this study. The

HOMO and LUMO levels are obtained from Ref. 93,98,110,111. .................................................... 61

Figure 5.12 (a) IV and LV characteristics of device A as well as its (b) EQE and power efficiency as a

function of luminance. The upper inset is the molecular structure of the emitter Ir(ppy)2(acac). The

lower inset is the corresponding EL spectra measured at various current densities........................... 63

Figure 5.13 (a) IV and LV characteristics of device A, B and C as well as (b) the corresponding EL

spectra measured at 5 mA/cm2. .......................................................................................................... 64

Figure 5.14 Current efficiency of device A, B and C. The insets are the enlarged EL spectra (by 30 times

in the range of 400-490 nm) that are measured at 5 and 50 mA/cm2. ................................................ 66

Figure 6.1 Schematic diagram of the dipole plane with vertical and horizontal dipoles. .......................... 70

Figure 6.2 Schematic diagram of a multilayer structure with n+1 layers. ................................................. 72

Figure 6.3 Schematic diagram of a multilayer structure with embedded source plane. ............................. 75

Figure 6.4 Experimental EL spectra of OLEDs with different thickness of CBP measured normal to the

substrate (open symbols) as well as the corresponding theoretical calculations (solid lines). The PL

spectrum of C545T doped Alq3 used in the calculation is also shown for comparison as dashed line.

........................................................................................................................................................... 78

x

Figure 6.5 Schematic OLED device structure with flexible plastic substrate. ........................................... 80

Figure 6.6 Calculated enhancement ratio of the Ta2O5/Au/MoO3 electrode relative to ITO as a function of

the thickness of both Au and Ta2O5. .................................................................................................. 82

Figure 6.7 (a) External quantum efficiency (EQE) and (b) Power efficiency (PE) of the device structure

optimized for Ir(ppy)2 (acac) as a function of luminance. ................................................................. 83

Figure 6.8 Current density as a function of average electric field of CBP single carrier hole only device

using Au/MoO3, ITO/MoO3 and Au anodes. The anode modified by MoO3 enables good hole

injection into CBP. The inset is the same data for Au/MoO3 and ITO/MoO3 plotted on a log-linear

scale. Clearly, the injection from Au/MoO3 is better than from ITO/MoO3. ..................................... 84

Figure 6.9 Photograph of a flexible OLED (50 mm × 50 mm) at high luminance. ................................... 85

xi

Abbreviations and symbols

OLED Organic light emitting diodes

ITO Indium tin oxide

IV Current-voltage

LV Luminance-voltage

CV Capacitance-voltage

HTL Hole transport layer

ETL Electron transport layer

EML Emissive layer

HOMO Highest occupied molecular orbital

LUMO Lowest unoccupied molecular orbital

VL Vacuum level

QCM Quartz crystal microbalance

UHV Ultra-high vacuum

UV Ultraviolet

ILC Injection limited current

SCLC Space charge limited current

TOF Time of flight

EL Electroluminescence

PL Photoluminescence

CNL

Charge neutrality level

S Interface slope parameter.

EF Fermi energy level

m and S Work function of metal and semiconductor respectively

E Energy level

EC Conduction band level

EV Valence band level

Δ Interface dipole

Electron affinity

xii

Bp and Be Injection barrier for holes and electrons

J Current density

V Voltage

d Thickness of the film

F Electric field

Mobility

0 Permeability of free space

Dielectric constant

e Electron charge

f Reduced electric field

A function of the reduced electric field

T Temperature

Bk Boltzmann constant

0N Density of chargeable sites in the organic film

p Total density of holes

P Normalized power density for dipoles

( )P Photopic response

E Optical electric field

n̂ Complex refractive index

L Layer matrix (Chapter 6)

ijI Interface matrix at i/j interface

M Total system transfer matrix

t Transmission

r Reflection

Wavelength

A Dipole source term

I Irradiance (the power per unit projected area)

Radiant power

xiii

S Projected area

Solid angle

Angle

vI Luminous intensity

Chapter 1 Introduction

1


In this chapter, a brief review on the major milestones in OLEDs developments will be

presented rather than a comprehensive review of the organic electronics technology. The

motivation of this thesis will also be discussed at the end.

1.1. Brief review on OLED developments

Organic light emitting diodes (OLEDs) have attracted considerable research interest due to

their potential to be used in next generation flat panel displays and low-cost solid state lighting.

The first electroluminescence (EL) in organic materials was observed by Helfrich and Schneider

from anthracene in National Research Council, Canada in 1965.1 However no practical

application of this technology was seen to be possible due to its extremely high operation

voltage (~ 100 V). In 1987, the first efficient low-voltage organic EL was demonstrated by Tang

et al2 in Kodak® using an organic heterostructure, i.e., an organic light emitting diode, which is

generally recognized as the most significant step towards the practical applications of OLEDs

technology (see Fig. 1.1).

Figure 1.1 Schematic structure of the first efficient OLED as well as the molecular structures. Adapted from

Ref. 2.


2

HOMO

LUMO

Metal

Cathode

Anode

(ITO)Hole Transport

Layer (HTL)

Electron Transport

Layer (ETL)

Emission layer (EL)

hv

1

2

1 2

3

3

4

Figure 1.2 Schematic energy diagram of an OLED with an organic heterostructure.

Such a “heterostructure” has become a standard practice in OLED design. The injected

electrons and holes from the cathode and anode respectively are accumulated at the

heterojunction. The probability of electron-hole recombination at this organic/organic interface

will be increased, which result in the reasonably high efficiency that was achieved. This

breakthrough also attracted lots of chemists and engineers to focus on the further development

of high performance OLEDs. In the last two decades, significant effort has been devoted to

maximizing device efficiency through the design and synthesis of new materials3-7

as well as

device engineering.8-10

For example, in 1989, Tang et al further proposed a guest-host system,11

in which charge transport and luminescence were separated into two different materials, i.e. host

and guest respectively. By using an emitter with high efficiency in photoluminescence (PL), the

internal quantum efficiency of the OLED was tremendously increased. This work was also well

recognized as another milestone in the OLED development.

In terms of electrode design, Mg was used as the cathode in early device design. However,

Mg is very sensitive to moisture and oxygen and hence can be easily oxidized, resulting in a

very short life time. Another breakthrough in OLEDs was achieved by Tohoku Pioneer® using

a Li compound (e.g. Li2O) as an electron injection layer to enable a more stable metal, Al, as the

cathode. At present, LiF/Al bi-layer cathode is the most commonly used cathode to inject


3

electrons into the devices.12,13

Indium tin oxide (ITO) is the de facto standard anode due to its

high conductivity and transparency. In fact, almost all of the device designs and optimizations

were based upon ITO. For example, different hole injection layers were proposed to inject

sufficient holes from the ITO to enhance the device performance.

Another major breakthrough in OLEDs was the electrophosphorescent device accomplished

by Forrest et al.,3 which significantly increased the internal quantum efficiency as compared to

the fluorescent devices. It is believed that by using a phosphorescent dopant in the guest-host

system, a nearly 100 % internal quantum efficiency can be achieved.3,14

Figure 1.3 shows a

schematic energy diagram of single (fluorescent) and triplet (phosphorescent) emission. Excited

states can be formed by both optical (photoluminescence) and electrical (electroluminescence)

excitation. In the electrical excitation, if singlets (excited states with total angular momentum

equal to zero in a two-particle system) and triplets (excited states with total angular momentum

equal to 1) are formed with equal probability, 25% of the excitons would be singlets and 75%

would be triplets. Therefore, the quantum efficiency of fluorescence (transition from single

states to ground states) has an upper limit of 25%. Usually, according to quantum mechanics,

the radiative relaxation from triplet states to ground states is generally forbidden (i.e. happens in

a very slow time scale). However, it has been shown that with the introduction of a

phosphorescent emitter in a host-guest system,3 the transition from the triplet states to ground

states become much faster than the non-radiative relaxation. Moreover, through the inter-system

crossing, the 25% excitons formed as single states can also be utilized (see Fig. 1.3) and

therefore it is theoretically possible to achieve a 100% internal quantum efficiency in

phosphorescent OLEDs.

S1

T1

S0

Inter-system

crossing

Fluorescent

(25%) Phosphorescence

Figure 1.3 Schematic energy diagram of single (fluorescent) and triplet (phosphorescent) emission.


4

Universal Display Corporation (UDC) is a company dedicated to the development of

phosphorescent OLEDs. Table 1.1 summarizes their most recent phosphorescent device

performance, which can also be considered as the state of the art in device performance reported

from industry.

Table 1-1 Phosphorescent OLEDs performance from UDC (adapted from Ref. 15)

Performance (at 1000 cd/m2)

1931 CIE color Coordinates

Luminous Efficiency (cd/A)

Operating Lifetime (Hrs, LT 95%)

DEEP RED (0.69, 0.31) 17 14,000

YELLOW (0.44, 0.54) 81 85,000

GREEN (0.31, 0.63) 85 18,000

LIGHT BLUE (0.18, 0.42) 50 700

Figure 1.4 Photograph of (left) the Samsung Galaxy S cell phone using 4 inch OLED display and (right) the

LG 55 inch OLED television prototype.


5

In terms of application, the first commercialization was realized by Pioneer® in 1997 -

producing the first generation OLED display. About 10 years later in 2008, SONY® launched

the first commercial OLED TV (11 inch). However, the retail price of this product was still too

high. Currently, the OLED display market is dominated by Samsung. More than 67 million 4”

OLED displays were sold by Samsung in 2010 according to DisplaySearch.16

Most recently, LG

Display launched the word’s first 55 inch OLED TV at Consumer Electronics Show 2012,

which may set the future direction of the OLED display market.

Another major potential application of OLED is lighting. In a recent issue of Nature17

,

editor Dr. Stefano Tonzani wrote an editorial commentary titled “Lighting technology: Time to

change the bulb”. As was identified by Tonzani, OLEDs would be the ultimate in lighting

technology. When mature, OLED lighting will offer: energy savings (less energy will be

converted into heat); non-toxicity (mercury free); adaptability to any location due to thinness

and flexibility; cost savings requiring no large semiconductor fab for epitaxy; no filaments to

break; complete controllability in color temperature; instant start; and operability in low

temperatures, etc..

The requirements for lighting are different from display applications as lighting mainly

focuses on efficiency at high luminance. For example, a luminance of > 5×103

cd/m2 for

lighting is generally needed as compared with ~ 102

cd/m2 for most display applications such as

liquid crystal display (LCD). The goal set by the U.S. Department of Energy (DOE) for OLED

lighting in 2015 is 125 lm/W with an illuminance of 10,000 lm/m2.18

To achieve this goal, a low

driving voltage and high power efficiency device is needed. The p-i-n structure, i.e. a structure

with p-type and n-type doped hole transport layer and electron transport layer respectively to

increase the carrier density, that was proposed by Karl Leo19

greatly reduced the driving voltage

of OLEDs, resulting in a much increased power efficiency. Recently, OLEDs with efficiencies

exceeding that of standard incandescent and fluorescent lighting have been demonstrated in the

p-i-n design.20

However, usually the p-i-n structure has a challenge in the lifetime, e.g. a lot of

the materials for the p-type and n-type doping are not thermally stable. However, for lighting a

thermally stable device is needed since considerable heat will be created at high luminance.


6

1.2. Motivations

Despite recent advances, the device physics behind these breakthroughs is still not yet clear.

A lot of the improvements are in fact based on a trial and error method which is time

consuming and cost intensive. Our ultimate goal is device design and engineering, which

requires a solid understanding of device physics, such as the energy level alignment and charge

carrier injection at electrode-organic interfaces. Most organic semiconductors contain almost no

intrinsic charge carrier due to their weak intermolecular coupling. Therefore proper interface

design will help us to increase the injection of carrier charge from the electrode which will result

in enhanced device performance. Another example is exciton formation. Nearly all existing

device designs are still based on the original OLED structure introduced by Tang et al.2; the

conventional bi-layer heterojunction design of the exciton formation zone has remained

relatively unchanged. There has therefore been recent interest in improving the conventional

design of the exciton formation zone, as any improvement has the potential to improve the

device performance of nearly all OLEDs. Moreover, the efficiency of the entire device depends

not only on the internal quantum efficiency, but also on how much light can be extracted out of

the substrate (external quantum efficiency). It has been shown that by using a high index glass

to match the refractive index of organic materials, more than two times amount of light can be

extracted.21

Therefore, by tuning the optical properties of the materials to reduce the total

reflection in the stratified structure can boost the efficiency significantly.

The overall object of this thesis is to advance the state of the art in high performance OLED

devices. This work focuses on both the electrical and optical design of OLEDs. In terms of the

electrical design, energy level alignment and charge injection at metal/organic interfaces was

systematically studied. High work function anodes were developed to inject sufficient amount of

charge carriers into the OLEDs. Studies on how to extract charge carrier mobility, one of the

most important parameters for organic materials, were accompanied with the study of charge

injection. The focus was on the design of simple device structure that was enabled by the high

work function anodes as well as the design of emission zone to reduce the exciton quenching at

organic/organic interfaces. In terms of the optical design, the focus was on understanding the


7

optical electric field distribution across the OLED device so as to reduce the light trapping in

different modes, e.g. substrate mode and glass mode.

1.3. Outline

This thesis is organized as follows: Chapter 2 is about the experimental methods. Chapter 3

discusses the energy level alignment at metal/organic interfaces. Chapter 4 details the analysis

of charge injection characteristics at electrode-organic interfaces. Chapter 5 presents

enhancements in OLED performance by redesigning the exciton formation zone. Chapter 6

discusses the development of an optical model based on dipole emission. Chapter 7 is the

summary of the thesis as well as the future work.

Chapter 2 Experimental methods

8


This chapter gives an overview of the experimental designs and fabrication processes in this

thesis which includes the designs of single carrier devices and OLED devices as well as the

device characterization.

2.1. Device designs

2.1.1. Single carrier devices

Experimentally, single carrier devices with a structure of Anode / Organic / Cathode are

commonly used to study both injection and transport properties. By blocking one type of

carriers (holes or electrons) from one side (anode or cathode) with a much larger injection

barrier, the other type of carrier is dominant and a single carrier device is achieved. Figure 2.1 is

the energy diagram for a hole only single carrier device, i.e., holes are the dominant carriers (at

least 2-3 orders of magnitude higher than electrons) and electrons are blocked from the cathode

due to its high injection barrier. The corresponding schematic device structure is shown in Fig.

2.2. To reduce the impact of the built-in potential on the current-voltage (IV) characteristics, a

relatively thick organic layer is needed (~500 nm).22

FmE

LU M O

VL

VL

Metal Organic Metal

Barrier too high to

overcome

H O M O

Figure 2.1 Energy diagram of a single carrier hole-only device with a structure of metal/organic/metal.


9

Glass

Anode

Organic layer

(~ 500 nm)

Cathode

Figure 2.2 Schematic device structure of a single carrier device.

Substrate

Anode

HTL

EL

ETL

Cathode

Figure 2.3 Schematic device structure of a typical multi-layer OLED.


10

2.1.2. OLED devices

OLED is the most important device in this study as the whole purpose here is to develop an

OLED structure of extremely high performance. Figure 2.3 shows the typical multi-layer OLED

structure, which consists of hole transport layer (HTL), electron transport layer (ETL) and

emissive layer (EML). Sometimes, some devices also require buffer layers to facilitate the

injection of holes or electrons and they are called injection layers. In this study, the most

commonly used standard singlet OLED structure (from Pioneer®) for reference is: ITO/ copper

phthalocyanine (CuPc) (25 nm)/ N,N'-diphenyl-N,N'-bis-(1-naphthyl)-1-1'-biphenyl

-4,4'-diamine (α-NPD) (45 nm)/ tris(8-hydroxy-quinolinato)aluminum (Alq3) (45 nm)/LiF (1

nm)/Al (100 nm).

2.2. Device fabrication

Sample preparation consists of substrate treatments and thin film deposition on substrates.

In this study, mainly two types of substrate were used: Corning 1737 glass substrates and

commercially patterned ITO coated glass with a sheet resistance less than 15 / (see Fig.

2.4). Substrates were ultrasonically cleaned with a standard regiment of Alconox®, acetone, and

methanol followed by Ultraviolet (UV) ozone treatment for 15 minutes. The size of the substrate

is 50 mm × 50 mm, which can fit as many as 32 devices (see Fig. 2.4). By using different

masks, up to eight different structures (4 devices each to eliminate possible run-to-run

variability) can be fabricated on one substrate. A schematic structure of a device is shown in

Fig. 2.5. Pixels are defined by the intersection of the top and bottom electrode. The active area

for all devices was 2 mm2.

All devices were fabricated in a Kurt J. Lesker LUMINOS® cluster tool (Fig. 2.6). Metal

electrodes were also used in the study with the same pattern as shown in Fig. 2.4 (as bottom

electrode), and were defined using a stainless shadow mask. Ni, Cu, Co, Ag and Au were

thermally deposited in a metallization chamber with a base pressure of ~ 10-8

Torr from alumina

coated molybdenum boats. Al was also deposited in the same metallization chamber from a

pyrolytic BN crucible. Mg was deposited in another dedicated magnesium deposition chamber

with a base pressure of ~ 10-8

Torr from a BN crucible in a high temperature Knudsen cell.


11

Figure 2.4 Layout of patterned ITO used in this study.

Figure 2.5 Schematic structure of the devices (single carrier devices or OLED).


12

Figure 2.6 Picture of Kurt J. Lesker LUMINOS® cluster tool.

Different oxides were deposited on the substrate as surface modification layer (~ 1 nm) on

top of different anode. Tungsten trioxide (WO3) and molybdenum trioxide (MoO3) were

thermally evaporated from tungsten boats and alumina coated molybdenum boats respectively.

Vanadium pentoxide (V2O5) was deposited using radio frequency (rf) magnetron sputtering in a

dedicated sputtering chamber with an rf power of 150 W in 1 mTorr Ar.

It is noted that different surface treatments were used for different oxides. WO3, MoO3 and

V2O5 were UV ozone treated for 30 minutes after the deposition from the respective oxide

powder on the substrates (ITO for example). In contrast, the NiOx, CoOx, and CuO films were

fabricated from pure Ni, Co and Cu films respectively by ex situ oxidation using UV ozone for

30 minutes. For the NiOx, CoOx, and CuO, similar results were obtained from in situ oxidized

films using O2 plasma (without breaking vacuum), which indicates that any possible

atmospheric contaminants play a negligible role in device performance.


13

Various organic molecules (see table 1.1) were deposited from alumina crucibles in an

organic chamber with a base pressure of ~ 10-8

Torr. The entire procedure was finished without

breaking the vacuum, i.e., the sample was transferred from one chamber to another through the

central distribution chamber with a base pressure of ~ 10-9

Torr. Film thicknesses were

monitored using a calibrated quartz crystal microbalance (QCM). The error of the QCM can be

up to 10% due to the variance over its life time. Since the thickness of the organic layers is

critical in the studies, it was further verified (for each device) using both a stylus profilometer

(KLA Tencor P-16+) and capacitance-voltage (CV) measurements (Agilent 4294A).

Table 2-1 Molecular structures of the organic molecules used in this study

Chemical name and abbreviation Formula Molecular

Weight Chemical structure

4,4’,4’’-tris(N-3-methylphenyl-N-p

henyl-amino) triphenylamine

(m-MTDATA)

C57H48N4 789.02

gm/mol

N,N’-diphenyl-N,N’-bis-(1-naphth

yl)-1-1’-biphenyl-4,4’-

Diamine

(α-NPD)

C44H32N2

588.74

gm/mol


14

4,4',4"

-Tris(N-(2-naphthyl)-N-phenyl-am

ino)triphenylamine

(2T-NATA)

C66H48N4 789.02

g/mole

4,4'-Bis(carbazol-9-yl)biphenyl

(CBP)

C54H36N4 740.89

g/mole

Tris(8-hydroxy-quinolinato)alumin

ium

(Alq3)

C27H18AlN3O

3

459.43

g/mole

N

O

N

O

N

O

Al

Phthalocyanine

(CuPc)

C32H16N8Cu 576.07

g/mole


15

2,2',2"

-(1,3,5-Benzinetriyl)-tris(1-phenyl-

1-H-benzimidazole)

(TPBi)

C45H30N6 654.76

g/mole

Tris(2,4,6-trimethyl-3-(pyridin-3-y

l)phenyl)borane

(3TPYMB)

C42H42N3B 599.61

g/mole

Tris(2-phenylpyridine)iridium(III)

[Ir(ppy)3]

C33H24IrN3 654.78

g/mole

Bis(2-phenylpyridine)(acetylaceton

ate)iridium(III)

[Ir(ppy)2(acac)]

C27H23IrN2O2 599.70

g/mole

N

Ir

2

O

O


16

2,3,6,7-Tetrahydro-1,1,7,7,-tetrame

thyl-1H,

5H,11H-10-(2-benzothiazolyl)quin

olizino-[9,9a,1gh]coumarin

(C545T)

C26H26N2O2S 430.56

g/mole

Fullerene

(C60)

C60 720.64

g/mole

2.3. Device characterization

IV and luminance-voltage (LV) characteristics for OLED were measured using an HP4140B

pA meter and a Minolta LS-110 Luminance meter respectively in ambient air. The luminous

flux for calculating the EQE and power efficiency were measured using an integrating sphere

with a silicon photodiode with NIST traceable calibration.23

Error analysis of the IV and LV

measurement can be found in Apendix A. EL spectra were measured by a USB4000 miniature

fiber optic spectrometer, which couples a linear CCD-array detector ranging from 350 nm from

1100 nm. For single carrier devices, the IV characteristics were measured in a home-made

closed-loop He3 cryostat (Fig. 2.7) with a base pressure of ~ 10-6

Torr at room temperature. The

operating pressure of the cryostat was significantly less at low temperature (i.e. ~ 10-8

Torr) due

to the cryo-pumping effect of the cryostat cold finger (i.e. residual gases condense onto the cold

finger). The glass substrates were mounted on the Cu block cold finger of the cryostat using

Apiezon N cryogenic grease for good thermal contact. The temperature of the substrates was

monitored using a calibrated chromel-alumel (Type K) thermocouple bonded directly to the top


17

of the substrate using the same cryogenic grease. The temperature was stable to within 0.1 K

over all temperature ranges.

Figure 2.7 Home-made closed-loop He3 low-temperature cryostat.

Ultraviolet photoelectron spectroscopy(UPS) characterization was performed using a PHI

5500 Multi-Technique system attached to a Kurt J. Lesker multi-access chamber ultra high

vacuum (UHV) cluster tool (base pressure of ~ 10-10

Torr). Organic molecules were deposited in

a dedicated organic chamber from a homemade transfer-arm evaporator cell (TAE-cell).24

The

spectrometer (hemispherical analyzer) was calibrated using monochromatic Al Kα (hν = 1486.7

eV) as per ISO 15472.25

Chapter 3 Energy level alignment at metal/organic interfaces

18


In this chapter, energy level alignment at metal/organic interfaces will be discussed. The

interface dipole theory will be used to describe the energy level alignment at metal/organic

interfaces. As an example, the metal/ C60/organic interfaces will be shown to illustrate the Fermi

level pinning at the metal/organic interfaces. The content of this chapter was published as Appl.

Phys. Lett. 95, 043302 (2009).

3.1. Introduction

Figure 3.1 (a) shows an idealized energy diagram for a particular metal and intrinsic

inorganic semiconductor before making contact. The vacuum level (VL) is used as a reference

level (zero energy level). Above the VL the electron can escape the solid. The difference

between VL and Fermi level EF, is called the work function. Here m and S denote the work

functions of the metal and the semiconductor respectively. The conduction band minimum EC is

a distance χ below the VL, where χ is the electron affinity. The difference between the valence

band level EV and EC is the band gap of the semiconductor. For an intrinsic semiconductor the

Fermi level is located approximately at the middle of the band gap. Similar to EV and EC, in an

organic semiconductor (see Fig. 3.1 b) there is the highest occupied molecular orbital (HOMO)

and the lowest unoccupied molecular orbital (LUMO). The major difference is that the states are

highly localized in an organic semiconductor while they are continuous in an inorganic

semiconductor. The Gaussian distribution model is commonly used to describe the energy

disorder in organic semiconductors.26

When the metal and organic form a contact, before any possible charge transfer takes place,

the energy levels are still aligned at the vacuum level. After the charge transfer takes place, to

balance the charge transfer, an interfacial dipole is formed within the first few mono-layers (see

Fig. 3.1 c). This is different from the traditional inorganic semiconductor since there is almost

no intrinsic charge carrier in the organic semiconductor, thus a depletion zone near the junction

is not expected. The formation of the interfacial dipole will result in a change of injection


19

barrier, i.e. the injection barrier would be different from the difference between the work

function of metal and the LUMO (e.g. for the injection of electron) of organic.

Metal Semiconductors

VLVL

m

Interface

s

CE

VE

FE

FE

Metal Organic

Interface

FE

FE

LUMO

HOMO

(a)

(b)

gE

Interfacial

dipole

(c)

be

VL

FE

FE

Figure 3.1 Energy diagrams of (a) metal and inorganic semiconductors and (b) metal and organic

semiconductors before making contact.

Unlike the traditional inorganic semiconductor, a lot of properties of organic semiconductor

are still unknown and are needed to be investigated. For example, since there are no dopants in

many applications of organic semiconductors, the Fermi level is not easily determined. The

argument that how the Fermi level aligns is still under debate.27

Mönch28

proposed that interface

dipole theory,29

originally developed to describe the interfacial dipole at Schottky contacts,


20

could also be applied to metal/organic interfaces. From interface dipole theory, the dipole at a

metal/organic interface is given by,29

1 m CNLS , (3.1)

where CNL is the charge neutrality level of the organic and S is the interface slope

parameter. S is commonly used to describe the degree of Fermi level ( FE ) pinning at a

metal/semiconductor contact30

,

F mS dE d . (3.2)

For vacuum level alignment S = 1 (Schottky-Mott limit) and for Fermi level pinning S = 0.

Here, metal/C60 is taken as an example to study the energy level alignment. Despite the

broad applications of C60, i.e. OLEDs,31-34

organic photovoltaics (OPVs),35,36

organic thin film

transistors (OTFTs)37

and organic memory devices (OMDs),38

C60 still remains poorly

understood in devices physics. Recently, there have been several reports of C60 as efficient hole

injection layer at the anode in OLEDs (e.g., Au/C60).31,32,34,39

Considering the fact that C60 is an

electron transporting molecule with a deep HOMO, it seems unlikely that C60 can enhance hole

injection, rather it should serve as hole blocking layer. In the following subsection, the

energy-level alignment and charge injection properties at metal/C60 interfaces will be studied

using UPS as well as IV measurements of metal/C60/organic/metal single carrier devices.

3.2. Results and discussion

In order to elucidate the unusual charge injection properties at metal/C60/organic interfaces,

single carrier devices with Au and Ag anodes and m-MTDATA, 2T-NATA and α-NPD organic

layers were fabricated. The device structure is: anode/C60(0, 3 nm)/organic (500 nm)

/Au(cathode). Figure 3.2 shows the room temperature (297 K) IV characteristics for the case of

Au. For all three organics the C60 interlayer dramatically improves hole injection by nearly 3

orders of magnitude (reduced injection barrier height). Similar improvements in the IV


21

characteristics were also found for Ag (which is not shown here). The seemingly universal

improvement in the IV characteristics is extremely surprising, given that the HOMO of C60 (6.3

eV) is much deeper than that of m-MTDATA (5.1 eV), 2T-NATA (5.1 eV) or α-NPD (5.4 eV).

Indeed, the injection barrier height from Au and Ag to C60 is expected to be ~ 1 eV and ~ 2 eV

respectively (HOMO to Fermi level offset).

0.01 0.110

-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Au/C60

/m-MTDATA

Au/m-MTDATA

Au/C60

/-NPD

Au/-NPD

Au/C60

/2T-NATA

Au/2T-NATA

Cu

rren

t D

ensity (

A/c

m2)

Electric Field (MV/cm)

0.5

Figure 3.2 Current density with Au and Au/C60 anodes at room temperature as a function of electric field for

single carrier devices. The average electric field (F) is taken as F = V/d, where V is the applied voltage and d

is the device thickness.

Since the C60 interlayer is very thin (~ 3 nm) it can be treated as a surface modification

layer on the metal surface. This assumption is reasonable given that C60 monolayers tend to be

metalized (i.e., take on metallic character) by the interaction with clean metal surfaces.40-43

It is

also well known44

that molecular adsorbates can significantly change the work function of a

clean metal surface. It is therefore likely that the C60 interlayer reduces the hole injection barrier

height by modifying the work function of the underlying metal, resulting in a more favorable

energy-level alignment for hole injection. Therefore photoemission measurements were

conducted to confirm this hypothesis. Figure 3.3 (a) shows the He Iα (hν = 21.22 eV) secondary

electron cut-off spectra for sputter cleaned Au and Ag with and without a 3 nm thick layer of

C60 (in situ deposited).


22

The extracted work function values from these spectra are summarized in Fig. 3.4 and are in

excellent agreement with other UPS and Kelvin probe measurements reported in literature

(shown as open points in Fig. 3.4). Surprisingly, the work function of both metals is found to

converge to ~ 4.7 eV, which indicates that there is a strong interfacial dipole ( ) between the

metals and C60 that acts to pin the Fermi level. The effective work function ( ,m eff ) of the C60

modified metals is then determined by the difference between the pristine metal work function

m（）and , as shown in the inset of Fig. 3.4. From Eq. (3.1) we can deduce that the Fermi

level is thus pinned to the charge neutrality level ( ) of C60, namely 4.7 eV.

17 16 15

17 16 15 2 1 0 -1

(b)

60Ag/C

60Au/C

Ag

Au

Ag/C60

Au/C60

No

rma

lize

d I

nte

nsity (

a.u

.)

(a)

60Au/C

Au/α-NPD

60Au/C /α-NPD

Au

Au/-NPD

Au/C60

/-NPD

Binding Energy (eV)

Bp

Bp

EF

HOMO

Figure 3.3 He Iα (hν = 21.22 eV) spectra of (a) secondary electron cut-off (SEC) of sputter cleaned Au and Ag

with and without 3 nm of C60, and (b) SEC and valence band of Au, Au/α-NPD (3 nm) and Au/C60 (3

nm)/α-NPD (3 nm); SEC for Au/C60 (dashed line) is also shown for reference. Interfacial dipoles ( ) and

hole injection barrier heights ( Bp ) are as indicated.

CNL


23

However, if C60 pins the Fermi level near 4.7 eV, then Au, with a higher work function of ~

5.1 eV, should still perform better than Au/C60 as anode. From Fig. 3.2 this is clearly not the

case for any of the Au/C60/organic interfaces. This discrepancy is due to the interfacial dipole at

the metal/organic interface (e.g., Au/α-NPD) that drastically reduces the effective work function

of Au, same as for the Au/C60 interface. In the case of m-MTDATA, 2T-NATA and α-NPD the

charge neutrality level is significantly lower than that of C60 resulting in a lower net value of

,45,46

and hence a higher injection barrier height.

4.0 4.5 5.0 5.53.0

3.5

4.0

4.5

5.0

5.5

Metal C60

Vacuum Level

HOMO

,m eff LUMO

Bp

Kelvin Probe (from Ref.)

UPS (from Ref.)

UPS (this work)

metal/C60

/NPD (transport)

metal/C60

/m-MTDATA (transport)

m

,eff

(eV

)

m (eV)

S ~ 0

Metal / C60

Figure 3.4 Effective work function ,m eff as a function of the pristine anode metal work function m . The

data shown with open points is obtained from Ref. 47,48, while the data shown with solid points is from this

work. The inset is the schematic energy level diagram of the band alignment at Metal/C60 interface. The Fermi

level of the metal is pinned by the adsorbed C60 molecules.

To illustrate this point the difference in energy-level alignment between Au/α-NPD and

Au/C60/α-NPD was also measured using UPS. Figure 3.3 (b) shows the He Iα secondary

electron cut-off spectra (left panel) and the He Iα valence band spectra (right panel) for

Au/α-NPD (3 nm) and Au/C60 (3 nm)/α-NPD (3 nm). As indicated in the figure (left panel) the

interfacial dipole ( ) is much larger for Au/α-NPD than for Au/C60/α-NPD resulting in a lower

,m eff


24

value of . From the valence band spectra (right panel), it is clear that there is also a larger

offset between the HOMO and Fermi level for Au/α-NPD resulting in a much higher injection

barrier height ( Bp 1.3 eV) than for Au/C60/α-NPD (Bp 0.9 eV). This finding is consistent

with the larger interfacial dipole for Au/α-NPD discussed above (i.e. /Au NPD >60/ /Au C NPD ).

Au

Au

/Au Organic

Vacuum Level

HOMO

(a)

LUMO

Organic Au Organic

HOMO

(b)

LUMO

Vacuum Level

Bp

60/

4.6 eV

eff

Metal C

C60

HOMO

LUMO

60/Au C

Bp

d

Ag

Ag

/Ag OrganicVacuum Level

HOMO

(c)

LUMO

Organic Ag Organic

HOMO

(d)

LUMO

Vacuum Level

Bp

C60

HOMO

LUMO

60/Ag C

Bp

d

60/ /Au C organic

60/ /Ag C organic

'

Bp

'

Bp

60/

4.6 eV

eff

Metal C

Figure 3.5 Schematic energy-level diagram of the band alignment at (a) Au/organic, (b) Au/C60/organic, (c)

Ag/organic and (d) Ag/C60/organic interfaces. The symbol of d indicates the thickness of the C60 layer.

,m eff


25

This finding can also be confirmed by comparing the injection barrier extracted from the IV

characteristics. The details of how to extract injection barrier from different injection models

will be discussed in the next chapter. If the effective work function of C60 modified metal is

calculated using the relation,m eff BpHOMO , for example for Ag/C60(3 nm)/α-NPD, a value

of ~ 4.55 eV can be found. However, this simple calculation is based on assumption that the

dipole between the C60 modified metal and the next organic layer is zero. As shown previously

in Fig. 3.3 (b), there is a small dipole between C60 modified metal and α-NPD. By applying Eq.

(3.1) using the values of and for different molecules reported elsewhere45

the

extracted effective work functions can be further corrected for the interfacial dipole between the

C60 modified metal (e.g., Ag/C60) and the next organic layer (see Fig. 3.5). After applying this

correction the effective work function was found to be ~ 4.7 eV, in excellent agreement with the

UPS results (see the results summarized in Fig. 3.4 for the other cases).

The findings discussed above are summarized in the energy-level diagrams shown in Fig.

3.5 for Au and Ag with and without a surface modification layer of C60. These energy diagrams

summarize three major points, which are the key findings of this work. First, clean metals, such

as Au and Ag, have a strong interfacial dipole with commonly used hole transporting molecules,

resulting in unfavorable energy-level alignment for hole injection. Second, by inserting a thin

interlayer of C60 the net interfacial dipole is reduced due to pinning of the electrode Fermi level

to the charge neutrality level of C60 (i.e., ~ 4.7 eV). Since the C60 interlayer acts as a surface

modification layer, holes are directly injected into the next organic layer, bypassing the HOMO

of C60, i.e. the injection barrier in this case is '

Bp (see Fig. 3.5 b and d). However, when the C60

layer is too thick, i.e. larger than 5 nm, the current density of the metal/C60/organic interface

becomes too small to be measured (beyond the range of the equipment, i.e. near the noise floor

of 10-10

A/cm2). In this case holes must be injected into the C60 layer, resulting in a large

injection barrier height of ~1.7 eV (6.3 eV – 4.6 eV). The C60 thickness of 3 nm is the optimized

value, i.e. it is thick enough to pin the Fermi level, but is thin enough to permit charge tunneling.

Finally, the previous two points do not negate the fact that there is a small dipole between the

C60 modified metal (e.g., Au/C60) and the next organic layer, which will also modify the

injection barrier at the metal/C60(3 nm)/organic interface. The net dipole (60/ /metal C organic ) as

CNL S


26

shown in Fig. 3.5 (b) and (d) is found to be less than the intrinsic dipole at the metal/organic

interface (without C60).

3.3. Summary

In summary, it was shown that interface dipole theory accurately describes the energy level

alignment at metal/C60/organic interfaces. The Fermi level at metal/C60/organic interfaces was

found to be pinned to the charge neutrality level of C60 (~ 4.7 eV). This phenomena was

attributed to the C60 interlayer disrupting the interfacial dipole at the metal/organic interfaces,

resulting in more favorable energy-level alignment for hole injection. As a result holes were

injected directly into the organic layer, bypassing the deep HOMO of C60.

Chapter 4 Analysis of charge injection characteristics at electrode-organic interfaces

27

Chapter 4 Analysis of charge injection characteristics at

electrode-organic interfaces

In this chapter, the hole injection from different metal oxides into α-NPD will be

systematically studied in single carrier hole-only devices. A criterion that defines Ohmic,

quasi-Ohmic, and injection limited contacts will be quantified based on a time-domain

simulation of charge transport across α-NPD single carrier devices. A barrier-thickness-voltage

“phase” diagram that defined the regions of SCLC, quasi-Ohmic and ILC for α-NPD will also

be presented. The content of this chapter was published as Phys. Rev. B 80, 235325 (2009).

4.1. Introduction

As mentioned previously, organic semiconductors contain almost no intrinsic charge

carriers due to their weak intermolecular coupling. In order to enhance the performance of

OLEDs, one has to increase the extrinsic carrier concentration. Therefore, improving the

injection of charge from the electrodes is of great significance to achieve highly efficient

OLEDs. A commonly used approach to reduce the barrier height for charge injection is to

increase (decrease) the work function of the anode (cathode), towards the ultimate goal of

achieving Ohmic contacts. In the previous chapter, the reason why a thin (3 nm) C60 interlayer

can enable Au as anode in green OLED32

has been shown, where the thin C60 interlayer pins the

work function of the anode to be ~4.7 eV,49

resulting a more favorable injection barrier than

pure metal anode. However, the effective work function of 4.7 eV still limits the selection of

hole transport layer (HTL), i.e. the effective work function is not yet high enough.

Recently, transition metal oxides, such as CuO, WO3, MoO3 and V2O5 have been shown to

be promising candidates to replace the previous generation of organic hole injection layers

(HILs) at the anode, due to their stability, low cost and their high work function.50-53

It is this

high work function (~ 5 - 6 eV) in particular that has attracted the most attention, since it

suggests the possibility of forming an Ohmic contact for holes with many of the commonly used


28

hole transport layers (HTLs). However, a high work function does not guarantee an Ohmic

contact, or even good hole injection. For example, as studied in the previous chapter, Au with a

high work function of ~ 5.1 eV will form a strong interfacial dipole at the metal/organic

interface, which significantly increases the injection barrier for holes. Therefore not only do we

have to consider the work function of the oxides, but we also have to take into account the

energy-level alignment, for example, interfacial dipole effects. In order to better understand the

physics that governs the injection process at such interfaces, particularly for the case of an

Ohmic contact, a systematic study of different oxides used in devices is needed. However, few

comparative studies have been conducted to study the injection properties of different oxides in

devices, i.e. single carrier hole-only devices. There may be a large variation in device

performance from study-to-study and different processing methods and the organic materials

used may introduce a large variation. It is therefore difficult to compare these results.

Several transition metal oxides have also been assumed to form an Ohmic contact with

commonly used hole transporting molecules, such as α-NPD54

. With the assumption of Ohmic

contacts, mobility of several organic semiconductors was conveniently extracted by modeling

the IV characteristics using the space charge limited current (SCLC).54,55

As will be shown in the

following text, however, without clear criteria to distinguish between injection limited current

(ILC) and SCLC, the data may easily be misinterpreted. For example the square law, i.e. the

Mott-Gurney law 2

0 3

9

8

VJ

d , has been commonly used to analyze the IV characteristics of

devices and has also been commonly cited as a criterion for SCLC.56-60

However, such a

criterion becomes questionable if the field dependent mobility is unknown, as one can simply

compare the IV characteristics to the quadratic relation. It is also well known that other effects

such as non-uniform emission from a “patchy” interface,61

the distribution of trap states in the

bulk of the organic and the built-in potential62

, will all change the shape of the IV characteristics.

It is therefore extremely difficult to distinguish ILC from SCLC in organic devices from the IV

characteristics alone. Thus, any IV data analysis to extract parameters, such as the energy

disorder and mobility of organics, are dubious. This work will show a comparative study

between the most commonly used transition metal oxides in organic electronic devices. These

diverse experimental results enabled a thorough analysis of the various models relating to the


29

injection properties at electrode/organic interfaces. To aid the analysis, time-domain simulations

were used to establish a criterion to distinguish ILC from SCLC for α-NPD. As will be shown

below there is no clear boundary between ILC and SCLC, but rather a large region in between,

which will be referred to as a “quasi-Ohmic” regime for the convenience of discussion. This

intermediate quasi-Ohmic regime requires special consideration in terms of device modeling.

4.2. Theory

4.2.1. Space charge limited current

The maximum current that an organic semiconductor can sustain in the bulk (i.e., the

amount of carriers in thermal equilibrium) is called the SCLC. An Ohmic contact is therefore an

interface capable of injecting enough charges to sustain SCLC. One significant feature of SCLC

is that the spatial distribution of electric field is 1/2( )F x x , where x is the distance from the

charge-injecting contact.63

Therefore the electric field at an Ohmic contact is equal to zero,

which as will be shown is an important criterion for distinguishing SCLC from ILC. Based on

traditional semiconductor device physics, the SCLC for uni-polar transport (i.e., single carrier

devices) in a perfect insulator (no intrinsic carriers) without traps is given by the Mott-Gurney

law,63

2

0 3

9

8

VJ

d , (4.1)

where V is the applied voltage, d is the thickness of the film, and μ is the field-independent

mobility. With a further consideration of an exponential tail of trap states, the IV characteristics

based on Eq. (4.1) follows,63

1

2 1

l

l

VJ

d

, (4.2)

where l is a parameter derived from the trap distribution. However, it is well known that the

mobility for most organic semiconductors is field-dependent. Also, in disordered organic

materials, it is believed that all electronic states are localized and participate in conduction


30

through thermally activated hopping, which yields a Poole-Frenkel like field dependence of the

mobility, i.e. 0( ) exp( )F F . Under the assumption of the Poole-Frenkel dependence, an

approximation to the SCLC for a field dependent mobility is given by,64

2

0 0 3

9exp 0.89

8SCLC

V VJ

d d

. (4.3)

Indeed, it has already been demonstrated that the above equation mathematically describes well

the IV characteristics of many organic semiconductors.65-67

4.2.2. Injection limited current

When the current in an organic semiconductor is limited by the injection of charge from the

electrode rather than the bulk properties of the material, it is called ILC. Under ILC conditions

the spatial electric field distribution is assumed to be uniform, i.e. ( ) /F x V d ; whereas for

SCLC the value of V/d only gives the average value of the electric field. This suggests that

SCLC can be distinguished from ILC by the electric field at the charge-injecting contact [i.e.,

( 0) 0F x for SCLC and ( 0) /F x V d for ILC].

The most commonly used injection models for ILC are derived based on

Richardson-Schottky (RS) emission,68

Fowler-Nordheim (FN) tunneling69

and the hopping

model.70

It has been pointed out that RS emission is not strictly correct71

for systems where the

electron mean free path is very short, such as in organic semiconductors. Although the hopping

model includes the discrete (molecular) nature of organic semiconductors, it has also been

pointed out that RS emission provides a solid basis for the analysis of the charge injection

characteristics of organic semiconductors since there is little quantitative difference between RS

emission and the hopping model.72,73

In this work the RS emission based model for ILC

proposed by J. C. Scott will be used. The model considers the scattering and diffusion effect by

solving the drift-diffusion equation in the depletion zone of an amorphous semiconductor,74

2 1/2

04 exp exp( )BILC

B

eJ N e F f

k T

. (4.4)


31

In Eq. (4.4), 0N is the density of chargeable sites in the organic film, B is the injection

barrier height, F is the electric field at the charge-injecting contact, is the electric field

dependent (Poole-Frenkel) carrier mobility, Bk is the Boltzmann constant, T is temperature, e

is the electron charge and is a function of the reduced electric field ( 3 2 2/ 4 Bf e F k T ),

2/12/112/11 )21( ffff . (4.5)

4.2.3. In between SCLC and ILC (quasi-Ohmic)

The two conditions discussed above, SCLC and ILC, correspond to the upper and lower

limits respectively. However, what about the case in between SCLC and ILC, when the electric

field at the electrode contact is 0 ( 0) /F x V d ? This indicates that there is in fact no clear

boundary between SCLC and ILC, but rather an intermediate regime in between. In other words,

just because an IV characteristic is not ILC does not guarantee that it is SCLC, and vice versa.

Since the intermediate regime exhibits characteristics of both SCLC and ILC, it is incorrect to

apply the models for either case [e.g., Eq. (4.1) – (4.5)]. This can easily be understood since all

models for SCLC require that ( 0) 0F x at the charge-injecting contact, while all models for

ILC require that ( 0) /F x V d at the charge-injecting contact. Hence, for the region in

between SCLC and ILC [i.e., 0 ( 0) /F x V d ] neither type of model can be applied since

the boundary conditions at the charge-injecting contact is not met. Therefore a new modeling

approach is required in order to properly deal with the case between SCLC and ILC. From now

on we will refer to this intermediate region as “quasi-Ohmic”, i.e. the current in the organic

semiconductor is limited by both the injection at the electrode and by the bulk properties of the

material.

Here, it is important to discuss the above definition of quasi-Ohmic. The term

“quasi-Ohmic” has been misused in literature to describe the injection from a contact that is

close enough to Ohmic, so that the IV characteristics might be approximated by SCLC.

However, this loose definition has often lead to physically meaningless analyses of IV

characteristics using Eq. (4.1) or Eq. (4.3). Therefore, it is necessary to clearly define injection


32

from a quasi-Ohmic contact in terms of the electric field at the charge-injecting contact [i.e.,

0 ( 0) /F x V d ]. Based on this clear definition, the quasi-Ohmic regime will be shown to

be much larger (i.e., covers a greater range of injection barrier height) than has previously been

expected. Moreover, it is important to note that the boundaries of the quasi-Ohmic regime are

dependent on the field-dependent carrier mobility μ, the injection barrier height B , the applied

voltage V and the device thickness d. As a result the same electrode contact (i.e., anode/organic

interface) may display characteristics of ILC, quasi-Ohmic and SCLC in different ranges of

applied bias. This has significant implications for device modeling since most metal/organic

interfaces used in real devices are found to fall into the quasi-Ohmic regime at typical operating

voltages.

As discussed above, the quasi-Ohmic regime in between SCLC and ILC cannot be

described using traditional models for SCLC or ILC. As a result a new approach that includes

both the injection at the interfaces and the transport in the bulk of the organic is required to deal

with this special case. Based on the transport models developed for inorganic semiconductors,

i.e. the drift-diffusion and Poisson equations, a time domain simulation can be conducted to

study the distribution of electric field and carriers at steady state. This method has been

employed by J. C. Scott et al.75,76

to study the charge injection and transport properties in

single-layer OLEDs. Here, based on a similar theoretical framework, a simulation of single

carrier hole-only devices22

was conducted to study the quasi-Ohmic injection and to define a

clear criterion to distinguishing SCLC from quasi-Ohmic and ILC.

In the simulation, space x and time t are discrete. The injection current density described by

Eq. (4.4) and (4.5) serves as a boundary condition at 0, 0x t (i.e., the charge-injecting

contact). The time-dependent continuity equation follows,

( , ) 1 ( , )p x t J x t

t e x

, (4.6)

where p is the total density of holes and J is the conduction current density. The relation

between the electric field and the charge density can be expressed by the Poisson equation as,


33

0 ( , )( , )r F x t

p x te x

. (4.7)

In Eq. (4.7) r and 0 are the dielectric constant of the organic and the dielectric permittivity

in vacuum respectively. The conduction current density can be calculated through the

drift-diffusion equation,

( , )( , ) ( , ) ( )

p x tJ x t ep x t F F eD

x

, (4.8)

where D is the diffusion coefficient, which can be obtained from Einstein’s relation (as a

function of the field-dependent mobility),

Bk TD

e

. (4.9)

The other boundary conditions for Eq. (4.6) – (4.9) are,

0

( , 0) /

( , 0) 0

( , 0) 0

d

V F dx

F x t V d

p x t

J x t

, (4.10)

where d is the thickness of the film and V is the applied voltage. It is noted that the transient

current density tJ is contributed by the displacement current and the response of the charge

carrier density as,

0

( , )( , ) ( , )t r

F x tJ x t J x t

t

. (4.11)


34

From Eq. (4.11) the total transient current at x d can be calculated until the steady state

is reached. Also, the spatial distribution of electric field at steady state can be obtained from the

simulation. From these simulation results the boundaries of the quasi-Ohmic regime (i.e., the

lower limit of SCLC and the upper limit of ILC) can then be defined in terms of the electric

field at the charge-injecting contact. Clearly, the boundaries of the quasi-Ohmic regime are

dependent on the field-dependent carrier mobility μ, the injection barrier height B , the applied

voltage V and the device thickness d.

Finally, it is noted that based on the results of Monte Carlo simulations, a criterion has been

previously proposed to distinguish SCLC from ILC in organic semiconductors.77

However this

criterion does not take the field dependent mobility into consideration, i.e. the criterion should

be dependent on the applied electric field and should be different for different organic

semiconductors. Moreover, the dependence of the film thickness has not been taken into account

either. It is well known that the thicker the organic layer, the easier the injection will reach the

SCLC since less charge can be sustained in the bulk. Here a criterion for defining the boundaries

of the quasi-Ohmic regime will be presented as a function of both the injection barrier height

and film thickness for α-NPD by conducting a time domain simulation under Eq. (4.4) – Eq.

(4.11).

4.3. Results and discussion

4.3.1. IV characteristics

Figure 4.1 compares the IV characteristics of α-NPD single carrier hole-only devices with

different oxide anodes and ITO at room temperature (297 K). The device structures are:

anode/α-NPD (~ 500 nm)/Au. Here Au is used as the cathode to block electron injection.22

In

order to eliminate any possible run-to-run difference in the organic layer thickness, the organic

layer thickness was measured for each device and the IV characteristics were plotted as a

function of electric field. Also, the device thickness was chosen as 500 nm to minimize any

effects of built-in potential at high electric field. The average electric field is taken as F = V/d,

where d is the device thickness. The results for a pure Ni metal anode with high injection barrier

height78

is also shown for comparison. The device performance can be divided into three distinct


35

groups: A) CuO, Co3O4, Ni2O3, MoO3; B) V2O5, WO3; and C) ITO, Cu2O. As can be seen from

the figure, the first group performs the best; the current density is almost one order of magnitude

higher than for the second group and nearly three orders of magnitude higher than the ITO

device.

0.01 0.110

-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

SCLC

CuO

Co3O

4

Ni2O

3

MoO3

V2O

5

WO3

Cu2O

ITO

Ni

C

B

C

urr

en

t D

ensity (

A/c

m2)


A

Figure 4.1 Energy Current density (J) as a function of average electric field (F = V/d) for single carrier

hole-only devices with different metal oxide anodes at room temperature (297 K). The structure of the

devices is anode/α-NPD (~ 500 nm)/Au. The solid line is the calculated SCLC from Eq. (4.3) for α-NPD

using the field dependent mobility measured by the TOF technique and reported previously in Ref. 79. Notice

that the different oxides fall into three different groups in terms of the current-voltage (s) characteristics.

The theoretical SCLC is calculated using Eq. (4.3) and shown in Fig. 4.1 (solid black line)

as the upper limit of the IV characteristics, where the field dependent mobility was measured by

the time-of-flight (TOF) technique and reported previously in Ref. 79. Indeed, the group A

oxides approach this upper limit at high electric field, which would suggest that they might be

close to forming an Ohmic contact with α-NPD. As discussed in the Theory section, this case

has traditionally been dealt with by assuming that the injection is close enough to an Ohmic


36

contact that the current can be approximated by the SCLC. However, it will be shown that none

of the oxides studied in this work were found to form an Ohmic contact with α-NPD (over the

studied range of applied voltage), despite previous reports to the contrary. This discrepancy is

due to the erroneous application of the Mott-Gurney law [i.e., Eq. (4.1) or Eq. (4.3)] to model IV

characteristics in the quasi-Ohmic regime between SCLC and ILC. The quasi-Ohmic regime is

found to cover a significant range of injection barrier heights and is dependent on the applied

voltage, which will be discussed in details in Section 4.3.4.

4.3.2. Fitting IV characteristics and transport parameters

In traditional semiconductor physics the slope of the IV characteristics in the saturation

regime is often used to identify SCLC following the Mott-Gurney law (i.e., Eq. 4.1). This

criterion has also been commonly used (incorrectly albeit) for organic semiconductors. For

example, the device performance of MoO3 shown in Fig. 4.1 is consistent with the data reported

in Ref. 54. (i.e., the current density is equivalent at the same applied electric field), in which the

contact between MoO3 and α-NPD was reported to be Ohmic since a slope of 2 of the IV

characteristics was achieved. However, as discussed in the Theory section the Mott-Gurney law

given by Eq. (4.1) does not apply to organic semiconductors with a field dependent mobility,

such as α-NPD. Hence a slope of 2 of the IV characteristics does not indicate that SCLC has

been obtained.

Since the mobility of α-NPD is field dependent, the slope of the IV characteristics for SCLC

should not be exactly 2, but in fact larger [see Eq. (4.3)]. The theoretical SCLC for α-NPD was

calculated (black solid line in Fig. 4.1) based on the field dependent mobility data that have been

reported elsewhere measured by the TOF method79

; a slope of ~ 2.5 is obtained from the

calculated curve at high electric field. However, as will be shown below, even a slope of ~ 2.5

of the IV characteristics is still not sufficient to indicate that SCLC has been obtained. It should

be pointed out that the shape of the IV curve is influenced by many other effects such as traps,

the built-in potential and non-uniform “patchy” emission. The effects of built-in potential will

be discussed in detail below. In any event, due to the convolution of injection and transport

properties, combined with the other effects mentioned above, the IV characteristics are


37

insufficient to determine whether the current density is SCLC. In other words it is extremely

difficult if not impossible to identify an Ohmic contact from the IV characteristics alone.

200 400 600 800-36

-34

-32

-30

-28

~ 10-6 cm

2/Vs

Ni2O

3

V2O

5

ITO

~ 10-5 cm

2/Vs

~ 10-7 cm

2/Vs

ln[ J/F

2(A

/V2)

]

[F (V/cm)]1/2

Figure 4.2 Current density (J) as a function of average electric field (F = V/d) for ITO/α-NPD, V2O5/α-NPD

and Ni2O3/α-NPD plotted as ln(J/F2 ) vs. F

1/2. The organic layer thickness (d) in all cases is ~ 500 nm. The

linear fits are used to extract the apparent mobility from the IV characteristics using Eq. (4.3). However, since

a good fit is achieved for all three anodes, despite the significant difference in injection properties (see Fig.

4.1), the goodness of fit cannot be used to distinguish an Ohmic contact. What is more, the extracted mobility

values are significantly different for each anode, and deviates significantly from the value measured by time

of flight (TOF). This serves as an example that transport parameters cannot be extracted from IV

characteristics, without verifying that a true Ohmic contact has been made using another technique.

Also, fitting the IV characteristics with an analytical equation for SCLC is another common

mistake used to distinguish an Ohmic contact, and has also been inappropriately used to extract

transport parameters, such as mobility. However, as discussed above the Mott-Gurney law given

by Eq. (4.1) is not applicable to organic semiconductors, since most have field dependent

mobility. Moreover, even the form of Eq. (4.3), which considers the field dependent mobility, is

only a necessary but not a sufficient condition for SCLC. In other words, a good mathematically

fit to the data using Eq. (4.3) does not guarantee that SCLC has been achieved. As a result,


38

transport parameters, such as mobility, extracted from merely fitting the IV characteristics are

physically meaningless. A quantitative example to emphasize this point is shown in Fig. 4.2.

Figure 4.2 shows the IV characteristics for Ni2O3 (group A), V2O5 (group B) and ITO along

with the fitting results for SCLC using Eq. (4.3) following the same method as in Ref. 55. The

data are linearized by plotting the IV characteristics in the form ln(J/F2

) versus F1/2

. For all three

oxides an excellent linear fit is achieved, despite the significant difference in injection properties

(see Fig. 4.1). Clearly, the goodness of mathematical fit cannot be used to distinguish SCLC.

This point is further illustrated by the extracted mobility values, which are ~10-7

cm2/Vs, ~10

-6

cm2/Vs and ~10

-5 cm

2/Vs extracted from ITO, V2O5 and Ni2O3 device respectively. Clearly

theses values are all incorrect, since a different value is obtained for the same organic

semiconductor, and all of which are orders of magnitude lower than the value measured by

TOF.79

Since the experiments were all conducted under the same well controlled conditions, the

possible system-to-system variation is eliminated. Furthermore, the thickness of the devices

shown in Fig. 4.1 is almost the same, ~ 500 nm. As a result, the difference in injection current is

not due to the apparent thickness dependent mobility proposed by other works.54,55

In any event,

the results of Fig. 4.2 clearly demonstrate that a simple fit of the IV characteristics is insufficient

to claim an Ohmic contact (SCLC), and furthermore any transport parameters extracted using

this technique are in general incorrect. As discussed in the Theory section, the reason behind

this incorrect analysis is the inappropriate assumption that the electric field at the

charge-injecting contact is equals to zero, i.e. ( 0) 0F x . All models for SCLC require that

( 0) 0F x as a boundary condition. However, as will be shown, for all of the examples in

Fig. 4.2 the value of the electric field at the charge-injecting contact is much greater than zero.

Clearly, the IV characteristics alone cannot be used to determine which models can be applied

(e.g., SCLC, quasi-Ohmic or ILC).

4.3.3. Built-in potential and device thickness dependence

Another common criterion used to identify SCLC is the thickness dependence of the IV

characteristics. For ILC the voltage V is proportional to the device thickness d (i.e., V d ) at a


39

fixed current density. While on the other hand, if the injection is bulk limited (SCLC), the

thickness dependence becomes (1 1.5)xV d x .80

However, without knowledge of the exact

value of the built-in potential it is difficult to distinguish between these two cases since the

built-in potential also introduces an additional thickness dependence. This point is illustrated in

Fig. 4.3, for Ni2O3 single carrier hole-only devices with different organic layer thickness.

0.01 0.1

10-7

10-6

10-5

10-4

10-3

10-2

10-1

0.01 0.110

-8

10-7

10-6

10-5

10-4

10-3

10-2

(b)

d = 625 nm

d = 443 nm

d = 278 nm

d = 98 nm

Cu

rre

nt

De

nsity (

A/c

m2)

(a)

1 1010

-5

10-4

10-3

10-2

10-1

Cu

rre

nt

De

nsity (

A/c

m2)

Voltage (V)

d = 625 nm

d = 443 nm

d = 278 nm

d = 98 nm


Figure 4.3 Current density (J) as a function of average electric field (F = V/d) for Ni2O3/α-NPD/Ag with

different organic layer thicknesses (d) of α-NPD, of which (a) is before the subtraction of built-in potential

(Vbi) and (b) is after subtraction of an estimated ~ 0.9 eV built-in potential. The inset of (a) is the current

density as a function of voltage in the case of (a). Ag was chosen as cathode so as to increase the built-in

potential.


40

Figure 4.3 (a) shows current density as a function of average applied electric field (i.e., F =

V/d); the inset shows current density as a function of voltage. Clearly, the IV characteristics are

different for each organic layer thickness, implying that the voltage is not proportional to the

thickness, and hence the current is not ILC. However, if an estimated built-in potential (Vbi) of ~

0.9 eV is subtracted and Fig. 4.3 (a) can be re-plotted as shown in Fig. 4.3 (b), where the IV

characteristics (i.e., J vs. (V-Vbi)/d) are in good agreement for each organic layer thickness (i.e.,

V d ). This would suggest that the current is in fact ILC, in contrast to the previous analysis.

It is well-known that the built-in potential cannot be simply calculated by the difference of

work function between anode and cathode,62

and it must be measured using other techniques,

such as electroabsorption81

and photovoltaic62

measurements. For example, even a symmetric

device with identical anode and cathode (e.g. Au/α-NPD/Au) does not necessarily have a zero

built-in potential due to difference in the energy-level alignment between organic deposited on

metal, and metal deposited on organic. Regardless, without directly measuring the value of the

built-in potential the thickness dependence of the IV characteristics is insufficient to claim an

Ohmic contact. It is also noted that even experimental determination of the exact value of the

built-in potential in organic devices remains controversial.

Alternatively, one strategy to reduce the influence of the built-in potential is to increase the

layer thickness of the organic, so as to increase the applied bias for a given electric field

strength. In this work all of the single carrier devices were fabricated with a relatively thick (~

500 nm) organic layer. As we have previously shown29

this thickness is sufficient for α-NPD to

negate any influence of the built-in potential at high electric field (the region where fitting is

performed).

4.3.4. Criterion for SCLC, quasi-Ohmic and ILC

As demonstrated in the previous sections, simple analytical equations cannot be directly

applied to describe the IV characteristics of organic semiconductors without careful

consideration. As of yet it is unclear what regime (i.e., SCLC, quasi-Ohmic or ILC) the data

shown in Fig. 4.1 falls into for each of the studied oxides. As discussed above the difficulty

arises from the treatment of the quasi-Ohmic regime in between SCLC and ILC. We will


41

therefore begin by defining the boundaries of the quasi-Ohmic regime (in terms of barrier

height, device thickness and applied voltage). In most cases, the quasi-Ohmic region is the most

elusive (and most mistreated) region as one cannot use simple analytical equations for either the

bulk limited current (SCLC) or ILC. For example, one cannot use the Mott-Gurney given by Eq.

(4.1) or Eq. (4.3) to analyze the IV characteristics since the electric field at the charge-injecting

contact does not equal to zero [i.e., the Mott-Gurney law requires that ( 0) 0F x ]. Also, the

data cannot be analyzed using the ILC models in this region either, since /F V d at the

interface, and the injection current has to be treated as one of the boundary conditions. A time

domain simulation that takes into account the dynamic nature of charge injection and transport

is thus needed.

1 10

10-4

10-3

10-2

10-1

100

101

Group B

ITO

ILC

quasi-Ohmic

Cu

rre

nt D

en

sity (

A/c

m2)

Voltage (V)

Simulation (0.55 eV)



ILC (0.55 ev)

SCLCGroup A

Figure 4.4 Current density (J) as a function of voltage (V) for different injection barrier heights ( Bp ). The

solid symbols correspond to the time-domain simulation results. The solid line is the calculated SCLC by Eq.

(4.3) using the field dependent mobility measured by the TOF technique and reported previously in Ref. 79.

The dashed line is the injection limited current (ILC) calculated using Eq. (4.4). The current density at 10 V

(i.e., F = V/d = 0.2 MV/cm) for group A oxides, group B oxides and ITO (see Fig. 4.1) is also shown as solid

star symbols for comparison.


42

Figure 4.4 shows the simulated current density (solid symbols) as a function of applied

voltage for a 500 nm thick device. The dashed line is the ILC calculated using Eq. (4.4). The

solid line is the calculated SCLC [from Eq. (4.2)] which defines a perfect Ohmic contact. As

shown in the figure, the upper boundary of the simulated quasi-Ohmic regime with a 0.25 eV

barrier height (solid triangles) converges with the Ohmic SCLC (black solid line). The lower

boundary (~ 0.55 eV) of the simulated current is taken as the convergence of the ILC (dashed

line) from Eq. (4.4) and the simulation results (solid squares). When the barrier height is larger

than ~ 0.55 eV, the current density becomes strictly ILC, i.e. the bulk can support all of the

injected charge, and hence the simulated current density and calculated ILC from Eq. (4.4)

(dashed line and solid squares, respectively) are in excellent agreement. Therefore models for

ILC can be directly applied to analyze the IV characteristics. It is clear that the quasi-Ohmic

regime covers a significant range of current densities, and in fact includes both the group A and

B oxides. The experimental results at an applied bias of 10 V (F = V/d = 0.2 MV/cm) for the

group A oxides, group B oxides and ITO are also indicated in the figure. Clearly, ITO is the

only example that yields ILC under these conditions.

As discussed above, the boundaries of the quasi-Ohmic regime were taken as the

convergence of the simulation results with the SCLC from Eq. (4.3) and the ILC from Eq. (4.4).

These boundaries represent the strict limits for SCLC and ILC in terms of the electric field at the

charge-injecting contact [i.e., ( 0) 0F x for SCLC and ( 0) /F x V d for ILC] as

discussed in the Theory section. Figure 4.5 shows the calculated electric field at the

charge-injecting contact (i.e., anode/α-NPD interface) as a function of the injection barrier

height for two α-NPD layer thicknesses (same average electric field F = V/d) . As shown the

figure, the interfacial electric field depends strongly on the barrier height in the quasi-Ohmic

regime. The interfacial electric field approaches zero as the barrier height is reduced to below ~

0.25 eV, as expected. For high barrier > 0.55 eV, the interfacial electric field reaches the

average value (i.e., F = V/d) indicating the ILC regime. Figure 4.5 also shows that the transition

amongst Ohmic, quasi-Ohmic, and ILC regimes depends on the device thickness. It is therefore

critical to evaluate the effect of device thickness for various barrier heights.


43

0.2 0.3 0.4 0.5 0.6 0.70.0

0.1

0.2

0.3

0.4

0.5

d = 100 nm

d = 1000 nm

F(x

= 0

) (M

V/c

m)

Injection Barrier (eV)

F = V/d = 0.5 MV/cm

Figure 4.5 The calculated electric field at the charger-injecting contact as a function of barrier height ( Bp )

for α-NPD with an organic layer thickness (d) of 100 nm and 1000 nm respectively. The average electric field

(F = V/d) for each case is 0.5 MV/cm. Notice that the electric field at the interface converges to the average

value for increasing barrier height, and tends to zero for decreasing barrier height; the region in between

defines a quasi-Ohmic contact.

Figure 4.6 is the calculated thickness-barrier height “phase” diagram for α-NPD devices at

two nominal applied electric fields (i.e., F = V/d). Three regions can be defined on the figure,

i.e. SCLC, quasi-Ohmic and ILC. The solid symbols correspond to an average electric field of

0.5 MV/cm (the typical working electric field for OLEDs) while the open symbols to an average

electric field of 0.1 MV/cm. Surprisingly, the quasi-Ohmic regime covers a significant portion

of the phase diagram, and in fact encompasses the typical working range of injection barrier

heights in real devices, such as OLEDs. It is also important to note that the boundaries of the

quasi-Ohmic regime are strongly dependent on the average electric field (i.e., applied bias) as

well. This is not surprising since the mobility and charge injection are dependent on electric

field. As shown in Fig. 4.6 when the electric field increases from 0.1 to 0.5 MV/cm, the

boundaries of the quasi-Ohmic regime significantly expand.


44

0.1 0.2 0.3 0.4 0.5 0.6 0.70

200

400

600

800

1000

B ITOA

quasi-OhmicSCLC

Film

Thic

kn

ess (

nm

)

Injection Barrier (eV)

ILC

-NPD

Figure 4.6 The calculated injection “phase” diagram for α-NPD indicating the boundaries of the quasi-Ohmic

regime (i.e., the criterion for ILC and SCLC) as a function of the injection barrier height ( ) and the organic

layer thickness (d). The first region (left side) defines the criterion for an Ohmic contact (SCLC), the second

region (middle) is for a quasi-Ohmic contact and the third region (right side) is for an injection limited contact

(ILC). The boundaries for these regimes are dependent on applied bias. The solid symbols correspond to an

average electric field of 0.5 MV/cm while the open symbols to an electric field of 0.1 MV/cm. The results for

the group A and group B oxides as well as ITO are also shown for comparison and are obtained from the

experimental results shown in Fig. 4.1.

Since both the group A and B oxides fall into the quasi-Ohmic regime (for the device

thickness and range of applied bias considered in this work) the time domain simulation must be

used to extract the injection barrier height (or mobility) by fitting the IV characteristics. The

barrier heights for the group A and group B oxides as well as ITO are indicated in Fig. 4.6.

Using this method the injection barrier height is estimated to be ~ 0.42 eV for the group A

oxides (such as Ni2O3), ~0.50 eV for group B oxide (such as WO3). On the other hand, the

injection from ITO into α-NPD is obviously injection limited; the barrier is estimated to be ~

0.56 eV using Eq. (4.4). These values agree well with values independently extracted from UPS

Bp


45

measurements, the details of which will be discussed elsewhere. Here it is also noted that the

exact values of the barrier height may vary slightly for different injection models used as the

boundary condition in the simulation. Also, although the time domain simulation can describe

the IV characteristics across all three regimes (i.e., SCLC, quasi-Ohmic and ILC), simple

analytic equations are preferable and convenient to describe SCLC and ILC due to the

computational complexity of the time domain simulation.

4.4 Summary

In summary, the hole injection from different metal oxides into α-NPD has been

systematically studied in single carrier hole-only devices yielding a IV database for variable

injection barrier heights. The device performance data was found to aggregate into three distinct

groups: A) CuO, Co3O4, Ni2O3, MoO3; B) V2O5, WO3; and C) ITO, Cu2O. Based on the

experimental results several key finds have been made.

First, it was found that none of the metal oxides studied in this work form a true Ohmic

contact (SCLC) to α-NPD (over the practical range of applied bias used in devices), despite

previous reports to the contrary. This discrepancy was attributed to incorrect data analysis in

previous studies as a result of merely applying simply analytical equations for SCLC (e.g.,

Mott-Gurney) to evaluate the IV characteristics. Without prior knowledge of the field-dependent

mobility, IV characteristics cannot be used to identify an Ohmic contact. A recent study also

confirms the finding in this chapter.82

Second, it was found that there was no clear boundary between SCLC and ILC conditions,

but rather a large intermediate regime, namely quasi-Ohmic, which includes characteristics of

both. As a result, the IV characteristics in the quasi-Ohmic regime cannot be simply analyzed by

either the bulk transport models (e.g., Mott-Gurney law) or the injection models. The boundaries

of the quasi-Ohmic regime were defined by the electric field at the electrode contact (i.e.,

0 ( 0) /F x V d ).

Third, it was found that the quasi-Ohmic regime is surprisingly large and covers a wide

range of barrier heights. Using a time-domain simulation of the transport of charge carriers


46

across an organic semiconductor the boundaries of the quasi-Ohmic regime were evaluated as a

criterion to distinguish SCLC from quasi-Ohmic and ILC. It was found that the IV

characteristics for most electrode/organic contacts fall into the quasi-Ohmic regime. It was also

found that the boundaries of the quasi-Ohmic regime have a strong dependence on the thickness

of the organic layer and the applied bias.

Fourth, it was found that the built-in potential can significantly distort the thickness

dependence of the IV characteristics, particularly for organic layer thicknesses < 100 nm.

Without measuring the exact value of the built-in potential, a thicker organic layer of > 500 nm

is required to minimize the effects of any built-in potential on the IV characteristics.

Finally, an injection “phase diagram” for α-NPD has been shown as a case study to clearly

demonstrate the above mentioned effects. The group A and group B oxides discussed above

were found to fall within the quasi-Ohmic regime, while ITO was found to be purely injection

limited. Using the time domain simulation the injection barrier height for the various oxides has

been deduced to be in the range of ~ 0.4 eV for group A oxides, ~ 0.5 eV for group B oxides,

and ~ 0.6 eV for ITO. For other organic semiconductors with different field-dependent mobility

a new phase diagram should be calculated.

Chapter 5 Organic/organic interface designs of OLEDs

47


In this chapter, it will be shown that the accumulated space charges at the exciton formation

interface have a negative impact on the device performance due to the exciton quenching. It will

be demonstrated that using a hole transport layer with very large HOMO level such as CBP can

reduce the exciton quenching and thus increase the device performance. Not only the device

efficiency of both fluorescent and phosphorescent OLEDs can be enhanced by this new device

design concept, the device structure is also highly simplified because no additional “injection

layer” and “blocking layer” are needed. The content of this chapter was published as Appl. Phys.

Lett. 96, 043303 (2010), J. Appl. Phys. 108, 024510 (2010) and Appl. Phys. Lett. 98, 073310

(2011).

5.1. Introduction

In the previous two chapters, the energy level alignment and charge carrier injection at

metal/organic interfaces were systematically studied. Although charge carrier injection into

organic semiconductors is of great significance for OLED, in an actual OLED, the carrier

transport through organic/organic interfaces is also critical to the device performance.

Despite significant advances in device performance since the first OLED was reported in

the 1980s,2 much of the device physics of even the simplest OLED structures is still not well

understood.13,49

For example, hole injection layers (HILs), such as m-MTDATA,83,84

have

traditionally been thought to provide an intermediate energy-level to facilitate hole injection

from the ITO anode to the hole transport layer (HTL), typically α-NPD. However, it has

recently been demonstrated that the most significant role of m-MTDATA in Alq3 OLEDs is to

reduce exciton quenching by limiting accumulated holes at the α-NPD/Alq3 interface; space

charges, such as α-NPD+ radical cations, accumulated near the emission layer (EML) are known

to quench excitons.83

This recent finding suggests that exciton quenching from accumulated

space charges may play a far more significant role in device performance than has previously

been thought. Eliminating exciton quenching from accumulated space charges remains a


48

significant challenge in existing device designs. For example, in the case of α-NPD/Alq3

OLEDs, the offset between the HOMOs creates an injection barrier that will always tend to

accumulate holes, regardless of what HILs are used. Thus, exciton quenching from α-NPD+

radical cations cannot be eliminated in traditional device designs. In this chapter a simple

method will be shown to control exciton quenching due to accumulated radical cations at the

exciton formation interface. Eventually, a highly simplified device structure that can enable

exceptionally high efficiency is demonstrated for both fluorescent and phosphorescent OLEDs.

5.2. CBP interlayer to reduce exciton quenching

The OLED structure is: ITO/CuPc (25 nm)/α-NPD (45 - x nm)/interlayer (x = 0, 3, 10

nm)/Alq3 (45 nm)/LiF (1 nm)/Al (100 nm). The two interlayers used were CBP and TPBi. To

reduce charge accumulation at the α-NPD/Alq3 interface, the injection barrier at the interface

should be eliminated, for example by inserting different interlayers between α-NPD and Alq3.

Figure 5.1(a) shows the current efficiency and power efficiency as a function of luminance of

OLEDs with and without a thin (~ 3 nm) CBP and TPBi interlayer inserted between α-NPD and

Alq3. The efficiency of the TPBi device is significantly lower than the reference device.

However, the current efficiency and power efficiency are increased by ~ 20 % in the CBP

device, despite similar HOMO energy-levels to TPBi. What is more, the maximum current

efficiency of the CBP device is 5.2 cd/A, which is amongst the highest device performance for

Alq3 based OLEDs with n-type or p-type doping.19,85,86

Figure 5.1(b) shows the IV characteristics of the same OLEDs. The CBP interlayer has no

impact on the IV characteristics, contrary to traditional wisdom, which suggests CBP should act

as a hole blocking layer since it has a much deeper HOMO than Alq3. For the TPBi interlayer,

however, the driving voltage is increased by nearly 0.8 V at a given current density of 100

mA/cm2. This finding suggests that the TPBi interlayer acts as a traditional hole blocking layer,

which is interesting given that the HOMO energy-levels of CBP and TPBi are similar. If TPBi

indeed acts as a hole blocking layer, a shift in the emission zone towards the α-NPD/TPBi

interface and hence blue emission from α-NPD in the EL spectra should be expected.


49

102

103

104

0

1

2

3

4

5

0 2 4 6 8 100

50

100

150

200

(b)

(I) -NPD/ Alq3

(II) -NPD/ CBP (3nm)/ Alq3

(III) -NPD/ TPBi (3nm)/ Alq3

Pow

er

Effic

iency (

lm/W

)

Cu

rren

t E

ffic

ien

cy (

cd/A

)

Luminance (cd/m2)

(a)

0

1

2

3

4

5

6

Cu

rren

t D

ensity (

mA

/cm

2)

Voltage (V)

Figure 5.1 Efficiency and (b) IV characteristics of the OLED devices with the following structures: (I)

α-NPD/Alq3 (standard reference); (II) α-NPD/CBP (3nm)/Alq3 and (III) α-NPD/TPBi (3nm)/Alq3.

Figure 5.2 shows the EL spectra of OLEDs with and without a thin (3 nm) and thick (10

nm) CBP and TPBi interlayer. For the TPBi interlayer blue emission from α-NPD is observed in

the EL spectrum (~ 430 nm),87,88

which explains the lower device efficiency in Fig. 5.1 (i.e., due

to excitons loss to the α-NPD). For a 10 nm thick TPBi interlayer the EL spectrum is dominated

by emission from α-NPD. This finding suggests that TPBi acts as a traditional hole-blocking

layer, shifting the emission zone towards the α-NPD/TPBi interface. A similar effect has

previously been reported for 2,9-dimethyl-4,7-diphenyl-phenanthroline (BCP) interlayers.89-91

However, the EL spectrum of the α-NPD/CBP/Alq3 device is identical to the standard reference


50

device (i.e., EL emission only from Alq3), indicating that the emission zone is unaffected. Even

for a CBP interlayer thickness of 10 nm only EL emission from Alq3 is observed. This finding

contradicts the exciton dissociation theory, proposed by Song et al.91

to explain the role of wide

band gap interlayers at the α-NPD/Alq3 interface. The theory predicts that any wide band gap

interlayer inserted at the α-NPD/Alq3 interface should result in blue emission from α-NPD.

Clearly, this is not the case since no blue emission from α-NPD was observed, even for a thick

CBP interlayer.

400 500 600 700 800

0.0

0.2

0.4

0.6

0.8

1.0 (a) -NPD/Alq

3

(b) -NPD/CBP (3nm)/Alq3

(c) -NPD/CBP (10nm)/Alq3

(d) -NPD/TPBi (3nm)/Alq3

(e) -NPD/TPBi (10nm)/Alq3

Inte

nsity (

a.u

.)

Wavelength (nm)

-NPD

Figure 5.2 Normalized EL spectra of the OLED devices with different thickness (0, 3, 10 nm) interlayer of

CBP and TPBi.

To elucidate the effect of CBP and TPBi on the hole accumulation at the various

organic/organic interfaces UPS measurements were employed to study the injections barrier.

Figure 5.3 shows the He Iα (hν = 21.22 eV) valence band spectra for the in situ subsequent

deposition of: (a) CuPc/α-NPD/Alq3; (b) α-NPD/CBP/Alq3; and (c) α-NPD/TPBi/Alq3. From the

UPS spectra, the measured HOMOs of the various molecules are: 4.8 eV (CuPc), 5.4 eV

(α-NPD), 5.75 eV (Alq3), 6.1 eV (CBP), 6.2 eV (TPBi). Figure 5.4 shows the energy-level

diagrams for the different device structures deduced from the UPS measurements; the injection

barriers and interfacial dipoles are summarized in Table I. In the reference device [see Fig.


51

5.4(a)] holes are accumulated at the CuPc/α-NPD (0.6 eV) and α-NPD/Alq3 (0.55 eV) interfaces

due to the hole injection barriers. When a thin CBP or TPBi interlayer is inserted between

α-NPD and Alq3, the hole injection barrier, and hence the accumulation of holes, is shifted from

the α-NPD/Alq3 interface to the α-NPD/CBP or α-NPD/TPBi interface, as shown in Fig. 5.4(b)

and (c) respectively. As mentioned previously, accumulated α-NPD+ radical cations at the

α-NPD/Alq3 interface will quench excitons. Since the HOMO of CBP is deeper than that of

Alq3, holes cannot accumulate at the CBP/Alq3 interface, thus reducing α-NPD+ radical cation

quenching.

CuPc

-NPD

Alq3

-NPD

CBP

Alq3

Inte

nsity (

a.u

.)

3 2 1 0

-NPD

TPBi

Alq3

Binding Energy (eV)

Figure 5.3 He Iα (hν = 21.22 eV) valence band spectra for: (a) CuPc/α-NPD/Alq3; (b) α-NPD/CBP/Alq3; and

(c) α-NPD/TPBi/Alq3. In (b) and (c) the CuPc/α-NPD interface is not shown for clarity since it is identical to

(a).


52

Table 5-1 Hole injection barrier heights (Bp ) and interfacial dipoles ( ) at different organic/organic

interfaces extracted from the UPS spectra.

CuPc/

α-NPD

α-NPD/

Alq3

α-NPD/

CBP

α-NPD/

TPBi

CBP/

Alq3

TPBi/

Alq3

Bp (eV) 0.60 0.55 0.70 1.00 -

-

(eV) 0 0.20 0 0.20 0.20 0

Based on the UPS measurements the hole injection barrier for TPBi (1.0 eV) is larger than

for CBP (0.7 eV), due to a combination of its relatively deep HOMO and the interfacial dipole

at the α-NPD/TPBi interface (see Table I). Therefore, although the TPBi interlayer is very thin

(~ 3 nm), the voltage drop across the interlayer is not negligible, which results in a higher

driving voltage [Fig. 5.1.(b)] . Also, the interfacial dipole at the α-NPD/TPBi interface shifts

down the LUMO level of TPBi [see Fig. 5.4 (c)] resulting in a more favorable LUMO

energy-level alignment for electron injection from Alq3 to TPBi. The high injection barrier for

holes, combined with the easier injection pathway for electrons into α-NPD results in formation

of exciton in and emission from α-NPD and thus explains the EL spectra in Fig. 5.2.

Arguably the higher injection barrier for TPBi than for CBP could account for the

difference in device performance shown in Fig. 5.1 (i.e., TPBi blocks more holes), as commonly

believed in literatures. However, this simple explanation is problematic. For example, the

barrier at the α-NPD/CBP interface is still larger than that at the α-NPD/Alq3, but no increase in

driving voltage or blue emission from α-NPD was observed. Since excitons will tend to form at

the interface between an HTL and electron transport layer (ETL),2 the impact of the carrier

transport characteristics of the interlayer and the location of excitons must also be considered.

An ETL interlayer (e.g., TPBi or BCP) will tend to shift the exciton formation zone towards the

α-NPD/interlayer interface, while an HTL interlayer (e.g., CBP) will shift the exciton formation

zone towards the interlayer/Alq3 interface. Since HTL molecules are typically donors and ETL

molecules acceptors, a natural donor-acceptor interface will tend to form at the exciton

formation zone (HTL/ETL interface). Indeed, the interfacial dipole measured at the various

organic/organic interfaces (indicative of a donor-acceptor interface)44

coincide with the


53

HTL/ETL interfaces (see Fig. 5.4), and are consistent with the position of the emission zone as

indicated by the EL spectra (see Fig. 5.2).

-NPD 3AlqCuPc -NPD 3

AlqCuPc CBP

-NPD 3AlqCuPc TPBi

electron accumulation

hole accumulation

interfacial dipole

(a) (b)

(c)

excition formation

Vacuum

level

Vacuum

level

Vacuum

level

LUMO

HOMO

LUMO

HOMO

LUMO

HOMO

Figure 5.4 Schematic energy diagram for the device structure: (a) CuPc/α-NPD/Alq3; (b) CuPc / α-NPD / CBP

/ Alq3; and (c) CuPc/α-NPD/TPBi/Alq3. The LUMOs are estimated from cyclic voltammetry measurements.92

Clearly the traditional notion of confining holes and electrons at the emission interface is

not an optimal device design concept. By controlling the energy-level alignment at the

HTL/EML interface, radical cation accumulation and the associated exciton quenching can be

prevented. To demonstrate the universality of this approach, singlet-emitter doped OLEDs with

fluorescent dye molecule, C545T were also fabricated, with a structure: ITO/CuPc (25

nm)/α-NPD (45 nm)/CBP(0, 3 nm)/Alq3:C545T (1 wt %, 30 nm)/Alq3 (15 nm)/LiF (1 nm)/Al

(100 nm). A consistent ~ 25% improvement at high luminance was achieved with the thin CBP

interlayer without sacrificing driving voltage, similar to the pure Alq3 devices discussed above.


54

The maximum current efficiency was increased from 15 cd/A to 19 cd/A at high luminance

(~104 cd/m

2).

5.3. Deep HOMO HTL: enable simple structure with high efficiency

Although the CBP interlayer prevents the accumulation of holes at the CBP/Alq3 interface

due to its deeper HOMO (6.1 eV) than that of Alq3 (5.75 eV), which results in an improved

device performance. However, using such an interlayer introduces additional complexity into

the device structure. Also it does not completely eliminate exciton quenching from radical

cations due to the numerous other organic/organic interfaces for holes to accumulate at (e.g., the

α-NPD/CBP interface). In this section, a simple method will be demonstrated to eliminate

exciton quenching due to accumulated radical cations by using deep HOMO HTL,CBP, as a

direct drop-in replacement of the most commonly used HTL, α-NPD. The fluorescent OLEDs

were all doped with fluorescent dye molecule C545T (1 wt.%) that gives a higher internal

quantum efficiency than the un-doped devices, which also aims at showing the universality of

this approach.

(a)

Glass

ITO

CuPc (25 nm)

α-NPD (45 nm)

Alq3 :C545T(1 wt.%, 30 nm)

Alq3 (15 nm)

LiF(1nm)/Al (100 nm)

(b)

Glass

ITO/Oxides (1 nm)

CBP (50 nm)

Alq3 :C545T(1 wt.%, 30 nm)

Alq3 (15 nm)

LiF(1nm)/Al (100 nm)

Figure 5.5 Device structure of (a) standard reference device, and (b) device with non-blocking exciton

formation zone.


55

Since the HOMO of CBP (6.1 eV) is much deeper than that of α-NPD (5.4 eV),93

direct

hole injection from the ITO anode into the deep HOMO of CBP requires an anode buffer layer

with high work function, such as a transition metal oxide (see Chapter 3). Figure 5.5 shows the

device structure of a non-blocking exciton formation zone with CBP as HTL in comparison to a

standard reference device. For simplicity we will refer to the device with a non-blocking exciton

formation zone (Fig. 5.5 b) as the CBP device. As we will show that removing the hole blocking

barrier in the traditional exciton formation zone design eliminates exciton quenching due to

accumulated radical cations, resulting in significantly improved device performance.

2 4 6 8 10 1210

-1

100

101

102

103

104

105

101

102

103

104

105

4

8

12

16

20

(b)

ITO / CuPc / a-NPD

ITO / WO3 / CBP

ITO / CBP

Lum

inance (

cd/m

2)

Voltage (V)

(a)0

50

100

150

200

250

300

Curr

ent D

ensity (

mA

/m2)

Curr

ent E

ffic

iency (

cd/A

)

Luminance (cd/m2)

0

4

8

12

16

20

24

28

Pow

er

Effic

iency (

lm/W

)

Figure 5.6 (a) Luminance-Voltage and Current-Voltage characteristics and (b) efficiency of the OLED devices

with the following structures: ITO/CuPc/α-NPD (square); ITO/WO3/CBP (circle); ITO/CBP (triangle).

Fig. 5.6 shows the device performance using CBP as HTL with and without a thin interlayer

(1 nm) of WO3, in comparison with a standard reference device using α-NPD as the HTL with a

25 nm thick CuPc HIL. Without the WO3 interlayer the driving voltage of the CBP device is


56

extremely high (and the efficiency very low) due to the poor hole injection from ITO into the

deep HOMO of CBP (6.1 eV).93

WO3 has previously been shown to improve hole injection into

other molecules with deep HOMOs, such as 4,4′,4″-tris(N-carbazolyl)-triphenylamine (TCTA)

with a HOMO of 5.9 eV.94

Remarkably, the driving voltage of the CBP device with WO3

interlayer is nearly identical to the standard reference device, indicating that WO3 might also

form a good injection contact with CBP. Moreover, the luminance of the WO3/CBP based

device is higher than the CuPc/α-NPD reference device at any given voltage, due to the

elimination of exciton quenching caused by radical cation accumulation near the emissive layer

(EML). As a result, the current efficiency of the WO3/CBP device is much higher than the ~ 15

cd/A for the CuPc/α-NPD based reference device, reaching ~ 20 cd/A, even at high luminance

(>105 cd/m

2). More importantly, the power efficiency is also dramatically increased for the

WO3/CBP device, which confirms that the improvement is due to the higher quantum

efficiency. This is the highest efficiencies reported for C545T doped bottom emitting OLED

with Alq3 as host.9 It is important to note that this significant improvement is achieved with a

simplified device structure (two layers) and without the use of an optical microcavity (e.g.,

using metal anode or top emission).

Clearly, from Fig. 5.6 in the device with CBP as HTL, the WO3 interlayer is necessary to

reduce the driving voltage due to the deep HOMO of CBP. Transition metal oxides HILs have

been extensively studied with α-NPD as HTL,50,95,96

but not for molecules with much deeper

HOMOs, such as CBP. To elucidate the low driving voltage of the WO3/CBP device the

energy-level alignment at the anode interface was measured using UPS. Fig. 5.7 shows the He

Iα (hν = 21.22 eV) valence band spectra and secondary electron cut-off of UV ozone treated

ITO and ITO/WO3 with and without a 3 nm layer of CBP. The measured work function of the

ITO and ITO/WO3 substrates were 5.5 eV and 5.8 eV, respectively. From the UPS spectra the

hole injection barrier are 0.69 eV for ITO and 0.52 eV for ITO/WO3, consistent with the work

function of the substrates (i.e., higher work function yields a lower barrier). Although the

injection barrier for ITO/WO3 is only ~ 0.2 eV lower than for ITO, the exponential dependence

of the injection current on the barrier height, makes even a ~ 0.2 eV difference in barrier height

significant in terms of device performance. Clearly, WO3 provides favorable energy-level

alignment for hole injection into CBP, despite its deep HOMO. Based on our previous work


57

with α-NPD,96

other transition metal oxides, such as V2O5 and MoO3, should also have good

injection into CBP. Indeed, as shown in Fig. 5.7 the barrier heights measured using UPS for

ITO/MoO3 and ITO/V2O5 are 0.50 and 0.54 eV respectively, very close to the value for

ITO/WO3. This suggests that the V2O5 and MoO3 should also work in an actual OLED device.

4 2 0

ITO/CBP

ITO

WO3

100

100

WO3

Binding Energy (eV)

ITO

HOMO

Anode/CBP

16 15

(c)(b)

No

rmaliz

ed Inte

nsity (

a.u

.)(a)

2 1 0

WO3/CBP

~ 0.2 eV

Figure 5.7 He Iα (hν = 21.22 eV) valence band spectra of ITO and ITO/WO3 with a 3 nm thick layer of CBP

showing (a) the secondary electron cut-off, (b) the valence band, and (c) the HOMO of CBP.

Figure 5.8 shows the device performance using CBP as HTL with and without a thin

interlayer (1 nm) of WO3, V2O5 and MoO3. Clearly, both the current efficiency and power

efficiency are nearly identical for the three different oxide interlayers. This demonstrates that the

significant improvement in efficiency for the device with CBP as HTL is not due to some

unique properties of the WO3 interlayer but due to the CBP (i.e., non-blocking exciton

formation zone). To further confirm this, another series of devices with the CBP replaced by

α-NPD with exactly the same thickness (50 nm) was also fabricated. Clearly all of the α-NPD

based devices (see Figure 5.8) have consistently lower efficiencies than the CBP devices. It is

well known that holes are already the dominant carrier in the α-NPD/Alq3 device structure. The

device performance is therefore not limited by the injection of holes, but rather by exciton

quenching due to accumulated excess holes at the α-NPD/Alq3 interface.83

Hence, improving the

injection of holes into α-NPD (using an oxide interlayer) should not improve the device


58

performance. In fact, all of the α-NPD devices perform worse than the CuPc/α-NPD reference

device (shown again in Figure 5.8 for clarity). The additional barrier at the CuPc/α-NPD

interface has been shown to reduce exciton quenching (and hence boost efficiency) by limiting

the number of excess holes that tend to accumulate at the α-NPD/Alq3 interface.83

Also, the

electroluminescence spectra of the various devices (see Figure 5.9) are all identical, which

suggests that the emission zone is unaffected by the CBP layer. These results provide strong

evidence that the significant improvement in device performance is due to the CBP layer (i.e.,

non-blocking exciton formation zone), and not due to a particular oxide interlayer or optical

effect.

8

12

16

20

24

100

101

102

103

104

105

8

12

16

20

24

(b)

Cu

rrent D

enstiy (

cd/A

)

(a)

ITO / MoO3 / CBP

ITO / V2O

5 / CBP

ITO / WO3 / CBP

ITO / MoO3 / -NPD

ITO / V2O

5 / -NPD

ITO / WO3 / -NPD

ITO / CuPc / -NPD

Pow

er

Effic

iency (

lm/W

)

Luminace (cd/m2)

Figure 5.8 (a) Current efficiency and (b) power efficiency of the OLED devices with the following structures:

ITO/MoO3, V2O5, WO3 (1 nm)/CBP(50 nm); ITO/MoO3, V2O5, WO3 (1 nm) /α-NPD(50 nm) and ITO/CuPc

(25 nm)/α-NPD(45 nm).


59

400 500 600 700

0.0

0.2

0.4

0.6

0.8

1.0

Wavelength (nm)

ITO / MoO3 / CBP

ITO / V2O

5 / CBP

ITO / WO3 / CBP

ITO / MoO3 / -NPD

ITO / V2O

5 / -NPD

ITO / WO3 / -NPD

ITO / CuPc / -NPDE

L (

a.u

.)

Figure 5.9 Electroluminescence (EL) spectra of the devices shown in Fig. 5.9.

ITO

ITO

3

2 5

3

MoO

V O

WO

-

-LiF/Al

LiF/Al

NN

HIL

(CuPc)

HTL

(α-NPD)

ETL

(Alq3)

HTL

(CBP)

ETL

(Alq3)

(a)

(b)

Figure 5.10 Schematic energy level diagram for the device structure: (a) CuPc/α-NPD/Alq3; (b) CBP/Alq3.

The energy offsets were obtained from UPS measurements.


60

Figure 5.10 shows the schematic energy-level diagram for the different device structures

deduced from the UPS measurements. The energy-level alignment for the CuPc/α-NPD

reference device and CBP/Alq3 interface were taken from our previous study.93

As discussed

above, holes will accumulate in the reference device at the CuPc/α-NPD and α-NPD/Alq3

interfaces, resulting in exciton quenching in the Alq3 emissive layer (see Fig. 5.10 a). In the

previous section, a thin (~ 3 nm) interlayer of CBP inserted at the α-NPD/Alq3 interface (i.e.,

α-NPD/CBP/Alq3) is shown to reduce the hole accumulation near the Alq3 layer, but not

eliminate entirely due to the accumulation at the α-NPD/CBP interface.93

However, it has been

shown that by using an oxide interlayer at the anode interface to efficiently inject holes directly

into CBP, the additional CuPc and α-NPD transport layers can be eliminated. As shown in Fig.

5.10 (b), such a device has no hole blocking interfaces. Therefore, exciton quenching due to

accumulated radical cations (i.e., holes) is eliminated, dramatically boosting device efficacy.

These findings suggest that removing the electron blocking barrier in the exciton formation

zone, for example by designing new HTL molecules with deeper LUMO (and HOMO) than that

of the ETL, may provide a further boost in device efficacy.

5.4. Deep HOMO HTL for Phosphorescent OLEDs

Phosphorescent organic light emitting diodes (PHOLEDs) that have the potential to achieve

an internal quantum efficiency close to 100% have attracted considerable research interest.3,97-100

Although a high external quantum efficiency (EQE) has been realized for green PHOLEDs at

low luminance (e.g., 100 cd/m2), it is still a significant challenge to obtain a similarly high

efficiency at high luminance (e.g., 10,000 cd/m2) due to various non-radiative recombination

processes that quench the long lived triplet excitons.100,101

For example, at high exciton densities

the triplet excitons will self-quench through triplet-triplet annihilation, greatly reducing the

efficiency.102-105

At high current densities triplet excitons may also quench with accumulated

polarons (charged molecules) at the various organic heterojunctions in the device.102-104

There

has therefore been significant effort devoted towards reducing the effect of these various

quenching processes at high luminance.99,106-109

However, many of these strategies significantly

increase the device complexity, which is undesirable for manufacturing. As a result, it remains a

challenge to realize a simplified device structure which maintains high efficiency at high


61

luminance.

CBP (3 nm)

CBP (35 nm)

α - NPD (35 nm)

CBP:Ir(ppy)2(acac) (15 nm)

TPBi (65 nm)

LiF/Al (100 nm)

Device B


TPBi (65 nm)

LiF/Al (100 nm)

Device A

Glass substrate

ITO/MoO3

α - NPD (32 nm)


TPBi (65 nm)

LiF/Al (100 nm)

Device C

α-NPD TPBi

Ir(ppy)2(acac)

6.1 eV 6.2 eV

5.6 eV

2.4 eV2.7 eV

3.0 eV

CBP

α-NPD TPBi

Ir(ppy)2(acac)

6.1 eV 6.2 eV

5.6 eV

2.4 eV2.7 eV

3.0 eV

CBP

2.8 eV

5.4 eV

2.8 eV

5.4 eV

TPBi

Ir(ppy)2(acac)

6.1 eV 6.2 eV

5.6 eV

2.7 eV

3.0 eV

CBP

2.8 eV

Glass substrate

ITO/MoO3

Glass substrate

ITO/MoO3

Figure 5.11 Schematic device structures and energy-level diagrams of the devices in this study. The HOMO

and LUMO levels are obtained from Ref. 93,98,110,111.


62

In this section, it will be shown that the approach that has been applied to achieve a simple

structure and high efficiency fluorescent OLED (see Section 5.3) can also be used to enable high

performance in phosphorescent OLEDs. In particular, a simplified bi-layer PHOLED with a

high EQE of >20% over a wide luminance range from 10 cd/m2 to >10,000 cd/m

2 is realized.

The simplified device structure is also shown to have comparable power efficiency to state of

the art p-i-n PHOLEDs.

The detailed device structures are depicted in Fig. 5.11, where the name of different

molecules have been summarized in Table 2-1. Ir(ppy)2(acac) was used as the phosphorescent

emitter and it was purified by gradient sublimation prior to use. The doping concentration of

Ir(ppy)2(acac) in CBP is 8 wt.%. Similarly (see Section 5.3), 1 nm MoO3 was thermally

evaporated on ITO to achieve a high work function for direct hole injection into CBP.

Figure 5.12(a) shows the IV and LV characteristics of the simplified bi-layer OLED

proposed in this work (device A). The EL spectra as a function of current density are shown in

the inset. The IV and LV curves increase rapidly after the onset, indicating efficient carrier

injection and transport in the CBP and TPBi layers. The device exhibits a low operating voltage

of 3.65 V at 100 cd/m2 and 4.55 V at 1,000 cd/m

2. Since there is no change in the EL spectra

with increasing current density, it can be concluded that the triplet excitons are well confined to

the EML.

Figure 5.12(b) shows the EQE and power efficiency of the same PHOLED (device A). The

EQE reaches 23.4% (81 cd/A) at 100 cd/m2, 24.5% (85 cd/A) at 1,000 cd/m

2, and 21.9% (76

cd/A) at 10,000 cd/m2. Even at an ultra-high luminance of 100,000 cd/m

2 the EQE is still as

high as 13% (45 cd/A). Due to the high EQE of our device over a broad range of luminance, the

power efficiency is also quite high and is equivalent to that of state of the art p-i-n

PHOLEDs,99,112,113

reaching 78.0 lm/W at 100 cd/m2, 65.0 lm/W at 1,000 cd/m

2, and 42.8 lm/W

at 10,000 cd/m2. The power efficiency can be further enhanced if the mobility of both the

electron and hole transport layer (e.g. CBP and TPBi herein) is increased.


63

0 2 4 6 810

-5

10-4

10-3

10-2

10-1

100

101

101

102

103

104

0

20

40

60

80

100

Voltage (V)

Cu

rren

t D

ensity (

mA

/cm

2)

100

101

102

103

104

N

Ir

O

O

CH 3

CH 32

Ir(ppy)2(acac)

(b)

Lu

min

an

ce (

cd/m

2)

(a)

Pow

er

Effic

iency (

lm/W

)

Luminance (cd/m2)

0

10

20

30

400 500 600 700 800

0.0

0.2

0.4

0.6

0.8

1.0

EL (

a.u

.)

Wavelength (nm)

0.1 mA/cm2

1 mA/cm2

10 mA/cm2

100 mA/cm2

23.4% @ 100 cd/m2

23.2% @ 5000 cd/m

21.9% @ 10000 cd/m

EQ

E (

%)

Figure 5.12 (a) IV and LV characteristics of device A as well as its (b) EQE and power efficiency as a function

of luminance. The upper inset is the molecular structure of the emitter Ir(ppy)2(acac). The lower inset is the

corresponding EL spectra measured at various current densities.

The high efficiency of our simplified PHOLED over a broad luminance range suggests that

the triplet exciton quenching processes at high luminance can be suppressed. Due to the unique

design of device A there are no energetic barriers in the device for charge carriers to accumulate

at (see Fig. 5.11). Using CBP as both the HTL and host for the phosphorescent emitter

eliminates any barrier at the HTL/EML interface. Furthermore, the energy-levels of CBP and the

TPBi electron transport layer (ETL) are nearly identical, resulting in almost no barrier at the

EML/ETL interface. Suppression of charge carrier accumulation may therefore account for the

low efficiency roll-off of our simplified PHOLED design (device A). To further investigate the


64

physics behind the low efficiency roll-off, PHOLEDs with a traditional device structure using

α-NPD as the HTL was fabricated for comparison (see device B in Fig. 5.11).

2 3 4 5 6 70

10

20

30

40

50

400 500 600 700 800

0.0

0.2

0.4

0.6

0.8

1.0

(b)

Voltage (V)

Curr

ent D

ensity (

mA

/cm

2) Device A

Device B

Device C

(a)

100

101

102

103

104

Lum

inance (

cd/m

2)

Device A

Device B

Device C

EL

(a

.u.)

Wavelength (nm)

Figure 5.13 (a) IV and LV characteristics of device A, B and C as well as (b) the corresponding EL spectra

measured at 5 mA/cm2.

In device B α-NPD is used as the hole transport layer (HTL) and CBP is used as the host for

the phosphorescent emitter. Since the HOMO of α-NPD (5.4 eV) is significantly shallower than

that of CBP (6.1 eV) a significant barrier should exist at the α-NPD/CBP interface. However, the

phosphorescent dopant can contribute to charge transport in the EML,98,114

which would greatly

reduce the barrier at the HTL/EML interface; the HOMO of Ir(ppy)2(acac) (5.6 eV) is very close


65

to that of α-NPD. To isolate the influence of direct charge injection from α-NPD into the

Ir(ppy)2(acac) dopant, a device with a thin un-doped layer of CBP (3 nm) inserted between the

α-NPD layer and the EML (see device C in Fig. 5.11) was also fabricated as a control device.

Figure 5.13(a) shows the IV and LV characteristics of the three devices. The operating

voltage of device C is significantly higher than that of device B, which confirms that the thin

un-doped layer of CBP does indeed limit direct charge injection from α-NPD into the

Ir(ppy)2(acac) dopant (i.e., there is a much larger energetic barrier at the α-NPD/CBP interface

in device C). It is worth noted that this is different from the fluorescent device as discussed in

the Section 5.2 (see Fig. 5.1), where the driving voltage of the device with the 3 nm CBP

interlayer does not change. The major difference is that in the fluorescent device, there is

already a very large hole injection barrier at the α-NPD/Alq3 interface (without CBP interlayer)

and therefore the introduced barrier at α-NPD/CBP interface in the device with CBP interlayer

has little impact on the driving voltage.

Figure 5.13(b) shows the EL spectra at 5 mA/cm2

of the three devices. Although the EL

spectra are nearly identical the efficiency roll-off of the three devices is markedly different (see

Fig. 5.14). Figure 5.14 shows the current efficiency of the three devices as a function of current

density. The EL spectra at different current densities are shown as the inset; the spectra are

normalized to the Ir(ppy)2(acac) peak at ~ 521 nm and have been enlarged by 30 times in the

region of 400-490 nm to highlight the fluorescence from α-NPD. Although the efficiency of the

three devices is similar at low current density (< 0.5 mA/cm2) the efficiency roll-off of devices

with α-NPD as the HTL (device B and C) is markedly worse than device A. Although this trend

might indicate reduced triplet-triplet annihilation and triplet-polaron quenching in device A, it

has been shown that these quenching processes take effect only at much higher current densities

(~1000 mA/cm2).

104

Also, the fluorescence from α-NPD observed from B and C at 50 mA/cm2 (see the inset of

Fig. 5.14) may suggest that the roll-off is due to the exciton “loss” to α-NPD, as the electrons

“leak” through the EML. However, no fluorescence from α-NPD is observed at 5 mA/cm2

where significant efficiency roll-off already takes place in B and C, which rules out exciton

“loss” to α-NPD as the major cause of the efficiency roll-off at high luminance. Recently, it has


66

been shown that loss of charge balance at high current density results in significant efficiency

roll-off.104

The fluorescence from α-NPD supports this hypothesis as it indicates excess

electrons “leaking” through the EML, a surefire indication of electron-hole imbalance. The

origin of this imbalance is most likely due to the blocking of holes at the α-NPD/CBP interface

in device B and C with a large energetic barrier. The efficiency roll-off is therefore less in device

A since there is no barrier between the doped and un-doped CBP regions for holes to be blocked

at. In fact, in the case of device A, the EQE increases from 0.5 mA/cm2 to 2.0 mA/cm

2

suggesting that the charge balance is actually improving over this range. This improvement in

charge balance therefore helps to suppress the efficiency roll-off.

10-3

10-2

10-1

100

101

102

0

20

40

60

80

100

120

140

160

400 440 480

EL

(a

.u.)

Wavelength (nm)

@ 50 mA/cm2

400 440 480

Wavelength (nm)

EL

(a.u

.)

Device A

Device B

Device C

Cu

rre

nt E

ffic

ien

cy (

cd

/A)

Current Density (mA/cm2)

@ 5 mA/cm2

Figure 5.14 Current efficiency of device A, B and C. The insets are the enlarged EL spectra (by 30 times in the

range of 400-490 nm) that are measured at 5 and 50 mA/cm2.

In summary, a simplified high efficiency bi-layer green PHOLED with EQE of > 20% up to

a high luminance of > 10,000 cd/m2 is demonstrated. This result was achieved without using any


67

discrete blocking layers or electrical doping, which greatly simplifies the device design. The

simplified devices are also shown to have comparable power efficiency to state of the art p-i-n

PHOLEDs. The high efficiency and low roll-off of the devices are attributed to the reduced

charge carrier accumulation and thus improved electron-hole balance, which is a result of using

CBP as both the HTL and host for the phosphorescent emitter. This simplified device design

strategy represents a pathway towards high efficacy OLEDs and should be applicable to other

phosphorescent emitters as well as white OLEDs.

5.4. Summary

In summary, it was demonstrated that exciton quenching due to accumulated space charges

at the exciton formation interface remains a significant loss of efficiency in existing device

designs. It was found that exciton quenching from α-NPD+ radical cations can be reduced by

using a wide band gap hole-transporting interlayer at the α-NPD/EML interface, resulting in

increased device efficacy. It was also found that an interfacial dipole at the interface between an

HTL and ETL molecule correlates well with the position of the exciton formation zone. The

findings of this work should have broad implications in materials selections in the design of

EML structures. It was further demonstrated that using CBP as a single layer HTL eliminates

the hole blocking barriers, preventing hole accumulation anywhere in the device. Exciton

quenching due to accumulated radical cations is thus eliminated, resulting in significantly

improved device performance in both fluorescent and phosphorescent OLEDs. It is envisioned

that this simple non-blocking design of the exciton formation zone represents a simple pathway

to achieve high performance OLEDs.

Chapter 6 Optical designs of OLEDs

68


In this chapter, an optical model describing the optical electric field of an OLED will be

presented. The developed model will be used to design high efficiency OLEDs with state of the

art performance. The key feature of the design is the replacement of ITO anode with an

oxide/metal/oxide electrode stack, which enables the use of low cost flexible plastic substrates.

The content of this chapter was published as J. Appl. Phys. 109, 053107 (2011) and Nature

Photon. 5, 753 (2011).

6.1. Introduction

It is generally believed that only ~20% of the light can be out-coupled from a standard bottom

emission OLED, resulting in a much smaller external quantum efficiency compared with the

internal efficiency. Most of the light in an OLED is suppressed by the wave guide modes in the

substrate and organic layers. Recently, various approaches have been proposed to increase the

out-coupling factor, e.g. using high-index substrates,21

top emission OLEDs taking advantage of

the micro-cavity effect,115-117

introducing distributed-Bragg-reflectors (DBR),118

and using micro

lenses or grids119,120

and photonic crystals.121

In order to optimize the optical out-coupling of light

to achieve maximum efficiency it is important to get an in-depth understanding of the optics

occurring in OLEDs. In this chapter a model using a dipole emission source term to describe the

optical electric field distribution inside and out of the stratified device structure is shown in detail.

The optical electric field in a stratified structure of homogeneous and isotropic media is described

by 2 × 2 matrices. Parameters such as reflectivity, transmittance and Poynting vector flux can be

obtained numerically, such that the emission pattern and optical out-coupling factor can be

calculated as a function of the refractive index and the thickness of different layers. Furthermore,

OLEDs with a weak micro-cavity are used to evaluate the model. A flexible OLED with very

high efficiency will be also shown as a case study of the optical model at the end of this chapter.


69

6.2. Optical model of OLEDs

Typically, an OLED is made up of a number of organic thin films sandwiched between two

electrodes, e.g. ITO and Al. An OLED may therefore be considered as an emission source

embedded in a Fabry-Perot microcavity. The theoretical spectrum for light emission normal to

the plane of the device has previously been approximated following Deppe’s approach.118,122

Since this model only describes emission normal to the plane of the device it cannot be used to

calculate the angular dependence. Also, this method is over simplified for OLEDs as it does not

consider the dipole emission characteristic of organic emitters. Although dipole emission can be

well described by both semi-classical and quantum mechanical approaches, for example using

the Hertz vector and Sommerfeld integrals,123,124

these methods are too complicated. A more

convenient method to describe the source term for dipole emission was proposed by Benisty et

al, 125

based on the earlier work by Lukosz126-128

. In this secton, a simple but accurate model

using a dipole emission source term combined with the transfer matrix method will be presented

to address the exciton distribution in OLEDs. Also, details on the application of this theory will

also be discussed for OLEDs with an example (experimental data).

6.2.1. Theory

a. Source term for dipole emission

The spontaneous emission of an OLED is modeled by vertical and horizontal dipoles

embedded in a homogeneous and isotropic media. Lukosz proposed that the normalized power

density of the s- and p- polarized light emitted in direction k, for the vertical and horizontal

dipoles respectively, is,128

23sin

8

pP

, (6.1)

2

/ /

3cos

16

pP

, (6.2)

0sP , (6.3)


70

/ /

3

16

sP

. (6.4)

In Eq. (6.1) – (6.4), is the emission angle, i.e. angle between the surface normal and the

wave vector k. Figure 6.1 shows the schematic diagram of a plane of dipoles embedded in a

homogeneous and isotropic media. With these power densities we can construct the source term

and use the standard matrix transfer method to describe the optical electric field distribution in

the stratified structure.

z

x

yz

x

y

Plane of dipolesPlane of dipoles

Figure 6.1 Schematic diagram of the dipole plane with vertical and horizontal dipoles.

b. Optical electric field distribution

Figure 6.2 shows the simplest case without an embedded source. A plane wave is incident

from layer j = 0 into the stratified structure. The thickness of layer j (j=1,2,…n) is dj. The

complex refractive index in each layer is j j jn n i . With the complex refractive index we

can unify the expression for an evanescent wave (e.g., propagation in the metal electrode with

strong damping). The optical electric field that is incident in and out of the system can be

written by using the total system transfer matrix as,


71

0 1

1

00

n

n

E EM

E

, (6.5)

where the + and – superscripts denote the right- and left- going waves respectively, and M is the

product of the layer matrix Li (phase change) and interface matrix Iij (refraction),

1 01 1 12 2 1k k kM I L I L L I

. (6.6)

In Eq. (6.6), the interface matrix can be obtained from the reflection (ij

r ) and transmission

(ij

t ) coefficients at interface i/j,

1/ /

/ 1/

ij ij ij

ij

ij ij ij

t r tI

r t t

. (6.7)

The reflection and transmission for s- and p- polarized light can be calculated using the Fresnel

equations as,

2 2

2 2

ˆ ˆ ˆ ˆcos cos

ˆ ˆ ˆ ˆcos cos

j i i i j jp

ij

j i i i j j

n n n nr

n n n n

, (6.8)

2

2 2

ˆ ˆ2 cos cos

ˆ ˆ ˆ ˆcos cos

i j i jp

ij

j i i i j j

n nt

n n n n

, (6.9)

ˆ ˆcos cos

ˆ ˆcos cos

j j i is

ij

j j i i

n nr

n n

, (6.10)

ˆ2 cos

ˆ ˆcos cos

s i i

ij

j j i i

nt

n n

. (6.11)


72

0 1 j n

0 0( , )E E

0k

jk

jk

0 j

1n

1d jd nd

'x

( '), ( ')j jE x E x

n+1

n

Index:

0k

1( ,0)nE

1nk

x

Figure 6.2 Schematic diagram of a multilayer structure with n+1 layers.

The phase matrix in Eq. (6.6) can be written as,

2ˆ cos

2ˆ cos

0

0

i i i

i i i

i n d

ii n d

eL

e

, (6.12)

where is the wavelength of the light. The optical electric field distribution inside the

stratified structure is generally of greater interest and can also be obtained by the matrix transfer

method. For example, as shown in Fig. 6.2, when the layer j is considered, the electric field at x

can be expressed as follows,

2ˆ cos

0

2ˆ cos

0

( )0

( )0

j j

j j

i n x

j

ji n x

j

E xE eM

E xEe

. (6.13)

Now that we have addressed the optical electric field distribution with the transfer matrix

method, we must now consider how to deal with the embedded emission source. We firstly

assume that the emission is confined to be at one interface, i.e. at the hole transport layer

(HTL)/electron transport layer (ETL) interface, and ignore the exciton distribution. Therefore at

the emission interface, the optical electric field is discontinuous (see Fig. 6.3 a) and can be


73

obtained by adding the source term as,

a b

a b

E EA

E A E

. (6.14)

In Eq. (6.14) ,

aE

and ,

bE

can be obtained from,

( )

0

0aa

a

EM

E E

, (6.15a)

1( )

0

b nb

b

E EM

E

, (6.15b)

where ( )aM and

( )bM are the system matrices on the left and right side of the emission

interface (see Fig. 6.3), i.e.,

( ) ( )

11 12( )

( ) ( )

21 22

a a

a

a a

m mM

m m

, (6.16a)

( ) ( )

11 12( )

( ) ( )

21 22

b b

b

b b

m mM

m m

. (6.16b)

From Eq. (6.14) to (6.16), with simple algebra, we can derive,

( ) ( ) ( ) ( )

( ) 22 11 11 21

0 21 ( )( )( ) ( ) 1112

21 11 ( )

12

a a b b

a

aab b

b

m m m A m AE m

mmm m

m

, (6.17)

where the source term ,A

can be expressed by the normalized power density P shown in Eq.

(6.1)-(6.4) (for s- or p- polarized) as,

,/ / ,/ /A P

, (6.18a)


74

,/ / ,/ /A P

. (6.18b)

In an actual OLED device, however, there will be an exciton distribution and the

distribution can be incorporated by using superposition. To model the optical electric field, we

discretize the source distribution into H blocks with a thickness of dx=demission/H (see Fig. 6.3

b), where demission is the total thickness of the exciton distribution region. At the bh sub-layer

( 1,2,...,h H ), the source term can be substituted by

, ,

( ) ,/ / ,/ /( ' )

hA A D x hdx

, (6.18)

where ( ')D x is the exciton distribution function. To reflect the discrete nature of organic

materials, we use 1dx nm, the typical size of a small molecule (or sites of a polymer).

Similar to Eq. (6.16a) and (6.16b), the transfer matrix on the left and right side of the bh/bh+1

interface becomes respectively

( ) ( )

11 12( )

( ) ( )

21 22

a a

h ha

h a a

h h

m mM

m m

, (6.19a)

( ) ( )

11 12( )

( ) ( )

21 22

b b

h hb

h b b

h h

m mM

m m

. (6.19b)

Similar to Eq.(6.17), we have,

( ) ( ) ( ) ( )

( ) 22 11 11 21

( )0 21 ( )( )( ) ( ) 1112

21 11 ( )

12

a a b b

a h h h h

h h aab b hh

h h b

h

m m m A m AE m

mmm m

m

, (6.20)

Therefore, with Eq.(6.20) we can calculate the total emission from the contribution of

excitons at different location with different emission strength. It is noted that the one important

assumption behind the superposition is that the sources are not coherent. However it does not

negate the ability of coherent interference between the source and its reflections.129

Moreover,

the study of the location and distribution of the emission zone is another interesting scientific


75

issue, which is highly dependent on the carrier concentration, i.e. the distribution of holes

injected from the anode and electrons injected from the cathode, and how they recombine, and

how the excitons diffuse. A simple uniform distribution130,131

of exciton across the doping region

will be employed in the calculation to describe the experimental EL spectra in the next section.

Since in a guest-host system, the dopants (guest), for example C545T in this work, may also

serve as hole traps,132,133

it is still a reasonable approximation to assume the uniform

distribution, which also gives a reasonably good description of the experimental EL spectra.

However, consideration of different exciton distributions can be easily accomodated by

changing the distribution function ( ')D x .

source

0

0

E

1

0

nE

a

a

E

E

b

b

E

E

( )aM( )bM

(a)

source

0

0

E

1

0

nE

( )aM ( )cM

a

a

E

E

c

c

E

E

1b 2b

1

1

b

b

E

E

1

1

'

'

b

b

E

E

2

2

b

b

E

E

2

2

'

'

b

b

E

E

(b)

hb

bh

bh

E

E

'

'

bh

bh

E

E

x

x

Figure 6.3 Schematic diagram of a multilayer structure with embedded source plane.


76

c. EL spectrum and device efficiency

Experimentally, we do not directly measure the optical electric field but the power instead.

Therefore, in order to relate the emission pattern and the efficiency of an OLED with the

calculations, we have to express the output power in the form of the dipole source terms. The

irradiance (the power per unit projected area) can be calculated by the Poynting vector flux as,

2~ | |d

I n EdS

, (6.21)

where is the power and S is the projected area.

The EL spectrum is proportional to the radiant intensity R

I (the power per solid angle).

Both the projected area and the solid angle will change when the light is propagating through the

stratified structure. Therefore, to extract the emission pattern, we have to consider all these

changes,

2

,

2

,

ˆ | | cos

ˆ | | cos

out

R out out out out active out out out active

activeR active active active out active active active out

active

d

I d I dS d n E d

dI I dS d n E d

d

, (6.22)

where the subscript “active” and “out” correspond to the active layer (i.e., emission zone) and

outside environment respectively. The change of solid angle has been derived in Ref. 127 as,

2

2

ˆ cos

ˆ cos

active out out

out active active

d n

d n

. (6.23)

Substituting Eq. (6.23) into (6.22), we can obtain,

3 2 2

,

3 2 2

,

ˆ cos | |

ˆ cos | |

R out out out out

R active active active active

I n E

I n E

. (6.24)

The emission power has contributions from both the vertical and horizontal dipoles. For an

isotropic source,125,128


77

/ /2

3 3total

P PP . (6.25)

This ratio should be applied to Eq. (6.24) when the normalized emission pattern is calculated.

Also, as mentioned previously, when a detailed exciton distribution is considered, the

contribution from excitons at different locations and different intensity should also be summed

up. The ,R active

I in Eq. (6.24) can be considered as the PL spectrum of the emitter in the OLED. It

is noted that the assumption behind such calculation is that the electrical properties do not

change when bias is applied, e.g. the band gap does not change at different bias. Also, the

typical width of OLEDs emission is ~ 75 nm which corresponds to a coherence length of ~ 103

nm.129

Given that the thickness of the substrate, e.g. glass, is ~106 nm, orders of magnitude

larger than the coherence length, the calculation here ignores the far field interference in the

glass, i.e. assumes that the electric field exiting the glass/organic interface is the same as the one

incident at the glass/air interface.

d. Normalized OLED efficiency

The last step is to consider the photopic response of the human eye, i.e., to convert the

radiance to luminance, which can be simply accomplished by integrating the photopic response

over wavelength,

0

( )v e

L C L P d

, (6.26)

where P is the photopic response function, e

L and v

L denotes the radiance and luminance

respectively and0

C is a constant (683 lm/W). Therefore, with Eq. (6.26) and (6.24) the

normalized current efficiency of an OLED can be calculated. In other words, the optical

out-coupling of different device structures can be calculated. Also, to calculate the normalized

power efficiency, the luminance should be integrated over emission angle as,

/2

02 sin

vW I d

, (6.27)

where Iv is the luminoux intensity (luminous flux per unit solid angle).


78

6.2.2. Evaluation of the model

Here, OLEDs with weak micro-cavity are used to evaluate the theoretical model, and the

structure of which is: Au (21 nm) / MoOx (1 nm) / CBP (x nm) / Alq3:C545T (33 nm) / Alq3 (33

nm) / LiF (1 nm) / Al (100 nm). Figure 6.4 shows the EL spectrum of the OLED devices with

different thicknesses of CBP HTL. The symbols are the experimental measurements while the

lines are the corresponding theoretical calculations using the present model. The calculation is

conducted by assuming that the excitons are uniformly distributed in the C545T doped region

(thickness of 33 nm). The calculation shows excellent agreement with the experiment results.

From the figure, we can see that when the CBP thickness increases, the EL spectrum is

broadened and red shifted.

400 500 600 700

CBP Thickness

25 nm

33 nm

44 nm

57 nm

68 nm

EL (

a.u

.)

Wavelength (nm)

Figure 6.4 Experimental EL spectra of OLEDs with different thickness of CBP measured normal to the

substrate (open symbols) as well as the corresponding theoretical calculations (solid lines). The PL spectrum

of C545T doped Alq3 used in the calculation is also shown for comparison as dashed line.

Although it is generally believed that the thickness of the ETL is much more sensitive than

that of the HTL to the device performance. However, it is only true when the microcavity effect


79

is very weak, e.g., in devices with ITO anode. In this study, the reflectivity of Au anode is much

larger than ITO and the device with Au anode exhibits a much stronger cavity effect (although it

is still considered a weak microcavity9). Therefore, the thickness of CBP is also very sensitive to

the EL spectrum. This also implies that the out-coupling factor of the OLED with weak

microcavity is also highly sensitive to the device thickness, i.e. at some wavelengths it may be

enhanced while at others it may be suppressed due to the wavelength dependent interference. It

is believed118

with a strong microcavity, such as a DBR mirror, that along the substrate normal

the emission spectrum may be much narrowed and the emission can be enhanced by several

times. However, at different observation angle, the enhancement may be reduced.

6.3. Flexible OLEDs

The research community has focused on using high refractive index (n ≥ 1.8) substrates in

place of standard glass to enhance the out-coupling of this trapped light.20,21,134

However, one of

the major advantages of OLEDs is the use of lightweight flexible plastic substrates,135

which

unfortunately have a low refractive index (n ≤ 1.6) similar to that of standard glass. In this

section, a novel out-coupling enhancement approach that does not rely on high-index substrates

and is fully compatible with low cost flexible plastic substrates as well as regular glass will be

shown as an example on how the optical model developed can be used to design OLED devices.

Using this out-coupling enhancement a flexible green OLED on plastic with an exceptionally

high efficiency is demonstrated.

The key feature of this out-coupling strategy is the replacement of de facto standard ITO

anode with an oxide/metal/oxide anode stack, which altogether eliminates the need for a high

refractive index substrate and hence enables the use of flexible plastic substrates (see Fig 6.5).

The electrode stack consists of a semi-transparent Au thin film sandwiched between a layer of

Ta2O5 and MoO3 (i.e., Ta2O5/Au/MoO3). In the electrode stack the Au layer serves as the

conductive channel for charge transport. Although metal thin films have been extensively

studied as alternative anodes in OLEDs, most research has focused on top-emission

architectures, which incorporate a strong optical micro-cavity between the highly reflective

metal anode and Al cathode.136

Although a strong optical micro-cavity can be used to enhance


80

the out-coupling of light in the forward direction, the EL emission spectrum is typically

narrowed and also exhibits a strong dependence on viewing angle,118

which is undesirable for

many applications. Therefore, in this electrode stack a thin semi-transparent layer of Au is

adopted, which forms only a weak optical micro-cavity with the Al cathode, and hence does not

narrow the EL spectrum or contribute to an angle dependent EL emission.9

Figure 6.5 Schematic OLED device structure with flexible plastic substrate.

Although semi-transparent metal thin films, such as Au, are well recognized as ideal

candidates to replace ITO, particularly for flexible devices on plastic substrates,9 there remain

two key challenges to overcome: i) injecting charge into organic materials from metals is

problematic due to a strong interfacial dipole formed at the metal/organic interface, and ii)

optimizing the weak optical micro-cavity formed between the semi-transport metal anode and

highly reflective Al cathode is extremely difficult without altering the electrical properties of the

device. In our electrode stack the MoO3 and Ta2O5 layers are employed to solve these two key

challenges, respectively.

It is well known that a strong interfacial dipole is formed at most metal/organic

interfaces.44

This dipole typically hinders the injection of charge from the metal into the organic,

resulting in very high operating voltages in devices. Many different strategies have been

proposed to reduce the interfacial dipole between Au and commonly used hole transport layers

(HTLs), such as N,N'-diphenyl-N,N'-bis-(1-naphthyl)-1-1'-biphenyl-4,4'-diamine (α-NPD), for


81

example by introducing a layer of C60 at the interface,39

(i.e., Au/C60) or by oxidizing the surface

of the metal9 (i.e., Au/AuOx). However, these techniques are only effective for molecules like

α-NPD with a very shallow HOMO of only 5.4 eV. In Chapter 5, it has been demonstrated that

using a deep HOMO HTL such as CBP can significantly enhance the performance of OLEDs.

Clearly, a new strategy to enhance the injection of charge from Au into the deep HOMO of CBP

is thus required in order to make use of this state of the art device structure (i.e., to maximize the

IQE). In Chapter 4 and Chapter 5, it has been shown to use a thin layer of a transition metal

oxide, such as MoO3, to enhance the work function of anode. The knowledge gained in Chapter

4 and Chapter 5 is used here in the electrode stack to overcome this challenge, i.e. directly

injecting charge into CBP.

The traditional method of tuning the micro-cavity in an OLED is to change the thickness

of the organic layers. However, such tuning is problematic since changing the thickness of the

organic layers will also simultaneously impact the electrical characteristics of the device (e.g.,

the electron-hole balance reaching the emissive layer which directly influences the IQE). The

use of p- and n-doped transport layers has been shown as an effective method of decoupling the

optical and electric performance of OLEDs,20,21

allowing fine tuning of the optical modes

without significantly influencing the electrical characteristics of the device. However, using

doped transport layers significantly complicates the device fabrication and requires numerous

additional blocking layers. We thus elected to use a high refractive index layer Ta2O5 in the

electrode stack to tune the optical micro-cavity (i.e., to maximize the out-coupling) without

having to alter the structure of the organic layers or influence the electrical characteristics of the

device.

The device structure is then: Substrate / Anode / CBP (40 nm) / CBP: Ir(ppy)2(acac) (8

wt%, 15 nm) / TPBi (10 nm) / 3TPYMB (60 nm) / LiF (1 nm) / Al (100 nm). The normalized

luminance using the Ta2O5/Au/MoO3 stacked electrode was simulated based the optical model

presented in Section 6.2. Figure 6.6 shows the enhancement in the out-coupling of light from a

device utilizing the Ta2O5/Au/MoO3 electrode, relative to a reference device with MoO3

modified ITO electrode. A peak enhancement in the out-coupling of ~2 times is found for a Au

thickness of d Au = 18 nm, which implies that the device with the Ta2O5/Au/MoO3 electrode has


82

the potential to be nearly twice as efficient as a standard ITO based device. Note that the organic

layer thicknesses for both devices were identical and were based on the optimized structure for

the ITO device.

Figure 6.6 Calculated enhancement ratio of the Ta2O5/Au/MoO3 electrode relative to ITO as a function of the

thickness of both Au and Ta2O5.

An OLED device was fabricated on a flexible polycarbonate substrate according to

optimization from the calculation (i.e., 70 nm Ta2O5 and 18 nm Au). Figure 6.7 a and b

summarize the measured EQE and power efficiency of the device fabricated in comparison to a

reference device fabricated on ITO coated glass. Remarkably the Ta2O5/Au/MoO3 device on

flexible plastic reaches a high EQE of ~40% EQE without the use of a high refractive index

substrate. However, the out-coupling enhancement is lower than that predicted by the optical

model, which is likely due to the slightly different hole injection from the Au/MoO3 electrode

than from ITO/MoO3, which is clearly shown in the different IV characteristics in the single

carrier hole only device as shown in Fig. 6.8. A similar phenomenon has also been reported for

En

ha

nce

nm

en

t ra

tio

dAu

(nm)d

Ta2O5 (nm)


83

other metal/oxide/organic interfaces that the work function of the underneath material (for

example the ITO and Au here) has an impact on the effective work function of the electrode.45,95

In other words, the discrepancy between the simulation and experimental results is most likely

due to a subtle difference in the electrical properties (i.e., the IQE) of the devices, which is not

accounted for in the optical simulation.

Figure 6.7 (a) External quantum efficiency (EQE) and (b) Power efficiency (PE) of the device structure

optimized for Ir(ppy)2 (acac) as a function of luminance.

1 10 100 1000 10000

20

40

60

80

100

Glass/ITO

Plastic/Ta2O

5/Au

Plastic/Ta2O

5/Au (out-coupling)

E

QE

(%

)

Luminance (cd/m2)

(a)

1 10 100 1000 1000010

100

Pow

er

Effic

iency (

lm/W

)

Luminance (cd/m2)

Glass/ITO

Plastic/Ta2O

5/Au

Plastic/Ta2O

5/Au

(out-coupling)(b)


84

Figure 6.8 Current density as a function of average electric field of CBP single carrier hole only device using

Au/MoO3, ITO/MoO3 and Au anodes. The anode modified by MoO3 enables good hole injection into CBP.

The inset is the same data for Au/MoO3 and ITO/MoO3 plotted on a log-linear scale. Clearly, the injection

from Au/MoO3 is better than from ITO/MoO3.

Not only does the novel Ta2O5/Au/MoO3 electrode design increases the efficiency, it also

enables the unique flexible form factor in OLEDs. Figure 6.9 shows a photograph of a large area

(50 mm × 50 mm) working device on flexible plastic at high luminance. Although the electrode

enhances the out-coupling of light, a significant fraction of the emitted photons are still trapped

in the substrate modes (i.e., the plastic substrate). To further improve the EQE of the device a

lens-based structure can be used to out-coupling the light trapped in the substrate modes,

resulting in a further enhancement of the efficiency by a factor of 1.55. As a result, the highest

EQE and power efficiency reaches 63% and 290 lm/W respectively (see Fig. 6.7), which is

equivalent to an enhancement of ~2.5 times over the ITO reference device. Although the further

extraction of light from the substrate modes using a lens is not fully compatible with flexible

devices, this issue can be easily overcome by using a plastic substrates with pre-molded half

sphere micro-lenses.3,27

0.2 0.4 0.6

0

10

20

30

40

0.2 0.3 0.4 0.5 0.6

10-1

100

101

Cu

rre

nt

De

nsity (

mA

/cm2)


Au/MoO3

ITO/MoO3

Au

Cu

rre

nt D

ensity (

mA

/cm

2)



85

Figure 6.9 Photograph of a flexible OLED (50 mm × 50 mm) at high luminance.

6.4. Summary

In summary, an optical model describing the optical electric field of an OLED has been

developed based on the dipole emission model and the matrix transfer method. Experimentally,

OLEDs with weak microcavity were fabricated. The theoretical calculation was in good

agreement with experimental results. At the end, by using the model, high efficiency OLEDs

with state of the art performance were designed and achieved on flexible plastic substrates. The

unique out-coupling strategy developed here may help to unlock the full potential of OLEDs on

flexible plastic and to enable the low cost mass-production of flexible OLEDs using roll-to-roll

processing for next generation flexible displays and solid state lighting.

Chapter 7 Summary and future work

86


7.1. Summary

In Chapter 3, energy level alignment at metal/organic interfaces was systematically studied.

It was shown that interface dipole theory can accurately describe the energy level alignment at

metal/organic interfaces. Furthermore, examples at metal/ C60/organic interfaces were shown

and the Fermi level at metal/C60/organic interfaces was found to be pinned to the charge

neutrality level of C60 (~ 4.7 eV). This phenomena was attributed to the C60 interlayer disrupting

the interfacial dipole at the metal/organic interfaces, resulting in more favorable energy-level

alignment for hole injection. As a result holes were injected directly into the organic layer,

bypassing the deep HOMO of C60.

In Chapter 4, hole injection from different metal oxides into α-NPD was systematically

studied in single carrier hole-only devices yielding a IV database for variable injection barrier

heights. It was found that the quasi-Ohmic regime is much larger (i.e., covers a greater range of

injection barrier height) than was previously expected. A criterion that defined Ohmic,

quasi-Ohmic, and injection limited contacts was quantified based on a time-domain simulation

of charge transport across α-NPD single carrier devices. This criterion included the effects of the

electric field dependent mobility, organic layer thickness and charge injection barrier height.

The effects of the built-in potential on the IV characteristics were also evaluated. A

barrier-thickness-voltage “phase” diagram that defined the regions of SCLC, quasi-Ohmic and

ILC for α-NPD was presented.

In Chapter 5, it was demonstrated that exciton quenching due to accumulated space charges

at the exciton formation interface was the cause of the significant loss of efficiency in existing

device designs. It was found that exciton quenching from α-NPD+ radical cations can be reduced

by using a wide band gap hole-transporting interlayer at the α-NPD/EML interface, resulting in

increased device efficacy. It was further demonstrated that using CBP as a single layer HTL

eliminates the hole blocking barriers, preventing hole accumulation anywhere in the device.

Exciton quenching due to accumulated radical cations is thus eliminated, resulting in


87

significantly improved device performance in both fluorescent and phosphorescent OLEDs. The

essence of the design concept proposed in this chapter was the elimination of any redundant

“injection layer” and “blocking layer”, contrary to the traditional device design concepts, which

significantly simplified the device structure and enhance the device performance.

Finally, in Chapter 6, an optical model describing the optical electric field of an OLED was

developed. Theoretical calculations from the developed model were compared with the device

performance and were in good agreement with experimental results. By using the model, high

efficiency OLEDs with state of the art performance were designed and realized. The key feature

of the design was the replacement of an ITO anode with an oxide/metal/oxide electrode stack,

which enabled the use of low cost flexible plastic substrates.

7.2. Future work

Various findings of this thesis have attracted considerable interest from major flat-panel

display manufacturers and are being commercialized by a spin-off company, OTI Lumionics Inc.

Although the findings have been seen to be useful in display application, continuous work has to

be conducted for real device applications, not only in flat-panel display, but also solid state

lighting (SSL), another important application of OLEDs.

The most significant contribution of my thesis is the achievement of a highly simplified

OLED structure with an exceptionally high EQE at high luminance. However, only a single

color (green) has been demonstrated to achieve such high efficiency. OLEDs of other colors

such as white with similarly high efficiency are yet to be demonstrated. In fact, white OLEDs

are of greater significance for both display (i.e. white light plus color filter) and lighting

applications. More investigations have to be done to test if the same device design concept in

this thesis can be used in white OLEDs. For example, the triplet energy level of CBP is smaller

than most phosphorescent blue emitters, i.e., the CBP host may not be capable to confine the

excitons formed in the blue emitters, resulting in a lower efficiency in both blue and white

OLEDs. There are two possible ways to solve this problem. One is to use another host material

that is capable to transport holes with even deeper HOMO (larger triplet energy level) to replace

CBP as both host and HTL. The challenge is that an anode with even higher work function is


88

needed to inject sufficient holes into this host material. The second is to recycle the “unconfined

excitons”. For example, the device can be designed to transfer the energy to the lower energy

emitters such as green and red, so that the overall efficiency of the white OLEDs can remain

high.

Another direction of the future work would be the further development of semitransparent

metal anodes. Although Au anode has a relatively low absorption window for green emission,

the absorption in both red and blue is much larger, i.e. it is challenging to achieve a similarly

high efficiency in red and blue by using Au anode. Moreover, generally the adhesion of Au on

glass and other plastic substrate is poor. Some interlayer such as a thin layer of Ni may be

needed to increase the adhesion. Such interlayer will change the optical properties of the Au

anode and the device structure may have to be re-optimized.

References

90

References

1 W. Helfrich and W. G. Schneider, Phys. Rev. Lett. 14, 229 (1965).

2 C. W. Tang and S. A. VanSlyke, Appl. Phys. Lett. 51, 913 (1987).

3 M. A. Baldo, D. F. O'Brien, Y. You, A. Shoustikov, S. Sibley, M. E. Thompson, and S. R.

Forrest, Nature 395, 151 (1998).

4 S.-J. Su, Y. Takahashi, T. Chiba, T. Takeda, and J. Kido, Advanced Functional Materials 19,

1260 (2009).

5 B. X. Mi, P. F. Wang, Z. Q. Gao, C. S. Lee, S. T. Lee, H. L. Hong, X. M. Chen, M. S. Wong,

P. F. Xia, K. W. Cheah, C. H. Chen, and W. Huang, Adv. Mater. 21, 339 (2009).

6 S.-J. Su, T. Chiba, T. Takeda, and J. Kido, Adv. Mater. 20, 2125 (2008).

7 Z. Q. Gao, M. Luo, X. H. Sun, H. L. Tam, M. S. Wong, B. X. Mi, P. F. Xia, K. W. Cheah,

and C. H. Chen, Adv. Mater. 21, 688 (2009).

8 K. R. Choudhury, J. Lee, N. Chopra, A. Gupta, X. Jiang, F. Amy, and F. So, Advanced

Functional Materials 19, 491 (2009).

9 M. G. Helander, Z.-B. Wang, M. T. Greiner, Z.-W. Liu, J. Qiu, and Z.-H. Lu, Adv. Mater.

22, 2037 (2010).

10 Z. W. Liu, M. G. Helander, Z. B. Wang, and Z. H. Lu, Appl. Phys. Lett. 94, 113305 (2009).

11 C. W. Tang, S. A. VanSlyke, and C. H. Chen, J. Appl. Phys. 65, 3610 (1989).

12 L. S. Hung, C. W. Tang, and M. G. Mason, Appl. Phys. Lett. 70, 152 (1997).

13 M. G. Helander, Z. B. Wang, L. Mordoukhovski, and Z. H. Lu, J. Appl. Phys. 104, 094510

References

91

(2008).

14 C. Adachi, M. A. Baldo, S. R. Forrest, and M. E. Thompson, Appl. Phys. Lett. 77, 904

(2000).

15 http://www.universaldisplay.com/default.asp?contentID=604

16 http://www.osa-direct.com/osad-news/474.html

17 T. Stefano, Nature 459, 312 (2009).

18 Bardsley Consulting, Navigant Consulting Inc., Radcliffe Advisors Inc., SB Consulting,

Solid State Lighting Services Inc., Solid-State Lighting Research and Development:

Multi Year Program Plan (2011)

19 J. Huang, M. Pfeiffer, A. Werner, J. Blochwitz, K. Leo, and S. Liu, Appl. Phys. Lett. 80,

139 (2002).

20 S. Reineke, F. Lindner, G. Schwartz, N. Seidler, K. Walzer, B. Lussem, and K. Leo, Nature

459, 234 (2009).

21 S. Mladenovski, K. Neyts, D. Pavicic, A. Werner, and C. Rothe, Opt. Express 17, 7562

(2009).

22 M. G. Helander, Z. B. Wang, M. T. Greiner, J. Qiu, and Z. H. Lu, Rev. Sci. Instrum. 80,

033901 (2009).

23 S. Forrest, D. Bradley, and M. Thompson, Adv. Mater. 15, 1043 (2003).

24 M. T. Greiner, M. G. Helander, Z. B. Wang, and Z. H. Lu, Rev. Sci. Instrum. 80, 125101

(2009).

25 M. P. Seah, Surf. Interface Anal. 31, 721 (2001).

26 H. Bässler, Phys. Status Solidi B 175, 15 (1993).

http://www.universaldisplay.com/default.asp?contentID=604

http://www.osa-direct.com/osad-news/474.html

References

92

27 J. C. Scott, J. Vac. Sci. Technol. A 21, 521 (2003).

28 W. Mönch, Appl. Phys. Lett. 88, 112116 (2006).

29 A. M. Cowley and S. M. Sze, J. Appl. Phys. 36, 3212 (1965).

30 J. Tersoff, Phys. Rev. B 32, 6968 (1985).

31 M. G. Helander, Z. B. Wang, and Z. H. Lu, Appl. Phys. Lett. 93, 083311 (2008).

32 Y. Y. Yuan, S. Han, D. Grozea, and Z. H. Lu, Appl. Phys. Lett. 88, 093503 (2006).

33 X. D. Feng, C. J. Huang, V. Lui, R. S. Khangura, and Z. H. Lu, Appl. Phys. Lett. 86,

143511 (2005).

34 J. Y. Lee, Appl. Phys. Lett. 88, 073512 (2006).

35 P. Peumans and S. R. Forrest, Appl. Phys. Lett. 79, 126 (2001).

36 S. Uchida, J. Xue, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett. 84, 4218 (2004).

37 G. J. Matt, T. B. Singh, N. S. Sariciftci, A. M. Ramil, and H. Sitter, Appl. Phys. Lett. 88,

263516 (2006).

38 A. Kanwal and M. Chhowalla, Appl. Phys. Lett. 89, 203103 (2006).

39 S. Han, Y. Yuan, and Z. H. Lu, J. Appl. Phys. 100, 074504 (2006).

40 S. J. Chase, W. S. Bacsa, M. G. Mitch, L. J. Pilione, and J. S. Lannin, Phys. Rev. B 46,

7873 (1992).

41 A. Tamai, A. P. Seitsonen, F. Baumberger, M. Hengsberger, Z. X. Shen, T. Greber, and J.

Osterwalder, Phys. Rev. B 77, 075134 (2008).

42 B. W. Hoogenboom, R. Hesper, L. H. Tjeng, and G. A. Sawatzky, Phys. Rev. B 57, 11939

(1998).

References

93

43 R. Nouchi and I. Kanno, J. Appl. Phys. 97, 103716 (2005).

44 H. Ishii, K. Sugiyama, E. Ito, and K. Seki, Adv. Mater. 11, 605 (1999).

45 M. G. Helander, Z. B. Wang, J. Qiu, and Z. H. Lu, Appl. Phys. Lett. 93, 193310 (2008).

46 H. Vazquez, W. Gao, F. Flores, and A. Kahn, Phys. Rev. B 71, 041306 (2005).

47 N. Hayashi, H. Ishii, Y. Ouchi, and K. Seki, Journal of Applied Physics 92, 3784 (2002).

48 S. C. Veenstra, A. Heeres, G. Hadziioannou, G. A. Sawatzky, and H. T. Jonkman, Applied

Physics A: Materials Science & Processing 75, 661 (2002).

49 Z. B. Wang, M. G. Helander, M. T. Greiner, J. Qiu, and Z. H. Lu, Appl. Phys. Lett. 95,

043302 (2009).

50 G. B. Murdoch, M. Greiner, M. G. Helander, Z. B. Wang, and Z. H. Lu, Appl. Phys. Lett.

93, 083309 (2008).

51 C.-W. Chu, S.-H. Li, C.-W. Chen, V. Shrotriya, and Y. Yang, Appl. Phys. Lett. 87, 193508

(2005).

52 H. Kanno, R. J. Holmes, Y. Sun, S. Kena-Cohen, and S. R. Forrest, Adv. Mater. 18, 339

(2006).

53 Y. Han, D. Yanfeng, Z. Zhiqiang, and M. Dongge, J. Appl. Phys. 101, 026105 (2007).

54 M. Toshinori, K. Yoshiki, and M. Hideyuki, Appl. Phys. Lett. 91, 253504 (2007).

55 T.-Y. Chu and O.-K. Song, Appl. Phys. Lett. 90, 203512 (2007).

56 C. Ganzorig, M. Sakomura, K. Ueda, and M. Fujihira, Appl. Phys. Lett. 89, 263501 (2006).

57 N. Huby, L. Hirsch, G. Wantz, L. Vignau, A. S. Barriere, J. P. Parneix, L. Aubouy, and P.

Gerbier, J. Appl. Phys. 99, 084907 (2006).

References

94

58 A. Ruchi, K. Pramod, G. Subhasis, and M. A. Kumar, Appl. Phys. Lett. 93, 073311 (2008).

59 P. W. M. Blom, C. Tanase, D. M. de Leeuw, and R. Coehoorn, Appl. Phys. Lett. 86, 092105

(2005).

60 M. Kiy, P. Losio, I. Biaggio, M. Koehler, A. Tapponnier, and P. Gunter, Appl. Phys. Lett. 80,

1198 (2002).

61 C. Herring and M. H. Nichols, Rev. Mod. Phys. 21, 185 (1949).

62 G. G. Malliaras, J. R. Salem, P. J. Brock, and J. C. Scott, J. Appl. Phys. 84, 1583 (1998).

63 M. A.Lampert and P. Mark, Current Injection in Solids. (Academic Press, 1970).

64 P. N. Murgatroyd, J. Phys. D: Appl. Phys. 3, 151 (1970).

65 S. W. Tsang, S. K. So, and J. B. Xu, J. Appl. Phys. 99, 013706 (2006).

66 C. H. Cheung, K. C. Kwok, S. C. Tse, and S. K. So, J. Appl. Phys. 103, 093705 (2008).

67 S. K. So, S. C. Tse, and K. L. Tong, J. Display Technol. 3, 225 (2007).

68 S. M. Sze, Physics of semiconductor devices. (Wiley-Interscience, 1981).

69 R. H. Fowler and L. Nordheim, Proc. R. Soc. Lond. A 119, 173 (1928).

70 V. I. Arkhipov, E. V. Emelianova, Y. H. Tak, and H. Bässler, J. Appl. Phys. 84, 848 (1998).

71 J. G. Simmons, Phys. Rev. Lett. 15, 967 (1965).

72 V. I. Arkhipov and H. Bassler, Appl. Phys. Lett. 77, 2758 (2000).

73 U. Wolf, V. I. Arkhipov, and H. Bässler, Phys. Rev. B 59, 7507 (1999).

74 J. C. Scott and G. G. Malliaras, Chem. Phys. Lett. 299, 115 (1999).

75 G. G. Malliaras and J. C. Scott, J. Appl. Phys. 85, 7426 (1999).

References

95

76 G. G. Malliaras and J. C. Scott, J. Appl. Phys. 83, 5399 (1998).

77 U. Wolf, S. Barth, and H. Bassler, Appl. Phys. Lett. 75, 2035 (1999).

78 Z. B. Wang, M. G. Helander, S. W. Tsang, and Z. H. Lu, Phys. Rev. B 78, 193303 (2008).

79 S. W. Tsang, M. W. Denhoff, Y. Tao, and Z. H. Lu, Phys. Rev. B 78, 081301 (2008).

80 M. A. Baldo and S. R. Forrest, Phys. Rev. B 64, 085201 (2001).

81 I. H. Campbell, T. W. Hagler, D. L. Smith, and J. P. Ferraris, Phys. Rev. Lett. 76, 1900

(1996).

82 C. E. Small, S.-W. Tsang, J. Kido, S. K. So, and F. So, Advanced Functional Materials, n/a

(2012).

83 H. Wang, K. P. Klubek, and C. W. Tang, Appl. Phys. Lett. 93, 093306 (2008).

84 S.-F. Chen and C.-W. Wang, Appl. Phys. Lett. 85, 765 (2004).

85 J. Kido and T. Matsumoto, Appl. Phys. Lett. 73, 2866 (1998).

86 K. R. Choudhury, J.-h. Yoon, and F. So, Adv. Mater. 20, 1456 (2008).

87 Y. T. Tao, E. Balasubramaniam, A. Danel, B. Jarosz, and P. Tomasik, Appl. Phys. Lett. 77,

1575 (2000).

88 L. C. Palilis, A. J. Makinen, M. Uchida, and Z. H. Kafafi, Appl. Phys. Lett. 82, 2209

(2003).

89 Y. Divayana, X. W. Sun, B. J. Chen, G. Q. Lo, C. Y. Jiang, and K. R. Sarma, Appl. Phys.

Lett. 89, 173511 (2006).

90 H.-T. Lu, C.-C. Tsou, and M. Yokoyama, Journal of Crystal Growth 277, 388 (2005).

91 Q. L. Song, C. M. Li, M. B. Chan-Park, M. Lu, H. Yang, and X. Y. Hou, Phys. Rev. Lett. 98,

References

96

176403 (2007).

92 P. I. Djurovich, E. I. Mayo, S. R. Forrest, and M. E. Thompson, Organic Electronics 10,

515 (2009).

93 Z. B. Wang, M. G. Helander, Z. W. Liu, M. T. Greiner, J. Qiu, and Z. H. Lu, Appl. Phys.

Lett. 96, 043303 (2010).

94 J. Meyer, S. Hamwi, T. Bulow, H. H. Johannes, T. Riedl, and W. Kowalsky, Appl. Phys.

Lett. 91, 113506 (2007).

95 M. G. Helander, Z. B. Wang, M. T. Greiner, J. Qiu, and Z. H. Lu, Appl. Phys. Lett. 95,

083301 (2009).

96 Z. B. Wang, M. G. Helander, M. T. Greiner, J. Qiu, and Z. H. Lu, Phys. Rev. B 80, 235325

(2009).

97 Y. Kawamura, K. Goushi, J. Brooks, J. J. Brown, H. Sasabe, and C. Adachi, Appl. Phys.

Lett. 86, 071104 (2005).

98 C. Adachi, M. A. Baldo, M. E. Thompson, and S. R. Forrest, J. Appl. Phys. 90, 5048

(2001).

99 S. Reineke, T. C. Rosenow, B. Lüssem, and K. Leo, Adv. Mater. 22, 3189 (2010).

100 S. J. Su, H. Sasabe, Y. J. Pu, K. i. Nakayama, and J. Kido, Adv. Mater. 22, 3311 (2010).

101 D. Tanaka, H. Sasabe, Y.-J. Li, S.-J. Su, T. Takeda, and J. Kido, Jpn. J. Appl. Phys. 46, L10.

102 M. A. Baldo, C. Adachi, and S. R. Forrest, Phys. Rev. B 62, 10967 (2000).

103 S. Reineke, K. Walzer, and K. Leo, Phys. Rev. B 75, 125328 (2007).

104 N. C. Giebink and S. R. Forrest, Phys. Rev. B 77, 235215 (2008).

References

97

105 Z. D. Popovic and H. Aziz, J. Appl. Phys. 98, 013510 (2005).

106 F. X. Zang, T. C. Sum, A. C. H. Huan, T. L. Li, W. L. Li, and F. Zhu, Appl. Phys. Lett. 93,

023309 (2008).

107 S. Y. Kim, W. S. Jeon, T. J. Park, R. Pode, J. Jang, and J. H. Kwon, Appl. Phys. Lett. 94,

133303 (2009).

108 S. Reineke, G. Schwartz, K. Walzer, and K. Leo, Appl. Phys. Lett. 91, 123508 (2007).

109 W. S. Jeon, T. J. Park, S. Y. Kim, R. Pode, J. Jang, and J. H. Kwon, Appl. Phys. Lett. 93,

063303 (2008).

110 V. I. Adamovich, S. R. Cordero, P. I. Djurovich, A. Tamayo, M. E. Thompson, B. W.

D'Andrade, and S. R. Forrest, Organic Electronics 4, 77 (2003).

111 Z. Gao, C. S. Lee, I. Bello, S. T. Lee, R.-M. Chen, T.-Y. Luh, J. Shi, and C. W. Tang, Appl.

Phys. Lett. 74, 865 (1999).

112 G. He, M. Pfeiffer, K. Leo, M. Hofmann, J. Birnstock, R. Pudzich, and J. Salbeck, Appl.

Phys. Lett. 85, 3911 (2004).

113 S. Watanabe, N. Ide, and J. Kido, Jpn. J. Appl. Phys. 46, 1186 (2007).

114 W. H. Choi, C. H. Cheung, and S. K. So, Organic Electronics 11, 872 (2010).

115 L. H. Smith, J. A. E. Wasey, and W. L. Barnes, Appl. Phys. Lett. 84, 2986 (2004).

116 Q. Wang, Z. Deng, J. Chen, and D. Ma, Opt. Lett. 35, 462 (2010).

117 Q. Wang, Z. Deng, and D. Ma, Opt. Express 17, 17269 (2009).

118 A. Dodabalapur, L. J. Rothberg, R. H. Jordan, T. M. Miller, R. E. Slusher, and J. M.

Phillips, J. Appl. Phys. 80, 6954 (1996).

References

98

119 S. Moller and S. R. Forrest, J. Appl. Phys. 91, 3324 (2002).

120 Y. Sun and S. R. Forrest, Nature Photon. 2, 483 (2008).

121 Y.-C. Kim, S.-H. Cho, Y.-W. Song, Y.-J. Lee, Y.-H. Lee, and Y. R. Do, Appl. Phys. Lett. 89,

173502 (2006).

122 D. Deppe, C. Lei, C. Lin, and D. Huffaker, Journal of Modern Optics 41, 325 (1994).

123 R. R. Chance, A. Prock, and R. Silbey, The Journal of Chemical Physics 60, 2744 (1974).

124 G. W. Ford and W. H. Weber, Physics Reports 113, 195 (1984).

125 H. Benisty, R. Stanley, and M. Mayer, J. Opt. Soc. Am. A 15, 1192 (1998).

126 W. Lukosz and R. E. Kunz, J. Opt. Soc. Am. 67, 1607 (1977).

127 W. Lukosz and R. E. Kunz, J. Opt. Soc. Am. 67, 1615 (1977).

128 W. Lukosz, J. Opt. Soc. Am. 69, 1495 (1979).

129 A. Epstein, N. Tessler, and P. D. Einziger, Quantum Electronics, IEEE Journal of 46, 1388

(2010).

130 V. Bulovicacute, V. B. Khalfin, G. Gu, P. E. Burrows, D. Z. Garbuzov, and S. R. Forrest,

Phys. Rev. B 58, 3730 (1998).

131 H. Riel, S. Karg, T. Beierlein, W. Riess, and K. Neyts, J. Appl. Phys. 94, 5290 (2003).

132 Y. Luo and H. Aziz, Advanced Functional Materials 20, 1285 (2010).

133 E. W. Forsythe, D. C. Morton, Q. T. Le, L. Yan, F. Nuesch, C. W. Tang, and Y. Gao, San

Jose, CA, USA, 1999 (unpublished).

134 A. Mikami and T. Koyanagi, SID Symposium Digest of Technical Papers 40, 907 (2009).

References

99

135 S. R. Forrest, Nature 428, 911 (2004).

136 M. Thomschke, R. Nitsche, M. Furno, and K. Leo, Appl. Phys. Lett. 94, 083303 (2009).

Appendix A. Error analysis

100


The major cause of device-to-device performance variation on the same substrate is

non-uniformities in the device structure. For example, the thicknesses of the organic layers may

vary with position on the substrate. Therefore the substrate was rotated during the deposition

process to average out the non-uniformities. However, the thickness of each organic layer may

still have a subtle difference across a very large substrate (2 inch by 2 inch). Figure A1 shows

the device performance of different devices, with the same device structure, fabricated on the

same substrate. The current efficiency of these devices at 100 cd/m2 is 83.0 0.7 cd/A, which

implies that the variation is within a 1 % range. Errors due to other processing steps during the

device fabrication are also within this range (< 1%).

Film thicknesses were monitored using a calibrated quartz crystal microbalance (QCM).

The error in the film thickness reading from the QCM can be up to 10% due to the build-up of

material on the QCM (i.e., change of its lifetime). Therefore, to reduce the run-to-run variation,

the QCMs were replaced when 4% of the maximum lifetime was reached. Note that the QCM

was changed considerably more often than the manufacturer's recommendation.

The IV curves shown in this thesis represent one measurement cycle (i.e., measuring the

device only once). The standard deviation for many measurement cycles is < 1%. Since the

organic materials used in this study are very sensitive to moisture and oxygen, proper

encapsulation is needed to increase the device stability. For example, if a device is measured

without encapsulation in ambient air, the variation of different measurement cycles of the same

device may be much larger (up to 5 %) depending on the nature of the materials (i.e., some

molecules are more sensitive to ambient air than others). An extreme example is OLED devices

with phosphorescent dopants, which are notorious for their poor ambient stability. Also due to

the unstable nature of phosphorescent emitters, the measured luminance of the phosphorescent

devices in this study (even with proper encapsulation or in situ measurement in vacuum)

dropped significantly after one full measurement cycle. Each measurement cycle typically spans


101

a device brightness from zero up to > 20,000 cd/m2. For example, the measured luminance of a

standard Ir(ppy)2(acac) based OLED during the second measurement cycle was already reduced

by more than 10% compared to the first cycle. Therefore, the I-V-L measurements reported in

this thesis are for the first measurement cycle only. However, as shown in Fig. A2, the results

are still highly reproducible since the device-to-device variation and the run-to-run variation are

small.

1 ------> 16

Device

Al

ITO

0 1 2 3 4 5 6 710

-6

10-5

10-4

10-3

10-2

10-1

100

101

Cu

rre

nt D

en

sity (

mA

/cm

2)

Voltage (V)

9

10

11

12

13

14

15

16

1

2

3

4

5

6

7

8

(B)

100

101

102

103

104

0

20

40

60

80

100

(C)

9

10

11

12

13

14

15

16

Cu

rre

nt E

ffic

ien

cy(c

d/A

)

Luminance (cd/m2)

1

2

3

4

5

6

7

8

(A)

Figure A1. (A) Picture of real substrate with 32 devices. (B)Current density and (C) current efficiency of

device 1 to 16 as labeled in (A).


102

In terms of the error in the efficiency measurement, one of the major systematic errors

comes from the luminance meter that was used to measure the efficiency. Luminance is a

parameter describing the power of light corrected for the spectral response of the human eye

(photopic response). Usually, a photopic filter is used in regular luminance meters to correct this

response. However, the spectral response of most photopic filters differs slightly from the actual

photopic response of the human eye. Figure A2 shows the photopic response of the luminance

meter used in this study (Minolta Luminance Meter LS-110) in comparison with the actual CIE

photopic luminosity response curve.[1] Clearly from Fig. A2, the error is dependent on the

wavelength, and will therefore vary depending on the emission spectrum of different emitters.

For example, the emission peak of an Ir(ppy)2(acac) green OLED is in the range of 500-540 nm.

The reading of the luminance (and calculated efficiency) from the LS-110 may be only 1%

over-estimated in this spectral range. However, for a blue OLED device with emission in the

range of 450 nm, the luminance and corresponding efficiency might be up to ~ 30%

under-estimated.

Figure A2. Dash line: CIE photopic luminosity response curve; solid line: spectral response of Minolta

Luminance Meter LS-110.[1]


103

In addition to the LS-110, the efficiency was also measured using an integrating sphere with

a silicon photodiode with NIST traceable calibration.[2] Compared to the luminance meter

efficiency measurements, this method does not require the assumption that the OLED emission

is Lambertian, since the total luminous flux can be measured directly. However, the

measurement setup used in this study has a low sensitivity in the low luminance range (< 100

cd/m2) resulting in an error in the measured total luminous flux of up to 5 – 10%. Another

source error of the integrating sphere with photodiode may arise if the emission spectrum of the

OLED changes as a function of voltage, i.e., the color is not stable with driving voltage. For

example, many white OLEDs are not color stable. If the quantum efficiency is calculated based

on one typical spectrum (e.g., at 1000 cd/m2) for a white OLED, there may be a very large error

in the calculated efficiency at other voltages due to significant differences in the spectrum. The

best solution is to replace the photodiode with a spectrometer with high sensitivity.

Detector

Baffle

SubstrateOLED

Integrating Sphere

Figure A3. Schematic diagram of the measurement geometry using an integrating sphere.

Reference

[1] Manual of Minolta Luminance Meter LS-110.

[2] Tanaka, I. & Tokito, S. Precise Measurement of External Quantum Efficiency of Organic

Light-Emitting Devices. Jpn. J. Appl. Phys. 43, 7733.

Appendix B. List of publications related to this thesis

104

Appendix B. List of publications related to this thesis

1. Z.B. Wang, M.G. Helander, M.T. Greiner, and Z.H. Lu, "Energy-level alignment and

charge injection at metal/C60/organic interfaces", Appl. Phys. Lett. 95, 043302 (2009).

2. Z.B. Wang, M.G. Helander, M.T. Greiner, J. Qiu, and Z.H. Lu, "Analysis of

charge-injection characteristics at electrode-organic interfaces: Case study of

transition-metal oxides",Phys. Rev. B 80, 235325 (2009).

3. Z.B. Wang, M.G. Helander, Z.W. Liu, M.T. Greiner, J. Qiu, and Z.H. Lu, "Controlling

carrier accumulation and exciton formation in organic light emitting diodes", Appl. Phys.

Lett. 96, 043303 (2010).

4. Z.B. Wang, M.G. Helander, J. Qiu, Z.W. Liu, M.T. Greiner, and Z.H. Lu, "Direct hole

injection in to 4,4'-N,N'-dicarbazole-biphenyl: A simple pathway to achieve efficient

organic light emitting diodes", J. Appl. Phys. 108, 024510 (2010).

5. Z.B. Wang, M.G. Helander, J. Qiu, D.P. Puzzo, M.T. Greiner, Z.W. Liu, and Z.H. Lu,

"Highly simplified phosphorescent organic light emitting diode with >20% external

quantum efficiency at >10,000 cd/m2", Appl. Phys. Lett. 98, 073310 (2011).

6. Z.B. Wang, M.G. Helander, X.F. Xu, D.P. Puzzo, J. Qiu, M.T. Greiner, and Z.H. Lu,

"Optical design of organic light emitting diodes", J. Appl. Phys. 109, 053107 (2011).

7. Z. B. Wang, M. G. Helander, J. Qiu, D. P. Puzzo, M. T. Greiner, Z.M. Hudson, S. Wang,

Z.W. Liu, and Z. H. Lu, “Unlocking the full potential of organic light-emitting diodes on

flexible plastic”, Nature Photon. 5, 753 (2011).

design of high performance organic light emitting diodes · organic light emitting diodes (oleds)...

Documents