Enhancing quantum-dot luminescence in visible and infrared light emitting devices Geoffrey James

Enhancing quantum-dot luminescencein visible and infrared light emitting devices

by

Geoffrey James Sasajima Supran

B.A., Trinity College, University of Cambridge (2009)

Submitted to the Department of Materials Science and Engineeringin partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Materials Science and Engineering

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

June 2016

c Massachusetts Institute of Technology 2016. All rights reserved.

Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Department of Materials Science and Engineering

April 29, 2016

Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vladimir Bulović

Professor of Electrical Engineering and Computer ScienceThesis Supervisor

Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Professor Donald R. Sadoway

Chair, Departmental Committee on Graduate Students

2

Enhancing quantum-dot luminescence

in visible and infrared light emitting devices

by

Geoffrey James Sasajima Supran

Submitted to the Department of Materials Science and Engineeringon April 29, 2016, in partial fulfillment of the

requirements for the degree ofDoctor of Philosophy in Materials Science and Engineering

Abstract

We investigate how the external quantum efficiency (EQE) of colloidal quantum-dotlight emitting devices (QD-LEDs) can be enhanced by addressing in situ QD photo-luminescence (PL) quenching mechanisms occurring with and without applied bias.QD-LEDs promise efficient, high colour-quality solid-state lighting and displays, andour cost analysis of industrial-scale QD synthesis suggests they can be cost competi-tive. Efficiency ‘roll-off’ at high biases is among the most enduring challenges facingall LED technologies today. It stands in the way of high efficiencies at high brightness,yet it has not previously been studied in QD-LEDs. Simultaneous measurements ofQD electroluminescence (EL) and PL in an operating device allow us to show forthe first time that EQE roll-off in QD-LEDs derives from the QD layer itself, andthat it is entirely due to a bias-driven reduction in QD PL quantum yield. Using thequantum confined Stark Effect as a signature of local electric fields in our devices,the bias-dependence of EQE is predicted and found to be in excellent agreementwith the roll-off observed. We therefore conclude that electric field-induced QD PLquenching fully accounts for roll-off in our QD-LEDs. To investigate zero-bias PLquenching, we fabricate a novel near-infrared (NIR)-emitting device based on core-shell PbS-CdS QDs synthesised via cation exchange. QDs boast high PL quantumyield at wavelengths beyond 1 𝜇m, making them uniquely suited to NIR applicationssuch as optical telecommunications and computing, bio-medical imaging, and on-chipbio(sensing) and spectroscopy. Core-shell PbS-CdS QDs enhance the peak EQE ofcore-only PbS control devices by 50- to 100-fold, up to 4.3 %. This is more than doublethe efficiency of previous NIR QD-LEDs, making it the most efficient thin-film NIRlight source reported. PL measurements reveal that the efficiency enhancement is dueto passivation of the PbS core by the CdS shell against a non-radiative recombinationpathway caused by a neighboring conductive layer within the device architecture.

Thesis Supervisor: Vladimir BulovićTitle: Professor of Electrical Engineering and Computer Science

3

4

Acknowledgments

"What’s cookin’ good lookin’?!" No words better sum up Prof. Vladimir Bulović’s

indefatigable curiosity, passion, and good humour, than his own. It has been an

honour - and more fun than I could have imagined - to work under the supervision

of such a smart, energetic, and kind mentor, and a pioneer of our field. When the

President of the United States paid our lab a visit during my first year, I knew I was in

for an adventure in ONELab (Organic and Nanostructured Electronics Laboratory),

and throughout the years, Vladimir has continually nurtured an incredible playground

of talented colleagues, amazing equipment, and world-class ideas. I am particularly

grateful to Vladimir for his patience, interest, and support as I have explored my

passion for climate change work during this defining period of my career. He has

stood unwaveringly behind me during the hardest of times, and been there to cut

cake and celebrate with me during the best of them. Thank you, Vladimir, for these

unforgettable and formative years!

I am also deeply indebted to Prof. Moungi Bawendi and Prof. Marc Baldo, whose

doors have always been open. It has been a privilege to draw on their unparalleled

expertise during much of my work, and a pleasure to strike up fruitful collaborations

with their students and postdocs. I also sincerely thank Prof. Silvia Gradeçak and

Prof. Geoff Beach, who have kindly served on my thesis committee and provided

valuable guidance over the years.

My work was made possible thanks to the financial support of the DOE Excitonics

Center (an Energy Frontiers Research Center funded by the U.S. Department of

Energy) and an MIT Energy Initiative Fellowship. I also gratefully acknowledge QD

Vision, Inc., who have provided materials and technical support.

I have been so lucky that in my colleagues I have found such good friends. From

day one of grad school, and for my first several years in ONELab, Yasu Shirasaki

was my QD-LED partner-in-crime. I am very grateful to him for taking the time

to show me the ropes. Together we made our first QD-LED, inevitably reinvented

some wheels, and in the process enjoyed a tremendous collaboration on the roll-off

5

project. I look back fondly on our years sitting side-by-side in 13-3150. A little while

later, we were joined in the office by Katherine Song, whose many extreme injuries

had me convinced there was a curse on our office! Thank you to Katherine for being

such a great friend and labmate, without whom the IR QD-LED project would have

been neither as successful nor as fun. That project was also made possible only with

the synthetic skills of Gyuweon Hwang in the Bawendi group, to whom I am very

grateful. Most recently, it has been my pleasure to collaborate with Sihan (Jonas)

Xie: QD-LEDs are in your hands now, Jonas!

Beyond specific project collaborators, there are several others I must thank. No

one could have asked for nicer, smarter lab mates than Ni Zhao and Tim Osedach,

who not only mentored us newbies in the early years, but brought such levity to

days and nights in lab. I’ll never forget the time Ni took us to listen to Chinese

opera...argh! Patrick Brown and I joined ONELab together, and ours continues to

be amazingly eclectic friendship, working on everything from transfer line repairs

to climate activism. Thanks immensely for introducing me to fossil fuel divestment,

Patrick - as you know, it changed everything for me! Joel Jean has also become a great

friend and climate comrade, and is by far the most happy, least injured marathon

runner I’ve ever had the fortune to meet. Thank you also to Apoorva Murarka for

all the fun office chats, and cheers to Gleb Akselrod for getting my British sense of

humour better than most.

My heartfelt thanks, as well, to Prof. Will Tisdale, Wendi Chang, Trisha Andrew,

Farnaz Niroui, Deniz Bozyigit, Annie Wang, Sam Stranks, Andrea Maurano, Dong

Kyun Ko, Alexi Arango, and LeeAnn Kim. Indeed, to all collaborators and all past

and present members of ONELab with whom I’ve had the chance to work: it’s been

a privilege and a pleasure, thanks for all the help, and for all the memories.

I once read that Richard Feynman had a habit of turning down invitations to

receive prestigious awards, only to say "yes" to the local science club. It is in this

spirit that I will be forever grateful to those professors who said "yes" to letting a

precocious high schooler and undergrad tinker in their labs. Without their generosity

I may never have discovered my love of science. Thank you to Profs. Jeremy Baum-

6

berg (Southampton University, now at Cambridge), Sir Richard Friend (Cambridge),

Paul Alivisatos (UC Berkeley), Alan Heeger (UC Santa Barbara); Ray Goldstein

(Cambridge); and the late, great David MacKay (Cambridge).

Beyond the lab, climate advocacy and activism has become a huge part of my

life during grad school, and I am so thankful for the passionate, supportive, and

selfless friends I’ve had the pleasure to make. Helping to lead Fossil Free MIT in

our years of campaigning to divest MIT’s endowment from fossil fuel companies has

been one of the most fulfilling, transformative, and stressful(!) experiences of my

life. Thank you to all the brave students, staff, faculty, and alumni who continue to

speak truth to power, including but in no way limited to Ploy Achakulwisut, Ioana

Aron, Patrick Brown, Jane Conner, Ben Franta, Jason Jay, Daniel Mascoop, Jeremy

Poindexter, Ben Scandella, and Profs. Ian Condry, Kerry Emanuel, Charlie Harvey,

Naomi Oreskes, Jim Recht, Kieran Setiya, and John Sterman. I will never forget our

116 day #ScientistsSitIn! A big thank you, as well, to SustainUS and 350.org, and

to Bill McKibben in particular.

Last but not least, thank you to those who are always beside me. To Ploy: your

love, support, patience, and bravery - through all the ups and downs of grad school,

climate activism, and life - are all I could ever ask for. To my brother Jon: thanks for

keeping me grounded by literally being a rock star back in London, and for providing

the soundtrack to my Ph.D. And to Miko and David Supran, a.k.a. Mum and Dad:

There aren’t enough words to recount all the love, support, and opportunities you

have given me, which have carried me here and carry me on. Thanks, simply, for

being Mum and Dad - for being there "always and always, no matter what happens".

If you can fill the unforgiving minute

With sixty seconds’ worth of distance run,

Yours is the Earth and everything that’s in it,

And—which is more—you’ll be a Man, my son!

From "If—", by Rudyard Kipling.

7

Dedicated to my family - Mum, Dad, and Jon -

and to a Fossil Free MIT.

8

Contents

1 Introduction 19

1.1 Thesis Organisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

1.2 Energy efficiency in lighting and displays . . . . . . . . . . . . . . . . 22

2 Quantum-Dots as Luminophores 31

2.1 Benefits for solid state lighting and displays . . . . . . . . . . . . . . 31

2.1.1 Tunable and pure colours . . . . . . . . . . . . . . . . . . . . 31

2.1.2 Bright emission . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.1.3 Solution processable . . . . . . . . . . . . . . . . . . . . . . . 37

2.1.4 Stable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.1.5 Cost competitive . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.2 Quantum-dot light emitting devices . . . . . . . . . . . . . . . . . . . 44

2.2.1 Evolution of QD-LEDs . . . . . . . . . . . . . . . . . . . . . . 44

2.2.2 Beyond Cd-based QDs: infrared and non-toxic QD-LEDs . . . 50

3 Colloidal Quantum-Dots 55

3.1 Colloidal versus epitaxial QDs . . . . . . . . . . . . . . . . . . . . . . 55

3.2 Colloidal QD synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.3 Optical properties of QDs . . . . . . . . . . . . . . . . . . . . . . . . 58

4 Origin of efficiency roll-off in QD-LEDs 63

4.1 Role of QD luminescence in QD-LED efficiency . . . . . . . . . . . . 63

4.2 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

9

Contents 10

4.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 68

5 Bias-driven QD luminescence quenching in QD-LEDs 73

5.1 Bias-dependent QD quenching mechanisms . . . . . . . . . . . . . . . 73

5.2 Electric field-induced quenching in QD-LEDs . . . . . . . . . . . . . . 78

5.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

5.2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

5.2.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 80

5.2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6 Zero-Bias QD Luminescence Quenching in Infrared QD-LEDs 91

6.1 Bias-independent QD quenching mechanisms . . . . . . . . . . . . . . 91

6.2 Efficient infrared QD-LEDs using core-shell QDs . . . . . . . . . . . . 94

6.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

6.2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

6.2.3 Methods summary . . . . . . . . . . . . . . . . . . . . . . . . 97

6.2.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 99

6.2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

6.2.6 Detailed Experimental Methods . . . . . . . . . . . . . . . . . 106

7 Conclusion 111

7.1 Thesis summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

7.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Appendix A Supplementary information to Chapter 6 121

A.1 Calculation and error estimate of EQE and radiance . . . . . . . . . . 132

A.1.1 EQE calculation . . . . . . . . . . . . . . . . . . . . . . . . . 132

A.1.2 Radiance calculation . . . . . . . . . . . . . . . . . . . . . . . 133

A.1.3 EQE measurement error . . . . . . . . . . . . . . . . . . . . . 134

A.1.4 Device-to-device EQE variation . . . . . . . . . . . . . . . . . 134

Contents 11

A.2 QD film thickness calibration . . . . . . . . . . . . . . . . . . . . . . 134

12

List of Figures 13

List of Figures

1-1 Growth dynamics of carbon dioxide emissions. . . . . . . . . . . . . . 24

1-2 Joint influence of carbon intensity and energy intensity improvements

for limiting global warming. . . . . . . . . . . . . . . . . . . . . . . . 25

1-3 Evolution, approaches, and challenges of nitride-based LEDs. . . . . . 28

2-1 Tunable and pure colour light emission from colloidal quantum dots. . 33

2-2 Optical advantages of colloidal QDs for display and solid-state lighting

applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

2-3 Cost analysis for large-scale QD synthesis, example flowchart for syn-

thesis of core-shell PbS-CdS QDs. . . . . . . . . . . . . . . . . . . . . 40

2-4 Progression of orange/red-emitting QD-LED performance over time in

terms of peak EQE and peak brightness. . . . . . . . . . . . . . . . . 45

2-5 QD Excitation mechanisms. . . . . . . . . . . . . . . . . . . . . . . . 47

2-6 Modern QD-LED architecture and operation. . . . . . . . . . . . . . 50

2-7 State-of-the-art QD-LEDs. . . . . . . . . . . . . . . . . . . . . . . . . 52

3-1 Schematic of colloidal QD synthesis. . . . . . . . . . . . . . . . . . . . 56

3-2 Photoluminescence spectra and quantum yield during PbS-CdS QD

cation exchange reaction. . . . . . . . . . . . . . . . . . . . . . . . . . 59

3-3 Absorption spectra of PbSe QDs as a function of nanocrystal diameter. 61

4-1 Efficiency roll-off in different LED technologies. . . . . . . . . . . . . 66

4-2 Simultaneous EL-PL measurement setup. . . . . . . . . . . . . . . . . 69

4-3 Absorption spectra of QD-LED components. . . . . . . . . . . . . . . 70

List of Figures 14

4-4 EQE roll-off in a QD-LED and correlation with QD PL. . . . . . . . 71

4-5 Bias-induced QD PL quenching, normalised by optical excitation in-

tensity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5-1 Heat-induced QD PL quenching. . . . . . . . . . . . . . . . . . . . . . 74

5-2 Zero-bias and bias-driven QD PL quenching mechanisms. . . . . . . . 76

5-3 Thermal image of QD-LED under constant current bias. . . . . . . . 79

5-4 QD-LED EL spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

5-5 QD PL electric field-dependence measurement technique. . . . . . . . 81

5-6 QD-LED EL measurement setup. . . . . . . . . . . . . . . . . . . . . 82

5-7 Comparison of QD PL spectra and QD EL spectra at corresponding

peak emission energies, for three different biases. . . . . . . . . . . . . 84

5-8 Electric field-dependence of QD-LED PL and EL. . . . . . . . . . . . 85

5-9 Measured EQE and predicted EQE as a function of voltage. . . . . . 86

5-10 Transient PL of QDs under electric fields in a QD-LED. . . . . . . . . 88

5-11 Influence of QD electronic confinement on bias-driven PL quenching

mechanisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

6-1 Influence of QD surface passivation and electronic confinement on zero-

bias and bias-driven PL quenching mechanisms. . . . . . . . . . . . . 93

6-2 Infrared QD PL, as seen by eye and with an infrared camera. . . . . . 96

6-3 QD synthesis and QD-LED design. . . . . . . . . . . . . . . . . . . . 98

6-4 QD-LED performance using core-only versus core-shell QDs. . . . . . 101

6-5 Progression in efficiencies over time of visible- and near-infrared-emitting

QD-LEDs, and comparison with this work and with commercial near-

infrared LEDs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

6-6 Correlation of QD-LED EQE with QD PL quantum yields. . . . . . . 104

6-7 Transient spectroscopy of shell-dependent QD PL quenching. . . . . . 105

7-1 Progression of blue-emitting LED EQEs over time. . . . . . . . . . . 114

7-2 QD-LED operating lifetimes. . . . . . . . . . . . . . . . . . . . . . . . 120

List of Figures 15

A-1 Raw EQE-current density data with measurement errors and device-

to-device variations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

A-2 Current density-voltage behavior. . . . . . . . . . . . . . . . . . . . . 124

A-3 Test-to-test variations in peak EQE. . . . . . . . . . . . . . . . . . . 125

A-4 Day-to-day variations in raw EQE-current density data. . . . . . . . . 126

A-5 Radiance-current density characteristics. . . . . . . . . . . . . . . . . 127

A-6 Enhancement in radiant intensity efficiency. . . . . . . . . . . . . . . 128

A-7 EQE-current density characteristics of anomalous ‘champion’ QD-LEDs.129

A-8 Dependence of QD-LED EQE-shell thickness correlation on current

density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

A-9 Transient photoluminescence of QDs on glass. . . . . . . . . . . . . . 131

A-10 QD film thickness-concentration calibration curves on zinc oxide. . . . 136

16

List of Tables 17

List of Tables

2.1 Calculated large-scale chemical synthesis costs for red- and near-infrared-

emitting QDs and for two archetypal organic dyes. . . . . . . . . . . . 43

A.1 Summary of reported NIR QD-LED peak external quantum efficiencies

and power efficiencies. . . . . . . . . . . . . . . . . . . . . . . . . . . 123

A.2 PL lifetimes of spin-cast films of core-only (PbS) and core-shell (PbS-

CdS) QDs on ZnO, calculated from Fig. 6-7. . . . . . . . . . . . . . . 131

18

Chapter 1

Introduction

The generational challenges of climate change and global sustainable development

compel technological innovations in both renewable energy and energy efficiency. High

performance lighting is a key target for energy efficiency improvements and solid state

light emitting diodes are already revolutionising this field. In the quest for low cost,

large area, high colour-quality lighting, thin-film light emitting devices (LEDs) based

on nanostructured materials have emerged, and have already established themselves

as multi-billion dollar display markets. This thesis explores the use of visible and

infrared colloidal nanocrystal "quantum-dots" (QDs) as bright, wavelength-tunable

luminophores in optically- and electrically-driven LEDs.

Near the onset of this work, we conducted two comprehensive literature reviews

of the quantum-dot light emitting device (QD-LED) field [1, 2]. Our conclusion was

that from a device efficiency perspective, the three key challenges facing QD-LEDs

today are QD photoluminescence (PL) quenching, poor photon outcoupling, and a

limited understanding of the fundamental operating mechanisms contributing to the

recombination of electrons and holes, yielding light, in QD-LEDs. We also found

that the operational lifetime and cost of QD-LEDs must be addressed if they are to

become a commercial reality.

Improving device shelf life and reducing the cost of materials and manufacture are

surely the greatest hurdles to the commercialisation of existing QD-LEDs. Enhancing

the efficiency with which generated light is outcoupled from QD-LEDs is also a critical

19

Chapter 1. Introduction 20

step, yet it is expected that this may largely be a matter of extending to QD-LEDs

some of the many structural modifications that have been successfully implemented

in organic LEDs (OLEDs) [3, 4] and nitride-based LEDs [5]. Relatively speaking, all

three present predominantly manufacturing-based challenges on the R&D chain.

In contrast, better understanding carrier recombination and QD PL quenching in

QD-LEDs compels relatively fundamental research. As our overview of past QD-LED

research in Chapter 2 shows, substantial attention has been devoted to improving the

efficiency of carrier recombination - bringing electrons and holes together onto QD

emitters - through the engineering of four distinct generations of device architectures.

In this thesis, we instead focus on the last step of the light generation process, which

to date has received less dedicated attention: once an exciton has formed on a QD,

what processes dictate its probability of recombining to emit a photon? In particular,

we seek to characterise and, where possible, mitigate the influence of zero-bias and

bias-driven QD PL quenching on the performance of QD-LEDs. Indeed, the external

quantum efficiency (EQE) of QD-LEDs is directly proportional to QD PL quantum

yield (𝜂PL = number of photons emitted per photon absorbed), and in some devices it

is now the limiting factor in efficient light generation [6]. As we shall see, controlling

𝜂PL presents opportunities for engineering tailored materials solutions, allowing us to

make progress towards QD-LEDs with higher efficiency and brightness.

1.1 Thesis Organisation

Chapter 1 outlines the need for more energy efficient lighting technologies to help

decarbonise the global energy system and describes the challenges faced by white

LEDs and OLEDs. Chapter 2 introduces colloidal QDs and the benefits they bring to

lighting and display markets, and traces the evolution of QD-LEDs, from invention

through to state-of-the-art. It includes a quantitative cost analysis of large-scale

QD synthesis, which suggests QDs are cost-competitive with the U.S. Department of

Energy (DOE)’s solid state lighting (SSL) cost targets for OLEDs. Chapter 3 presents

a rudimentary background on the chemical approaches to QD synthesis relevant to this


thesis and the quantum mechanical origins of QDs’ fundamental optical properties.

Experimental methods are summarised at the start of each relevant chapter, and

where necessary, further details are presented at their end. We note that, expect

where explicitly described, the fabrication and characterisation tools employed in this

thesis are standard in the field, and the interested reader is directed to ref. [7–10],

which provide a basic contemporary overview of all of these techniques as used in our

laboratory.

Chapter 4 begins with an overview of the different processes that may contribute

to quenching of QD PL in QD-LEDs. This leads us to broadly categorise PL quench-

ing mechanisms as either ‘zero-bias’ (occurring even without bias) or ‘bias-driven’

(dependent on applied bias). While these quenching pathways have previously been

identified for isolated QDs, here, in situ device measurements allow us for the first

time to directly observe the detrimental impact of these mechanisms in operating

QD-LEDs, and to engineer solutions to mitigate these effects.

In the second part of Chapter 4, we focus on bias-driven PL quenching in QD-

LEDs. By simultaneously measuring the PL and EL of operating red-emitting QD-

LEDs, we demonstrate for the first time that the so-called ‘roll-off’ in efficiency typ-

ically observed at high applied biases in QD-LEDs can be wholly explained by a

simultaneous, bias-driven, reversible reduction in QD 𝜂PL. This understanding is

crucial to designing devices with high efficiencies at high brightness.

This leads us in Chapter 5 to explore different possible mechanisms for this bias-

dependent QD PL quenching. By measuring the electric field dependence of QD PL,

we are able to quantitatively predict the EQE roll-off in our devices and therefore to

deduce that it is almost entirely governed by electric field-induced QD PL quenching,

and not carrier leakage or QD charging, as is often the conventional wisdom. In situ

transient PL measurements of our QD-LEDs suggest that the cause of quenching

is either a decrease in radiative exciton recombination rate (for example, due to a

decrease in the overlap of electron and hole wavefunctions) [11] or a decrease in the

efficiency of thermalized-exciton formation [12,13].

Chapter 6 turns to zero-bias PL quenching in QD-LEDs. Using near-infrared


(NIR)-emitting QD-LEDs as a testbed of relatively unfulfilled technological potential,

we show that quenching of QD PL due to the QDs’ local device environment makes

𝜂PL a limiting factor in EQE. To address this, we fabricate a novel NIR QD-LED

architecture based on surface-passivated NIR-emitting QDs synthesised by overcoat-

ing core-only QDs with a wide band-gap shell. We show that this shell passivation

enhances in situ 𝜂PL by almost two orders of magnitude by mitigating quenching

caused by a neighbouring metal oxide charge transport layer. The correlated increase

in EQE, to more than double the previous record value, yields the most efficient

thin-film NIR light emitting device ever reported.

Finally, Chapter 7 presents a summary of this work, as well as an outlook for

QD-based light emitting technologies. Note that Chapter 2 draws substantially from

our work published in ref. [1, 2]. Chapters 4 and 5 appear in ref. [14]. Chapter 6 is

based on our work published in ref. [15] (for patent application see ref. [16]).

1.2 Energy efficiency in lighting and displays

A few months before the submission of this thesis, at the end of 2015, the world’s

nations adopted the Paris Agreement [17]. As part of the United Nations Frame-

work Convention on Climate Change, this treaty commits countries to "Holding the

increase in the global average temperature to well below 2 ∘C above pre-industrial

levels and to pursue efforts to limit the temperature increase to 1.5 ∘C above pre-

industrial levels, recognizing that this would significantly reduce the risks and impacts

of climate change". Just a few months earlier, the United Nations General Assembly

also adopted the global Sustainable Development Goals, which include "Ensur[ing]

access to affordable, reliable, sustainable and clean energy for all" and "Tak[ing] ur-

gent action to combat climate change and its impacts" [18]. As it happened, 2015 was

also the United Nations’ International Year of Light and Light-Based Technologies,

just months after the 2014 Nobel Prize in Physics was awarded "for the invention of

efficient blue light-emitting diodes which has enabled bright and energy-saving white

light sources" [19]. The prospect of designing and deploying high efficiency lighting


technologies to improve energy efficiency motivates the - necessarily fundamental -

research of novel QD-based light-emitting devices described in these pages.

Climate policy targets intending to limit the global mean temperature increase to

2 ∘C will likely require that greenhouse gas emissions peak within the next decade,

and then fall rapidly to zero (or below) well before the end of this century. While

much focus rightly falls on supply-side strategies for reducing the carbon intensity

of energy production technologies, the Kaya Identity (which relates greenhouse gas

emissions to the complex, self-reinforcing interactions between carbon intensity, en-

ergy intensity, population, and affluence) makes clear that demand-side energy effi-

ciency also has a vital role to play [20]. Fig. 1-1 shows how the exponential rise in

global greenhouse gas emissions since the Industrial Revolution has been driven by

an even faster exponential growth in energy use, only partially offset by a gradual

decline in carbon intensity [21]. Indeed, it is an inescapable theme of decarboniza-

tion scenarios that three fundamental energy system transformations are required:

fuel-switching (electrification); decarbonization of electricity; and energy efficiency

and conservation [20,22–27]. As Fig. 1-2 shows, ensuring a reasonable probability of

holding global warming below 2 ∘C requires on the order of a doubling in the rate

of energy intensity improvement in tandem with improvements in carbon intensity.

Energy efficiency improvements play an even more vital role in lower (1.5 ∘C) sta-

bilisation scenarios [28]. The Intergovernmental Panel on Climate Change projects

that efficiency improvements will be especially important in the near term, prior to

roughly 2030, compared to supply-side measures [27]. Buildings, in particular, have

the largest low-cost potential for improved energy efficiency [27]. Lighting, which is

the second largest user of building energy, and which accounts for roughly 19 % of

global electricity consumption (conservatively expected to grow by 80 % by 2030) [29]

and 6.5 % of the world’s greenhouse gas emissions [30], has therefore been identified

as a key area for early energy intensity reductions [26]. The U.S. DOE, for example,

is coordinating a major lighting research effort, and is directed by the Energy Policy

Act of 2005 to systematically accelerate SSL development [31]. Pertaining to the

U.N.’s Sustainable Development Goals, another key driver for lighting innovation is


Figure 1-1: Growth dynamics of carbon dioxide emissions. Global anthro-pogenic carbon dioxide (CO2) emissions, U (1015 g C yr−1) from fossil fuel andland-use change, with exponential trend ln𝑈 = 𝛼(𝑡 − 𝑡1); (𝛼 = 0.0179 ± 0.0006yr−1, 𝑡1 = 1883 ± 1.7). Global primary energy use, E (1018 J yr−1) with trendln𝐸 = 𝜂(𝑡 − 𝑡1); (𝜂 = 0.0238 ± 0.0008 yr−1, 𝑡1 = 1775 ± 3.5). Carbon intensity ofglobal primary energy, 𝑐 = 𝑈/𝐸 (g C 106 J−1) with trend ln𝑐 = (𝛼 − 𝜂)(𝑡 − 𝑡1).Adapted from ref. [21].

the fact that 1.6 billion people today lack access to reliable lighting. Nearly all of

them instead rely on kerosene lamps, which account for 1 % of global lighting yet

20 % of lighting-based CO2 emissions [32], and which are responsible for 1.5 million

deaths per year [33].

To this end, almost all OECD (Organisation for Economic Co-operation and De-

velopment) governments have implemented policies aimed at phasing out incandes-

cent lighting [35], which converts electrical energy to light with an efficiency of just 5

%, and with an efficacy of roughly 15 lm/W [32]. Fluorescent tubes (efficiency ∼25

%, efficacy ∼60-100 lm/W), and increasingly compact fluorescent lamps (efficiency

∼20 %, efficacy ∼35-80 lm/W), have come to provide the bulk of global lighting,

but owing to their mercury content and inferior efficacy relative to LEDs (discussed

below), they are perceived by many as a stop-gap measure to replace incandescent

lamps and their derivatives (halogen and high-intensity discharge lamps) until more

efficient, non-toxic white light sources become available and cost effective [29,30].


Figure 1-2: Joint influence of carbon intensity and energy intensity improve-ments for limiting global warming. The plot, adapted from ref. [34], shows thejoint influence of these two factors on limiting warming to below 2 ∘C during thetwenty-first century, in a large illustrative set of scenarios (𝑛 > 100). Average globalrates are for the period 2010 to 2050. The yellow star indicates the historically ob-served rate between 1971 and 2005, and six-pointed stars indicate the values foundin the SRES (Special Report on Emissions Scenarios - see ref. [34]) scenarios. SRESmarker scenarios are highlighted in red. The arrow illustrates the direction of in-creasing climate protection. Other symbols are colour-coded as a function of theirprobability of limiting warming to below 2 ∘C, and the shape of the symbols reflectsthe base level of future energy demand assumed in the scenarios (diamond: high;circle: intermediate; square: low). Note that whereas all scenarios in ref. [34] as-sume a consistent evolution of climate mitigation over the course of the full century,the SRES scenarios represent baseline scenarios without climate mitigation. Thelong-term development in the SRES scenarios might thus differ from their short-termtrends.


SSL based on inorganic LEDs continues to emerge as the most likely contender.

Akin to Moore’s Law for transistors, SSL development has progressed exponentially

for more than 40 years according to Haitz’ Law; each decade exhibiting a ten-fold

drop in cost-per-lumen and a twenty-fold rise in light generated per LED package

(for a given wavelength of emission) [30]. Sales of high-brightness LEDs have had a

compound annual growth rate of over 46 % since 1995 [29]. Whereas CFL efficacy

has lately remained flat at ∼35-80 lm/W, in just the last part of 2014, the aver-

age efficacy of LEDs increased from ∼70 lm/W up to almost 100 lm/W [36]. And

unlike incumbent technologies, LEDs are not limited by fundamental physics-based

limitations that prevent more than incremental improvements (for example, incan-

descents are limited by the laws governing a blackbody radiator) [30]. For SSL to

realise decisive energy efficiency benefits, however, even higher performance levels

must be achieved. The U.S. DOE’s Multi-Year Program Plan for SSL aims to re-

duce lighting-based energy use by 40-60 % by 2030 (2,216-3,900 TWh), equivalent to

roughly $220-380 billion in energy savings [37]. To do so, a stretch goal of 200 lm/W

by 2030 has been set, which envisions LED lighting virtually eliminating the use of

high intensity discharge sources (e.g. sodium lamps) and reducing the installed base

of linear fluorescent lamps to one-third of its current share. LED lighting sales (in

lumen-hours) would need to increase in the U.S., for example, from less than 4 % in

2014 to 88 % in 2030 [31].

Yet in most regions of the world right now, even with government policy sup-

port, fewer than 10 % of existing lighting installations (installed base) use SSL prod-

ucts [31]. A major research challenge of reaching the above goals is that white LEDs

(WLEDs) based on SSL must achieve such high efficacies while simultaneously deliv-

ering exceptional colour quality at low cost. Fig. 1-3b highlights three approaches

that are being explored to achieve these aims. The single-chip phosphor-conversion

approach based on a blue LED (InGaN, for example) backlight and a yellow or green

(and sometimes red, too; Fig. 1-3b, left) down-converting phosphor [38] are presently

the commercial technology of choice, but suffer from a tradeoff between efficacy and

colour quality (quantified by colour rendering index, CRI, explained in Section 2.1.1)


(Fig. 2-2b). This is due to the absorption and emission losses of the phosphor(s) [39],

as well as a 20-25 % energy loss because of the Stokes shift from a blue pump to lower

energy phosphor(s) [30]. The optimisation of phosphors, especially narrow-band red-

emitting ones with less thermal instability, is a key step to achieving efficacies truly

surpassing those of fluorescent lamps while maintaining high CRI [39,40]. Ultimately,

however, the Stokes shift may preclude the realisation of very high efficacy (∼200

lm/W) WLEDs [29]. An alternative avenue for achieving higher CRI, illustrated in

Fig. 1-3b, right, is the use of multi-chip red, green, blue (and sometimes yellow, too:

RYGB) WLEDs. But this approach is simultaneously made complicated and costly

by various aspects of assembly, colour mixing, and feedback drive circuitry required to

continuously balance the different colour emitters that age at different rates [38, 39].

Critically, RYGB WLEDs also face the decade-long challenge of improving the low

efficiencies of LEDs in the ‘green-yellow gap’ wavelengths [40,41]. To attain long-term

SSL performance targets, high LED internal quantum efficiencies of > 90 % will need

to be achived at particular green and yellow wavelengths [39]. Another LED chal-

lenge, relevant to both phosphor conversion and multi-chip approaches, is efficiency

‘droop’, which is the fall in efficiency of (nearly all GaN-based) LEDs of even 70 %

or more at desired high operating currents. This phenomenon is discussed in detail

in Chapter 4.

Meanwhile, OLEDs have emerged as a potential complimentary white light tech-

nology owing to their prospects as flexible, thin, lightweight, large-area, and diffuse

light sources. For example, ultrathin, large OLED panels can provide the same light

output as a compact GaN-based LED while reducing the brightness per unit area, re-

sulting in a glare-free light source [42]. However, despite research progress, in practice,

the development of OLEDs for lighting has lagged behind that of nitride-based LEDs.

White OLEDs for lighting are only just nearing commercial availability because it has

not been possible so far to simultaneously achieve high brightness at high efficiency

with long lifetime [29]. For example, the relatively broad line-width of red OLED

emission makes it difficult to achieve excellent colour quality and high efficacy at the

same time [31]. Another well-known roadblock to the commercialisation of OLEDs is


Figure 1-3: Evolution, approaches, and challenges of nitride-based LEDs.a, Haitz’ Law evolution of LEDs: (filled) Exponential rise in the (highest) flux-per-lumen of commercially available red and cool-white LEDs between 1968 and 2010;(unfilled) exponential fall in the (lowest) original equipment manufacturers (OEM)cost-per-lumen of commercially available red and cool-white LEDs between 1973 and2010. Adapted from ref. [30]. b, Approaches for generating white light from LEDsand their corresponding power spectra: (left) single-chip blue LED pumps red andgreen phosphors; (right) multi-chip RYGB WLED. Adapted from ref. [39]. c, EQEsof various state-of-the-art LEDs based on two material systems, demonstrating the‘green-yellow’ efficiency gap. InGaN: (1) Thin-Film Flip-Chip LEDs; (2) InGaNvertical thin-film LED; and (3) conventional chip LED. AlGaInP: (4) Truncated-inverted-pyramid LEDs. 𝑉 (𝜆) is the luminous response curve of the human eye.Reproduced from ref. [39].


the particular challenge of achieving deep blue electrophosphorescence with combined

high efficiency, brightness, and long-term operational stability [43]. Deep-blue emis-

sion is crucial to realizing a wide color gamut for higher colour quality SSL. In terms

of efficiency, while fluorescent organic dyes can certainly achieve deep blue light, they

show comparatively much lower efficiencies than their phosphorescent counterparts.

Conversely, phosphorescent dyes enable far higher efficiencies, but with only sky-blue

or bluish-green emission [44, 45]. Likewise, achieving high brightness in deep blue

phosphorescent OLEDs leads to a commensurate roll-off in efficiency [43]. And unlike

red- and green-emitting OLEDs, achieving industry-standard device lifetimes in blue

devices currently requires the use of only inefficient fluorescent emitters [42].

Overall, there has been a gradual market penetration of inorganic LEDs from

mobile applications to displays to general lighting [46]. This is because many of the

challenges faced by WLED lighting are similar to those in developing display tech-

nologies - both requiring high brightness, multi-colour emitters. As Haitz et al. have

observed in mapping the evolution of SSL, applications like displays provide critical

"stepping stone" markets that help drive the innovation and cost reductions necessary

for ultimately penetrating the general illumination sector [30]. OLED displays, too,

are already an established multi-billion-dollar market reality [42]. This is in part be-

cause of more aggressive, profit-driven research funding in the display market [31], but

also because of their low energy consumption, wide viewing angles and high contrast

values (in comparison to back-lit liquid crystal displays (LCDs)), and compatibility

with high-resolution patterning.

Colloidal QDs offer a possible solution to some of the challenges currently fac-

ing the design and manufacture of LEDs for lighting and display applications, as

we describe in the next chapter. These solution-processed semiconducting nanocrys-

tals boast unique size-dependent optical properties that have motivated increasingly

active research aimed at applying them in the next generation of optoelectronic and

bio(medical) technologies. Since the first directed QD synthesis three decades ago, use

of QD thin-films has been demonstrated in a range of optoelectronic devices includ-

ing LEDs [47–50], photovoltaics (PVs) [51], photodiodes [52], photoconductors [53],


and field-effect transistors [54], while QD solutions have been used in a myriad of

in vivo and in vitro imaging, sensing, and labelling techniques [55]. In passing,

we note that the research of QD-LEDs therefore not only extends the frontiers of

lighting and display technologies, but can offer fundamental insights into the photo-

physical properties of QDs, which can help inform sister technologies like QD PVs.

The infrared-emitting QD-LED architecture described in this thesis, for example, is

extremely similar to that used by some of today’s most efficient QD PVs [56].

Chapter 2

Quantum-Dots as Luminophores

2.1 Benefits for solid state lighting and displays

2.1.1 Tunable and pure colours

Colloidal quantum-dots (QDs) are solution-processed nanoscale crystals of semicon-

ducting materials. They comprise a small inorganic semiconductor core (1-10 nm

in diameter), often a wider-bandgap inorganic semiconductor shell, and a coating

of organic passivating ligands (Fig. 2-1b, insets). They emit bright, pure and tun-

able colours of light, making them excellent candidates for colour centers in next-

generation display and SSL technologies. The greatest asset of QDs for light-emitting

applications is their tunable bandgap, governed by the quantum size effect. Confine-

ment of electron-hole pairs (excitons) on the order of the bulk semiconductor’s Bohr

exciton radius (5.6 nm for CdSe, a common QD material) leads to quantisation of

bulk energy levels, resulting in atomic emission-like spectra. Another result of this

confinement is that as its size decreases, the QD’s bandgap increases, leading to a

blue shift in emission wavelengths [57]. This is shown in Figure 2-1a, which also illus-

trates how this spectral tunability can be extended through changes in QD chemical

compositions and stoichiometries [58, 59]. Such systematic and precise spectral tun-

ability of efficient emission, even in the near-infrared (NIR) region, is a distinguishing

and significant technological advantage of QDs over organic dyes, which we exploit

31

Chapter 2. Quantum-Dots as Luminophores 32

in engineering the NIR QD-LEDs described in this thesis. The quantum mechanical

origins of this colour tunability are discussed further in Chapter 3. CdSe-based core-

shell QDs are currently the material of choice in visible QD-LEDs [58,60–62] and lead

chalcogenide QDs dominate NIR devices [63,64]. In the visible, the spectrally-narrow

emission of QDs (see Figure 2-1b; full width half maximum (FWHM) ∼30 nm for

CdSe) [65] compared with those of inorganic phosphors (FWHM ∼50-100 nm) [66]

identify QDs as outstanding luminescent sources of saturated emission colour.

This high colour quality can be quantified using the Commission International

de l’Eclairage (CIE) chromaticity diagram (Figure 2-2a), which maps colours visible

to the human eye in terms of hue and saturation. By combining the emission of

three light sources, such as red, green, and blue (RGB) emissive display pixels, a set

of apparent colours can be generated corresponding to the colours enclosed by the

triangle on the CIE diagram defined by the coordinates of the three pixels. Figure

2-2a shows that, with the highly saturated colours of QD emission, it is possible

to select RGB QD-LED sources whose subtended colour gamut is larger than that

required by high-definition television (HDTV) standards (dashed line) [67].

Broad spectral tunability also allows a more controlled combination of colours,

such that higher quality white light, with a precisely tailored spectrum, can be gener-

ated. The white light’s quality can be measured in terms of correlated colour temper-

ature (CCT) and colour rendering index (CRI), which compare LED emission with

that from the Sun (the ‘ideal’ white light source, with a CRI of 100). Conventional

white LEDs, comprising a blue inorganic LED backlight coated with a yellow phos-

phor optical down-converter (see Fig. 1-3), typically exhibit a cool bluish emission,

characteristic of high CCTs (>5000 K) and low CRIs (mostly between 80 and 85), as

shown in Figure 2-2b. For lower CCT lights (for example, 2700 K) it is particularly

hard to simultaneously maintain high luminous efficacy and high colour quality be-

cause the required red luminophores must have relatively narrow emission spectra so

as to avoid photon loss as infrared emission. The emission spectra of conventional red

phosphors is unfortunately too broad (>60 nm FWHM) to avoid this loss. In contrast,

the narrow spectral emission (∼30 nm FWHM) of QDs in QD Vision Inc.’s Quan-


Figure 2-1: Tunable and pure colour light emission from colloidal quantumdots. a, Solutions of colloidal QDs of varying size and composition, exhibiting PLunder optical (ultraviolet) excitation [58]. b, PL spectra of CdSe-ZnS and PbS-CdScore-shell colloidal QDs [67, 68]. The upper inset shows a schematic of a typicalcore-shell colloidal QD. The lower inset is a high-resolution transmission electron mi-croscope image of a CdSe QD (scale bar, 1.5 nm). a demonstrates the size- andcomposition-dependent tunability of QD emission colour, whereas b shows the exten-sion of this narrowband emission into the near-infrared. QD-LED electroluminescencetypically closely matches corresponding PL spectra.


tum LightTM optic offer supplementary and more selective optical down-conversion

of some of the backlight’s bluer emission (generated by Nexxus Lighting, Inc. LED

lighbulbs) into redder light, leading to a CRI >90 % and a superior CCT of 2700 K,

while maintaining a very high 65 lm/W efficacy [67] (see Figure 2-2b for a comparison

with other LED light sources). QDs thus enable higher colour quality and, accord-

ingly, lower power consumption in SSL sources. Through analogous approaches, QDs

can be utilised as backlights in high-colour-quality liquid crystal displays (LCDs). In

fact, Sony’s 2013 line of Triluminos LCD televisions use edge-mounted red and green

QDs from QD Vision, Inc. to optically downconvert some of the television’s blue LED

backlight (absorbing some of the blue light and re-emitting it as red and green light),

optimising its colour balance so that it fulfills >100 % of the NTSC colour television

standard gamut, compared with ∼70 % for conventional LCD screens [69, 70]. The

result is a television picture with colour quality comparable to that of organic LED

(OLED) screens, but achieved at the cost of an LCD display [2].

2.1.2 Bright emission

Overcoating QDs with wider bandgap inorganic semiconductor shell(s) (Figure 2-

1b, inset) has been shown to dramatically enhance their PL quantum yield (𝜂PL)

and photostabilities. Overcoating passivates surface non-radiative recombination sites

more effectively than organic ligands alone, and, simultaneously, shifts the electron

wavefunction, confining excitons to the QD core, away from surface trap states [71–

73]. For example, solutions of CdSe-ZnS core-shell QDs can be routinely synthesised

with 𝜂PL of between 30 % and 95-plus %, almost one order of magnitude greater than

those of native CdSe cores [74]. As noted above, the extendibility of QD emission

into the NIR is of significant technological interest. Whereas organic molecules have

negligible optical activity beyond wavelengths of 𝜆=1 𝜇m (𝜂PL is <5 % at these

wavelengths) and exhibit poor chemical and photostability, QDs are relatively stable

and retain 𝜂PL >50 % throughout the NIR [74–78]. As with visible-emitting QDs, this

high NIR brightness is owing in large part to improvements in 𝜂PL with overcoating

[75].


Figure 2-2: Optical advantages of colloidal QDs for display and solid-statelighting applications. a, CIE chromaticity diagram showing that the spectralpurity of QDs enables a colour gamut (dotted line) larger than the high-definitiontelevision standard (dashed line). b, Plot showing the luminous efficacy and colourrendering index (CRI) of various lighting solutions. The first commercial QD-basedsolid-state light source, developed by QD Vision and Nexxus Lighting, consists ofsheets of red QDs backlit by a blue LED with a yellow phosphor coating, resultingin a high CRI without compromising high luminescence efficacy. Recently, Philips’A-Style LED, which employs remote phosphors, has demonstrated even more energy-efficient lighting. There is evidently an emerging market for high-quality opticaldownconverters, such as QDs.


Since QD-LEDs often comprise neat films of QDs, it is their 𝜂PL in this close-

packed form that then dictates maximum device efficiencies. For core-only QDs in

solution, 𝜂PL is typically reduced by one to two orders of magnitude when deposited

as thin films [79]. Evidence suggests that this ‘self-quenching’ results from efficient

non-radiative Förster Resonant Energy Transfer (FRET) of excitons within the in-

homogeneous size distribution of QDs to non-luminescent sites [6,80–82], where they

recombine non-radiatively [83]. It follows from the very strong inter-dot spacing de-

pendence of FRET efficiency (it decreases as spacing increases) that QD ligand length

and shell thickness can profoundly impact the degree of QD ‘self-quenching’. Thin

films of core-shell CdSe-ZnS QDs with long oleic acid ligands, for example, typically

retain 𝜂PL of 10-20 %, which directly affects the external quantum efficiency (EQE)

of QD-LEDs containing those QD films. The EQE of a QD-LED is defined as the

ratio of the number of photons emitted by the LED in the viewing direction to the

number of electrons injected. It may be expressed as:

EQE = 𝜂r𝜒𝜂PL𝜂oc (2.1)

where 𝜂r is the fraction of injected charges that form excitons in QDs, 𝜒 is the fraction

of these excitons whose states have spin-allowed optical transitions, 𝜂PL is the QD

PL quantum yield resulting from these transitions, and 𝜂oc is the fraction of the

emitted photons which are coupled out of the device. The internal quantum efficiency

(IQE) is the efficiency of the charge recombination process, independent of 𝜂oc (i.e.

IQE = EQE/𝜂oc). A full description of how EQE is defined and measured in our

experiments can be found in Appendix A, Section A.1.

It is also technologically significant that for CdSe QDs, for example, 𝜒 ≈ 1,

identical to that of the most efficient organic phosphors, which are used in high-

efficiency organic LEDs (OLEDs) [84]. In CdSe QDs, the high 𝜒 is a result of the

small energetic separation (<25 meV) of the ‘bright’ and ‘dark’ band-edge excitonic

states [85], which have spin-allowed and spin-forbidden transitions to the ground

state, respectively. Thermal mixing at room temperature enables efficient crossing of


excitons from dark states to higher energy bright states, leading to a high effective 𝜒.

With 𝜒 ≈ 1, IQE is then limited by 𝜂r𝜂PL. The greatest hurdle to achieving high

efficiency QD-LEDs is the simultaneous maximisation of 𝜂r and 𝜂PL. QD-LEDs have

been reported that appear to demonstrate PL-normalised IQEs (that is, IQE/𝜂PL)

approaching (at least within a factor of two) 100 %, and that are therefore limited

only by 𝜂PL [6,58,63,86,87]. This motivates the focus of this thesis on understanding

and controlling the quenching of QD PL in QD-LEDs.

2.1.3 Solution processable

QD surface ligands confer solubility in a variety of organic solvents. This enables the

use of low cost QD deposition techniques such as spin-coating [88], mist coating [89],

inkjet printing [90,91], and microcontact printing [92,93]. Ligands can also be chosen

[50] (or cross-linked post-deposition [94, 95]) to enable the deposition of subsequent

materials in orthogonal solvents. These methods have led to, for example: organic-

QD hybrid structures [96]; molecular length-scale control of dot-to-dot separation

[97]; QDs deposited on curved surfaces [98]; QD monolayers [88]; QD multilayer

superstructures [99]; and one-dimensional chains [100].

2.1.4 Stable

It is commonly attested that the photostability of QDs exceeds that of organic chro-

mophores, and that this advantages their application in LEDs. Yet oxidation of QDs

has been seen to cause spectral diffusion (blue-shifting) and PL quenching in both

single QDs [101, 102] and ensembles of QDs [103]. Exposure to light generally exac-

erbates these effects through photo-oxidation and photo-bleaching [102], though sub-

stantial photo-brightening (increased 𝜂PL following exposure to light) has also been

observed [104–106]. Beyond the presence of oxygen, these phenomena have been

found to be critically dependent on an array of factors including humidity [107,108],

QD film geometry [109], and the duration [103, 106], intensity, and wavelength [109]

of illumination.


Nevertheless, QD shells markedly improve photostability [102] by passivating sur-

face traps, by confining excitons to QD cores, and by hindering the diffusion of oxygen,

for example, into QD cores. Moreover, thick inorganic multishells [110,111], surface-

passivating ligands [112], and radially graded alloyed shells [113] have been shown to

heavily attenuate, and even entirely suppress the PL intermittency phenomenon of

CdSe QDs known as ‘blinking’. Reductions in blinking are relevant to QD-LEDs be-

cause they translate to higher ensemble 𝜂PL [114]. Talapin et al. recently synthesised

QDs with inorganic molecular metal chalcogenide ligands [115] and with metal-free

ionic ligands [116], relieving QDs of instabilities associated with photo-damage of or-

ganic ligands [110]. We caution that many of the above studies, however, are based

on single-QD spectroscopy at cryogenic temperatures.

Overall, at least as bioanalytical labels, QDs are proving to be more photostable

than organic dyes [117]. Whether this comparison could hold good in LEDs, however,

is yet unclear. Tremendous opportunities exist to better understand and improve the

longevity of QDs in QD-LEDs by investigating the chemistry and photo-physics of

films of QDs under operating conditions [118].

2.1.5 Cost competitive

From a manufacturing standpoint, QD-LEDs may be approximated as QD-enhanced

OLEDs. The manufacturing cost of QD-LEDs can be broadly divided into the cost of

raw materials and the fabrication costs of processing these materials. The similarity

of the constituent materials of QD-LEDs and OLEDs means that they are fabri-

cated using a similar toolbox of thin-film processing techniques, so that QD-LED

commercialisation would benefit from the manufacturing infrastructure and expertise

developed for OLED production. Aside from the QDs themselves, the materials typ-

ically employed in QD-LEDs (metals, metal oxides and organic small molecules) are

also very similar to those found in OLEDs. Their materials costs should therefore be

commensurate with those that are enabling the growth of OLED markets, and would

benefit from their economies of scale.

To estimate the materials costs of QDs typical in light-emitting applications when


produced in large quantities, we perform a quantitative analysis based on published

small-scale synthetic procedures, which colleagues have previously used to asses the

commercial viability of organic materials for photovoltaics [119]. We consider a few of

the most common and promising QD-LED materials and synthetic preparations: red-

emitting ‘legacy’ CdSe QDs [65], ‘modern’ CdSe QDs [120], ‘legacy’ CdSe-ZnS core-

shell QDs [72], and ‘modern’ CdSe-CdS core-shell QDs [120] (all with trioctylphos-

phine oxide (TOPO) ligands); NIR-emitting PbS-CdS core-shell QDs (with oleic acid

ligands); and, lastly, PbS QDs (with oleic acid ligands), which are not only commonly

used in NIR QD-LEDs but have also garnered tremendous interest as an active ma-

terial in QD-based solar cells. The ‘legacy’ and ‘modern’ labels refer to the synthetic

recipe evaluated, as discussed below. As detailed in ref. [119], our cost analysis takes

into account all of the material inputs to these procedures in order to estimate the

total material costs for each type of QD (based on our assembled database of quota-

tions from major chemical suppliers for each of the input materials). As an example,

our model for the synthesis of PbS-CdS is represented graphically as a flowchart in

Figure 2-3. The first box in the flowchart represents the starting material, lead (II)

oxide. Red arrows indicate reagents, green arrows indicate solvents, and blue ar-

rows indicate additional materials required for workup and purification (‘crash out’).

The indicated quantities of input materials and waste are calculated to produce one

kilogram of product.

The materials cost results from our models for each synthetic procedure are sum-

marised in Table 2.1. We note that the materials costs that we consider represent

only one component of the overall cost to produce these materials. In the case of

pharmaceutical drugs, for example, materials only account for 20-45 % of the cost

of drug synthesis. The balance includes contributions for labor, capital, utilities,

maintenance, waste treatment, taxes, insurance, and various overhead charges [119].

We find that the materials costs of visible-emitting ‘legacy’ CdSe QDs ($569-660

g-1; lower value is without workup, upper value is with workup) exceed those of ‘mod-

ern’ CdSe QDs ($58-59 g-1) by an order-of-magnitude as a consequence of similarly

sizeable differences in the costs of reagents, solvents, and workup materials. This


Figure 2-3: Cost analysis for large-scale QD synthesis, example flowchartfor synthesis of core-shell PbS-CdS QDs. The flowchart describes the synthesisof 1 kg of PbS-CdS core-shell QDs. The requisite quantities of (red arrow) reagent,(green arrow) solvents, and (blue arrow) workup (‘crash out’) materials are indicatedfor this single-step process. Note that a quantitative yield is assumed, as discussedin the caption of Table 2.1.

reflects the 20-year evolution in synthetic procedures that has led to the use of signif-

icantly smaller quantities of more economical input materials, notably a significantly

cheaper and air-stable source of Cd (cadmium oxide replaces dimethyl cadmium).

These advances carry forward to the ‘modern’ CdSe-CdS core-shell QDs ($61-65 g-1),

which cost only fractionally more than their core-only equivalents, again owing to

the use of an economical source of Cd for the shell. In contrast, the ‘legacy’ CdSe-

ZnS core-shell QDs ($1884-1996 g-1) inherit the ten-fold higher costs of their starting

CdSe QDs and require the use of an expensive source of zinc for their shell (replacing

dimethyl zinc is key to lower costs). It is possible that reports of QD costs of up

to $10,000 g-1 (ref. [69]) may result from evaluation of antiquated ‘legacy’ syntheses

rather than more economical state-of-the-art approaches. The materials costs of NIR-

emitting QDs (PbS, Method 1: $18-29 g-1; PbS, Method 2: $45-68 g-1; and PbS-CdS:


$68-97 g-1) are roughly commensurate with those of the ‘modern’ visible-emitting

QDs.

Current target prices for QDs synthesised via scaled-up continuous processes are

∼$10 g-1 (ref. [121]). As a guide, the materials cost of Alq3 - an archetypal organic

dye used in OLEDs since the 1980s, and therefore subject to considerable economies

of scale - is ∼$4 g-1 (ref. [119]). However, most heavy-metal-based phosphors found in

high-performance OLEDs are considerably more expensive. A representative example

is Ir(ppy)2(acac) (ref. [122]), for which we calculate a materials cost of $658-1297 g-1.

Nevertheless, direct comparison of the materials costs of QDs with those of organic

dyes is complicated by the specifics of a given application, which determine how much

material is consumed. Considerations include whether it is used at a neat film or as a

dopant dispersed in a host matrix, the thickness of such a film, and the wastefulness of

the deposition technique employed. As is to be expected, the first QD-based products

address optical down-conversion by using a compact edge-mounted geometry [70] that

requires relatively small amounts of QDs (often dispersed to maximize 𝜂PL).

One way to try to assess our results is to translate them into approximate materials

costs-per-unit-area (Table 2.1). The per-area costs are based on the assumption that

a typical QD-LED might comprise a 25 nm film of hexagonally close packed QDs

separated by ∼0.5 nm due to their surrounding organic ligands. As expected, we

obtain similar values for ‘modern’ CdSe ($3 m-2) and CdSe-CdS ($4 m-2) QDs as

for PbS (Method 1: $1-2 m-2; Method 2: $3-5 m-2) and PbS-CdS ($4-6 m-2) QDs.

‘Legacy’ CdSe and CdSe-ZnS QDs are significantly more expensive ($35-41 m-2 and

$90-95 m-2, respectively). Especially given the likelihood that QD syntheses will

be subject to some economies-of-scale [69], this is competitive, for example, with

the DOE’s 2020 SSL target cost of organic materials of $10-15 m-2 for OLEDs [123].

Moreover this is just one possible metric, which does not necessarily reflect the higher

materials costs that luxury items such as displays might be able to shoulder.

One significant assumption that has so far been made, however, is that we can

deposit QDs with 100 % efficiency. In reality, the spin-casting technique (and therefore

microcontact printing, which in published studies involves a spin-casting step) that


has so far dominated laboratory demonstrations of QD-LEDs wastes ∼95 % of the

starting solution [124]. Unless the QD waste is recyclable, the associated 20-fold

increase in QD materials costs could render some applications economically unviable.

This points to the importance of developing low-waste QD-deposition techniques.

For example, as mentioned earlier, inkjet printing of multi-coloured pixel arrays of

QDs for both down-conversion [125] and RGB QD-LED [124] technologies has been

demonstrated, but further refinements in film quality and device performance are

required.


Tabl

e2.

1:C

alcu

late

dla

rge-

scal

ech

emic

alsy

nthes

isco

sts

for

red-

and

nea

r-in

frar

ed-e

mit

ting

QD

san

dfo

rtw

oar

chet

ypal

orga

nic

dye

s.

Com

pou

nd

Ref

.St

eps

Rea

gent

sSo

lven

tW

orku

pTo

tal(

w/o

wor

kup)

Tota

l(w

/w

orku

p)

($g-

1 )($

g-1 )

($g-

1 )($

g-1 )

($m

-2)

($g-

1 )($

m-2)

Vis

ible

QD

s

‘Leg

acy’

CdS

e(T

OP

O)

QD

[65]

129

8.22

271.

1490

.82

569.

3635

.35

660.

1840

.99

‘Mod

ern’

CdS

e(T

OP

O)

QD

[120

]1

42.4

115

.30

1.62

57.7

13.

3559

.34

3.44

‘Leg

acy’

CdS

e-Zn

S(T

OP

O)

QD

[72]

11,

613.

1327

1.14

111.

481,

884.

2710

5.37

1,99

5.76

111.

60‘M

oder

n’C

dSe-

CdS

(TO

PO

)Q

D[1

20]

145

.23

16.1

23.

2461

.34

4.11

64.5

94.

33IR

QD

s

PbS

(OA

)Q

D,M

etho

d1

[126

]1

14.6

63.

2511

.08

17.9

01.

2929

.98

2.09

PbS

(OA

)Q

D,M

etho

d2

[75]

136

.00

8.99

22.5

144

.99

3.24

67.5

04.

86P

bS-C

dS(O

A)

QD

*[7

5]1

49.6

218

.39

29.2

668

.01

4.27

97.2

76.

11O

rgan

ics

Alq

3[1

27]

10.

440.

003.

900.

44-

4.34

-Ir

(ppy

) 2(a

cac)

[128

]1

621.

6635

.86

639.

2665

7.52

-1,

296.

78-

Cos

t-pe

r-gr

amfo

r‘le

gacy

’C

dSe

QD

s(w

ith

trio

ctyl

phos

phin

eox

ide

(TO

PO

)lig

ands

);‘m

oder

n’C

dSe

QD

s(T

OP

Olig

ands

;w

urtz

ite-

CdS

esy

nthe

-si

s);

‘lega

cy’

CdS

e-Zn

Sco

re-s

hell

QD

s(T

OP

Olig

ands

);‘m

oder

n’C

dSe-

CdS

core

-she

llQ

Ds

(TO

PO

ligan

ds;

wur

tzit

e-C

dSe

synt

hesi

s);

PbS

(via

two

met

hods

)an

dP

bS-C

dSco

re-s

hell

QD

s(o

leic

acid

ligan

ds);

fluor

esce

ntdy

e,tr

is(8

-hyd

roxy

quin

olin

ato)

alum

iniu

m(A

lq3);

and

phos

phor

esce

ntdy

e,ac

etyl

acet

onat

obis

(2-p

heny

lpyr

idin

e)ir

idiu

m(I

r(pp

y)2(a

cac)

).T

heco

stan

alys

isac

coun

tsfo

ral

lmat

eria

linp

uts

(rea

gent

s,so

lven

ts,a

ndw

orku

p[‘c

rash

out’

]),y

ield

ing

tota

lcos

ts-p

er-g

ram

both

wit

hout

and

wit

hw

orku

p.Fo

rth

eQ

Ds,

thes

eha

vebe

enco

nver

ted

into

cost

s-pe

r-ar

eaas

sum

ing

aQ

Dfil

mof

25nm

thic

knes

s,as

deta

iled

inth

ete

xt.

*In

the

abse

nce

oflit

erat

ure

yiel

ds,w

eas

sum

equ

anti

tati

veyi

elds

;alt

houg

hth

isis

clea

rly

asl

ight

over

-es

tim

ate,

itis

are

ason

able

appr

oxim

atio

ngi

ven

the

synt

heti

cre

finem

ents

and

was

tere

cycl

ing

that

will

sure

lyac

com

pany

scal

e-up

sin

QD

synt

hesi

s.


2.2 Quantum-dot light emitting devices

A typical electrically driven colloidal QD-LED comprises two electrodes, which inject

charge into a series of active layers sandwiched between them (see Fig. 3a, for exam-

ple). Since their invention in 1994 [48], their performance has improved dramatically.

Fig. 2-4 summarizes this progress for the case of orange/red-emitting (almost always

CdSe-based) QD-LEDs in terms of two metrics: (a) peak EQE; and (b) peak bright-

ness. EQE is directly proportional to power conversion efficiency, a key metric for

SSL and displays, and brightness values of 103 − 104 cd m-2 and 102 − 103 cd m-2 are

required for SSL and display applications, respectively.

These achievements have in part been a result of evolutions in device architecture,

which in two recent reviews of the field, we classified into the four device ‘types’ de-

picted in the inset of Fig. 2-4a [1,2]. It can be seen that these four types have evolved

roughly chronologically. Despite the scattered data, we observe steady increases in

both EQE and brightness, with values approaching those of (phosphorescent) OLEDs.

In the following sections, we outline the distinguishing features of each type of QD-

LED.

2.2.1 Evolution of QD-LEDs

Type-I: QD-LEDs with polymer charge transport layers

These earliest QD-LEDs, pioneered in the early 1990s, were a natural progression

from polymer LEDs (PLEDs). Devices comprised a CdSe core-only QD-polymer

bilayer or blend [130], sandwiched between two electrodes. QD electroluminescence

(EL) was achieved but EQEs were extremely low (<0.01 % EQE, ∼100 cd m-2), in

part due to the low 𝜂PL of QDs without shells (10 % in solution). The low brightness

was a consequence of the very low current densities that could be reached while using

insulating QDs as both charge transport and emissive materials. Core-shell CdSe QDs

were later employed in type-I structures to take advantage of their higher 𝜂PL [131],

and EQEs of up to 0.22 % (max. 600 cd m-2) were reported using CdS shells [132].

However, these devices still exhibited significant parasitic polymer EL, indicative of


Figure 2-4: Progression of orange/red-emitting QD-LED performance overtime in terms of peak EQE and peak brightness. a, Peak EQE. b, Peakbrightness. QD-LEDs (a substantial but non-exhaustive selection from the literature)are classified into one of four ‘types’, as described in the text, and are comparedwith selected orange/red-emitting (phosphorescent) OLEDs. Solid lines connect newrecord values. Data for a are taken from references: (type-I) [129–132]; (type-II)[47, 58, 59, 87, 88, 133–136]; (type-III) [49, 137, 138]; (type-IV) [60, 61, 86, 94, 139–142];and (OLEDs) [84, 93, 118, 143–146]. Data for b are taken from references: (type-I) [48, 89, 129, 131, 132]; (type-II) [47, 62, 88, 135, 136]; (type-III) [137, 147]; (type-IV) [6, 60, 61,86,93,94,139–142]; and (OLEDs) [145,148–150].


inefficient exciton formation in QDs.

In these initial QD-LEDs, QD EL has been speculated to be driven by direct

charge injection [49] (Fig. 2-5b), FRET (Fig. 2-5c), or both, with the relative contri-

bution of these mechanisms remaining unclear [61,129,151,152]. In the case of direct

charge injection, an electron and a hole are injected from charge transport layers

(CTLs) into a QD, forming an exciton, which subsequently recombines via emission

of a photon. FRET is also a viable mechanism, unique to devices having luminescent

species, such as emissive polymers [153], small molecule organics [154], or inorganic

semiconductors [155,156], in close proximity to the QDs. In this scheme, an exciton is

first formed on a luminescent CTL. Thereafter, the exciton energy is non-radiatively

transferred to a QD via dipole-dipole coupling (Fig. 2-5f, blue arrows). The relative

contribution of these mechanisms, in all four types of QD-LEDs in fact, remains un-

clear and a better understanding of their roles, for example as a function of QD-LED

architecture, will be essential in designing more efficient and brighter devices.

Type-II: QD-LEDs with organic small molecule charge transport layers

Type-II QD-LEDs were first introduced by Coe et al. in 2002 and instead comprise

a monolayer of QDs at the interface of a bilayer OLED [47] (Fig. 2-6). These de-

vices demonstrated record EQEs of 0.5 %; an efficiency that has been augmented by

an additional factor-of-ten through optimisations. The enhanced performance was

attributed to the use of a single monolayer of QDs (enabled by the development of

spin-coating [47] and microcontact printing [92, 157, 158] techniques), which decou-

ples the luminescence process in the QDs from charge transport through the organic

layers [47,49,133,154,155,158].

Using the microcontact printing method, Anikeeva et al. showed a series of QD-

LEDs with emission tunable across the entire visible spectrum by varying the com-

position of QDs sandwiched between two organic CTLs (Fig. 2-6c) [58]. A maximum

EQE of 2.7 % was achieved for orange emission. The spectral purity and tunability of

the QD-LEDs reported in this work clearly demonstrates the potential that QD-LEDs

hold for EL displays. It has also been demonstrated that white-emitting QD-LEDs


Figure 2-5: QD Excitation mechanisms. There are four routes by which to gen-erate excitons in QDs that have been used in QD-LEDs. a, Optical excitation: anexciton is formed in a QD by absorbing a high-energy photon. b, Charge injection:an exciton is formed by injection of an electron and a hole from neighboring chargetransport layers. c, Energy transfer: an exciton is transferred to a QD via FRETfrom a nearby donor molecule. d, Ionisation: a large electric field ionises an electronfrom one QD to another, thereby generating a hole. When these ionisation eventsoccur throughout a QD film, generated electrons and holes can meet on the same QD,forming excitons. e, Energy band diagram of a typical type-II QD-LED that outlinesthe two suspected QD excitation mechanisms: charge injection and energy transfer.

can be fabricated by simply mixing different compositions [134] or sizes [159] of QDs.

A CRI of 86 was achieved by mixing red, green, and blue QDs. As shown in Fig.

2-2a, even higher CRIs should be achievable through the mixing of a greater variety

of colours of QDs, suggesting the realisability of white QD-LEDs as SSL sources. Fur-

thermore, compatibility of type-II QD-LEDs with flexible substrates is exemplified

by the flexible white-light-emitting QD-LED shown in Fig. 2-7c.

Studies in our group have indicated that, at least in certain type-II QD-LED

geometries, FRET is the dominant QD excitation mechanism [133]. Yet, for example,

the achievement of EQEs >2 % in QD monolayer-based devices comprising organic

donor materials with very low 𝜂PL [135] challenges the universality of the FRET

model.

Combining an OLED architecture with a monolayer of QDs was, nevertheless, a

significant step forward in demonstrating efficient QD-LEDs. These devices boast

all of the advantages of OLEDs, with the added benefits of enhanced spectral purity


and tunability, as illustrated by the tunable EL shown in Fig. 2-6b, c. However, the

use of organic layers introduces device instabilities upon air exposure [160, 161]. As

with OLEDs, commercialised QD-LEDs would then require protective encapsulation,

which adds to manufacturing costs and hinders certain applications, such as flexible

technologies. Furthermore, the relatively insulating nature of organic semiconductors

can limit the current densities achievable in QD-LEDs prior to device failure, there-

fore limiting their brightness.

Type-III: QD-LEDs with inorganic charge transport layers

Replacing the organic CTLs of type-II QD-LEDs with inorganic CTLs could lead to

greater device stability in air [160,161], and could enable the passage of higher current

densities and therefore brighter emission.

One such all-inorganic QD-LED (apart from the organic ligands) was made by

Mueller et al. by sandwiching a monolayer of QDs between epitaxially grown n-

and p-type GaN (ref. [49]). QD EL was observed, though at very limited efficiencies

(EQE <0.01 %). The epitaxial growth of GaN, however, diminishes the advantage

of using colloidal QDs to inexpensively fabricate large area devices. This necessitates

alternative approaches to developing QD-LEDs with inorganic CTLs.

One such alternative is to use sputtered metal oxides as CTLs. Like organic mate-

rials, metal oxide and chalcogenide thin films can be deposited at room temperature by

sputtering. The broad variety of metal oxide and chalcogenide compositions enables

fine-tuning of their energy bands, as needed for the optimal operation of QD-LEDs.

In addition, metal oxides can be more conductive than their organic counterparts,

with the conductivity of metal oxides tunable by controlling oxygen partial pressure

during thin film growth. Caruge et al. applied this technique in QD-LEDs comprising

zinc tin oxide and NiO as n- and p-type CTLs, respectively [137]. As expected, these

devices were able to pass higher current densities (up to 4 A cm-2) but the EQE was

less than 0.1 %.

This inefficiency was attributed to the damage of QDs during sputtering of the

overlying oxide layer, to carrier imbalance (due to a large hole injection barrier be-


tween the p-type metal oxide and the QDs), and to quenching of QD PL by the

surrounding conductive metal oxide [162]. To our knowledge, there has not been a

report of a type-III QD-LED with efficiencies comparable to those of type-II devices.

Aside from these DC-driven QD-LEDs, another sort of all-inorganic QD-LED that

operates by an altogether different excitation mechanism has emerged over the past

few years. These are capacitive structures consisting of two contacts sandwiching

a film of dielectric material, with a layer of QDs at its centre [162, 163]. High AC

voltages drive these devices, resulting in operation by field-assisted ionisation of QDs

to generate free carriers (Fig. 2-5d) [164]. This architecture eliminates the need for

CTLs and energy band alignment between different semiconductors.

Type-IV: QD-LEDs with hybrid organic-inorganic charge transport lay-

ers

The hybrid structure of type-IV QD-LEDs offers a compromise between type-II and

type-III QD-LEDs, often comprising an inorganic metal oxide electron transport layer

(notably solution processed ZnO [60, 61]) and an organic small molecule hole trans-

port layer [50,60,61,139,142]. A typical device structure and its corresponding EL are

shown in Fig. 2-6a, b. Significant efficiency gains have resulted, with recent visible-

emitting QD-LEDs reaching EQEs as high as 20 % [140, 142] (Fig. 2-4a and 2-6a);

approaching those of commercially mature phosphorescent OLEDs [118] as well as the

outcoupling-limited ceiling of ∼20-25 % [165]. The brightness of type-IV QD-LEDs

has also reached record levels of 218,800 cd m-2 (ref. [60]).

Type-IV devices can, if desired, be solution processed using colloidal metal ox-

ide nanoparticles as the electron transport layer [60, 61]. In particular, Qian et al.

have demonstrated red, green, and blue solution-processed (excluding electrodes) QD-

LEDs. The EQEs of these devices were 1.7 %, 1.8 %, and 0.22 %, with maximum

brightness values of 31,000 cd m-2, 68,000 cd m-2, and 4,200 cd m-2 for red, green, and

blue devices, respectively. These brightness values are within range of the highest

reported [61].

High-resolution microcontact printing of QD films (>1000 pixels per inch, 25 𝜇m


Figure 2-6: Modern QD-LED architecture and operation. a, Inverted hybridorganic-QD-inorganic type-IV QD-LED structure that has risen to prominence owingto its record efficiencies and brightness. It typically comprises a few monolayersof QDs sandwiched between an inorganic metal oxide electron transport layer andan organic hole transport layer. Photographs of tunable electroluminescence (EL)colours from: b, type-IV [2]; and c, type-II QD-LEDs (together with respective ELspectra) [58].

features) [92] (Fig. 2-7b) has already enabled the fabrication of full-colour 4-inch

QD-LED displays [93] and, by mixing different compositions [134] or sizes [159] of

QDs, white-emitting QD-LEDs that are highly amenable to SSL have been demon-

strated with excellent colour quality (CRI ∼90). White-emitting QD-LEDs on flexible

substrates (Fig. 2-7c) have also been realized at QD Vision, Inc. [1, 70].

The energy transfer scheme that is suspected to dominate QD excitation in type-

II QD-LEDs requires migration of one carrier type through the close-packed QDs of

a monolayer film, so as to form excitons in an adjacent donor material (Figure 2-

5e) [133]. Since type-III and type-IV QD-LEDs, in contrast with type-II QD-LEDs,

employ QD films thicker than one monolayer (up to ∼50 nm), the working mechanism

of type-IV QD-LEDs is more compatible with a charge injection model.

2.2.2 Beyond Cd-based QDs: infrared and non-toxic QD-LEDs

The paucity of high-𝜂PL NIR molecular and polymeric dyes provides a compelling

impetus to extend the EL of QD-LEDs from the visible into the NIR range (780-


2,500 nm). As EQEs of OLED and polymer LEDs emitting at 𝜆 > 1 𝜇m remain less

than 0.3 % [166], NIR-emitting QD-LEDs boast unique potential. For example, one

can envision solution-processable (potentially Si-compatibile) sources of EL in the 1.3-

1.55 𝜇m telecommunications band, which can be deposited on any substrate and at

lower cost than existing (usually epitaxially grown) IR-emitters, finding application

in optical telecommunications and computing [63, 77, 167–170]. Utilising biological

transparency windows between 800 nm and 1700 nm (ref. [171]), one can also imagine

deep tissue biomedical imaging and optical diagnostic [77,171] applications, as well as

on-chip bio(sensing) and spectroscopy usages [55,77,172] (for example, low cost NIR

QD-LEDs in microfluidic point-of-care devices [173]). Night-vision-readable displays

[174], night-time surveillance, and other security applications also present themselves.

Extension of the wet chemical methods previously discussed has readily enabled

the synthesis of a variety of efficient NIR-emitting colloidal QDs, which in the telecom-

munications band include PbE (E=S, Se, Te), InAs, and HgTe, as well as core-shell

QDs such as PbE-CdS and InAs-ZnSe. In many cases, high 𝜂PL have been achieved

(>50 %). There are a number of in-depth reviews [74, 76, 77, 177] for the interested

reader.

Most NIR QD-LEDs have been based largely on type-I architectures and core-

only NIR (lead chalcogenide) QDs [64, 151, 168, 169, 178, 179], with EQEs of up to

∼2 % reported [64] (Fig. 2b). At QD Vision, Inc., NIR QD-LEDs with active areas

of up to 4 cm2 (Fig. 2b, inset) and radiances of up to 18.3 W sr-1 m-2 have been

achieved; comparable to commercial IR LEDs and sufficient to serve as large-area IR

illuminators (Fig. 2-7d). In this thesis, we describe the realisation of devices with

efficiencies exceeding 4 % [15] - more than double the previous record - by transitioning

to a type-IV device structure and exploiting the enhanced passivation of core-shell

NIR QDs. Just as with visible-emitting QD-LEDs, it has been argued that electrical

excitation of QDs in these NIR devices occurs either by FRET [63, 170] or by direct

charge injection [168,169,180–182]; a balance between the two is likely in most cases.

Recently, Cheng et al. described the first Si QD-based LEDs, with very high

EQEs of 0.6 % (type-I) [180] and 8.6 % (type-II) [183], though with rather blue NIR


Figure 2-7: State-of-the-art QD-LEDs. a, EQE versus applied voltage for veryhigh performance (peak EQE of 18 %) red-emitting QD-LEDs (photograph in inset)[175]. b, The first demonstration of red-green-blue electroluminescence from (type-II) QD-LED pixels, patterned using microcontact printing [92]. c, Flexible white-emitting type-II QD-LED [176]. d, Large-area infrared illumination by an infrared-emitting QD-LED, as seen through an infrared camera [2].


EL of ∼850 nm. Although emission at wavelengths beyond 1 𝜇m may be difficult

to achieve with Si QDs (efficient blue and green emitters have also not yet been

demonstrated), this new breed of heavy-metal-free QD-LEDs nevertheless addresses

growing concerns regarding the risks that cations such as cadmium, lead, and mercury

pose to our health and to the environment [1]. The European Union’s Restriction of

Hazardous Substances Directive, for example, severely limits the use of these materials

in consumer electronics. Therefore, the likelihood of commercial success of QD-LEDs

will be greatly increased if these devices can be fabricated using heavy-metal-free

QDs.

As an aside though, we ask, how much cadmium (Cd) is actually in a QD-LED?

We can estimate this by assuming that a typical QD-LED might comprise a 5 cm × 5

cm film, 25 nm thick, of hexagonally close-packed CdSe-CdS core-shell QDs separated

by ∼0.5 nm due to their surrounding organic ligands. Then, the effective density of

CdSe-CdS QDs with a diameter of 4 nm and a shell thickness of 2 nm is 3.5 g cm-3,

which translates to a mass of 0.16 mg of CdSe-ZnS and therefore 63 𝜇g of Cd. This

is comparable to one’s daily Cd intake: the age-weighted mean Cd intake for males

in the United States is 0.35 𝜇g kg-1 day-1, or 24.5 𝜇g day-1 for a 70 kg male [184].

All the same, the commercial demand for novel down-converters in single-chip white

LEDs has specifically led to QDs being identified as likely candidate ‘phosphors’,

yet concerns over Cd compliance issues threaten to inhibit their use [40]. There are

therefore strong incentives to develop Cd-free QDs competitive with their traditional

counterparts.


Chapter 3

Colloidal Quantum-Dots

3.1 Colloidal versus epitaxial QDs

Quantum-dots (QDs) may be categorised by their synthetic route as either ‘colloidal’

or ‘epitaxial’ (or ‘self-assembled’). Whereas the latter derive from relatively high-

energy-input ‘dry’ methods of epitaxial growth from the vapour phase [185], colloidal

QDs are synthesised by ‘wet’ chemical approaches [74] and are the focus of this

thesis. The precise size and shape control, as well as the high monodispersity, spectral

purity and 𝜂PL afforded by the chemical synthesis of QDs, are unmatched by epitaxial

techniques. Colloidal QDs are freestanding and therefore amenable to a plethora

of chemical post-processing and thin-film assembly steps, in contrast to epitaxial

QDs, which are substrate-bound [99]. Additionally, the relatively inexpensive, facile,

and scalable solution-based conditions necessary for the synthesis of nearly defect-

free colloidal QDs have an impurity tolerance far exceeding that of the ultra-high

vacuum environments required for epitaxial growth. Moreover, only weak quantum-

confinement effects are observed in epitaxial QDs [186] owing to their relatively large

lateral dimensions (typically >10 nm) and difficulties in size control. This is in stark

contrast to the size-tunable emission of colloidal QDs, which are therefore favourable

as luminophores in LEDs [187].

55

Chapter 3. Colloidal Quantum-Dots 56

Figure 3-1: Schematic of colloidal QD synthesis. Adapted from ref. [9]

3.2 Colloidal QD synthesis

The benchmark preparation of colloidal quantum-dots (QDs, throughout), yielding

high quality and monodisperse (size variation <4 % r.m.s.) nanoparticles, involves py-

rolysis of organometallic precursors (such as dimethyl cadmium and trioctylphosphine

selenium for CdSe and, in this thesis, lead (II) oleate and hexamethyldisilathiane for

PbS) injected into a hot organic coordinating solvent (at temperatures of 120 to 360∘C) [65, 74, 188]. The experimental procedure is shown schematically in Fig. 3-1.

Thermally activated nucleation and growth of small crystallites from the precursors

ensues until arrested by cooling. Fine control of QD size (for example, from 1.5 nm

to 12 nm for CdSe QDs [65]) and size dispersion can thus be achieved through control

of reaction time and temperature, as well as precursor and surfactant concentrations.

QD size is monitored by measuring absorption spectra of aliquots extracted from

the growth solution during synthesis. Overcoating CdSe core-only QDs with ZnS

shells has been shown to enhance QD stability and brightness [71,72]. This is usually

achieved by injecting organometallic precursors to the shell material into a flask con-

taining the cores. The resulting QDs are dressed with organic ligands, which confer

solubility in a diversity of common non-polar solvents. Commercial red-emitting QDs

used in this thesis were obtained from QD Vision, Inc.

Post-synthesis size-selective precipitation can further increase monodispersity in


colloidal QD solutions. Typically, prior to use in devices, QDs are precipitated and

redissolved at least twice in order to eliminate excess ligands, then recast in a suitable

solvent for solution-based deposition, such as spin-casting. For QDs with oleic acid

capping ligands, as the case in this thesis, we precipitate our QD starting solution

using butanol and then methanol (roughly 1 part butanol and 8 parts methanol to

1 part QD solution, by volume), redisperse is hexane, precipitate a second time, and

finally redisperse the QDs in chloroform. Each round of precipitation is accelerated

by centrifuging the solution at 3,500 rpm for 5 minutes.

As shall be the focus of Chapter 6, NIR-emitting QDs such PbSe and PbS are

relatively unstable and prone to oxidation and PL quenching, and accordingly, display

relatively low 𝜂PL. Pietryga et al. have demonstrated a partial cation exchange

method to controllably overcoat lead chalcogenide cores to form more stable core-

shell QDs such as PbSe-CdSe, PbSe-CdSe-ZnS, and PbS-CdS [75]. As compared

to the conventional shell growth approach mentioned above, cation exchange relies

on the lability of Pb ions, which are susceptible to exchange with Cd ions when

immersed in a highly concentrated solution of cadmium oleate. Gyuweon Hwang in

Moungi Bawendi’s group used this technique to synthesise stable PbS-CdS QDs with

𝜂PL in solution of up to 53 % and emission peaks from 1.1 to 1.4 𝜇m. Synthesis

conditions were tailored to allow slower shell growth than normal by lowering the

reaction temperature to 80-100 ∘C and reducing the Cd-precursor concentration [189].

Fig. 3-2a shows PL spectra at various times throughout the cation exchange reaction

of two batches of starting PbS QDs. The gradual blue-shifting of PL peaks during

the reaction reflects the in-growth of the CdS shell during Pb-to-Cd cation exchange,

effectively shrinking the confining PbS core. Fig. 3-2b plots 𝜂PL as a function of

shell thickness for the second batch. The 72 % rise in 𝜂PL upon addition of a shell

attests to the passivation of surface traps on PbS cores. However, this initial increase

(corresponding to a shell thickness of ∼2.5 monolayers) is followed by a gradual

decline in 𝜂PL as shell thickness increases further, attributable to a rise in defect

density resulting from lattice mismatch-induced stress [189].

Scaling up these synthesis techniques so as to reduce the cost of QDs is a pre-


requisite for the commercialisation of QD technologies, and increases in yield, from

milligrams to kilograms per week, have been reported [121,190].

3.3 Optical properties of QDs

The size-tunable colour of QD PL and absorption spectra is arguably their most

unique and important property, and is governed by the quantum size effect. See, for

example, ref. [191–193] for in-depth discussions. The intrinsic material constituting

the QD confers a bulk crystal band-gap, but in a QD, electrons and holes are subject

to quantum confinement effects that quantise these bulk energy levels, resulting in

atomic absorption- and emission-like spectra [192]. When the radius of a QD is less

than the material’s Bohr radius (𝑎B = ~2𝜅/𝜇e2, where 𝜅 is the dielectric constant of

the semiconductor and 𝜇 is the effective mass of the electron-hole pair) the energy

levels of a QD can be analytically approximated by modeling the QD as a particle-in-

a-sphere with infinite potential walls. This applies, for example, to PbS QDs, whose

exciton Bohr radius is approximately 18 nm [9], versus typical QD radii diameters of

2-5 nm. This gives a confinement energy of,

𝐸𝑒,ℎ𝑙,𝑛 =

~2𝜑2𝑙,𝑛

2𝑚𝑒,ℎ𝑎2(3.1)

where 𝜑𝑙,𝑛 is the nth root of the spherical Bessel function of order l, 𝑗𝑙(𝜑𝑙,𝑛) = 0,

and 𝑚𝑒,ℎ is the effective mass of the electron or hole. This can be viewed as the

kinetic energy of a free particle, but with quantised wave vectors, 𝑘𝑙,𝑛. Since QDs

are typically larger than the lattice constant of their constituent material, one can

apply the effective mass approximation, modeling the conduction and valence bands

as parabolic and describing particle wavefunctions as linear combinations of Bloch

functions. It is then possible to derive the energy of the electron-hole pair states,

𝐸 = 𝐸𝑔 + 𝐸ℎ(𝑎) + 𝐸𝑒(𝑎) − 1.8𝑒2

𝜅𝑎(3.2)

The first term on the right-hand side of the equation is the fundamental band-gap of


Figure 3-2: Photoluminescence spectra and quantum yield during PbS-CdSQD cation exchange reaction. a, PL spectra of aliquots collected at representativetimes with increasing CdS shell thickness (right to left) throughout two separatecation exchange reactions. First reaction batch (dashed) shell thicknesses and reactiontimes: 0 nm (i.e. PbS cores); 0.57 nm (5 min); 0.75 nm (30 min); and 0.89 nm (2h). Second reaction batch (solid) shell thicknesses and reaction times: 0nm (i.e.PbS cores); 0.18 nm (5 min, 80 ∘C); 0.21 nm (5 min, 100 ∘C); 0.69 nm (2 h); 0.8nm; and 0.9 nm. The red-side of the red-most spectrum is clipped because of thespectral responsivity limit of the InGaAs array detector used to collect the data. b,PL quantum yield as a function of CdS shell thickness for the second batch (whichwe make use of in Chapter 6) of QDs in chloroform solution.


the QD material, 𝐸𝑔. The two middle terms are the quantum confinement energies

of holes and electrons, respectively, defined above, which explain the size-dependent

band-gap that we observe in QDs: as a QD is reduced in size, its confinement in-

creases, the energy of electron or hole wavefunctions increases, and electron-hole re-

combination will emit a higher energy (bluer) photon. The final term is a first order

Coulombic correction that accounts for the attraction between electrons and holes in

the QD.

The quantum confinement effect derived from the simple parabolic band approxi-

mation is elegantly manifested in spectral absorption measurements of QDs, wherein

the absorption band edges move to higher energies as QDs are reduced in size (Fig.3-

3). It also accounts for the correlation between QD size and PL energy. However,

given our focus in this thesis on QD PL, it is worth noting that a more detailed

multi-band effective mass approximation is in fact required to explain some of the in-

tricacies of QD PL spectra. These include their red Stokes shift with respect to their

first absorption peak and their unusually long radiative lifetimes (roughly an order

of magnitude longer than in the bulk [194]). These are explained by the fact that,

as mentioned in Section 2.1.2, the lowest energy exciton state of a QD is in fact an

optically inactive "Dark Exciton" state, which sits at lower energy than an optically

active "Bright Exciton" state [194]. Their splitting energy is on the order of 𝑘BT and

is therefore overcome at room temperature, effectively rendering QDs ‘phosphors’ at

this temperature.


Figure 3-3: Absorption spectra of PbSe QDs as a function of nanocrystaldiameter. Reproduced from ref. [9]


Chapter 4

Origin of efficiency roll-off in

QD-LEDs

4.1 Role of QD luminescence in QD-LED efficiency

Much of the focus of QD-LED research over the past two decades has been on under-

standing and improving the efficiency of free carrier recombination in these devices,

𝜂r. This is reflected in the engineering of the multiple generations of QD-LED archi-

tectures described in Chapter 2, and is of course motivated by the direct dependence

of EQE on 𝜂r by way of Eq. 2.1. This expression also shows that EQE is directly pro-

portional to the photoluminescence (PL) quantum yield (𝜂PL) of our QDs, yet this

factor has received less attention. In particular, while there have been substantial

efforts to elucidate the mechanisms of PL quenching in experimentally isolated QDs

(for example, single QDs or QD ensembles) and QDs under very specific experimental

conditions (for example, in electrochemical cells), much less scrutiny has been paid to

QD PL quenching in operational QD-LEDs; accounting for their unique in situ envi-

ronment, which includes neighbouring conductive layers, large electric fields, charge

current flow, and the associated device heating. We therefore begin by looking to

the earlier literature of isolated QDs to identify the different processes that may be

suspected to dictate the probability of QD PL in QD-LEDs. In this chapter and the

next, we limit our scope to the bias dependence of QD-LED efficiency and investigate

63

Chapter 4. Origin of efficiency roll-off in QD-LEDs 64

the contribution of QD PL quenching on this behaviour. Zero-bias (occurring even

at zero bias) contributions to PL quenching will be the focus of Chapter 6.

4.2 Overview

Even casual inspection of the typical EQE-versus-current density (or voltage) plot of

a QD-LED reveals a pronounced decline in EQE (Fig. 4-1c): from its maximum value

at relatively low currents (∼ 0.01 to 0.1 A cm-2 at ∼ 2 V) to roughly half this value as

current increases to ∼ 1 A cm-2 at ∼ 8 V. This so-called efficiency ‘roll-off’ or ‘droop’

in fact plagues all LED technologies to date, yet until now, it has not been studied in

QD-LEDs. Understanding its cause is essential to developing high-brightness, high-

efficiency devices. Does a bias-driven reduction in QD 𝜂PL account for this efficiency

roll-off in QD-LEDs? If so, what mechanism(s) are responsible for this quenching?

These are the questions we set out to address in this chapter and the next.

4.3 Introduction

Nitride-based LEDs find wide and growing commercial application, and OLEDs and

QD-LEDs have the potential to capitalise on the vast solid state lighting (SSL) and

display markets. For all lighting technologies, however, efficiency roll-off presents a

considerable challenge. Indeed, roll-off is among the most significant and enduring

hurdles facing GaN-based LEDs and SSL, severely limiting the efficiencies achievable

with high-brightness devices. Today’s commercial blue GaInN LEDs, for example,

often suffer more than a 40 % loss in EQE at desired high current densities (∼35

A cm-2) compared to their peak EQEs at <10 A cm-2 [195] (Fig. 4-1a). Roll-off

raises SSL costs by requiring the use of multiple LEDs in a single lamp to keep ef-

ficiencies high. It adds constraints on optics and design, and it increases thermal

loads [196]. All major LED companies reportedly have active ongoing research into

roll-off effects [197]. According to the U.S. DOE’s latest assessment, addressing roll-

off is among the highest priorities for LED-based lighting R&D [31]. Roll-off has also


been the topic of substantial investigation in efficient phosphor-based OLEDs over

the past two decades. Like QD-LEDs, OLED efficiencies fall rapidly beyond ∼ 0.01

to 0.1 A cm-2 (Fig. 4-1b). In a comprehensive review of white OLEDs, Reineke et

al. recently proposed that the largest improvement factor needed for enhancing the

efficiencies of this technology may come from reducing roll-off [198]. Importantly,

the exciton annihilation processes that have been found to be responsible for roll-off

in OLEDs (outlined briefly below) may also be a source of material degradation, be-

cause the associated high excitation energies may break the chemical bonds conferring

functionality to some organic materials [199].

Despite all of this, confirming the origins of efficiency roll-off in GaN-based LEDs

continues to be an active area of experiment and theory. Various mechanisms have

been proposed, most prominently thermionic electron leakage from the light-emitting

active region and Auger recombination (described in the next chapter) within this re-

gion, but also carrier delocalisation and poor hole injection [31,195,202]. Meanwhile,

efficiency roll-off in OLEDs has been attributed to mechanisms including triplet-

triplet annihilation, triplet-polaron quenching, and electric field-induced exciton dis-

sociation [198,203,204]. Among these three, the first two (especially the former) are

considered to play larger roles [201].

Surprisingly, efficiency roll-off and its origins have never previously been inves-

tigated closely in QD-LEDs. To do so, we begin by reexamining the expression for

EQE in Eq. 2.1. Based on our physical understanding of each term in the equa-

tion, as well as inspection of the commonly cited mechanisms in the GaN and OLED

literature, discussed above, we expect only two of the terms, 𝜂PL and 𝜂r, to have

a strong bias-dependence (we neglect outcoupling efficiency, which may strictly also

have some bias dependence, for example due to dipole reorientation under applied

electric fields). That is, EQE (V) = 𝜂r(V)𝜒𝜂PL(V)𝜂oc ∝ 𝜂r𝜂PL. Efficiency roll-off can

therefore be approximated as attributable to the bias dependence of 𝜂PL, of 𝜂r, or of

some combination of the two. Falling 𝜂PL would signify changes to the QD emitter

itself, and the numerous potential mechanisms responsible for bias-driven QD PL

quenching will be discussed in Chapter 5. In contrast, a reduction in 𝜂r would re-


Figure 4-1: Efficiency roll-off in different LED technologies. External quantumefficiency (EQE) is seen to fall dramatically at high current densities in: a, GaN-based LEDs [200]; b, OLEDs [201]; and c, QD-LEDs (a typical red-emitting devicefabricated in our laboratory and investigated in this thesis). Note that a exhibitsroll-off due to both current density and thermal stress, a distinction discussed brieflyin Chapter 5.


sult from one or more forms of carrier leakage, such as those for GaN-based LEDs

above, whereby electrons and/or holes are not confined to the QD region and leak

through their respective blocking layers in a QD-LED. Unsurprisingly, distinguishing

between 𝜂PL and 𝜂r as the origin of EQE bias-dependence in GaN-based LEDs and

OLEDs has been key to uncovering the mechanisms discussed above. For example, in

GaInN/GaN multi-quantum well (MQW) LEDs, Kim et al. measured EQE and PL

as a function of excitation power and found no correlation, concluding that roll-off

must be due to recombination outside of the MQWs [205]. They deduced that electron

leakage was instead the dominant cause of roll-off. In contrast, Reineke et al. [204]

and Kalinowski et al. [206] both observed good correlations between EL and PL in

OLEDs, leading to the advancement of models for roll-off based on PL quenching via

triplet-triplet annihilation. In this chapter, we take a similar approach to quantifying

the contribution of PL quenching (versus carrier leakage) to roll-off in QD-LEDs.

4.4 Methods

The device structure we investigate is a type-IV QD-LED (see Section 2.2) with

organic-inorganic hybrid charge transport layers that has recently attracted attention

owing to its record high EQEs and brightness [2]. This red-emitting QD-LED is also

more conveniently compatible with the visible-wavelength spectroscopic tools at our

disposal, which we make use of in this chapter and the next. The device was fabricated

on a glass substrate coated with indium tin oxide (ITO) and has the structure: ITO

(150 nm) / zinc oxide (ZnO) (50 nm) / QDs (50 nm) / 4, 4-bis(carbazole-9-yl) biphenyl

(CBP) (100 nm) / molybdenum oxide (MoO3) (10 nm) / aluminium (Al) (100 nm).

ZnO was radio-frequency sputtered, QDs were spin-cast out of chloroform, and CBP,

MoO3, and Al were thermally evaporated. We used CdSe-ZnCdS core-shell QDs with

a peak PL wavelength of 610 nm, provided by QD Vision, Inc. The energy band

diagram of the device is shown in the inset of Fig. 4-4 and is based on literature

values [58,93,207,208].

Based on our discussion in the previous section, the decrease in EQE at high


biases may be the result of either charge carriers leaking out of the QD layer or a

reduction in QD luminescence efficiency. To identify which of these two mechanisms

dominates, we perform a simultaneous EL-PL experiment and monitor the relative

PL efficiency of the QDs as the device bias is swept. The experimental setup is shown

in Fig. 4-2. To isolate the PL contribution from total luminescence, we modulate

the PL excitation source (𝜆 = 530 nm LED) at 1 kHz and send the combined EL-

PL signal (collected using a Si photodiode and a current preamplifier) to a lock-in

amplifier. The PL intensity is intentionally kept low (PL/EL <0.001 % at 13 V) to

avoid significantly increasing the charge density within the QD layer. An excitation

wavelength of 530 nm ensures that the QD layer is excited without exciting the

surrounding wider band gap charge transport layers, as absorption spectra of the

QD-LEDs constituent materials confirm (Fig. 4-3).

4.5 Results and discussion

Current density and normalised EQE measured for a typical device as described above

are shown in Fig. 4-4a. The EQE peaks at 2 % for 4 V applied bias and rolls off by 50

% by 8 V, quite typical of a modern QD-LED. The results of the simultaneous EL-PL

experiment are shown in Fig. 4-4b, with EQE and QD PL intensity normalised at 4

V applied bias. Below 4 V, the PL intensity remains constant (voltage independent).

In contrast, above 4 V, the PL intensity decreases monotonically with increasing

bias, tracking the decrease in EQE of the QD-LED. The correspondence between

the decreasing PL intensities and EQE with applied bias identifies the change in the

QD 𝜂PL to be sufficient to explain the QD-LED roll-off behavior. Importantly, this

is the first time it has been demonstrated that efficiency roll-off in QD-LEDs can

be ascribed to the QD emitters themselves, allowing us to discount extrinsic charge

leakage effects as the cause, and to focus our investigation specifically on bias-driven

QD PL quenching mechanisms.

To verify that the optically generated excitons used to probe our QD-LED do

not interfere with the device’s intrinsic exciton dynamics, we conduct the same PL


Figure 4-2: Simultaneous EL-PL measurement setup. Relative EL intensity iscollected by a photodetector mounted in front of the biased QD-LED. Simultaneously,QD PL is induced by excitation of the QD-LED with 530 nm light from an LEDpulsed at 1 kHz passed through a microscope objective. The resulting combined ELand PL signal is collected by another photodetector and fed into a lock-in amplifierthat separates the EL (d.c.) and PL (a.c.) components, enabling relative PL intensityto be isolated.


Figure 4-3: Absorption spectra of QD-LED components. Absorption spectraof CdSe-ZnCdS QDs (red), CBP (dark blue), and ZnO (light blue) confirm that atthe excitation wavelength of 530 nm, only the QDs absorb significantly.

measurements at multiple excitation intensities, from 15.6 𝜇W to 44.5 𝜇W (Fig. 4-

5). Normalising each PL signal by its excitation intensity, we confirm that the PL

quenching trend is independent of this variable.

In the context of the discussion in Section 4.3, unlike some GaN-based LEDs

but similarly to some OLEDs, we therefore conclude that we can neglect charge

leakage effects in accounting for efficiency roll-off in QD-LEDs. We note that another

possibility - that roll-off could be ascribed to permanent or semi-permanent thermal

degradation of our device - can be neglected given the reproducibility of the above

experiment when repeated on a one second timescale. In the next chapter, we probe

the mechanistic origins of the bias-driven QD PL quenching deduced here.


Figure 4-4: EQE roll-off in a QD-LED and correlation with QD PL. a, Currentdensity-voltage and EQE-voltage characteristics of the QD-LED under investigation.Inset: Energy band diagram of the device, with indicated energy values referencedto the vacuum level. b, Simultaneous EQE and QD PL intensity of the QD-LED(normalised at 4 V, when the peak EQE is 2 %) as a function of voltage. Roll-off ofthe EQE above 4 V is seen to reflect reduced QD PL efficiency at high biases.


Figure 4-5: Bias-induced QD PL quenching, normalised by optical excitationintensity. Relative QD PL intensities under 530 nm excitation at multiple intensitiesconfirms the excitation probe does not disturb the device’s intrinsic exciton dynamics:(black) 15.6 𝜇W; (dark grey) 33.4 𝜇W; (light grey) 44.5 𝜇W.

Chapter 5

Bias-driven QD luminescence

quenching in QD-LEDs

Our measurements of visible-emitting QD-LEDs in Chapter 4 have identified bias-

driven QD photoluminescence (PL) quenching to be responsible for the roll-off in

efficiency of our QD-LEDs at high current densities. In this chapter, we conduct

in situ measurements of QD-LED electroluminescence (EL) and PL spectra, and by

comparing the two, reveal that strong electric fields are responsible for the reduced QD

PL quantum yield (𝜂PL). The quantum confined Stark effect is found to consistently

explain the observed phenomena.

5.1 Bias-dependent QD quenching mechanisms

Potential mechanisms responsible for bias-dependent QD PL quenching are encapsu-

lated by Eq. 5.1, which expresses QD 𝜂PL (and therefore external quantum efficiency,

EQE) in terms of radiative (𝑘r) and non-radiative (𝑘nr) recombination rates:

EQE(V) ∝ 𝜂PL(V) =𝑘r(V)

𝑘r(V) + 𝑘nr(V)(5.1)

We have explicitly reflected the potential bias-dependence (V) of each term. First

consider 𝑘nr. As we shall see in Chapter 6, one contribution to the non-radiative

73

Chapter 5. Bias-driven QD luminescence quenching in QD-LEDs 74

Figure 5-1: Heat-induced QD PL quenching. PL spectra of a CdSe-ZnCdS QDspin-cast thin film at temperatures ranging from room temperature up to 100 ∘C.

recombination rate comes from surface traps and the local environment (Fig. 5-2,

left). But trap-assisted quenching may also be exacerbated under bias as the device

heats up, for example due to thermally-activated charge transfer from a QD core to a

defect state. Heat-induced PL quenching can be seen by monitoring the PL spectrum

of a film of our QDs on a temperature-controlled substrate. As plotted in Fig. 5-

1, a temperature rise of just a few tens of degrees Celsius is enough to cause QD

𝜂PL to begin to fall. Indeed, the impact of thermal efficiency droop on the EQE of

GaN-based LEDs can be as detrimental as that of current density-related efficiency

droop [200] (Fig. 4-1a).

Studies of single QDs and QD ensembles also demonstrate another bias-dependent

contribution to 𝑘nr; namely, free charge carriers (Fig. 5-2, middle). For example, with

current densities in QD-LEDs routinely exceeding the order of 1 A cm-2, imbalanced

charge injection can lead to charge accumulation on a QD, increasing Auger non-

radiative recombination, whereby an exciton dissipates its recombination energy as


kinetic energy to a third charge rather than as a photon [6, 12, 67, 87, 136, 138, 209].

Because the rate of Auger recombination is orders of magnitude faster than the rate of

radiative recombination (few to hundreds of picoseconds, versus nanoseconds), PL is

completely quenched in charged QDs. Over the years, numerous specific mechanisms

by which Auger recombination proceeds in QDs have been shown. These include

coupling between an exciton and a carrier that is in a QD [12,209], on its surface [12],

or in a deep trap state [210], coupling between multiple charge-neutral excitons, such

as biexcitons and triexcitons [209], and coupling between a carrier and an exciton

with one of its charges in a deep trap state [211].

Application of bias in a QD-LED also imposes an electric field across the QD film

on the order of 1 MV cm-1, providing the third mechanism for PL quenching shown in

Fig. 5-2. The electric field spatially polarises the electron and hole wavefunctions of

the exciton, |𝜓e⟩ and |𝜓h⟩, respectively, to opposite hemispheres of the QD, reducing

their overlap integral. As a result, 𝑘r, which is proportional to the optical matrix

element of the electron and hole states (𝑘r ∝ |⟨𝜓e|p |𝜓h⟩|2, where p is the momentum

operator), is reduced by the field [11]. From Eq. 5.1, a decrease in 𝑘r can therefore

reduce 𝜂PL and with it, EQE. These polarisation effects are also seen, for example, in

InGaN LEDs, where piezoelectric and spontaneous polarisation leads to large electro-

static fields (> 1 MV cm−1) in the quantum well active regions, spatially separating

electron and hole wavefunctions and reducing 𝑘r. Interestingly though, this is not

thought to be the cause of efficiency roll-off, but rather is suspected as (partly) re-

sponsible for the ‘green-yellow gap’ in LED efficiencies [39,41]. Additionally in QDs,

the shift of the electron and hole wave functions from the QD’s center breaks its local

neutrality and increases the exciton’s interaction with polar optical phonons. This

interaction leads to significant phonon broadening of the PL line [213]. A closely

related characteristic of the electric field’s effect is the quantum confined Stark Effect

(QCSE) [214], which is a shift in the bound state energies of the QD leading to a

change in the effective band-gap of the material. Described in terms of second order

perturbation theory, the shift in effective band-gap of a QD in the presence of an

electric field, E, is given by [214],


Figure 5-2: Zero-bias and bias-driven QD PL quenching mechanisms.Schematic of an archetypal QD-LED under bias, with an electric field across theQD film and free charge carriers present. Even at zero bias, excitons in QDs are ex-posed to PL quenching pathways such as (box, leftmost) relaxation via surface defectstates and (box, second from left) energy or charge transfer to/from neighbouringtrap states (represented by the blue arrow), for example in conductive charge trans-port layer materials. As bias increases, PL quenching by free charge carriers alsobecomes possible via Auger recombination (box, second from right). In these threecases, QD 𝜂PL is reduced due to an increase in the non-radiate recombination rate,𝑘nr. However, according to Eq. 5.1, 𝜂PL can also be reduced due to a decrease in𝑘r, and this can occur under exposure to an electric field that polarises the exciton(box, rightmost). Alternative processes due to an electric field could include excitondissociation, phonon coupling, or coupling with surface state defects. In these casesquenching would again occur, but due instead to an increase in 𝑘nr. Adapted fromref. [212]


∆𝐸𝑛 = ⟨𝜓n| 𝑒r · E |𝜓n⟩ +∑𝑘 =𝑛

|⟨𝜓k| 𝑒r · E |𝜓n⟩|2

𝐸𝑛 − 𝐸𝑘

(5.2)

or, in simplified form,

∆𝐸𝑛 = 𝜇𝐸 +1

2𝛼𝐸2 (5.3)

where 𝜇 is the projection of the dipole moment of the QD in the direction of the

electric field, and 𝛼 is the QD’s polarisability in the same direction. In QD ensembles,

𝜇 averages to zero and one observes a quadratic red-shift in QD PL with increasing

electric field [214].

However, a reduction in QD 𝜂PL by electric field is not necessarily dominated by

the reduction in 𝑘r described above [215]. An electric field may also reduce 𝜂PL by

enhancing phonon coupling [213], exciton coupling to surface trap states [12,216], or

exciton dissociation [135,164,217–220]. In all three cases, 𝜂PL is reduced by an increase

in 𝑘nr. For example, the dissociation of excitons can occur when the Coulombic

binding strength of the electron-hole pair is overcome by the polarising electric field,

causing one of the charges to be ionised out of the QD [221]. The expected (roughly

exponential) field-dependence of the exciton dissociation rate given by the Onsager-

Braun model would result in a proportional increase in 𝑘nr [11]. Analogously, one

prominent explanation for roll-off in nitride-based LEDs is that polarisation fields

drive electron leakage out of the active region of these devices [39,205].

In this chapter, we seek to determine which of the three bias-dependent quenching

mechanisms - heating, Auger recombination, or electric-field-inducing quenching - is

primarily responsible for the QD PL quenching found to be responsible for EQE

roll-off in the previous chapter.


5.2 Electric field-induced quenching in QD-LEDs

5.2.1 Introduction

As discussed above, reduction of PL efficiency in QD thin films has been previously

measured when QDs are heated [222], charged (Auger recombination) [219], or placed

under a strong electric field [216,217]. We preliminarily eliminate temperature effects

on QD 𝜂PL, as measurement of the operating temperature of our QD-LEDs with an

infrared camera shows a change of no more than a few degrees (∼5-6 ∘C rise over

∼10 seconds at 0.1 A cm-2, corresponding to ∼8 V, as shown in Fig. 5-3). Consulting

Fig. 5-1, this temperature rise is not sufficient to affect 𝜂PL enough to explain the

roll-off, especially on the short timescales (on the order of a second) of each of our

measurements. In QD-LEDs, like in OLEDs, the device temperature will likely rise

by several tens of degrees Celsius, but this temperature rise has a time constant on

the order of tens to hundreds of seconds [223] - incompatible with the fast, repeatable

nature of our measurements. Tentatively speaking, charging effects also seem less

than likely because QDs generally have long charge retention times (on the order of

minutes to hours [219]), whereas the efficiency roll-off curve (Fig. 4-4) is measured

within seconds and was reproduced multiple times successively, with the peak EQE

unchanged. Moreover, we consistently observe a red-shift in EL with increasing biases

applied to our QD-LEDs (Fig. 5-4), qualitatively consistent with the red-shift one

would expect from a QCSE. From these observations, we hypothesise that it is the

electric field associated with the applied bias that is quenching the QD PL at high

voltages and set about testing this hypothesis.

5.2.2 Methods

To characterize the QCSE in our QD-LED structure, we first measure the PL spectra

of the QD films in our devices under reverse bias. The experimental setup is shown

in Fig. 5-5b. Reverse biasing allows the effect of electric field on the QDs in the

QD-LED to be studied in situ, in absence of any charge injection. To avoid damaging


Figure 5-3: Thermal image of QD-LED under constant current bias. Col-lected with an infrared camera, the thermal image shows the temperature distributionacross a half-inch substrate (delineated by a large grey box) resulting from biasing ofa single QD-LED pixel (1.21 mm2, small grey box) at 0.1 A cm-2 for ∼10 seconds.

Figure 5-4: QD-LED EL spectra. Normalised spectra of QD-LED EL biased at:(black) 2 V; (orange) 5 V; and (red) 8 V exhibit a gradual red-shift, qualitativelyconsistent with the quantum confined Stark Effect.


the device by prolonged reverse biasing, we apply sawtooth-like voltage waveforms

(Fig. 5-5a) with a 500 Hz repetition rate. The sawtooth amplitude peaks at -18 V

and is followed by a duration of positive voltage (1.6 V) to reduce stress on the device

by minimising the average net applied voltage, but without turning on EL. QD PL

is induced by a 530 nm wavelength LED emitting 100 𝜇s long pulses synchronised

with the voltage waveform and QD PL spectra are collected with a spectrometer.

By sweeping the time delay (phase shift) between the voltage waveform and the

illumination pulse, PL spectra of the QDs under different electric-field strengths can

be collected while keeping all other conditions unchanged. To assess the degree to

which the QCSE occurs while the QD-LED is in operation, EL spectra are monitored

as the device is forward biased, again with all other experimental variables held

constant. The experimental setup for this measurement is shown in Fig. 5-6. This

combined approach allows the study of QD PL and EL from the same active device

structure.

5.2.3 Results and discussion

The resulting PL and EL spectra are normalized and their peak emission energies are

compared. Fig. 5-7 shows EL spectra obtained at 5, 11.6, and 13.8 V overlaid with PL

spectra with coincident peak energies (PL at 1.6, -8.6, and -16 V, respectively). Both

PL and EL spectra are approximately Gaussians at low biases and redshift at higher

biases. However, EL does not exhibit the same spectral broadening that is observed

in the PL. In particular, a shoulder begins to appear on the low energy side of the

PL spectra. The inset of each panel in Fig. 5-7 shows a double-Gaussian fit to each

asymmetrically broadened PL spectrum. We attribute the double-Gaussian profile

to emission from QD subpopulations that are subject to two different environments;

for example, a layer of QDs next to ZnO and a layer of QDs away from ZnO. QDs

placed adjacent to the ZnO are expected to exhibit energy levels that differ from that

of QDs placed adjacent to the CBP, which has a lower dielectric constant [224]. The

difference between EL and PL spectra (even when the peaks are matched) can be

explained by the fact that the electric-field distribution is generally different between


Figure 5-5: QD PL electric field-dependence measurement technique. Re-verse biasing our QD-LED allows measurement of PL spectra of QDs in situ in ourdevices, exposed to varying electric fields in absence of charge injection. a, The rel-ative timing (phase) of the 530nm LED optical excitation square wave pulse (green)and a reverse bias sawtooth-like voltage waveform (black), synchronised at 500 Hzrepetition rate, determine the magnitude of the field at which each PL spectrum isrecorded by the spectrometer setup shown in b.


Figure 5-6: QD-LED EL measurement setup. EL spectra from our QD-LED aremeasured as increasing increments of forward bias are applied.


forward and reverse biased diodes [225], thus affecting the two populations differently.

Each PL and EL spectrum is decomposed into two Gaussians, and their intensities

and peak energies as a function of device voltage are shown in Figs. 5-8a and b,

respectively. The black solid line in Fig. 5-8a is a fit to the PL intensity data

assuming a simplified version of the model described in Ref. [216]. In Fig. 5-8b, PL

and EL peak energies for both subpopulations redshift under high bias. In particular,

the PL peaks of subpopulation A show a quadratic dependence on voltage (black fit),

which is a signature of the QCSE. There is no clear fit for the EL peak shift because

the distribution of the electric field inside the diode is voltage dependent. The peak

energies are maximum when the device is slightly forward biased, indicating the

presence of a built-in electric field, which is expected in a diode structure.

Assuming that the EQE of the QD-LED is predominantly governed by the QCSE

at high forward biases, we should be able to predict the EQE by comparing the

forward-bias EL to the reverse-bias PL from Fig. 5-8a and b. A QD film exposed

to an electric field will undergo a QCSE, which is manifested as a shift in the QD

PL or EL emission spectra and a concomitant decrease in its PL or EL efficiency.

Because the emission spectrum is a function of the applied field, whenever the forward-

bias QD EL emission spectrum of subpopulation A (subpopulation B) matches the

reverse-bias QD PL spectrum of the same subpopulation, the QDs in subpopulation A

(subpopulation B) are experiencing the same local electric field under those particular

EL and PL biasing conditions. Therefore, for each subpopulation, the EL efficiency

at each forward bias can be predicted by finding the corresponding PL spectrum in

reverse bias with peak energy matching that of the EL peak, and assigning the PL

efficiency at that electric field to the EL efficiency. We emphasise that the choice of

physical model used to fit the data in Fig. 5-8a does not affect the predicted EQE,

which is calculated directly from the PL and EL spectra and the corresponding PL

intensities.

For example, subpopulation A (Fig. 5-8b, red) shows an EL peak shift from 1.990

to 1.984 eV between 5 and 10 V. This shift corresponds to a PL peak shift from -4.5

to -11.3 V and indicates that the luminous efficiency is reduced by about 37 % (Fig.


Figure 5-7: Comparison of QD PL spectra (black lines) and QD EL spec-tra (orange diamonds) at corresponding peak emission energies, for threedifferent biases. At high biases, the PL spectra exhibit a red shoulder that is notobserved at lower biases or in the EL spectra. Insets: PL spectra (black) are recon-structed (green) using two Gaussians, which correspond to emission from two QDsubpopulations, A and B (red and blue, respectively).


Figure 5-8: Electric field-dependence of QD-LED PL and EL. a, Relativeintensities of subpopulations A (red) and B (blue). The PL data are fitted to a sim-plified version of the model presented in ref. [216]. b, Peak energies of subpopulationA (red) and subpopulation B (blue). Quadratic fits (black lines) to the PL data aremade assuming that the shifts are due to the quantum confined Stark effect.


Figure 5-9: Measured EQE and predicted EQE as a function of voltage. EQEis predicted through the comparison of PL and EL data (Fig. 5-8) as described in thetext. The agreement between the data and the prediction shows that the quantumconfined Stark effect can self-consistently account for the QD-LED efficiency roll-off.

5-8a) for subpopulation A as a result of the QCSE. The relative number of excitons

formed on the two subpopulations (A and B) of QDs is calculated by dividing their

EL intensities in Fig. 5-8a by their respective PL efficiencies. The overall EQE is

then the weighted average of the PL efficiencies of the two subpopulations. This

analysis is applied to EL data between 2.5 and 14 V and the resulting predicted

EQE, which is scaled to match the maximum of the measured EQE, is shown in Fig.

5-9. Predicted and measured EQEs are in good agreement, with EQE rolling off by

up to 40 % at 13 V. The match between the EQE behavior predicted by the QCSE

and the experimentally observed efficiency roll-off is evidence that the electric field

strength alone - and not carrier leakage or QD charging (Auger recombination) - is

sufficient to model the efficiency roll-off.

To further understand the effect of the electric field on the QD PL efficiency,

we measured transient PL of the QDs in the QD-LED. Based on our discussion in

Section 5.1, it may be possible to distinguish between different possible microscopic


mechanisms behind electric field-induced quenching in our QD-LEDs by observing

whether the field causes an increase in 𝑘nr or a decrease in 𝑘r. Reduction of QD

𝜂PL has previously been attributed to a decrease in radiative exciton recombination

rate (e.g. reduced electron-hole wave function overlap [226, 227]), an increase in

non-radiative exciton recombination rate (e.g. exciton dissociation [221, 228]), or a

decrease in the probability of forming thermalized excitons (e.g. hot charge carrier

trapping by QD surface traps [12,13]).

We perform this measurement using the same reverse biasing scheme as in Fig.

5-5a, but with a 100 ps laser pulse train at 540 nm wavelength replacing the green

excitation LED. PL was detected with a Si avalanche photodiode and timing informa-

tion was obtained via a time-correlated single photon counting module. The resulting

transient PL at four different voltages reveals a lifetime of 4 ns for all of the voltages

applied while the initial intensity decreases with higher applied voltage (Fig. 5-10).

The inset indicates the times at which the QD PL intensity I has decreased from its

initial value of 𝐼0 so that 𝐼/𝐼0 = 𝑒−1 and 𝐼/𝐼0 = 𝑒−2 (𝜏𝑒−1 and 𝜏𝑒−2 , respectively).

Because 𝜂PL of our QD film is 8 %, PL lifetime is dominated by the non-radiative rate.

Therefore, the voltage independent PL lifetime observed suggests that the cause is ei-

ther a decrease in the radiative rate or a decrease in the thermalized-exciton formation

efficiency.

5.2.4 Conclusion

In conclusion, we have identified the electric field-induced PL quenching of QDs to be

responsible for the efficiency roll-off in QD-LEDs. We use the relationship between

PL peak shifts and PL quenching of QDs subject to the QCSE - observed while

reverse biasing a QD-LED - to predict the efficiency roll-off in forward bias. The roll-

off predicted by this analysis is in excellent agreement with our experimental data

and correctly traces an EQE reduction of nearly 50 %. Transient PL measurements

suggest that the reduced QD luminescence efficiency is not the result of an increased

non-radiative recombination rate.

This was the first study offering detailed insights into the efficiency roll-off in


Figure 5-10: Transient PL of QDs under electric fields in a QD-LED. Time-resolved PL spectra of QDs in our QD-LED, reverse biased at 0, -8, -12, and -16 V.Time constants of the decays (inset) are independent of the applied voltage, suggestingthat the non-radiative exciton recombination rate is independent of the electric field.


QD-LEDs - a must for designing high-brightness QD-LEDs. It demonstrated defini-

tively the contribution of electric field-induced QD PL quenching on QD-LED EQE,

a conclusion that has since been corroborated [11]. Subsequently, others have re-

ported evidence of roll-off dominated instead by Auger recombination in their QD-

LEDs [6,229,230]. While this apparent dichotomy has yet to be explicitly resolved in

detail, we hypothesise that it can probably be broadly interpreted as a manifestation

of variations in the electronic confinement of the QDs used in different experiments

by different groups. By "electronic confinement", we mean the degree of spatial con-

finement of an exciton’s electron and hole wavefunctions within the QD’s core. It is

dictated by the energy levels of the valence and conduction bands of the core relative

to the shell. For example, whereas overcoating a CdSe core with a ZnS shell tight-

ens an exciton’s confinement in the core, overcoating with a CdS shell reduces the

confinement [231]. In the case of CdSe-ZnS QDs, the Type-I energy offset between

CdSe and ZnS presents large barriers to both electrons and holes, confining both to

the core. In contrast, the quasi-Type-II band diagram of CdSe-CdS constrains the

hole to the CdSe core while presenting only a small energetic barrier to delocalisation

of the electron into the shell.

It has been shown that PL quenching by multicarrier Auger recombination is

very significant in highly confined QDs because strong spatial confinement promotes

carrier-carrier interactions, and that it is reduced in less confined QDs [232, 233].

As an analogous form of ‘confinement’, it has been seen experimentally [229, 233]

and theoretically [234] that ‘alloyed’ QDs (graded compositions of cores or of core-

shell interfaces) negate Auger recombination due to a ‘smoothing’ of confinement

potentials. These confinement effects were shown to affect the roll-off of some QD-

LEDs [6, 229, 230]. But there is a tradeoff between mitigation of charge- and elec-

tric field-induced PL quenching in QDs. Lower confinement simultaneously enables

electric fields to more easily polarise the exciton wavefunction [11], a phenomenon

particularly evident along the long (low-confinement) axis of nanorods, which exhibit

pronounced Stark shifts in electric fields [235–237]. This tradeoff is illustrated in Fig.

5-11, which organises the different QD PL quenching mechanisms discussed above in


Figure 5-11: Influence of QD electronic confinement on bias-driven PLquenching mechanisms. Schematic shows the qualitative dependence of electricfield-induced quenching and Auger recombination on QD electronic confinement. Theexample band diagrams show that whereas confinement increases in the Type-I energyalignment of CdSe-ZnS QDs, it is decreased by the quasi-Type-II offsets in CdSe-CdSQDs. Adapted from ref. [212]

terms of the degree of electronic confinement. It is likely that this tradeoff may be

responsible for the attribution of roll-off to electric field by some and to charging by

others. For example, observations of charge-induced roll-off in QD-LEDs are often

based on devices employing QDs with Type-I band alignments [6], which are therefore

more susceptible to Auger recombination than electric field effects. This is not always

the case [229], however. Additional, more systematic investigations to correlate QD

chemistry (which may, without intention, vary subtly but importantly with synthesis

procedures) with roll-off mechanisms are needed, as discussed in Section 7.2.

Chapter 6

Zero-Bias QD Luminescence

Quenching in Infrared QD-LEDs

Chapters 4 and 5 examined the effect on QD-LED performance of in situ QD photolu-

minescence (PL) quenching that occurs as a function of applied bias. In this chapter,

we turn our attention instead to zero-bias (occurring even at zero bias) contributions

to PL quenching and QD-LED performance. By chemically passivating the surface

of near-infrared (NIR)-emitting QDs, we are able to significantly mitigate in situ PL

quenching and realise devices of record efficiency.

6.1 Bias-independent QD quenching mechanisms

In the previous chapters, we have seen how bias can induce quenching both by a

reduction in the radiative recombination rate of QDs (𝑘r) and by an increase in their

non-radiative rate (𝑘nr), per Eq. 5.1. At zero bias, past QD spectroscopy work and

a limited number of QD-LED studies suggest that the primary contribution to PL

quenching is by way of an increase in 𝑘nr, via electronic trap states on QDs and their

neighbouring environment (Fig. 5-2, left). For example, if a defect state exists on the

surface of a QD, trap-assisted non-radiative recombination can occur, whereby either

the electron or hole of an exciton relaxes to the trap state, before recombining with the

other carrier without photon emission [238]. (One specific bias-dependent instance of

91

Chapter 6. Zero-Bias QD Luminescence Quenching in Infrared QD-LEDs 92

this, relevant to quenching in operating devices, is heat-assisted quenching, whereby

heat excites a carrier out of the QD into a surface trap state [222, 239]. This was

discussed in Section 5.1.) A diversity of non-radiative interactions between excitons

in QDs and neighbouring materials is also possible in QD-LEDs. For example, non-

radiative energy transfer of an exciton to surface plasmon modes or non-radiative

mid-gap states in adjacent transport layers are possible quenching pathways that

will increase 𝑘nr [139] (Fig. 5-2). Exciton dissociation at charge transport layer

(CTL)/QD interfaces is another possibility [207,240,241]. We note that the presence

of highly conducting substrates can result in zero-bias Auger recombination (discussed

in Chapter 5) due to equilibrium charge transfer [242], but this is unlikely to occur

with typical QD-LED materials.

In thinking about how materials engineering may help characterise and control PL

quenching processes, it is instructive to think about the influence of QD composition

[212]. In addition to the electronic confinement effects depicted in Fig. 5-11 in Section

5.2.4, we can also expect the extent of surface passivation of a QD to impact the degree

of trap-assisted non-radiative recombination, represented by an additional axis in Fig.

6-1. By "surface passivation", we refer to both: the chemical passivation of QD surface

defect states (for example, by treatment with passivating ligands); and the physical

separation of an exciton in a QD core from surface trap states [71, 120] and from

the QD’s neighbouring environment (again, using long-chain organic ligands and, for

example, by overcoating QD cores with a shell material). Several chemical families of

ligands have been tailored to reduce surface defect densities while improving charge

transport through QD films [243], though the tradeoff between defects and transport

is less severe in QD light emitting applications, which therefore tend to make use of

insulating long-chain surfactants (8-18 carbons). The strict distance-dependence of

the rate of Förster Resonant Energy Transfer from QD cores to neighbouring QDs [82]

and adjacent layers means that long ligands and thick shells help reduce the quenching

effect of local defect states on adjacent QDs (‘self-quenching’, see Section 2.1.2) and

on transport layer materials. Likewise, the rate of QD-to-QD carrier tunneling has

previously been shown to depend roughly exponentially on the width and height


Figure 6-1: Influence of QD surface passivation and electronic confinementon zero-bias and bias-driven PL quenching mechanisms. Schematic shows thequalitative dependence of trap-assisted quenching, electric field-induced quenching,and Auger recombination on both QD surface passivation and electronic confinement.The example shown is overgrowth of a CdSe core by a CdS or ZnS shell. Both shellsconfer passivation, but as the corresponding band diagrams show, whereas confine-ment increases in the Type-I energy alignment of CdSe-ZnS QDs, it is decreased bythe quasi-Type-II offsets in CdSe-CdS QDs. Adapted from ref. [212]

((∆𝐸)1/2) of the potential barrier separating adjacent QDs [244], such that ligands and

shells can also impede PL quenching via charge transfer to/from the QD [229]. Fig.

6-1 portrays the common practice of adding a wide band-gap shell of ZnS or CdS to a

CdSe core-only QD. The passivating effect is experimentally reflected by an increase in

𝜂PL of almost one order of magnitude compared to native CdSe cores [74]. Passivation

is also evidenced by the recent advent of "giant" shelled CdSe QDs [110, 245]; shell

thicknesses more than double the diameter of cores confer resilience to PL quenching

even upon removal of passivating ligands.

Fig. 6-1 delineates the two variables - confinement and passivation - and the

quenching processes we postulate to dominate in each regime. We note that the


quenching mechanisms may be distinguished by their bias-dependence: whereas Auger

recombination and electric field-induced quenching, by definition, require the appli-

cation of a device bias (‘bias-driven’), trap-assisted recombination exists even in its

absence (‘zero-bias’). They can also be distinguished by their contribution to PL

quenching: either increasing 𝑘nr (trap states, Auger recombination, and electric-field

induced phonon coupling, surface state coupling, or exciton dissociation) or decreasing

𝑘r (electric field-induced exciton polarisation).

In this chapter, we ask if improved QD passivation (left to right in Fig. 6-1) can

indeed mitigate zero-bias PL quenching induced by surface trap states and the QDs’

local environment. Can this strategy be used to enhance the performance of infrared-

emitting QD-LEDs, which until now have been based almost solely on core-only QDs?

6.2 Efficient infrared QD-LEDs using core-shell QDs

6.2.1 Overview

Near-infrared (NIR) light sources integrated at room temperature with any planar

surface could be realised by harnessing the broad spectral tunability, high brightness

and solution-processability of colloidal QDs. Yet the performance of NIR QD-LEDs

has remained low compared with visible-emitting devices until now [1]. One notable

difference to date is that while visible-emitting QD-LEDs have employed core-shell

QDs (QD cores overcoated by a wider band-gap inorganic shell) almost since their

inception [132, 246], NIR QD-LEDs have, with one exception [151, 179], been based

solely on core-only QDs. As we have discussed above, the use of QDs without shells

can compromise their intrinsic 𝜂PL, increase their susceptibility to extrinsic (zero-bias)

PL quenching processes, and also reduce their photostability [75]. In this chapter, we

show that PbS-CdS core-shell QDs enhance the peak EQE of PbS core-only control

devices by 50- to 100-fold, up to 4.3±0.3 %. ‘Turn-on’ voltages are lowered by 0.6±0.2

V and per-amp radiant intensities are increased by up to 150 times. Peak EQEs and

power conversion efficiencies are more than double those of previous QD-LEDs [64]


emitting at wavelengths beyond 1 𝜇m, and are comparable with those of commercial

NIR LEDs. We demonstrate that the primary origin of the performance enhancement

is passivation of PbS cores by CdS shells against in situ PL quenching, suggesting

that core-shell QDs may become a mainstay of high-efficiency NIR QD-LEDs.

6.2.2 Introduction

Electrically-driven and wavelength-selectable NIR and shortwave infrared (SWIR)

light sources that can be deposited on any substrate and at lower cost [1,2] than ex-

isting (usually epitaxially grown) IR-emitters have the potential to enhance existing

technologies and to trigger the development of new ones. Beneficiaries could include

optical telecommunications and computing [77,167]; bio-medical imaging [77,171,247];

on-chip bio(sensing) and spectroscopy [55, 77, 172]; and night-time surveillance and

other security applications. Among the candidate large-area planar light-emitting

technologies, colloidal QD-LEDs and organic LEDs (OLEDs) stand out, as they en-

able room-temperature processing and non-epitaxial deposition. Between them, QD-

LEDs should perform better at wavelengths beyond 𝜆 = 1 𝜇m because in this spectral

range the luminescent organic dyes used in OLEDs exhibit lower 𝜂PL (<5 %) [166])

than the colloidal QDs used in QD-LEDs. The PL emission wavelength of QDs can

be precisely tuned through quantum-confinement effects, with high 𝜂PL maintained

throughout the visible and NIR (𝜂PL > 95% [120] and 𝜂PL > 50%, respectively, in

solution). This can be appreciated directly in Fig. 6-2b, which is a false-colour

InGaAs camera image of six different PbS QD samples emitting at evenly spaced

wavelength intervals between 1.2 and 1.7 𝜇m under ultraviolet illumination. Fur-

thermore, solution processability of QDs enables their epitaxy-free integration with

Si-electronics [94]. NIR QD-LEDs with peak EQEs of 8.6 % at 𝜆 > 1 𝜇m [183] and 2

% at 𝜆 > 1 𝜇m [64] have so far been demonstrated, compared with 6.3 % [174] and

<0.3 % [166] for OLEDs, respectively.

In this work, we significantly raise the performance of 𝜆 > 1 𝜇m NIR QD-LEDs

by adopting strategies used in visible-emitting QD-LEDs (which have demonstrated

EQEs of over 20 % in the red [140]). Based on the discussion above, our primary


Figure 6-2: Infrared QD PL, as seen by eye and with an infrared camera. Sixvials of colloidal PbS QDs illuminated with an ultraviolet lamp (475 nm), as seen:a, by eye, captured with a regular camera; and b, with an infrared camera. Thesamples emit at roughly evenly spaced wavelengths intervals between 1.2 and 1.7 𝜇m.In b, three images were collected with an InGaAs camera (sensitive from 900-1700nm): one with a 1300 nm bandpass filter; a second with a 1475 nm bandpass filter; athird with a 1600 nm bandpass filter. The three images were each assigned a differentprimary colour and then overlaid. Courtesy of Jessica Carr.


strategy is to emulate visible-emitting QD-LEDs in their use of QDs overcoated by

wide band-gap shells [132, 246]. NIR QD-LEDs have, with one exception [151, 179],

employed solely core-only QDs, exposing the QDs to the quenching pathways out-

lined above. Secondly, the use of ‘type-IV’ (organic-QD-inorganic hybrid) QD-LED

architectures (as defined in Section 2.2), which have shown the highest EQEs, has

heretofore been restricted almost exclusively to visible-emitting devices [1]. Here we

report a novel NIR ‘type-IV’ QD-LED and demonstrate that the use of core-shell

PbS-CdS QDs results in an enhancement of peak EQEs of one to two orders of mag-

nitude over PbS core-only controls. This is shown to be a consequence of increased

in situ QD 𝜂PL accrued by virtue of the passivating CdS shell. The peak EQEs and

power conversion efficiencies obtained surpass those of any previously reported 𝜆 > 1

𝜇m NIR QD-LED [64].

6.2.3 Methods summary

Core-only (control) oleic acid-capped PbS QDs are synthesised using the procedure

described in Section 6.2.6 at the end of this chapter. The QDs are then subjected

to a partial Pb2+-to-Cd2+ cation-exchange reaction [75] (Fig. 6-3a, Section 3.2, and

Section 6.2.6), whereby the addition of a large excess of cadmium oleate yields core-

shell PbS-CdS QDs, also capped with oleic acid. Growing wide band-gap shells

like CdS on PbS cores by cation-exchange likely circumvents mismatch of crystal

structures encountered in traditional shell growth techniques.

The core-only PbS QDs have a mean diameter of 4.0 nm, deduced by transmission

electron miscroscopy and from the spectral position of the 1S excitonic absorption

peak. This corresponds to a peak PL wavelength of 𝜆 = 1315 nm in solution. By

cation-exchange we synthesised two types of thin-shelled PbS-CdS QDs and one type

of thicker-shelled PbS-CdS QDs. The thin-shelled QDs have core diameters of 3.6 nm

and shell thicknesses equivalent to 0.2 nm (assuming uniform shell coverage), but are

distinguished by their synthesis injection temperatures of 80 ∘C (PbS-CdS(0.2 nm,

Sample A)) and 100 ∘C (PbS-CdS(0.2 nm, Sample B)), respectively. The thicker-

shelled PbS-CdS QDs have a core diameter of 2.6 nm and a shell thickness equivalent


Figure 6-3: QD synthesis and QD-LED design. a, Schematic illustration of thecation-exchange reaction used to convert core-only PbS QDs into core-shell PbS-CdSQDs. b, Device architecture (left) and cross-sectional scanning electron microscope(SEM) image (right) of the ‘type-IV’ QD-LED based on these QDs. c, The QD-LED’s flat-band energy level diagram. Band energies are in eV (referenced to thevacuum level) and are taken from literature [122, 248–253]. Based on the model inref. [249], the electron affinity of PbS is tuned from approximately 3.8 eV to 3.9 eVby the reduction in core size (from 4.0 nm to 3.6 nm) accompanying cation-exchange.


to 0.7 nm (PbS-CdS(0.7 nm)) (again assuming uniform shell coverage). These dimen-

sions were determined by wavelength dispersive spectroscopy (see Section 6.2.6). 𝜂PL

of these QDs, measured in chloroform solution, are shown in Fig. 3-2b.

Using these QDs, we fabricate QD-LEDs comprising a zinc oxide (ZnO) sol-

gel-derived, solution-processed electron-transport layer (ETL) and an organic small

molecule hole-transport layer (HTL) of 4,4-bis(carbazole-9-yl)biphenyl (CBP), sand-

wiching a film of QDs in an inverted ‘type-IV’ device architecture (as defined in Sec-

tion 2.2). Thin films of indium tin oxide (ITO), gold (Au), and molybdenum oxide

(MoO3) serve as cathode, anode, and hole-injection layers [253, 254], respectively. A

schematic diagram of the QD-LED and a scanning electron microscope (SEM) image

of its cross-section are shown in Fig. 6-3b. Its corresponding energy band-diagram is

depicted in Fig. 6-3c. As previous reports have shown [254], MoO3 plays the vital role

of mediating efficient hole-injection from the Au anode into the deep highest occupied

molecular orbital (HOMO) of CBP.

6.2.4 Results and discussion

Fig. 6-4a presents EL spectra for the PbS and PbS-CdS QD-LEDs described above.

Their overall agreement with corresponding PL spectra from the same locations con-

firms pure NIR QD EL at a wavelength of ∼1.2 𝜇m. In the absence of a QD layer,

high voltages (>4 V) induce low-intensity visible EL characterized by two peaks at

𝜆 = 400 nm and 𝜆 = 670 nm (see Fig. 6-4a, Inset), resulting from CBP and/or ZnO

emission. However, upon addition of a QD layer, this parasitic emission is entirely

suppressed, indicative of efficient carrier recombination solely within the QDs. Fig. 6-

4a shows that a blue-shift in EL and PL of more than 100 nm is achieved by adding a

0.7 nm CdS shell to core-only PbS. This is indicative of increased exciton confinement

due to a reduction in core size during cation-exchange [75]. By fabricating similar

devices with other batches of PbS and with thicker CdS shells, we have so far tuned

EL peaks from PbS-CdS QDs between 𝜆 = 1163 nm and 𝜆 = 1341 nm. NIR EL and

current-dependent brightness is visualized with an infrared camera (Fig. 6-4d, Inset;

a corresponding video is available as online Supplementary material to the original


publication of this work, ref. [15]).

The effect of shell growth on NIR QD-LED performance is shown in Fig. 6-4b-d,

which compare core-only PbS QDs with PbS-CdS(0.2 nm, Sample B) QDs. Results

for the other core-shell QDs are summarised in Fig. 6-4b and in Appendix A, Fig.

A-1 to A-4. Fig. 6-4b shows that the current density-voltage (J -V ) behavior for both

core-only and overcoated QDs is described by 𝐽 ∝ 𝑉 𝑚+1, a signature of trap-limited

space-charge conduction [94, 255]. The corresponding radiance-voltage (L-V ) plot

(Fig. 6-4c) shows that replacing the PbS QDs with PbS-CdS(0.2 nm, Sample B) QDs

results in a reduction in ‘turn-on’ voltage (measured at 3 × 10-4 W sr-1 m-2) from

2.0 ± 0.1 V to 1.4 ± 0.1 V. This may be due to more efficient QD charge injection,

suggested by the simultaneous reduction in ‘turn-on’ current density of more than

an order of magnitude with increasing shell thickness [246] (Appendix A, Fig. A-

5). Taken together, the J -V and L-V data yield an average peak power conversion

efficiency of 2.4 %; more than double that of any previous 𝜆 > 1 𝜇m NIR QD-LED [64]

(Appendix A, Table A.1).

EQE-current density data is shown in Fig. 6-4d. Core-shell PbS-CdS(0.2 nm,

Sample B) QD-LEDs exhibit average peak EQEs of 4.3 ± 0.3 %. To our knowledge,

this surpasses that of any previously reported 𝜆 > 1 𝜇m NIR QD-LED [64] (Fig. 6-5

and Appendix A, Table A.1). This EQE is also commensurate with those typical of

commercial NIR LEDs (3 % to 12 %) [256], though at significantly lower emission

powers. In comparison, the average peak EQE of control core-only PbS QD-LEDs is

approximately 50 to 100 times lower: 0.05±0.01 %; consistent with the performance of

QD-LEDs previously reported using core-only PbS with similar long-chain carboxylic

acid ligands [64]. As shown in Fig. 6-6b (black columns), EQE is sizably enhanced for

all core-shell QDs investigated. A direct consequence of this is that radiant intensity

efficiencies (radiant intensity per amp) are increased at all current densities above

‘turn-on’ by between 35 and 150 times by switching from core-only to core-shell QDs

(Appendix A, Fig. A-5 and A-6). The grey columns in Fig. 6-6b represent the average

peak EQEs observed from an anomalous (not included in any averaged performance

data) ‘champion’ set of devices: up to 8.3 ± 1.0 % for PbS-CdS(0.2 nm, Sample B)


Figure 6-4: QD-LED performance using core-only versus core-shell QDs. a,Normalised EL (solid, 3 V applied bias) and PL (dashed) spectra of QD-LEDs con-taining: (black) PbS; (red) PbS-CdS(0.2 nm, Sample A); (green) PbS-CdS(0.2 nm,Sample B); and (blue) PbS-CdS(0.7 nm) QDs. Inset: Normalised EL spectra of oth-erwise identical LEDs (with the structure shown in Fig. 6-3) with and without a filmof QDs (8 V and 6 V applied biases, respectively). b, Average current density-voltage(J -V ) behavior of: (black) core-only PbS; and (green) core-shell PbS-CdS(0.2 nm,Sample B) QD-LEDs. The dashed grey lines are fits to 𝐽 ∝ 𝑉 𝑚+1, with 𝑚 = 0 and𝑚 > 1 corresponding to Ohmic and trap-limited space-charge conduction regimes,respectively. The arrows indicate EL ‘turn-on’, which coincides with the upturn inslope of the J -V curves. c, Radiance-voltage characteristics of these devices. Solidlines represent average radiances and dotted lines correspond to minimum and max-imum ‘turn-on’ voltages recorded for all measured devices. d, EQE-current densityperformance of these devices. Black and green error bars indicate experimental mea-surement errors while the standard deviation of device-to-device variations are shownin grey (Appendix A, Section A.1). Inset: photograph of an array of five QD-LEDstaken with an infrared camera, with the middle device turned on (active area of 1.21mm2) and emitting at a centre wavelength of 𝜆 = 1242 nm.


Figure 6-5: Progression in efficiencies over time of visible- and near-infrared-emitting QD-LEDs, and comparison with this work and with commercialnear-infrared LEDs. Visible-emitting QD-LED peak EQEs (a substantial but non-exhaustive selection from the literature, reproduced from our recent review [1]) arerepresented by filled black circles. Peak EQEs of near-infrared (NIR) QD-LEDs (𝜆 > 1𝜇m), represented by filled green circles, are taken from Table A.1 in Appendix A. Solidlines connect new record values. The 4.3 % average peak EQE of the core-shell NIRQD-LEDs reported here is shown by a green star. The range of EQEs of typicalcommercially-available NIR LEDs are also indicated.

QD-LEDs (Appendix A, Fig. A-7). Repeated measurements of these ‘champion’

devices manifested persistently stable operation, suggesting that NIR QD-LEDs with

EQE >10 % may be within reach.

Since the EQE of a QD-LED is proportional to 𝜂PL of its emitting QDs (Eq. 2.1),

we can better understand the dramatic increase in EQE that comes from replacing

core-only QDs with core-shell QDs by quantifying the correlation between changes

in EQE and in 𝜂PL of the QDs in our devices. As we discussed in Section 6.1, even

at zero-bias, 𝜂PL can be subject to in situ PL quenching mechanisms such as exciton

energy transfer to conductive charge-transport layers (CTLs) [137, 257, 258], exciton

dissociation at CTL/QD interfaces [207,240,241], and Auger recombination [242].

We therefore measure relative 𝜂PL of each type of QD in glass/ZnO/QD/CBP


and glass/ZnO/QD structures (Fig. 6-6b, Inset) that replicate the QDs’ in situ

environment in our QD-LEDs. Relative 𝜂PL is quantified by comparing the integrated

areas under the steady-state PL spectra of the QDs in these structures (see Section

6.2.6). The results for each type of QD are shown in Fig. 6-6b relative to 𝜂PL of

core-only PbS, which has been normalised to unity. Whereas shell growth produces

only a three-fold rise in 𝜂PL of QD-only (glass/QD) samples (blue in Fig. 6-6a,b)

(and similarly when measured in solution, Fig. 3-2b), relative 𝜂PL of these QDs in

situ (orange and green in Fig. 6-6b) increases by 30 to 100 times, indicating that the

CdS shell significantly passivates PbS QDs against in situ non-radiative pathways.

The similarity of the results for test structures with and without CBP (orange versus

green in Fig. 6-6b) implies that ZnO is mostly responsible for the PL quenching. The

generally good agreement between relative in situ 𝜂PL and average peak EQE (black

columns) indicates that the shell-induced increases in 𝜂PL predominantly explain the

efficiency enhancements achieved. This is further substantiated by the reproducibility

of this correlation for EQE data corresponding to all measured current densities;

examples at ∼10−4 A cm-2 and ∼10−2 A cm-2 are shown in Appendix A, Fig. A-8.

Time-resolved PL spectroscopy offers further insights into the shell-dependent QD

PL quenching by ZnO. Fig. 6-7 shows the normalised transient QD PL decay curves

for the glass/ZnO/QD samples. Compared with glass/QD samples (Appendix A,

Fig. A-9), PL lifetimes are reduced for all of the QD types in the presence of ZnO,

indicative of PL quenching; similar changes have previously been seen for PbS QDs

with other metal oxides [259]. The longer PL lifetimes of core-shell versus core-only

QDs (Appendix A, Table A.2) corroborate our conclusion that the former are more

effective at mitigating PL quenching [242,259,260].

What specific mechanisms, symbolised by the blue arrow in the inset of Fig. 6-7,

underpin the ZnO-induced QD PL quenching in our devices? Non-radiative energy

transfer from visibly-coloured QDs to surface plasmon modes in conductive metal

oxide CTLs of QD-LEDs has previously been postulated [137], but is unlikely here

given the low carrier density of our undoped ZnO [261, 262]. Indeed in general,

energy transfer from PbS QDs to ZnO - for example, to ZnO mid-gap states - seems


Figure 6-6: Correlation of QD-LED EQE with QD PL quantum yields. a,Absolute PL quantum yields (𝜂PL) of drop-cast films of core-only (PbS) and core-shell(PbS-CdS) QDs on glass as a function of shell thickness, measured using an integratingsphere (described in Section 6.2.6). Cation-exchange serves as a facile post-synthesisapproach to increasing 𝜂PL of NIR-emitting QDs to unprecedented values (𝜂PL = 33%in thin-film for a 0.7 nm shell). Error bars show standard deviations of the valuesmeasured at four locations per sample. b, Corresponding dependence of QD-LEDaverage peak EQE on QD shell thickness is indicated by black columns. These EQEscorrelate well with average relative in situ 𝜂PL of spin-cast films of these QDs (𝜂PL

of core-only QDs is normalised to unity) in: (green spots) glass/ZnO/QD; and (or-ange triangles) glass/ZnO/QD/CBP optical structures, depicted schematically in theinset. In contrast, relative intrinsic QD 𝜂PL (blue squares), taken from a, greatly un-derestimate the EQE enhancements. Average peak EQEs obtained from a ‘champion’set of devices are shown as grey columns. The EQE error bars for the main devices(black columns) and the ‘champion’ devices (grey columns) indicate device-to-devicevariations (see Section 6.2.6 and Section A.1 for details). The QD 𝜂PL error bars showstandard deviations between values measured at roughly five locations per sample.


Figure 6-7: Transient spectroscopy of shell-dependent QD PL quenching.Normalised time-dependent PL intensity of spin-cast films of: (black) PbS; (red)PbS-CdS(0.2 nm, Sample A); and (green) PbS-CdS(0.2 nm, Sample B) QDs, all onZnO. Inset: Illustrations depicting the passivation by CdS shells of ZnO-induced PbSPL quenching (blue arrow).

at odds with the very low extinction coefficients reported for sol-gel ZnO in the

NIR [262, 263]. Alternatively, electron transfer from PbS QDs to ZnO, which is

widely observed in photovoltaics [207], could be responsible. As has been noted in the

context of CdSe-CdS QD quenching by ZnO nanocrystals [142], however, the energetic

favourability of the ZnO/QD dissociative interface for electron transfer is expected

to be voltage-dependent (forward biasing mitigates electron transfer) whereas the

ZnO-induced PL quenching that we observe is, at least in order-of-magnitude terms,

voltage-independent (Appendix A, Fig. A-8). Hole transfer to ZnO mid-gap states

[142] is therefore more plausible, yet the challenge in reconciling either energy or

charge transfer models with the extremely short length scales on which PL quenching

occurs (∼0.2 nm of CdS shell) points to a second possibility; quenching by defect

states arising from the oxidation of PbS QDs by ZnO.

The roles of the QD core-shell structure and of the QD-LED ‘type-IV’ architec-


ture are inextricably linked. While the former largely explains the enhancement in

EQEs observed, the latter is probably central to their high absolute magnitudes.

The ‘type-IV’ design is conducive to efficient carrier recombination because it al-

lows the processes of charge-transport and light-emission to be decoupled. Strikingly,

the only two previously reported NIR QD-LEDs not possessing a ‘type-I’ structure

(a more rudimentary architecture in which QDs play dual roles of CTL and emis-

sion centre, see Section 2.2) have also been the highest performing [64,183]. With the

present demonstration we propose that together, cation-exchanged core-shell QDs and

a ‘type-IV’ device architecture offer a new paradigm for the design of high-efficiency

NIR QD-LEDs.

6.2.5 Conclusion

This work demonstrates that the efficiency of NIR QD-LEDs can be enhanced by up

to 100 times by replacing NIR core-only QDs with core-shell QDs, as a result of a

corresponding increase in in situ QD 𝜂PL. As the demand for high brightness - and

therefore high current density - QD-LEDs continues to necessitate the use of metal

oxide charge-transport materials [162] (that commonly quench QD PL, even at zero

bias), the use of core-shell QDs can serve as a general strategy by which to maximise

the performance of future NIR QD-LEDs.

6.2.6 Detailed Experimental Methods

QD and zinc acetate synthesis and characterisation

Based on previously published procedures, core-only PbS QDs were synthesised using

a large-scale method [264] and core-shell PbS-CdS QDs were synthesised by cation-

exchange [75, 265], as described below. QD shell thicknesses were deduced from ele-

mental analysis of drop-cast films on glass substrates by wavelength dispersive spec-

troscopy [265].

PbS: A lead precursor is prepared by heating 30 mmol of lead acetate trihydrate


(Sigma Aldrich), 125 mL of octadecene (Sigma Aldrich), and 175 mL of oleic acid

(TCI) in a round-bottomed flask for 12 hours under vacuum at 100 ∘C. A sulphur

precursor is prepared by adding 15 mmol hexamethyldisilithiane (Fluka) to 150 mL of

octadecene. The lead precursor is placed under nitrogen and heated to 150 ∘C before

injection of the sulphur precursor. After cooling to room temperature, this solution

is transferred into a nitrogen-filled glove box without air exposure. The synthesised

PbS QDs are purified by adding methanol and 1-butanol until the solution is turbid,

followed by centrifugation to precipitate the QDs. The supernatant is discarded and

the QDs are redissolved in hexane [266].

PbS-CdS: PbS-CdS core-shell QDs are prepared by cation-exchange [75, 265]. To

control shell thickness, exchange is performed at 80 to 100 ∘C for between 5 min and

2 hours under nitrogen using an excess of cadmium oleate. Also see Section 3.2.

Zinc acetate dihydrate solution: The ZnO precursor is prepared according to

a variant of well-established procedures [252, 267], dissolving zinc acetate dihydrate

(Zn(CH3COO)2·2H2O, Aldrich, 99.999 %, 12.56 g) and ethanolamine (NH2CH2CH2OH,

Aldrich, >99.5 %, 3.2 mL) in anhydrous 2-methoxyethanol (CH3OCH2CH2OH, Aldrich,

99.8 %, 76.8 mL) under agitation in an ultrasonic bath for 4 hours. The solution was

stored in a nitrogen-filled glovebox.

QD-LED and optical sample fabrication

Prior to device fabrication, the QDs undergo three rounds of precipitation and cen-

trifugation with methanol and 1-butanol and are then redissolved in chloroform at 40

mg mL-1 and filtered (0.45 𝜇m PTFE). Devices are fabricated on glass substrates that

are pre-patterned with a 150 nm thick film of indium-tin-oxide (ITO) (obtained from

Thin Film Devices) and cleaned with solvent and oxygen-plasma. First, a zinc acetate

dihydrate solution is spin-cast onto the ITO at 2000 rpm for 30 s in a nitrogen-filled

glovebox. The xerogel is then sintered on a hot-plate at 300 ∘C for 5 min in the

10-20 % relative humidity environment of a glove-bag continuously flushed with dry


air. This process results in the crystallisation of 55 ± 1 nm (by optical ellipsometry

(Gaertner Scientific)) thick ZnO films and mitigates the sensitivity of ZnO to varia-

tions in atmospheric conditions. QD films are then deposited by spin-coating ∼ 40𝜇L

of ∼ 5 mg mL-1 PbS (or PbS-CdS) QDs in chloroform at 1500 rpm for 60s; film

thicknesses of 8 ± 1 nm are obtained, as determined by thickness-concentration cal-

ibration curves measured by profilometer (Tencor P-16) (Appendix A, Section A.2).

This level of accuracy is necessary in order to allow the effect of shell thickness to be

studied independently from the effects of QD film thickness. A 150 nm thick HTL of

4,4-bis(carbazole-9-yl)biphenyl (CBP) (purified before use via thermal gradient sub-

limation), a 10 nm thick hole-injection layer of molybdenum oxide (MoO3), and a

100 nm thick Au anode (shadow masked) are then successively deposited via thermal

evaporation at a rate of ∼0.1 nm s-1 at a base pressure of 1 × 10-6 Torr. Overlap of

anode and cathode defines active area pixels of 1.21 mm2. The glass/ZnO/QD/CBP

and glass/ZnO/QD structures are fabricated by the same methods as the QD-LEDs.

QD-LED characterisation

Current density-voltage (J -V ) characteristics are recorded in a nitrogen-filled glove-

box using a computer-controlled Keithley 2636A current/voltage source meter. Simul-

taneously, front face EL power output through ITO is measured using a calibrated

Newport 818-IR germanium photodiode and recorded with a computer-interfaced

Newport Multi-Function Optical Meter 1835-C. The photodiode’s active area is aligned

with the emissive pixel and a diaphragm between the two prevents collection of waveg-

uided EL from the glass substrate. Total radiated power (from which radiance-voltage

(L-V ) characteristics and power efficiency are obtained) is then calculated by assum-

ing Lambertian emission and by accounting for the wavelength dependence (weighted

by the device’s EL spectrum) of the photodiode’s responsivity (Appendix A, Section

A.1.2). EQE is calculated as the ratio of the number of forward-emitted photons to

the number of injected electrons, per unit time (Appendix A, Section A.1). Power

conversion efficiency is defined as the radiated optical power divided by the input

electrical power. For each QD type, L-J -V and EQE data are taken from about


three pixels per device, and from devices fabricated on two to five separate runs; the

only exceptions are the ‘champion devices’, for which data comes from three to five

pixels on one ‘champion’ device (see Appendix A, Section A.1). In all cases, variations

between data for a given QD type are referred to as "device-to-device" variations in

this chapter.

Optical measurements

NIR EL and PL spectra, transient PL data, and 𝜂PL from the integrating sphere are

collected in air, before which active sample areas are hermetically encapsulated by a

glass coverslip in a nitrogen-filled glovebox using a solvent-free self-curing two-part

epoxy (Torr Seal) rated to 10-9 Torr.

NIR EL and steady-state PL data are collected through a 𝜆 = 850 nm long-pass

filter with a Princeton Instruments Spectra Pro 300i spectrometer coupled to a liquid

nitrogen-cooled Princeton Instruments OMA V InGaAs CCD array detector. The ex-

citation source for steady-state NIR PL is a 𝜆 = 655 nm diode laser (Thorlabs), which

ensures that in the relative 𝜂PL measurement only the QDs, and not ZnO or CBP, are

significantly excited. The fixed geometry of the setup maximises the comparability

of relative PL intensities. A fiber-coupled Ocean Optics SD2000 spectrometer is used

to monitor for visible EL.

Transient PL data is measured by exciting glass/ZnO/QD samples at 𝜆 = 633 nm

and collecting resulting PL using superconducting nanowire single photon detectors,

as previously reported [268].

For both steady-state and transient measurements of relative in situ QD 𝜂PL, care

was taken to record data in a low excitation-intensity regime (0.5 mW cm-2 for the

steady-state experiments) where PL intensity was not decaying with time (due to

irreversible photobleaching), and at many sample locations to confirm homogeneity.

To extract relative 𝜂PL, steady-state PL spectra were scaled to account for the spectral

responsivity of the InGaAs CCD and for differences in relative absorbance between

QD samples.

Absolute 𝜂PL measurements and calculations are performed using an integrating


sphere and calibrated reference detectors as described previously [266], except that

here we account for the wavelength dependence (weighted by the PL spectra of the

drop-cast QD films) of the photodiode’s responsivity using an approach analogous to

that for calculating EQE and radiance (Appendix A, Section A.1). It was necessary

to measure 𝜂PL for drop-cast (rather than spin-cast) films in order to detect sufficient

signal. 𝜂PL is measured at four locations per sample, giving the means and standard

deviations plotted in Fig. 6-6a.

NIR EL photographs are taken with a Xenics Xeva-2.5-320 HgCdTe short-wave

infrared camera.

Chapter 7

Conclusion

7.1 Thesis summary

The mainstream commercialisation of colloidal QDs for light-emitting applications

has already begun. Sony televisions emitting QD-enhanced colours have been on sale

since 2013 [2], and optical down-conversion in QDs has been used to manufacture

solid state light (SSL) bulbs with enhanced colour quality and reduced power con-

sumption [1]. The bright and uniquely size-tunable colours of solution-processable

semiconducting QDs also highlight the potential of electroluminescent QD-LEDs for

use in energy-efficient, high-colour-quality thin-film display and SSL applications.

Chapter 1 described how the threats of climate change and the drive for more sus-

tainable global development create an imperative for improvements in society’s energy

efficiency and carbon intensity, and explained why lighting technologies - including,

potentially, QD-LEDs - have an important role to play. Chapter 2 discussed the key

advantages of using QDs as luminophores in LEDs and outlined the two-decade evo-

lution of four ‘types’ of QD-LEDs that have seen efficiencies rise from less than 0.01

% to more than 20 %. Indeed, the efficiencies of state-of-the-art electrically driven

QD-LEDs now rival those of the most efficient molecular organic LEDs (OLEDs), and

full-colour QD-LED displays have now been realised, foreshadowing QD technologies

that will transcend the optically excited QD-enhanced products already available. A

cost-analysis of industrial-scale QD synthesis was also conducted, showing that QD-

111

Chapter 7. Conclusion 112

LEDs may be competitive with OLEDs in the SSL market. Chapter 3 provided an

overview of the synthesis of colloidal QDs and the quantum confinement effects that

dictate their remarkable size-tunable optical properties.

At the start of Chapter 4, we make the observation that whereas significant atten-

tion has been paid to engineering QD-LEDs with improved charge carrier recombina-

tion efficiencies, the influence of QD photoluminescence quantum yield (𝜂PL), which is

also directly proportional to a device’s EQE, has not been carefully addressed. Sub-

dividing QD PL quenching mechanisms into those that affect QD-LEDs even at zero

bias and those that are bias-dependent, Chapter 4 then focuses on bias-driven QD

PL quenching in QD-LEDs. We simultaneously measure the intensity of PL and EL

as a function of applied bias, demonstrating that the roll-off in EQE observed in QD-

LEDs at high biases can be fully explained by a commensurate reversible reduction

in QD 𝜂PL. This allows us to discount the possibility that roll-off results from charge

leakage in our devices. For two decades, efficiency roll-off has been - and continues

to be - one of the most persistent hurdles facing the realisation of high efficiencies

at high brightness in all LED technologies. Ours is the first dedicated study of this

effect in QD-LEDs, serving to narrow down the mechanisms that may be responsible.

Chapter 5 builds on the previous chapter by uncovering the specific process causing

QD PL quenching, and therefore efficiency roll-off, in our QD-LEDs. Preliminarily

ruling out heating and charge-induced Auger recombination in QDs, we hypothesise

that electric field may be the primary quenching pathway and test that hypothesis.

Using the shift in wavelength of QD PL from the quantum confined Stark effect

as a proxy for the QDs’ local electric field, we measure QD PL under reverse bias

(i.e. electric field without charge injection) to determine the field-dependent PL

quenching. This allows us to predict the forward bias-induced EQE roll-off in our QD-

LEDs. Excellent agreement with observed EQE confirms that electric field-induced

quenching is indeed sufficient to account for roll-off in our QD-LEDs - an important

guide to the engineering of future high efficiency, high brightness QD-LEDs.

In Chapter 6, we turn our attention to mitigating the influence of zero-bias quench-

ing via electronic trap states on QDs and their local environment in QD-LEDs. We


choose to use NIR-emitting QD-LEDs as a testbed, since these devices boast unique

potential. One can envision NIR light sources that can be deposited on any substrate

and at lower cost than existing (usually epitaxially grown) IR-emitters finding applica-

tion in optical telecommunications and computing (solution-processability may enable

Si-compatibility), bio-medical imaging (utilising biological transparency windows be-

tween 800 nm and 1700 nm), on-chip bio(sensing) and spectroscopy, and night-time

surveillance and other security applications. Yet to date, the performance of NIR-

emitting QD-LEDs has trailed far behind their visible counterparts. By overcoating

PbS core-only QDs with CdS shells via cation exchange, we demonstrate through

in situ steady-state and transient PL spectroscopy that even atom-thickness surface

passivation dramatically suppresses zero-bias PL quenching by the neighboring ZnO

electron transport layer in our devices. 𝜂PL is enhanced by two orders of magnitude,

yielding a record-performance NIR QD-LED with an average EQE of ∼4.3 % and

emission wavelengths in the technologically important 1.1-1.3 𝜇m range.

7.2 Outlook

This thesis has been dedicated to understanding the role of QD PL quenching on

QD-LED performance with and without bias. However, while our work on NIR QD-

LEDs yielded a solution to overcoming some of these challenges at zero bias, we have

not yet engineered a solution to the electric field-induced roll-off mechanism that

we have identified. Indeed, further work is required to deepen our understanding of

this process and its interplay with charge-induced Auger recombination in QD-LEDs,

which has since been observed by others to sometimes also be involved. Moreover,

based on our reviews of the QD-LED field and observation of its progress with respect

to the SSL and display markets [1, 2], key research hurdles facing QD-LEDs include

better understanding and optimising blue emission, outcoupling, and device lifetime.

We briefly discuss directions each of these topics could take.

∙ Understanding QD-LED efficiency roll-off through QD composition:

Based on our discussion in Section 5.2.4, our understanding of the interplay


Figure 7-1: Progression of blue-emitting LED EQEs over time. EQEs overtime of blue-emitting OLEDs (black) compared with QD-LEDs (blue). OLEDs aredistinguished as either PhOLEDs (circle) or TADF OLEDs (square). Filled symbolsshow peak EQEs whereas open symbols show EQEs as high brightness (1000 cd m-2),illustrating greater efficiency roll-off at high biases in OLEDs than QD-LEDs.

between electric field and Auger recombination in causing EQE roll-off in QD-

LEDs may be tested by investigating the effect of QD composition - specifically

electronic confinement - on roll-off. Klimov et al. have recently presented

evidence of roll-off due to Auger recombination and its correlation with QD shell

chemistry [6,229], yet similar efforts should be undertaken to better understand

and mitigate electric field-induced roll-off in our devices. We observe that, for

example, EQE roll-off is pronounced in our NIR-emitting QD-LEDs (Fig. 6-4)

but, based on an initial review of the literature, is apparently less so in blue-

emitting devices (Fig. 7-1). It is possible that the different conclusions we and

others have arrived at as to the cause of roll-off in QD-LEDs (electric field versus

charge) can be ascribed to subtle but systematic differences in the QDs used.

∙ Mitigating QD-LED efficiency roll-off: We propose three approaches to at

least partly address roll-off in QD-LEDs, but there are surely others.


Firstly, one could investigate the use of QDs that are optimised to negate

Auger recombination (for example, with an alloyed shell) [229,233,234] but

whose emission derives from a localised trap state that is less sensitive to

the quantum confined Stark Effect. For example, QD-LEDs might employ

QDs comprising a luminescent impurity like Mn+ [269–271] (note these

QDs have already been successfully employed in a type-III QD-LED [162]).

Another approach, proposed by Bozyigit et al. [212], is to reduce the elec-

tric field felt by the QDs in a QD-LED by embedding the QDs in a material

of high dielectric constant. For example, atomic layer deposition might be

used to in-fill the QD film. Simulations suggest that the use of a matrix

material such as TiO2 (𝜀 ≥ 86) could reduce the effective electric field by

a factor of 10 [212].

An alternative strategy could be to operate QD-LEDs in an altogether

different way so as to avoid the deleterious effects of charge and field, and

in fact use them to our advantage. For example, as we have previously

proposed [272], emission from a light emitting device might be based on

optical excitation of QD PL (a "PL-LED"), with electric field (in absence

of charge injection, by way of a capacitor structure, say) applied only to

intentionally dim or turn off QD PL emission. Particularly for applications

that do not require large ON-OFF ratios, this approach leverages the most

invaluable properties of QDs - their optical ones - while avoiding the dif-

ficulties and challenges of exciting them electrically. Since thin dielectric

films can then replace thicker, often absorbing charge transport layers, this

opens the door to realising a transparent light-emitting technology.

∙ Thermal efficiency roll-off in QD-LEDs: Our thermal imaging experiments

indicated insufficient short-term temperature rises to explain the roll-off in our

QD-LEDs (Fig. 5-3). However, just as with OLEDs, prolonged operation may

nevertheless lead to a substantial rise in temperature (many tens of degrees

Celsius) in QD-LEDs, threatening both efficiency and material integrity. Sys-


tematic QD-LED junction temperature measurements as a function of current

density, akin to those performed in OLEDs [223], may be informative of their

commercial high-brightness performance limits.

∙ Blue QD-LEDs: The performance of blue-emitting QD-LEDs has improved

dramatically over the past few years [273]. Fig. 7-1 shows the progression in

record efficiencies of hybrid deep-blue (CIE 𝑦 < 0.1) QD-LEDs, which are seen

to be quickly approaching those of state-of-the-art OLEDs using phosphorescent

and thermally-activated delayed fluorescence (TADF) emitters. As Fig. 7-1

also demonstrates, blue QD-LEDs exhibit much less severe efficiency roll-off

at high biases compared to OLEDs, such that QD-LED and OLED EQEs are

comparable at high brightness. While enabling true blue EL emission, these

QD-LEDs have sub-bandgap turn-on voltages of 2.5-3 V [60, 61, 140, 141, 273]

that are lower than those of corresponding OLEDs (3.3-5.4 V [274–276]); a key

factor in ensuring small energy losses during electron-photon conversion, and

therefore in achieving high power-efficiencies [277]. As we described in Section

1.2, efficient, stable blue-emission from OLEDs remains elusive, presenting an

impediment to OLED commercialisation as a white light source.

Blue-emitting QD-LEDs have received limited attention to date compared with

red-emitting devices, so we propose a course of research investigating the two

limiting factors to the performance of blue QD-LEDs: efficiency and lifetime.

By developing understanding in these two largely unexplored areas, it may be

possible to develop efficient and stable blue QD-LEDs with the potential to

enable a new generation of white SSL sources.

To our knowledge, there has been no systematic study to uncover the factors

limiting EQEs and lifetimes in blue QD-LEDs. The substantial rise in efficien-

cies of blue QD-LEDs in just the last few years (Fig. 7-1) has been largely

attributed to apparent advances in QD synthesis chemistry. At the same time,

it is most frequently cited that blue QD-LEDs exhibit inferior performance to

their red and green counterparts because of the larger hole injection barrier at


the QD/hole transport layer (HTL) interface resulting from the deeper highest

occupied molecular orbital (HOMO) level of the higher-energy QDs [60, 273].

This argument is certainly plausible, but has not yet been validated by system-

atic investigation of the impacts of hole injection and charge balance on EQE.

Building on successes with OLEDs [118,278], and with the suspicion that excess

electron injection may detrimentally impact today’s blue QD-LEDs, one might

investigate the impact of charge (im)balance on device performance through

the use of doped charge transport layers, conductive metal oxides, and charge

blocking layers.

∙ Outcoupling in QD-LEDs: While many of the structural modifications suc-

cessfully implemented in OLEDs and GaN-based LEDs to enhance the outcou-

pling of generated light may be translatable to QD-LEDs, it is emerging that

many of the outcoupling strategies reported in the academic literature may

not be compatible with commercial applications. Outcoupling techniques must

be low cost, scalable to large area devices and manufacturing, compatible with

OLED device architectures, and robust with respect to lifetime and environmen-

tal degradation [40]. For example, plasmon losses can most easily be mitigated

by increasing the distance between the emissive layer and metal cathode, yet

this can increase device operating voltages. Solutions to this specific tradeoff

may include doped charge transport materials or unconventional transport ma-

terials such as perovskites. Interestingly, ongoing investigations of large QD

PL enhancements (between ∼2- and 50-fold) through QD-metal nanostructure

interactions suggest that carefully tailored plasmonic templates may be used to

enhance the efficiency of QD-LEDs [93,279].

∙ Lifetimes of QD-LEDs: Intrinsic QD PL lifetimes of >14,000 hours and re-

spectable PL thermal stabilities (12 % fall-off at 140 ∘C) are reported by QD

Vision, Inc [280, 281]. These are comparable with inorganic phosphors and

render QDs already commercially viable in many optical down conversion ap-

plications. However, lifetimes of electrically driven QD-LEDs remain relatively


short. QD-LED half-lives, 𝑇50, from an initial luminance of 100 cd m-2, are

typically on the order of 102-104 hours, compared to 105-107 hours for today’s

OLEDs, which have undergone more than a decade of rigorous commercial-scale

refinement [282].

Fig. 7-2 depicts the evolution of QD-LED operational lifetimes, with all devices

compared to the same initial luminance of 100 cd m-2 [60–62, 93, 94, 131, 132,

140, 140, 141, 281, 283]. Red-emitting QD-LEDs, which are the colour most

commonly reported, clearly demonstrate a steady and substantial improvement

in lifetimes over the past several years. In fact, encouragingly, people have

recently reported red- and green-emitting QD-LEDs with half-lives of nearly

105 hours when operated at 100 cd m-2 initial brightness (see Fig. 7-2).

However, to date, discussion of the lifetime limitations of QD-LEDs has been

sparse - in almost every case a side note to other kinds of device studies. Indeed,

to our knowledge, there has been no systematic study of the degradation of sta-

tistically significant populations of QD-LEDs operating over extended periods

of time in ambient room conditions; for example, no detailed characterisation

of the physical parameters of degradation or attempts to assign specific mech-

anisms to this process. In contrast, in the more established field of OLEDs,

substantial effort over the past 30 years has gone into discovering and over-

coming their failures and degradation. Initial research topics in this area might

include:

Establishing consistent benchmarks and testing procedures for QD-LED

lifetime measurement and reporting: device encapsulation, testing con-

ditions, data fitting, and extrapolation are often not consistent between

studies.

Investigating whether QD-LED ambient stability is primarily limited by

intrinsic or extrinsic degradation processes. In particular, are non-emissive

‘dark spots’ the dominant factor in short QD-LED lifetimes, as was previ-

ously found to be the case in OLEDs?


Studying whether colloidal QDs in fact confer additional stability com-

pared to organic luminophores on overall light-emitting device perfor-

mance, as is commonly supposed. Is QD-LED lifetime primarily limited

by degradation of the QDs themselves, or by their surrounding materials?


Figure 7-2: QD-LED operating lifetimes. a, Progression in reported lifetimes(𝑇50) over time of electrically-driven visible-emitting QD-LEDs, all compared at thesame initial luminance of 100 cd m-2, driven by constant current. The red-emittingQD-LED reported in 2007 (hollow red triangle) was tested under low vacuum (5×10−3

Torr) [62], whereas all other devices were tested in ambient conditions. b, RequiredOLED/QD-LED luminance for certain applications. Overlaid are the reported life-times of several green phosphorescent OLEDs, which are compared to the long-livedgreen QD-LED developed by Yang et al. [141] (unfortunately this is the only instancein which QD-LED lifetimes have been measured for more than one initial brightness).In the cases where two lifetime data points are available, a straight line extrapola-tion offers a rough projection of the device half-life at low luminances, as previouslyshown [284]. It is widely accepted that for indoor display applications, the requiredbrightness for individual pixels (depending on size, color, and device architecture) isabout 200-600 cd m-2 [285, 286]. For other applications like outdoor display, lightingand window integrated transparent devices, higher brightness is required. The indus-trial community now demands a device lifetime as high as 20,000 h at 500 cd m-2

for displays, where the lifetime is defined to reach 80 % (𝑇80) or even 95 % (𝑇95) ofthe initial brightness; that is, exhibiting a drop of only 20 % and 5 %, respectively,instead of the 50 % (𝑇50) metric typically reported by researchers.

121

Appendix A. Supplementary information to Chapter 6 122

Appendix A

Supplementary information to

Chapter 6

Figure A-1: Raw EQE-current density data with measurement errors and

device-to-device variations. EQEs of individual QD-LED pixels using: a, PbS; b,

PbS-CdS(0.2 nm, Sample A); c, PbS-CdS(0.2 nm, Sample B); and d, PbS-CdS(0.7

nm) QDs are indicated by black lines, with blue dots showing their averages. Mea-

surement errors and device-to-device variations are represented by blue and grey error

bars, respectively.


Tabl

eA

.1:Sum

mar

yof

repor

ted

NIR

QD

-LED

pea

kex

tern

alqu

antu

meffi

cien

cies

and

pow

ereffi

cien

cies

.Pea

kpe

rfor

man

cem

etri

csfo

ra

num

ber

ofre

port

edne

ar-in

frar

edQ

D-L

ED

sar

eta

bula

ted

inch

rono

logi

calo

rder

.

Ref

.Y

ear

Pea

kEQ

E(%

)Pea

kPow

erC

onve

rsio

nEffi

cien

cy(%

)

Tess

ler

etal

.[17

9]20

000.

50.

02St

ecke

letal

.[17

0]20

030.

001

<0.

1K

okty

shet

al.[

181]

2004

<0.

001

-K

onst

anta

tos

etal

.[63

]20

050.

270.

01O

’Con

ner

etal

.[18

2]20

050.

021.

2×

10-4

Bou

rdak

oset

al.[

168]

2008

1.15

0.66

Cho

udhu

ryet

al.[

169]

2010

0.83

0.08

Che

nget

al.[

183]

(𝜆<

1𝜇m

)20

116.

7(3

nmQ

Ds)

;8.6

(5nm

QD

s)0.

4(3

nmQ

Ds)

;8(5

nmQ

Ds)

Sun

etal

.[64

]20

122.

01.

0M

aet

al.[

95]

2012

0.72

0.11

This

wor

k[1

5]20

144.

32.

4

All

peak

exte

rnal

quan

tum

effici

enci

es(E

QE

s)ar

equ

oted

dire

ctly

from

the

refe

renc

es.

Unl

ess

prov

ided

,pea

kpo

wer

effici

enci

esar

eap

prox

-im

ated

byco

mpu

ting

(𝑃EL/𝐼

𝑉)×100%

asa

func

tion

ofI

orV

.Her

e,𝑃

EL

ism

axim

umem

itte

dpo

wer

,whi

ch,u

nles

sgi

ven,

isap

prox

imat

edby

extr

acti

ngpl

otte

dda

ta(P

lot

Dig

itiz

er)

and

com

puti

ng𝑃

EL=

(𝐼·E

QE·ℎ

𝑐)/(|𝑒|·

𝜆)

(whe

re𝜆

ispe

akem

issi

onw

avel

engt

h)as

afu

nc-

tion

ofI

orV

soas

tofin

dit

sm

axim

um.

Das

hes

(-)

indi

cate

unkn

own

valu

esno

tca

lcul

able

byth

ese

met

hods

.N

ote

that

𝜆>

1𝜇m

for

all

refe

renc

esex

cept

Che

nget

al.


Figure A-2: Current density-voltage behavior. Average performance of QD-

LEDs based on: (black) PbS; (red) PbS-CdS(0.2 nm, Sample A); (green) PbS-CdS(0.2

nm, Sample B); and (blue) PbS-CdS(0.7 nm) QDs.


Figure A-3: Test-to-test variations in peak EQE. Peak EQEs of individual QD-

LED pixels (arbitrarily numbered) tested multiple times using: a, PbS; b, PbS-

CdS(0.2 nm, Sample A); c, PbS-CdS(0.2 nm, Sample B); and d, PbS-CdS(0.7 nm)

QDs. Peak EQEs are shown after multiple test cycles: (black) Test 1; (red) Test 2;

(green) Test 3; (blue) Test 4; and (turquoise) Test 5. Successive test cycles yield peak

EQE values between 84 % and 104 % of the initial values for all pixels and all QD

types.


Figure A-4: Day-to-day variations in raw EQE-current density data. EQEs

for: (black and red) two pixels of PbS-CdS(0.2 nm, Sample A) QDs; and for (dark

green, light green, blue) three pixels of PbS-CdS(0.2 nm, Sample B) QDs (all ‘cham-

pion’ devices). For each pixel, data was recorded one day after fabrication (Day 1,

open circles) and again (solid lines) either two (Day 2) or four (Day 4) days after

fabrication. We see that EQE either remains unchanged or increases slightly over the

course of at least a few days.


Figure A-5: Radiance-current density characteristics. Average radiance for

QD-LEDs based on: (black) PbS; (red) PbS-CdS(0.2 nm, Sample A); (green) PbS-

CdS(0.2 nm, Sample B); and (blue) PbS-CdS(0.7 nm) QDs. Upon shell growth,

‘turn-on’ current densities (measured at 3 × 10-4 W sr-1 m-2) are reduced by 8, 16

and 31 times for PbS-CdS(0.2 nm, Sample A), PbS-CdS(0.2 nm, Sample B), and

PbS-CdS(0.7 nm) QDs, respectively. For any given current density above core-only

QD-LED ‘turn-on’, the radiance of the core-shell QD-LEDs is between 35 and 150

times higher than that of the core-only QD-LEDs.


Figure A-6: Enhancement in radiant intensity efficiency. The enhancement

in radiant intensity efficiency (radiant intensity per amp) compared with that of the

control core-only PbS QD-LED is shown as a function of current density for QD-

LEDs using: (red) PbS-CdS(0.2 nm, Sample A); (green) PbS-CdS(0.2 nm, Sample

B); and (blue) PbS-CdS(0.7 nm) QDs. ‘Turn-on’ current densities are indicated by

dashed lines. Once the core-only PbS QD-LED (black dashed line) has turned on,

the enhancement is between 35 and 150 times for all core-shell devices.


Figure A-7: EQE-current density characteristics of anomalous ‘champion’

QD-LEDs. Record performance curves for one set of QD-LEDs using: (black) PbS;

(red) PbS-CdS(0.2 nm, Sample A); (green) PbS-CdS(0.2 nm, Sample B); and (blue)

PbS-CdS(0.7 nm) QDs. Peak EQEs are 1.7 %, 7.7 %, 9.4 % and 5.3 %, respectively,

as plotted in Fig. 6-6b.


Figure A-8: Dependence of QD-LED EQE-shell thickness correlation on

current density. Average EQE data measured at current densities of ∼10−4 A

cm-2 and ∼10−2 A cm-2, extracted from Fig. A-1, are represented by green and

blue columns, respectively, and compared with the average peak EQE data (black

columns) shown in Fig. 6-6b. The error bars indicate device-to-device variations in

EQE. For all data sets a similar correlation with shell-thickness is reproduced.


Table A.2: PL lifetimes of spin-cast films of core-only (PbS) and core-shell(PbS-CdS) QDs on ZnO, calculated from Fig. 6-7. 𝜏𝑒−1 and 𝜏𝑒−2 are the timesat which QD PL intensity, I, has decreased from its initial value of 𝐼0 to 𝐼 = 𝐼0/𝑒 and𝐼 = 𝐼0/𝑒

2, respectively.

PbS PbS-CdS(0.2 nm, A) PbS-CdS(0.2 nm, B)

𝜏𝑒−1(ns) 0.4 1.2 3.6𝜏𝑒−2(ns) 1.1 8.8 17.6

Figure A-9: Transient photoluminescence of QDs on glass. Normalised time-

dependent PL intensity of spin-cast films of: (black) PbS; (red) PbS-CdS(0.2 nm,

Sample A); and (green) PbS-CdS(0.2 nm, Sample B) QDs, all on glass.


A.1 Calculation and error estimate of EQE and ra-

diance

A.1.1 EQE calculation

EQE is calculated as the ratio, per unit time, of the number of forward-emitted

photons to the number of injected electrons, 𝐼(V)/ |𝑒|, where 𝐼(V) is the current

passing through the QD-LED at an applied bias, V. We can express this as,

EQE(V) [%] =𝑁phot(V)) |𝑒|

𝐼(V)· 𝑔 · 100 (A.1)

where 𝑁phot(V) is the number of forward-emitted photons actually collected by the

photodiode, and the geometric factor, g, accounts for the solid angle of the EL profile

(assumed to be Lambertian) subtended by the photodiode, Ω = 𝜋/𝑔:

𝑔 =𝑎2 + 𝐿2

𝑎2(A.2)

Here a is the diameter of the active area of the photodiode and L is the distance

between the emitting QD-LED pixel and this active area.

We calculate 𝑁phot(V) from the current output (‘photocurrent’, 𝐼measpc ) of the pho-

todiode in response to detected EL at each applied bias, accounting for the wavelength

dependence (weighted by the device’s EL spectrum) of the photodiode’s responsivity,

𝑅(𝜆).

To do this, first note that the area below a wavelength-resolved EL spectrum

measured from a QD-LED, 𝐸𝐿(𝜆), is proportional to the number of photons emit-

ted per unit time. Then, multiplying this by the energy per photon ℎ𝑐/𝜆 and by

𝑅(𝜆)[AW−1

], we obtain values proportional to measured photocurrent,

𝐼′

pc(𝜆) = 𝐸𝐿(𝜆) · ℎ𝑐𝜆

·𝑅(𝜆) (A.3)

Integrating between the wavelength limits of the EL spectrum (𝜆𝑖 and 𝜆𝑓 ), a figure

proportional to total measured photocurrent is then given by,


𝐼proppc (𝜆) =

∫ 𝜆𝑓

𝜆𝑖

𝐼′

pc(𝜆)d𝜆 (A.4)

Scaling 𝐸𝐿(𝜆) to satisfy the requirement that 𝐼proppc (𝜆) equals 𝐼meas

pc yields a revised

EL spectrum, 𝐸𝐿(𝜆)′, the area below which is now equal to the number of photons

emitted per unit time.

𝐸𝐿(𝜆)′ = 𝐸𝐿(𝜆) ·𝐼measpc

𝐼proppc (𝜆)

(A.5)

Thus, 𝑁phot =

∫ 𝜆𝑓

𝜆𝑖

𝐸𝐿(𝜆)′d𝜆 (A.6)

Repeating this procedure for all applied biases and substituting 𝑁phot into Eq. A.1

gives voltage (or current density)-dependent EQE, as plotted in Fig. 6-4d. We note

that in the limit - often assumed for relatively narrow QD EL spectra - that emission

is monochromatic, with wavelength 𝜆0, Eq. A.6 simplifies to 𝐼measpc 𝜆0/ℎ𝑐𝑅(𝜆0) and

therefore Eq. A.1 can be rewritten as,

EQE(V) [%] =𝐼measpc

𝑅(𝜆0)· 𝜆0ℎ𝑐

· |𝑒|𝐼(V)

· 𝑔 · 100 (A.7)

Eq. A.7 is the simplified expression for EQE often quoted [7].

A.1.2 Radiance calculation

Analogously, the ‘true’ emitted power is 𝑃EL =∫ 𝜆𝑓

𝜆𝑖𝐸𝐿(𝜆)′ · ℎ𝑐

𝜆d𝜆, giving radiance as,

𝐿(V) =𝑃EL(V)

𝐴 · Ω(A.8)

where A is the QD-LED pixel area. Again, this is evaluated for all applied biases,

leading to Fig. 6-4c.


A.1.3 EQE measurement error

Eq. A.1 indicates that EQE is subject to experimental measurement errors in 𝑁phot,

I and g, namely 𝜎𝑁phot, 𝜎𝐼 , and 𝜎𝑔, respectively. 𝜎𝐼 results from Keithley 2636A

equipment resolution and 𝜎𝑔 from geometric errors in a and L in Eq. A.2. From Eq.

A.6, 𝜎𝑁phot in turn arises from errors in 𝐼measpc , 𝐸𝐿(𝜆), and 𝑅(𝜆), namely 𝜎𝐼meas

pc, 𝜎𝐸𝐿,

and 𝜎𝑅, respectively. 𝜎𝑅 results from Newport 818-IR photodiode calibration error

whereas 𝜎𝐼measpc

is due to background noise recorded by the photodiode/Newport Multi-

Function Optical Meter 1835-C measurement setup. 𝜎𝐸𝐿 is taken to be negligible.

Total measurement errors, represented by black and green error bars in Fig. 6-4d,

are calculated by combining 𝜎𝐼measpc

, 𝜎𝑅, 𝜎𝐼 , and 𝜎𝑔 in quadrature, accounting for

multiplication factors implicit in Eq. A.1. Measurement errors for QD-LEDs based

on all QD types are shown in Fig. A-1, indicated by blue error bars.

A.1.4 Device-to-device EQE variation

Fig. A-1 shows raw EQE-current density data (from about three pixels per device,

and from devices fabricated on two to five separate runs) for QD-LEDs based on all

QD types. These device-to-device variations strictly arise due to systematic errors

because they exceed the measurement errors calculated above. However, as a useful

guide, device-to-device variations are approximated by the standard deviations of raw

data from their mean, shown as grey error bars in Figs. 6-4d, 6-6b, and A-1.

The EQE errors for the ‘champion’ QD-LEDs, indicated by grey error bars in

Fig. 6-6b, represent the standard deviations in EQE of three to five pixels on single

‘champion’ devices.

A.2 QD film thickness calibration

In order to separate the effect of QD shell thickness from that of QD film thickness it

is necessary to accurately control the latter. We do this by varying the concentration

of the spin-cast QD chloroform solution. QD film thickness-concentration calibration


curves are determined for each QD batch by spin-coating QD solutions of known con-

centrations onto ZnO (prepared identically to that in the QD-LEDs) and measuring

their film thicknesses with a profilometer. To do this, an edge of each QD film is

wiped-off with a chloroform-moistened polyurethane swab (VWR). Because this pro-

cedure does not detectably affect ZnO film thickness, the step-height measured by

profilometer corresponds to QD film thickness. Roughly 15 line-scans per sample (5

line scans (𝜇m separation) at 3 positions (mm separation)) yield the data plotted in

Fig. A-10. Orthogonal distance regressions, accounting for errors in both QD con-

centration and film thickness, are then used to calculate concentrations required for

the 8 ± 1 nm films used in all QD-LEDs.


Figure A-10: QD film thickness-concentration calibration curves on zinc

oxide. Black dots indicate average thicknesses (as measured by profilometer) of spin-

cast films of QDs as a function of the QD solution’s concentration in chloroform for:

a, PbS; b, PbS-CdS(0.2 nm, Sample A); c, PbS-CdS(0.2 nm, Sample B); and d, PbS-

CdS(0.7 nm) QDs. Horizontal error bars indicate uncertainties in concentrations and

vertical error bars show standard deviations in measured thicknesses from multiple

locations per sample. Solid lines are orthogonal distance regressions (Igor Pro 6.1,

WaveMetrics, Inc.) from which concentrations used in QD-LED fabrication were

determined.

Bibliography 137

Bibliography

[1] Shirasaki*, Y., Supran*, G. J., Bawendi, M. G., & Bulović, V. Emergence ofcolloidal quantum-dot light-emitting technologies. Nature Photonics 7 (2013),13.

[2] Supran, G. J., Shirasaki, Y., Song, K. W., Caruge, J.-M., Kazlas, P. T., Coe-Sullivan, S., Andrew, T. L., Bawendi, M. G., & Bulović, V. QLEDs for displaysand solid-state lighting. MRS Bulletin 38 (2013), 703.

[3] Patel, N. K., Cinà, S., & Burroughes, J. H. High-Efficiency Organic Light-Emitting Diodes. IEEE Journal on Selected Topics in Quantum Electronics 8(2002), 346.

[4] Saxena, K., Jain, V., & Mehta, D. S. A review on the light extraction techniquesin organic electroluminescent devices. Optical Materials 32 (2009), 221.

[5] Pimputkar, S., Speck, J. S., Denbaars, S. P., & Nakamura, S. Prospects forLED lighting. Nature Photonics 3 (2009), 180.

[6] Lim, J., Jeong, B. G., Park, M., Kim, J. K., Pietryga, J. M., Park, Y. S.,Klimov, V. I., Lee, C., Lee, D. C., & Bae, W. K. Influence of shell thickness onthe performance of light-emitting devices based on CdSe/Zn1-xCdxS core/shellheterostructured quantum dots. Advanced Materials 26 (2014), 8034.

[7] Wood, V. C. Electrical Excitation of Colloidally Synthesized Quantum Dots inMetal Oxide Structures. Ph.D. thesis, Massachusetts Institute of Technology(2010).

[8] Anikeeva, P. O. Physical Properties and Design of Light-Emitting Devices Basedon Organic Materials and Nanoparticles. Phd thesis, Massachusetts Instituteof Technology (2009).

[9] Osedach, T. Colloidal Quantum Dots and J-Aggregating Cyanine Dyes forInfrared Photodetection. Ph.D. thesis, Harvard University (2012).

[10] Shirasaki, Y. Efficient Forster Energy Transfer From Phosphorescent OrganicMolecules to J-aggregate Thin Film. M. eng. thesis, Massachusetts Institute ofTechnology (2008).

Bibliography 138

[11] Bozyigit, D., Yarema, O., & Wood, V. Origins of Low Quantum Efficiencies inQuantum Dot LEDs. Advanced Functional Materials 23 (2013), 3024.

[12] Galland, C., Ghosh, Y., Steinbrück, A., Sykora, M., Hollingsworth, J. A.,Klimov, V. I., & Htoon, H. Two types of luminescence blinking revealed byspectroelectrochemistry of single quantum dots. Nature 479 (2011), 203.

[13] Pandey, A. & Guyot-Sionnest, P. Hot Electron Extraction From Colloidal Quan-tum Dots. The Journal of Physical Chemistry Letters 1 (2010), 45.

[14] Shirasaki*, Y., Supran*, G. J., Tisdale, W. A., & Bulović, V. Origin of Ef-ficiency Roll-Off in Colloidal Quantum-Dot Light-Emitting Diodes. PhysicalReview Letters 110 (2013), 217403.

[15] Supran, G. J., Song, K. W., Hwang, G. W., Correa, R. E., Scherer, J., Dauler,E. A., Shirasaki, Y., Bawendi, M. G., & Bulović, V. High performance near-infrared light emitting devices using core-shell (PbS-CdS) colloidal quantum-dots. Advanced Materials 27 (2015), 1437.

[16] Supran, G. J., Song, K. W., Hwang, G. W., Correa, R. E., Scherer, J., Shirasaki,Y., Bawendi, M. G., & Bulović, V. Near-infrared light emitting device usingsemiconducting nanocrystals. US patent application no. 14/101,867 (2013).

[17] Adoption of the Paris Agreement (FCCC/CP/2015/L.9/Rev.1) (2015).

[18] Transforming our world: the 2030 Agenda for Sustainable Development(A/RES/70/1), Resolution adopted by the General Assembly of the UnitedNations (2015).

[19] The Nobel Prize in Physics 2014 - Press Release (2014).

[20] Hoffert, M. I., Caldeira, K., Jain, A. K., Haites, E. F., Harvey, L. D. D., Potter,S. D., Schlesinger, M. E., Schneider, S. H., G.Watts, R., Wigley, T. M. L., &Wuebbles, D. J. Energy implications of future stabilization of atmospheric CO2content. Nature 395 (1998), 881.

[21] Jarvis, A. J., Leedal, D. T., & Hewitt, C. N. Climate-society feedbacks and theavoidance of dangerous climate change. Nature Climate Change 2 (2012), 668.

[22] Deep Decarbonization Pathways Project, Pathways to deep decarbonization2015 report, SDSN - IDDRI. Technical report (2015).

[23] Williams, J. H., DeBenedictis, A., Ghanadan, R., Mahone, A., Moore, J., Mor-row III, W. R., Price, S., & Torn, M. S. The technology path to deep greenhousegas emissions cuts by 2050: the pivotal role of electricity. Science 335 (2012),53.

Bibliography 139

[24] Wei, M., Nelson, J. H., Greenblatt, J. B., Mileva, A., Johnston, J., Ting, M.,Yang, C., Jones, C., McMahon, J. E., & Kammen, D. M. Deep carbon re-ductions in California require electrification and integration across economicsectors. Environmental Research Letters 8 (2013), 014038.

[25] International Energy Agency, World Energy Outlook 2012. Technical report(2012).

[26] Nauclér, T. & Enkvist, P.-A. Pathways to a low-carbon economy: Version 2 ofthe global greenhouse gas abatement cost curve. McKinsey & Company (2009).

[27] Ürge-Vorsatz, D. & Metz, B. Energy efficiency: How far does it get us incontrolling climate change? Energy Efficiency 2 (2009), 87.

[28] Rogelj, J., Luderer, G., Pietzcker, R. C., Kriegler, E., Schaeffer, M., Krey, V., &Riahi, K. Energy system transformations for limiting end-of-century warmingto below 1.5 C. Nature Climate Change 5 (2015), 519.

[29] Humphreys, C. J. Solid-state lighting. MRS Bulletin 33 (2008), 459.

[30] Haitz, R. & Tsao, J. Y. Solid-state lighting: ‘The case’ 10 years after and futureprospects. Physica Status Solidi (A) Applications and Materials Science 208(2011), 17.

[31] U.S. Departament of Energy, Solid-State Lighting R & D Plan. Technical report(2015).

[32] Waide, P. & Tanishima, S. International Energy Agency, Light’s Labour’s Lost.Technical report (2006).

[33] United Nations 2015 International Year of Light and Light-Based Tech-nologies (http://www.light2015.org/Home/LightForDevelopment/Study-after-Sunset.html).

[34] Rogelj, J., Mccollum, D. L., & Riahi, K. The UN’s ‘Sustainable Energy for All’initiative is compatible with a warming limit of 2 C. Nature Climate Change 3(2013), 545.

[35] Waide, P. International Energy Agency: Energy Efficiency Series, Phase Outof Incandescent Lamps: Implications for International Supply and Demand forRegulatory Compliant Lamps. Technical report (2010).

[36] Comstock, O. LED light bulbs keep improving in efficiency and quality(http://www.eia.gov/todayinenergy/detail.cfm?id=18671) (2014).

[37] U.S. Department of Energy, Energy Savings Forecast of Solid-State Lighting inGeneral Illumination Applications (2014).

[38] Schubert, E. F. Light-Emitting Diodes, (Cambridge University Press,2006).

Bibliography 140

[39] Crawford, M. H. LEDs for solid-state lighting: Performance challenges andrecent advances. IEEE Journal on Selected Topics in Quantum Electronics 15(2009), 1028.

[40] U.S. Department of Energy, Joint Solid State Lighting Roundtables on ScienceChallenges. Technical report (2011).

[41] Nakamura, S. Current Status of GaN-Based Solid-State Lighting. MRS Bulletin34 (2009), 101.

[42] Reineke, S. Complementary LED technologies. Nature Materials 14 (2015),459.

[43] Lee, J., Chen, H.-F., Batagoda, T., Coburn, C., Djurovich, P. I., Thompson,M. E., & Forrest, S. R. Deep blue phosphorescent organic light-emitting diodeswith very high brightness and efficiency. Nature materials 15 (2015), 1.

[44] Lee, C. W. & Lee, J. Y. Above 30% external quantum efficiency in blue phos-phorescent organic light-emitting diodes using pyrido[2,3-b]indole derivatives ashost materials. Advanced Materials 25 (2013), 5450.

[45] Su, S. J., Gonmori, E., Sasabe, H., & Kido, J. Highly efficient organic blue- andwhite-light-emitting devices having a carrier- and exciton-confining structure forreduced efficiency roll-off. Advanced Materials 20 (2008), 4189.

[46] De Almeida, A., Santos, B., Paolo, B., & Quicheron, M. Solid state lightingreview - Potential and challenges in Europe. Renewable and Sustainable EnergyReviews 34 (2014), 30.

[47] Coe, S., Woo, W.-K., Bawendi, M. G., & Bulović, V. Electroluminescence fromsingle monolayers of nanocrystals in molecular organic devices. Nature 420(2002), 800.

[48] Colvin, V. L., Schlamp, M. C., Alivisatos, A. P. Light-emitting diodes madefrom cadmium selenide nanocrystals and a semiconducting polymer. Nature370 (1994), 354.

[49] Mueller, A. H., Petruska, M. A., Achermann, M., Werder, D. J., Akhadov, E. A.,Koleske, D. D., Hoffbauer, M. A., & Klimov, V. I. Multicolor light-emittingdiodes based on semiconductor nanocrystals encapsulated in GaN charge injec-tion layers. Nano Letters 5 (2005), 1039.

[50] Stouwdam, J. W. & Janssen, R. A. J. Red, green, and blue quantum dot LEDswith solution processable ZnO nanocrystal electron injection layers. Journal ofMaterials Chemistry 18 (2008), 1889.

[51] Pattantyus-Abraham, A. G., Kramer, I. J., Barkhouse, A. R., Wang, X., Kon-stantatos, G., Debnath, R., Levina, L., Raabe, I., Nazeeruddin, M. K., Gratzel,M., & Sargent, E. H. Depleted-Heterojunction Colloidal. ACS Nano 4 (2010),3374.

Bibliography 141

[52] Pal, B. N., Robel, I., Mohite, A., Laocharoensuk, R., Werder, D. J., & Klimov,V. I. High-Sensitivity p-n Junction Photodiodes Based on PbS NanocrystalQuantum Dots. Advanced Functional Materials 22 (2012), 1741.

[53] Konstantatos, G., Howard, I., Fischer, A., Hoogland, S., Clifford, J., Klem,E., Levina, L., & Sargent, E. H. Ultrasensitive solution-cast quantum dotphotodetectors. Nature 442 (2006), 180.

[54] Koh, W.-K., Saudari, S. R., Fafarman, A. T., Kagan, C. R., & Murray, C. B.Thiocyanate-Capped PbS Nanocubes: Ambipolar Transport Enables. NanoLetters 11 (2011), 4764.

[55] Medintz, I. L., Uyeda, H. T., Goldman, E. R., & Mattoussi, H. Quantum dotbioconjugates for imaging, labelling and sensing. Nature Materials 4 (2005),435.

[56] Chuang, C.-H. M., Brown, P. R., Bulović, V., & Bawendi, M. G. Improvedperformance and stability in quantum dot solar cells through band alignmentengineering. Nature materials 13 (2014), 1.

[57] Norris, D. J., Bawendi, M. G., Brus, L. E. Molecular Electronics: A "Chemistryfor the 21st Century" Monograph, Chapter 9, (Blackwell Science, Oxford,1997).

[58] Anikeeva, P. O., Halpert, J. E., Bawendi, M. G., & Bulović, V. Quantum dotlight-emitting devices with electroluminescence tunable over the entire visiblespectrum. Nano Letters 9 (2009), 2532.

[59] Niu, Y. H., Munro, A. M., Cheng, Y. J., Tian, Y. Q., Liu, M. S., Zhao, J. L.,Bardecker, J. A., Jen-La Plante, I., Ginger, D. S., & Jen, A. K. Y. ImprovedPerformance from Multilayer Quantum Dot Light-Emitting Diodes via ThermalAnnealing of the Quantum Dot Layer. Advanced Materials 19 (2007), 3371.

[60] Kwak, J., Bae, W. K., Lee, D., Park, I., Lim, J., Park, M., Cho, H., Woo, H.,Yoon, D., Char, K., Lee, S., & Lee, C. Bright and efficient full-color colloidalquantum dot light-emitting diodes using an inverted device structure. NanoLetters 12 (2012), 2362.

[61] Qian, L., Zheng, Y., Xue, J., & Holloway, P. H. Stable and efficient quantum-dotlight-emitting diodes based on solution-processed multilayer structures. NaturePhotonics 5 (2011), 543.

[62] Sun, Q., Wang, Y. A., Li, L. S., Wang, D., Zhu, T., Xu, J., Yang, C., & Li,Y. Bright, multicoloured light-emitting diodes based on quantum dots. NaturePhotonics 1 (2007), 717.

[63] Konstantatos, G., Huang, C., Levina, L., Lu, Z., & Sargent, E. H. Efficient In-frared Electroluminescent Devices Using Solution-Processed Colloidal QuantumDots. Advanced Functional Materials 15 (2005), 1865.

Bibliography 142

[64] Sun, L., Choi, J. J., Stachnik, D., Bartnik, A. C., Hyun, B.-R., Malliaras, G. G.,Hanrath, T., & Wise, F. W. Bright infrared quantum-dot light-emitting diodesthrough inter-dot spacing control. Nature Nanotechnology 7 (2012), 369.

[65] Murray, C. B., Norris, D. J., & Bawendi, M. G. Synthesis and Characterizationof Nearly Monodisperse CdE (E = S, Se, Te) Semiconductor Nanocrystallites.J. Am. Chem. Soc. 115 (1993), 8706.

[66] Lee, J., Sundar, V. C., Heine, J. R., Bawendi, M. G., Jensen, K. F. FullColor Emission from II-VI Semiconductor Quantum Dot-Polymer Composites.Advanced Materials 12 (2000), 1102.

[67] Wood, V. & Bulović, V. Colloidal quantum dot light-emitting devices. NanoReviews 1 (2010), 1.

[68] Shiang, J. J., Kadavanich, A. V., Grubbs, R. K., & Alivisatos, A. P. Symmetryof Annealed Wurtzite CdSe Nanocrystals: Assignment to the C3v Point Group.J. Phys. Chem. 99 (1995), 17417.

[69] Bourzac, K. Quantum dots go on display. Nature 493 (2013), 283.

[70] Coe-Sullivan, S., Liu, W., Allen, P., & Steckel, J. S. Quantum Dots for LEDDownconversion in Display Applications. ECS Journal of Solid State Scienceand Technology 2 (2012), R3026.

[71] Dabbousi, B. O., Mikulec, F. V., Heine, J. R., Mattoussi, H., Ober, R., Jensen,K. F., & Bawendi, M. G. (CdSe)ZnS Core-Shell Quantum Dots: Synthesisand Characterization of a Size Series of Highly Luminescent Nanocrystallites.Journal of Physical Chemistry B 101 (1997), 9463.

[72] Hines, M. A. & Guyot-Sionnest, P. Synthesis and Characterization of StronglyLuminescing ZnS-Capped CdSe Nanocrystals. J. Phys. Chem. 100 (1996), 468.

[73] Peng, X., Schlamp, M. C., Kadavanich, A. V., & Alivisatos, A. P. EpitaxialGrowth of Highly Luminescent CdSe/CdS Core/Shell Nanocrystals with Pho-tostability and Electronic Accessibility. J. Am. Chem. Soc. 119 (1997), 7019.

[74] Hollingsworth, J. A. & Klimov, V. I. "Soft" Chemical Synthesis and Manipu-lation of Semiconductor Nanocrystals. In Nanocrystal Quantum Dots, SecondEdition, chapter 1, (CRC Press, Boca Raton,2010), second edition.

[75] Pietryga, J. M., Werder, D. J., Williams, D. J., Casson, J. L., Schaller, R. D.,Klimov, V. I., & Hollingsworth, J. A. Utilizing the lability of lead selenide toproduce heterostructured nanocrystals with bright, stable infrared emission. J.Am. Chem. Soc. 130 (2008), 4879.

[76] Rogach, A. L., Eychmüller, A., Hickey, S. G., & Kershaw, S. V. Infrared-emitting colloidal nanocrystals: synthesis, assembly, spectroscopy, and applica-tions. Small 3 (2007), 536.

Bibliography 143

[77] Sargent, E. H. Infrared Quantum Dots. Advanced Materials 17 (2005), 515.

[78] Sommer, J. R., Farley, R. T., Graham, K. R., Yang, Y., Reynolds, J. R.,Xue, J., & Schanze, K. S. Efficient near-infrared polymer and organic light-emitting diodes based on electrophosphorescence from (tetraphenyltetranaph-tho[2,3]porphyrin)platinum(II). ACS Applied Materials & Interfaces 1 (2009),274.

[79] Chang, T., Maria, A., Cyr, P., Sukhovatkin, V., Levina, L., & Sargent, E. Highnear-infrared photoluminescence quantum efficiency from PbS nanocrystals inpolymer films. Synthetic Metals 148 (2005), 257.

[80] Kagan, C. R., Murray, C. B., Nirmal, M., & Bawendi, M. G. Electronic EnergyTransfer in CdSe Quantum Dot Solids. Physical Review Letters 76 (1996), 1517.

[81] Kagan, C., Murray, C., & Bawendi, M. Long-range resonance transfer of elec-tronic excitations in close-packed CdSe quantum-dot solids. Physical review. B,Condensed matter 54 (1996), 8633.

[82] Akselrod, G. M., Prins, F., Poulikakos, L. V., Lee, E. M. Y., Weidman, M. C.,Mork, A. J., Willard, A. P., Bulović, V., & Tisdale, W. A. Subdiffusive ExcitonTransport in Quantum Dot Solids. Nano letters 26 (2014).

[83] Xu, F., Ma, X., Haughn, C. R., Benavides, J., Doty, M. F., Cloutier, S. G.Efficient Exciton Funneling in Cascaded PbS Quantum Dot Superstructures.ACS Nano 5 (2011), 9950.

[84] Baldo, M. A. & Forrest, S. R. Highly efficient phosphorescent emission fromorganic electroluminescent devices. Nature 395 (1998), 151.

[85] Efros, A., Rosen, M., Kuno, M., Nirmal, M., Norris, D., & Bawendi, M. Band-edge exciton in quantum dots of semiconductors with a degenerate valenceband: Dark and bright exciton states. Physical review. B, Condensed matter54 (1996), 4843.

[86] Coe-sullivan, S. Quantum-Dot Light-Emitting Diodes for Near-to-Eye andDirect-View Display Applications (QD Vision, Inc.). In Session 12: Near-to-Eye and Head-Worn Display Applications, The Society for Information Display,Display Week 2011 (Los Angeles).

[87] Stouwdam, J. W. & Janssen, R. A. J. Electroluminescent Cu-doped CdS Quan-tum Dots. Advanced Materials 21 (2009), 2916.

[88] Coe-Sullivan, S., Steckel, J. S., Woo, W.-K., Bawendi, M. G., & Bulović, V.Large-Area Ordered Quantum-Dot Monolayers via Phase Separation DuringSpin-Casting. Advanced Functional Materials 15 (2005), 1117.

Bibliography 144

[89] Zhu, T., Shanmugasundaram, K., Price, S. C., Ruzyllo, J., Zhang, F., Xu, J.,Mohney, S. E., Zhang, Q., & Wang, a. Y. Mist fabrication of light emittingdiodes with colloidal nanocrystal quantum dots. Applied Physics Letters 92(2008), 023111.

[90] Haverinen, H. M., Myllylä, R. A., & Jabbour, G. E. Inkjet printing of lightemitting quantum dots. Applied Physics Letters 94 (2009), 073108.

[91] Wood, V., Panzer, M. J., Chen, J., Bradley, M. S., Halpert, J. E., Bawendi,M. G., & Bulović, V. Inkjet-Printed Quantum Dot-Polymer Composites forFull-Color AC-Driven Displays. Advanced Materials 21 (2009), 2151.

[92] Kim, L., Anikeeva, P. O., Coe-Sullivan, S. A., Steckel, J. S., Bawendi, M. G.,& Bulović, V. Contact printing of quantum dot light-emitting devices. NanoLetters 8 (2008), 4513.

[93] Kim, T.-H., Cho, K.-S., Lee, E. K., Lee, J. S., Chae, J., Kim, J. W., Kim, D. H.,Kwon, J.-Y., Amaratunga, G., Lee, S. Y., Choi, B. L., Kuk, Y., Kim, J. M.,& Kim, K. Full-colour quantum dot displays fabricated by transfer printing.Nature Photonics 5 (2011), 176.

[94] Cho, K.-S., Lee, E. K., Joo, W.-J., Jang, E., Kim, T.-H., Lee, S. J., Kwon,S.-J., Han, J. Y., Kim, B.-K., Choi, B. L., & Kim, J. M. High-performancecrosslinked colloidal quantum-dot light-emitting diodes. Nature Photonics 3(2009), 341.

[95] Ma, X., Xu, F., Benavides, J., & Cloutier, S. G. High performance hybridnear-infrared LEDs using benzenedithiol cross-linked PbS colloidal nanocrys-tals. Organic Electronics 13 (2012), 525.

[96] Kwak, J., Bae, W. K., Zorn, M., Woo, H., Yoon, H., Lim, J., Kang, S. W.,Weber, S., Butt, H.-J., Zentel, R., Lee, S., Char, K., & Lee, C. Characterizationof Quantum Dot/Conducting Polymer Hybrid Films and Their Application toLight-Emitting Diodes. Advanced Materials 21 (2009), 5022.

[97] Liu, Y., Gibbs, M., Puthussery, J., Gaik, S., Ihly, R., Hillhouse, H. W., & Law,M. Dependence of carrier mobility on nanocrystal size and ligand length inPbSe nanocrystal solids. Nano letters 10 (2010), 1960.

[98] Malko, A. V., Mikhailovsky, A. A., Petruska, M. A., Hollingsworth, J. A.,Htoon, H., Bawendi, M. G., & Klimov, V. I. From amplified spontaneousemission to microring lasing using nanocrystal quantum dot solids. AppliedPhysics Letters 81 (2002), 1303.

[99] Klar, B. T. A., Franzl, T., Rogach, A. L., & Feldmann, J. Super-EfficientExciton Funneling in Layer-by-Layer Semiconductor Nanocrystal Structures.Advanced Materials 17 (2005), 769.

Bibliography 145

[100] Tang, Z., Ozturk, B., Wang, Y., & Kotov, N. A. Simple Preparation Strategyand One-Dimensional Energy Transfer in CdTe Nanoparticle Chains. Journalof Physical Chemistry B 108 (2004), 6927.

[101] Nirmal, M., Dabbousi, B. O., Bawendi, M. G., Macklin, J. J., Trautman, J.K., Harris, T. D., Brus, L. E. Fluorescence Intermittency in single cadmiumselenide nanocrystals. Nature 383 (1996), 802.

[102] van Sark, W. G. J. H. M., Frederix, P. L. T. M., Bol, A. A., Gerritsen, H. C., &Meijerink, A. Blueing, Bleaching, and Blinking of Single CdSe/ZnS QuantumDots. ChemPhysChem 3 (2002), 871.

[103] Koberling, F., Mews, A., B. Oxygen-Induced Blinking of Single CdSe Nanocrys-tals. Advanced Materials 13 (2001), 672.

[104] Jones, M., Nedeljkovic, J., Ellingson, R. J., Nozik, A. J., & Rumbles, G. Pho-toenhancement of Luminescence in Colloidal CdSe Quantum Dot Solutions.Journal of Physical Chemistry B 107 (2003), 11346.

[105] Tice, D. B., Frederick, M. T., Chang, R. P. H., & Weiss, E. A. ElectronMigration Limits the Rate of Photobrightening in Thin Films of CdSe QuantumDots in a Dry N2 (g) Atmosphere. Journal of Physical Chemistry C 115 (2011),3654.

[106] Wang, X., Zhang, J., Nazzal, A., & Xiao, M. Photo-oxidation-enhanced cou-pling in densely packed CdSe quantum-dot films. Applied Physics Letters 83(2003), 162.

[107] Cordero, S. R., Carson, P. J., Estabrook, R. A., Strouse, G. F., & Buratto, S. K.Photo-Activated Luminescence of CdSe Quantum Dot Monolayers. Journal ofPhysical Chemistry B 104 (2000), 12137.

[108] Oda, M., Tsukamoto, J., Hasegawa, A., Iwami, N., Nishiura, K., Hagiwara, I.,Ando, N., Horiuchi, H., & Tani, T. Photoluminescence of CdSe/ZnS/TOPOnanocrystals expanded on silica glass substrates: Adsorption and desorptioneffects of polar molecules on nanocrystal surfaces. Journal of Luminescence119-120 (2006), 570.

[109] Uematsu, T., Maenosono, S., & Yamaguchi, Y. Photoinduced fluorescenceenhancement in mono- and multilayer films of CdSe/ZnS quantum dots: de-pendence on intensity and wavelength of excitation light. Journal of PhysicalChemistry B 109 (2005), 8613.

[110] Chen, Y., Vela, J., Htoon, H., Casson, J. L., Werder, D. J., Bussian, D. A.,Klimov, V. I., & Hollingsworth, J. A. "Giant" multishell CdSe nanocrystalquantum dots with suppressed blinking. Journal of the American ChemicalSociety 130 (2008), 5026.

Bibliography 146

[111] Mahler, B., Spinicelli, P., Buil, S., Quelin, X., Hermier, J.-P., & Dubertret, B.Towards non-blinking colloidal quantum dots. Nature materials 7 (2008), 659.

[112] Hohng, S. & Ha, T. Near-complete suppression of quantum dot blinking inambient conditions. Journal of the American Chemical Society 126 (2004),1324.

[113] Wang, X., Ren, X., Kahen, K., Hahn, M. A., Rajeswaran, M., Maccagnano-Zacher, S., Silcox, J., Cragg, G. E., Efros, A. L., & Krauss, T. D. Non-blinkingsemiconductor nanocrystals. Nature 459 (2009), 686.

[114] Spinicelli, P., Mahler, B., Buil, S., Quélin, X., Dubertret, B., & Hermier, J.-P.Non-blinking semiconductor colloidal quantum dots for biology, optoelectronicsand quantum optics. ChemPhysChem 10 (2009), 879.

[115] Kovalenko, M. V., Scheele, M., & Talapin, D. V. Colloidal nanocrystals withmolecular metal chalcogenide surface ligands. Science 324 (2009), 1417.

[116] Nag, A., Kovalenko, M. V., Lee, J.-S., Liu, W., Spokoyny, B., & Talapin, D. V.Metal-free Inorganic Ligands for Colloidal Nanocrystals: S2-, HS-, Se2-, HSe-,Te2-, HTe-, TeS32-, OH-, and NH2-as Surface Ligands. Journal of the AmericanChemical Society 133 (2011), 10612.

[117] Resch-Genger, U., Grabolle, M., Cavaliere-Jaricot, S., Nitschke, R., & Nann,T. Quantum dots versus organic dyes as fluorescent labels. Nature Methods 5(2008), 763.

[118] Meerheim, R., Scholz, S., Olthof, S., Schwartz, G., Reineke, S., Walzer, K., &Leo, K. Influence of charge balance and exciton distribution on efficiency andlifetime of phosphorescent organic light-emitting devices. Journal of AppliedPhysics 104 (2008), 014510.

[119] Osedach, T. P., Andrew, T. L., & Bulović, V. Effect of synthetic accessibilityon the commercial viability of organic photovoltaics. Energy & EnvironmentalScience 6 (2013), 711.

[120] Chen, O., Zhao, J., Chauhan, V. P., Cui, J., Wong, C., Harris, D. K., Wei,H., Han, H.-S., Fukumura, D., Jain, R. K., & Bawendi, M. G. Compact high-quality CdSe-CdS core-shell nanocrystals with narrow emission linewidths andsuppressed blinking. Nature Materials 12 (2013), 1.

[121] Oliver, J. Quantum Dots: Technologies and Global Markets. BCC Researchpaper NAN027C (2011).

[122] Helander, M. G., Wang, Z. B., Qiu, J., Greiner, M. T., Puzzo, D. P., Liu, Z. W.,& Lu, Z. H. Chlorinated indium tin oxide electrodes with high work functionfor organic device compatibility. Science 332 (2011), 944.

Bibliography 147

[123] U. S. Department of Energy, Solid-State Lighting Research and Development:Manufacturing Roadmap. Technical report (2012).

[124] Haverinen, H. M., Myllylä, R. A., & Jabbour, G. E. Inkjet Printed RGBQuantum Dot-Hybrid LED. Journal of Display Technology 6 (2010), 87.

[125] Panzer, M. J., Wood, V., Geyer, S. M., Bawendi, M. G., & Bulovic, V. TunableInfrared Emission From Printed Colloidal Quantum Dot/Polymer CompositeFilms on Flexible Substrates. Journal of Display Technology 6 (2010), 90.

[126] Zhao, N., Osedach, T. P., Chang, L.-Y., Geyer, S. M., Wanger, D., Binda,M. T., Arango, A. C., Bawendi, M. G., & Bulović, V. Colloidal PbS quantumdot solar cells with high fill factor. ACS Nano 4 (2010), 3743.

[127] Horihata, T., Okamoto, K., Hidaka, J., Einaga, H. Ligand-substitutionreaction kinetics of tris-(8-quinolinolato)- and bis-(2-methyl-8-quinolinolato)(8-quinolinato)aluminum(III) complexes with ethylenediamine-N,N,N’,N’-tetraacetic acid. Bull. Chem. Soc. Jpn. 61 (1988), 2999.

[128] M. Furugori, S. Igawa, J. Kamatani, S. Miura, H. Mizutani, T. Moriyama, S.Okada, T. Takiguchi, A. T. Metal Coordination Compound and Electrolumi-nescence Device (2003).

[129] Bae, W. K., Kwak, J., Lim, J., Lee, D., Nam, M. K., Char, K., Lee, C., & Lee,S. Multicolored light-emitting diodes based on all-quantum-dot multilayer filmsusing layer-by-layer assembly method. Nano Letters 10 (2010), 2368.

[130] Dabbousi, B. O., Bawendi, M. G., Onitsuka, O., & Rubner, M. F. Electrolumi-nescence from CdSe quantum-dot/polymer composites. Applied Physics Letters66 (1995), 1316.

[131] Mattoussi, H., Radzilowski, L. H., Dabbousi, B. O., Thomas, E. L., Bawendi,M. G., & Rubner, M. F. Electroluminescence from heterostructures ofpoly(phenylene vinylene) and inorganic CdSe nanocrystals. Journal of AppliedPhysics 83 (1998), 7965.

[132] Schlamp, M. C., Peng, X., & Alivisatos, A. P. Improved efficiencies in lightemitting diodes made with CdSe(CdS) core/shell type nanocrystals and a semi-conducting polymer. Journal of Applied Physics 82 (1997), 5837.

[133] Anikeeva, P. O., Madigan, C. F., Halpert, J. E., Bawendi, M. G., & Bulović,V. Electronic and excitonic processes in light-emitting devices based on organicmaterials and colloidal quantum dots. Physical Review B 78 (2008), 085434.

[134] Anikeeva, P. O., Halpert, J. E., Bawendi, M. G., & Bulović, V. Electrolumi-nescence from a mixed red-green-blue colloidal quantum dot monolayer. NanoLetters 7 (2007), 2196.

Bibliography 148

[135] Jing, P., Zheng, J., Zeng, Q., Zhang, Y., Liu, X., Liu, X., Kong, X., & Zhao,J. Shell-dependent electroluminescence from colloidal CdSe quantum dots inmultilayer light-emitting diodes. Journal of Applied Physics 105 (2009), 044313.

[136] Zhao, J., Bardecker, J. A., Munro, A. M., Liu, M. S., Niu, Y., Ding, I.-K., Luo,J., Chen, B., Jen, A. K.-Y., & Ginger, D. S. Efficient CdSe/CdS quantum dotlight-emitting diodes using a thermally polymerized hole transport layer. NanoLetters 6 (2006), 463.

[137] Caruge, J. M., Halpert, J. E., Wood, V., Bulović, V., & Bawendi, M. G. Col-loidal quantum-dot light-emitting diodes with metal-oxide charge transport lay-ers. Nature Photonics 2 (2008), 247.

[138] Wood, V., Panzer, M. J., Halpert, J. E., Caruge, J.-M., Bawendi, M. G., &Bulović, V. Selection of metal oxide charge transport layers for colloidal quan-tum dot LEDs. ACS Nano 3 (2009), 3581.

[139] Caruge, J.-M., Halpert, J. E., Bulović, V., & Bawendi, M. G. NiO as aninorganic hole-transporting layer in quantum-dot light-emitting devices. NanoLetters 6 (2006), 2991.

[140] Dai, X. L., Zhang, Z. X., Jin, Y. Z., Niu, Y., Cao, H. J., Liang, X. Y., Chen,L. W., Wang, J. P., & Peng, X. G. Solution-processed, high-performance light-emitting diodes based on quantum dots. Nature 515 (2014), 96.

[141] Yang, Y., Zheng, Y., Cao, W., Titov, A., Hyvonen, J., Manders, J. R., Xue,J., Holloway, P. H., & Qian, L. High-efficiency light-emitting devices based onquantum dots with tailored nanostructures. Nature Photonics (2015), 1.

[142] Mashford, B. S., Stevenson, M., Popovic, Z., Hamilton, C., Zhou, Z., Breen, C.,Steckel, J., Bulović, V., Bawendi, M., Coe-sullivan, S., & Kazlas, P. T. High-efficiency quantum-dot light-emitting devices with enhanced charge injection.Nature Photonics 7 (2013), 407.

[143] Chang, Y.-L., Wang, Z., Helander, M., Qiu, J., Puzzo, D., & Lu, Z. Enhancingthe efficiency of simplified red phosphorescent organic light emitting diodes byexciton harvesting. Organic Electronics 13 (2012), 925.

[144] O’Brien, D. F., Baldo, M. A., Thompson, M. E., Forrest, S. R. Improved energytransfer in electrophosphorescent devices. Applied Physics Letters 74 (1999),442.

[145] Tang, C. W., VanSlyke, S. A., & Chen, C. H. Electroluminescence of dopedorganic thin films. Journal of Applied Physics 65 (1989), 3610.

[146] Tsuzuki, T. & Tokito, S. Highly Efficient and Low-Voltage PhosphorescentOrganic Light-Emitting Diodes Using an Iridium Complex as the Host Material.Advanced Materials 19 (2007), 276.

Bibliography 149

[147] Bendall, J. S., Paderi, M., Ghigliotti, F., Li Pira, N., Lambertini, V., Lesnyak,V., Gaponik, N., Visimberga, G., Eychmüller, A., Torres, C. M. S., Welland,M. E., Gieck, C., & Marchese, L. Layer-by-Layer All-Inorganic Quantum-Dot-Based LEDs: A Simple Procedure with Robust Performance. Advanced Func-tional Materials 20 (2010), 3298.

[148] Chen, B., Lin, X., Cheng, L., Lee, C-S., Gambling, W. A., Lee, S.-T. Im-provement of efficiency and colour purity of red-dopant organic light-emittingdiodes by energy levels matching with the host materials. Journal of PhysicsD: Applied Physics 34 (2001), 30.

[149] Hamada, Y., Kanno, H., Tsujioka, T., Takahashi, H., & Usuki, T. Red organiclight-emitting diodes using mitting assist dopant. Applied Physics Letters 75(1999), 1682.

[150] Leung, M.-K., Chang, C.-C., Wu, M.-H., Chuang, K.-H., Lee, J.-H., Shieh, S.-J.,Lin, S.-C., & Chiu, C.-F. 6-N,N-Diphenylaminobenzofuran-Derived Pyran Con-taining Fluorescent Dyes: A New Class of High-Brightness Red-Light-EmittingDopants for OLED. Organic Letters 8 (2006), 2623.

[151] Solomeshch, O., Kigel, A., Saschiuk, A., Medvedev, V., Aharoni, A., Razin, A.,Eichen, Y., Banin, U., Lifshitz, E., & Tessler, N. Optoelectronic properties ofpolymer-nanocrystal composites active at near-infrared wavelengths. Journalof Applied Physics 98 (2005), 074310.

[152] Zhang, Y. Q. & Cao, X. A. Electroluminescence of green CdSe/ZnS quantumdots enhanced by harvesting excitons from phosphorescent molecules. AppliedPhysics Letters 97 (2010), 253115.

[153] Chang, T.-W. F., Musikhin, S., Bakueva, L., Levina, L., Hines, M. A., Cyr,P. W., & Sargent, E. H. Efficient excitation transfer from polymer to nanocrys-tals. Applied Physics Letters 84 (2004), 4295.

[154] Panzer, M. J., Aidala, K. E., Anikeeva, P. O., Halpert, J. E., Bawendi, M. G.,& Bulović, V. Nanoscale morphology revealed at the interface between colloidalquantum dots and organic semiconductor films. Nano Letters 10 (2010), 2421.

[155] Achermann, M., Petruska, M. A., Kos, S., Smith, D. L., Koleske, D. D., &Klimov, V. I. Energy-transfer pumping of semiconductor nanocrystals using anepitaxial quantum well. Nature 429 (2004), 642.

[156] Achermann, M., Petruska, M. A., Koleske, D. D., Crawford, M. H., & Klimov,V. I. Nanocrystal-based light-emitting diodes utilizing high-efficiency nonradia-tive energy transfer for color conversion. Nano Letters 6 (2006), 1396.

[157] Rizzo, A., Mazzeo, M., Palumbo, M., Lerario, G., D’Amone, S., Cingolani, R.,& Gigli, G. Hybrid Light-Emitting Diodes from Microcontact-Printing Double-Transfer of Colloidal Semiconductor CdSe/ZnS Quantum Dots onto OrganicLayers. Advanced Materials 20 (2008), 1886.

Bibliography 150

[158] Steckel, J. S., Snee, P., Coe-Sullivan, S., Zimmer, J. P., Halpert, J. E., Anikeeva,P., Kim, L.-A., Bulović, V., & Bawendi, M. G. Color-saturated green-emittingQD-LEDs. Angewandte Chemie 45 (2006), 5796.

[159] Li, Y. Q., Rizzo, A., Cingolani, R., & Gigli, G. Bright White-Light-EmittingDevice from Ternary Nanocrystal Composites. Advanced Materials 18 (2006),2545.

[160] Burrows, P. E., Bulović, V., Forrest, S. R., Sapochak, L. S., & Mccarty, D. M.Reliability and degradation of organic light emitting devices. Appl. Phys. Lett.65 (1994), 2922.

[161] Friend, R. H., Gymer, R. W., Holmes, A. B., Burroughes, J. H., Marks, R.N., Taliani, C., Bradley, D. D. C., Dos Santos, D. A., Bredas, J. L., Logdlund,M., Salaneck, W. R. Electroluminescence in conjugated polymers. Nature 397(1999), 121.

[162] Wood, V., Halpert, J. E., Panzer, M. J., Bawendi, M. G., & Bulović, V. Alter-nating current driven electroluminescence from ZnSe/ZnS:Mn/ZnS nanocrys-tals. Nano letters 9 (2009), 2367.

[163] Cho, S. H., Sung, J., Hwang, I., Kim, R. H., Choi, Y. S., Jo, S. S., Lee, T. W.,& Park, C. High Performance AC Electroluminescence from Colloidal QuantumDot Hybrids. Advanced Materials 24 (2012), 4540.

[164] Bozyigit, D., Wood, V., Shirasaki, Y., & Bulović, V. Study of field drivenelectroluminescence in colloidal quantum dot solids. Journal of Applied Physics111 (2012).

[165] Meerheim, R., Furno, M., Hofmann, S., Lussem, B., & Leo, K. Quantificationof energy loss mechanisms in organic light-emitting diodes. Applied PhysicsLetters 97 (2010), 253305.

[166] Qian, G., Zhong, Z., Luo, M., Yu, D., Zhang, Z., Wang, Z. Y., & Ma, D. Simpleand Efficient Near-Infrared Organic Chromophores for Light-Emitting Diodeswith Single Electroluminescent Emission above 1000 nm. Advanced Materials21 (2009), 111.

[167] Klem, E. J. D., Levina, L., & Sargent, E. H. PbS quantum dot electroabsorptionmodulation across the extended communications band 1200-1700 nm. AppliedPhysics Letters 87 (2005), 053101.

[168] Bourdakos, K. N., Dissanayake, D. M. N. M., Lutz, T., Silva, S. R. P., &Curry, R. J. Highly efficient near-infrared hybrid organic-inorganic nanocrystalelectroluminescence device. Applied Physics Letters 92 (2008), 153311.

[169] Choudhury, K. R., Song, D. W., & So, F. Efficient solution-processed hybridpolymer-nanocrystal near infrared light-emitting devices. Organic Electronics11 (2010), 23.

Bibliography 151

[170] Steckel, J., Coe-Sullivan, S., Bulović, V., & Bawendi, M. G. 1.3𝜇m to 1.55𝜇mTunable Electroluminescence from PbSe Quantum Dots Embedded within anOrganic Device. Advanced Materials 15 (2003), 1862.

[171] Lim, Y.T, Kim, S., Nakayama, A., Stott, N.E., Bawendi, M. G., Frangioni,J. V. Selection of quantum dot wavelengths for biomedical assays and imaging.Molecular Imaging 2 (2003), 50.

[172] Frasco, M. F. & Chaniotakis, N. Semiconductor quantum dots in chemicalsensors and biosensors. Sensors 9 (2009), 7266.

[173] Yager, P., Edwards, T., Fu, E., Helton, K., Nelson, K., Tam, M. R., & Weigl,B. H. Microfluidic diagnostic technologies for global public health. Nature 442(2006), 412.

[174] Borek, C., Hanson, K., Djurovich, P. I., Thompson, M. E., Aznavour, K., Bau,R., Sun, Y., Forrest, S. R., Brooks, J., Michalski, L., & Brown, J. Highly ef-ficient, near-infrared electrophosphorescence from a Pt-metalloporphyrin com-plex. Angewandte Chemie 46 (2007), 1109.

[175] Mashford, B. S., Kazlas, P. T., Stevenson, M., Popovic, Z., Hamilton, C., Zhou,Z., Breen, C., Steckel, J., Bulović, V., Bawendi, M. G., Coe-Sullivan, S. Low-voltage and high-efficiency quantum dot light emitting device via optimizedcharge injection. Nature Photonics (2013).

[176] Coe-Sullivan, S. OECD/NNI Symposium, presenting work from AFOSR grantnumber FA9550-07-C-0056 (2012).

[177] Kershaw, S. V., Harrison, M., Rogach, A. L., & Kornowski, A. Development ofIR-Emitting Colloidal II-VI Quantum-Dot Materials. IEEE Journal of SelectedTopics in Quantum Electronics 6 (2000), 534.

[178] Bakueva, L., Musikhin, S., Hines, M. A., Chang, T.-W. F., Tzolov, M., Scholes,G. D., & Sargent, E. H. Size-tunable infrared (1000-1600 nm) electrolumi-nescence from PbS quantum-dot nanocrystals in a semiconducting polymer.Applied Physics Letters 82 (2003), 2895.

[179] Tessler, N., Medvedev, V., Kazes, M., Kan, S., & Banin, U. Efficient near-infrared polymer nanocrystal light-emitting diodes. Science 295 (2002), 1506.

[180] Cheng, K.-Y., Anthony, R., Kortshagen, U. R., & Holmes, R. J. Hybrid sili-con nanocrystal-organic light-emitting devices for infrared electroluminescence.Nano Letters 10 (2010), 1154.

[181] Koktysh, D. S., Gaponik, N., Reufer, M., Crewett, J., Scherf, U., Eychmüller,A., Lupton, J. M., Rogach, A. L., & Feldmann, J. Near-infrared electrolumi-nescence from HgTe nanocrystals. ChemPhysChem 5 (2004), 1435.

Bibliography 152

[182] O’Connor, E., O’Riordan, A., Doyle, H., Moynihan, S., Cuddihy, A., & Red-mond, G. Near-infrared electroluminescent devices based on colloidal HgTequantum dot arrays. Applied Physics Letters 86 (2005), 201114.

[183] Cheng, K.-Y., Anthony, R., Kortshagen, U. R., & Holmes, R. J. High-efficiencysilicon nanocrystal light-emitting devices. Nano Letters 11 (2011), 1952.

[184] US Department of Health and Human Services, "Toxicological Profile for Cad-mium". Technical report (2012).

[185] Shchukin, V. A. & Bimberg, D. Spontaneous ordering of nanostructures oncrystal surfaces. Reviews of Modern Physics 71 (1999), 1125.

[186] Grundmann, M. The present status of quantum dot lasers. Physica E 5 (2000),167.

[187] Coe-sullivan, S. Quantum dot developments. Nature Photonics 3 (2009), 315.

[188] Murray, C. B., Kagan, C. R., Bawendi, M. G. Synthesis and Characterization ofMonodisperse Nanocrystals and Close-Packed Nanocrystal Assemblies. Annu.Rev. Mater. Sci 30 (2000), 545.

[189] Hwang, G. Surface Trap Passivation and Characterization of Lead SulfideQuantum Dots for Optical and Electrical Applications. Ph.D. thesis, Mas-sachusetts Institute of Technology (2015).

[190] Sanderson, K. Quantum dots go large. Nature 459 (2009), 760.

[191] Efros, Al. L., Rosen, M. The Electronic Structure of Semiconductor Nanocrys-tals. Annu. Rev. Mater. Sci 30 (2000), 465.

[192] Norris, D. J. Electronic Structure in Semiconductor Nanocrystals. In Semicon-ductor and Metal Nanocrystals: Synthesis and Electronic and Optical Proper-ties, (Marcel Dekker Ltd,2003).

[193] Yoffe, A. D. Semiconductor quantum dots and related systems: Electronic, op-tical, luminescence and related properties of low dimensional systems. Advancesin Physics 50 (2001), 1.

[194] Nirmal, M., Norris, D. J., Kuno, M., Bawendi, M. G., Efros, A. L., & Rosen,M. Observation of the "dark exciton" in CdSe quantum dots. Physical ReviewLetters 75 (1995), 3728.

[195] Cho, J., Schubert, E. F., & Kim, J. K. Efficiency droop in light-emitting diodes:Challenges and countermeasures. Laser & Photonics Reviews 7 (2013), 408.

[196] Weisbuch, C. Droop in LEDs: Origin and Solutions. In DOE SSL R&D Work-shop - Tampa, FL (2014).

Bibliography 153

[197] National Research Council, Assessment of Advanced Solid State Lighting. Tech-nical report (2013).

[198] Reineke, S., Thomschke, M., Lüssem, B., & Leo, K. White organic light-emitting diodes: Status and perspective. Reviews of Modern Physics 85 (2013),1245.

[199] Mesta, M., van Eersel, H., Coehoorn, R., & Bobbert, P. A. Kinetic Monte Carlomodeling of the efficiency roll-off in a multilayer white organic light-emittingdevice. Applied Physics Letters 108 (2016), 133301.

[200] Meyaard, D. S., Shan, Q., Cho, J., Fred Schubert, E., Han, S. H., Kim, M. H.,Sone, C., Jae Oh, S., & Kyu Kim, J. Temperature dependent efficiency droopin GaInN light-emitting diodes with different current densities. Applied PhysicsLetters 100 (2012), 8.

[201] Reineke, S., Walzer, K., & Leo, K. Triplet-exciton quenching in organic phos-phorescent light-emitting diodes with Ir-based emitters. Physical Review B -Condensed Matter and Materials Physics 75 (2007), 1.

[202] Chiang, Y. C., Lin, B. C., Chen, K. J., Chiu, S. H., Lin, C. C., Lee, P. T.,Shih, M. H., & Kuo, H. C. Efficiency and droop improvement in GaN-basedhigh-voltage flip chip LEDs. International Journal of Photoenergy 2014 (2014).

[203] Song, D., Zhao, S., Luo, Y., & Aziz, H. Causes of efficiency roll-off in phos-phorescent organic light emitting devices: Triplet-triplet annihilation versustriplet-polaron quenching. Applied Physics Letters 97 (2010), 2008.

[204] Reineke, S., Schwartz, G., Walzer, K., & Leo, K. Reduced efficiency roll-offin phosphorescent organic light emitting diodes by suppression of triplet-tripletannihilation. Applied Physics Letters 91 (2007), 6.

[205] Kim, M.-H., Schubert, M. F., Dai, Q., Kim, J. K., Schubert, E. F., Piprek,J., & Park, Y. Origin of efficiency droop in GaN-based light-emitting diodes.Applied Physics Letters 91 (2007), 183507.

[206] Kalinowski, J., Stampor, W., Mezyk, J., Cocchi, M., Virgili, D., Fattori, V.,& Di Marco, P. Quenching effects in organic electrophosphorescence. PhysicalReview B 66 (2002), 1.

[207] Brown, P. R., Lunt, R. R., Zhao, N., Osedach, T. P., Wanger, D. D., Chang, L.-Y., Bawendi, M. G., & Bulović, V. Improved current extraction from ZnO/PbSquantum dot heterojunction photovoltaics using a MoO3 interfacial layer. NanoLetters 11 (2011), 2955.

[208] Baldo, M. A., Lamansky, S., Burrows, P. E., Thompson, M. E., & Forrest, S. R.Very high-efficiency green organic light-emitting devices based on electrophos-phorescence. Applied Physics Letters 75 (1999), 4.

Bibliography 154

[209] Klimov, V. I., Mikhailovsky, A. A., McBranch, D. W., Leatherdale, C. A., &Bawendi, M. G. Quantization of multiple auger rates in semiconductor quantumdots. Science 287 (2000), 1011.

[210] Hartmann, T., Reineker, P., & Yudson, V. I. Auger release of a deeply trappedcarrier in a quantum dot. Physical Review B - Condensed Matter and MaterialsPhysics 84 (2011), 1.

[211] Cohn, A. W., Schimpf, A. M., Gunthardt, C. E., & Gamelin, D. R. Size-dependent trap-assisted auger recombination in semiconductor nanocrystals.Nano Letters 13 (2013), 1810.

[212] Bozyigit, D. & Wood, V. Challenges and solutions for high-efficiency quantumdot-based LEDs. MRS Bulletin 38 (2013), 731.

[213] Empedocles, S., Norris, D., & Bawendi, M. Photoluminescence Spectroscopy ofSingle CdSe Nanocrystallite Quantum Dots. Physical Review Letters 77 (1996),3873.

[214] Empedocles, S. A. & Bawendi, M. G. Quantum-Confined Stark Effect in SingleCdSe Nanocrystallite Quantum Dots. Science 278 (1997), 2114.

[215] Dujardin, F., Feddi, E., Assaid, E., & Oukerroum, a. Stark shift and dissociationprocess of an ionized donor bound exciton in spherical quantum dots. TheEuropean Physical Journal B 74 (2010), 507.

[216] Park, S.-J., Link, S., Miller, W. L., Gesquiere, A., & Barbara, P. F. Effect ofelectric field on the photoluminescence intensity of single CdSe nanocrystals.Chemical Physics 341 (2007), 169.

[217] Huang, H., Dorn, A., Nair, G. P., Bulović, V., & Bawendi, M. G. Bias-inducedphotoluminescence quenching of single colloidal quantum dots embedded inorganic semiconductors. Nano Letters 7 (2007), 3781.

[218] Wang, Z.-B., Zhang, H.-C., & Zhang, J.-Y. Quantum-Confined Stark Effect inEnsemble of Colloidal Semiconductor Quantum Dots. Chinese Physics Letters27 (2010), 127803.

[219] Woo, W.-K., Shimizu, K., Jarosz, M., Neuhauser, R., Leatherdale, C., Rubner,M., & Bawendi, M. Reversible Charging of CdSe Nanocrystals in a SimpleSolid-State Device. Advanced Materials 14 (2002), 1068.

[220] Sun, L., Bartnik, A., Hyun, B., Wise, F., Reed, J., Slinker, J., Malliaras, G.,Zhong, Y., Bao, L., & Abruna, H. PbS quantum dot photoluminescence quench-ing induced by an applied bias. 2008 Conference on Lasers and Electro-Optics(2008), 1.

Bibliography 155

[221] Leatherdale, C. A., Kagan, C. R., Morgan, N. Y., Empedocles, S. A., Kastner,M. A., & Bawendi, M. G. Photoconductivity in CdSe quantum dot solids.Physical Review B 62 (2000).

[222] Pugh-Thomas, D., Walsh, B. M., & Gupta, M. C. CdSe(ZnS) nanocompositeluminescent high temperature sensor. Nanotechnology 22 (2011), 185503.

[223] Bergemann, K. J., Krasny, R., & Forrest, S. R. Thermal properties of organiclight-emitting diodes. Organic Electronics: physics, materials, applications 13(2012), 1565.

[224] Leatherdale, C. & Bawendi, M. Observation of solvatochromism in CdSe col-loidal quantum dots. Physical Review B 63 (2001), 1.

[225] Campbell, I. H., Joswick, M. D., & Parker, I. D. Direct measurement of theinternal electric field distribution in a multilayer organic light-emitting diode.Applied Physics Letters 67 (1995), 3171.

[226] Dissanayake, A. D., Lin, J Y, Jiang, H. X. Quantum-confined Stark effects inCdSSe quantum dots. Physical Review B 51 (1995), 8.

[227] Wen, G. W., Lin, J. Y., Jiang, H. X., Chen, Z. Quantum-confined Stark effectsin semiconductor quantum dots. Physical Review B 52 (1995), 5913.

[228] Jarosz, M., Porter, V., Fisher, B., Kastner, M., & Bawendi, M. Photoconduc-tivity studies of treated CdSe quantum dot films exhibiting increased excitonionization efficiency. Physical Review B 70 (2004), 1.

[229] Bae, W. K., Park, Y.-S., Lim, J., Lee, D., Padilha, L. A., McDaniel, H., Robel,I., Lee, C., Pietryga, J. M., & Klimov, V. I. Controlling the influence of Augerrecombination on the performance of quantum-dot light-emitting diodes. Naturecommunications 4 (2013), 2661.

[230] Bae, W. K., Brovelli, S., & Klimov, V. I. Spectroscopic insights into the per-formance of quantum dot light-emitting diodes. MRS Bulletin 38 (2013), 721.

[231] Li, J. & Wang, L.-W. First principle study of core/shell structure quantumdots. Applied Physics Letters 84 (2004), 3648.

[232] García-Santamaría, F., Chen, Y., Vela, J., Schaller, R. D., Hollingsworth, J. A.,& Klimov, V. I. Suppressed Auger Recombination in "Giant" NanocrystalsBoosts Optical Gain Performance. Nano Letters 9 (2009), 3482.

[233] García-Santamaría, F., Brovelli, S., Viswanatha, R., Hollingsworth, J. A.,Htoon, H., Crooker, S. A., & Klimov, V. I. Breakdown of volume scaling inauger recombination in CdSe/CdS heteronanocrystals: The role of the core-shellinterface. Nano Letters 11 (2011), 687.

Bibliography 156

[234] Cragg, G. E. & Efros, A. L. Suppression of auger processes in confined struc-tures. Nano Letters 10 (2010), 313.

[235] Muller, J., Lupton, J. M., Lagoudakis, P. G., Schindler, F., Koeppe, R., Rogach,A. L., Feldmann, J., Talapin, D. V., & Weller, H. Wave function engineering inelongated semiconductor nanocrystals with heterogeneous carrier confinement.Nano letters 5 (2005), 2044.

[236] Kraus, R., Lagoudakis, P., Rogach, A., Talapin, D., Weller, H., Lupton, J., &Feldmann, J. Room-Temperature Exciton Storage in Elongated SemiconductorNanocrystals. Physical Review Letters 98 (2007), 3.

[237] Rothenberg, E., Kazes, M., Shaviv, E., & Banin, U. Electric field inducedswitching of the fluorescence of single semiconductor quantum rods. Nano letters5 (2005), 1581.

[238] Kuno, M., Lee, J. K., Dabbousi, B. O., Mikulec, F. V., & Bawendi, M. G. Theband edge luminescence of surface modified CdSe nanocrystallites: Probing theluminescing state. J. Chem. Phys. 106 (1997), 9869.

[239] Wu, Y.-h., Arai, K., & Yao, T. Temperature dependence of the photolumi-nescence of ZnSe/ZnS quantum-dot structures. Physical Review B 53 (1996),R10485.

[240] Tvrdy, K., Frantsuzov, P. A., & Kamat, P. V. Photoinduced electron transferfrom semiconductor quantum dots to metal oxide nanoparticles. Proceedings ofthe National Academy of Sciences 108 (2011), 29.

[241] Wise, F. W., Goodreau, J. D., Matthews, J. R., Leslie, T. M., & Borrelli, N. F.Electron Injection from Colloidal PbS Nanoparticles. ACS Nano 2 (2008), 2206.

[242] Song, N., Zhu, H., Liu, Z., Huang, Z., Wu, D., & Lian, T. Unraveling theExciton Quenching Mechanism of Quantum Dots on Antimony-Doped SnO2Films by Transient Absorption and Single Dot Fluorescence Spectroscopy. ACSNano 7 (2013), 1599.

[243] Wang, R., Shang, Y., Kanjanaboos, P., Zhou, W., Ning, Z., & Sargent, E. H.Colloidal quantum dot ligand engineering for high performance solar cells. En-ergy Environ. Sci. 9 (2016), 1130.

[244] Chandler, R. E., Houtepen, A. J., Nelson, J., & Vanmaekelbergh, D. Electrontransport in quantum dot solids: Monte Carlo simulations of the effects of shellfilling, Coulomb repulsions, and site disorder. Physical Review B - CondensedMatter and Materials Physics 75 (2007), 1.

[245] Mahler, B., Spinicelli, P., Buil, S., Quelin, X., Hermier, J.-P., & Dubertret, B.Towards non-blinking colloidal quantum dots. Nature Materials 7 (2008), 659.

Bibliography 157

[246] Pal, B. N., Ghosh, Y., Brovelli, S., Laocharoensuk, R., Klimov, V. I.,Hollingsworth, J. A., & Htoon, H. ‘Giant’ CdSe/CdS core/shell nanocrystalquantum dots as efficient electroluminescent materials: strong influence of shellthickness on light-emitting diode performance. Nano Letters 12 (2012), 331.

[247] Kim, S., Lim, Y. T., Soltesz, E. G., De Grand, A. M., Lee, J., Nakayama,A., Parker, J. A., Mihaljevic, T., Laurence, R. G., Dor, D. M., Cohn, L. H.,Bawendi, M. G., & Frangioni, J. V. Near-infrared fluorescent type II quantumdots for sentinel lymph node mapping. Nature Biotechnology 22 (2004), 93.

[248] Greiner, M. T., Helander, M. G., Tang, W.-M., Wang, Z.-B., Qiu, J., & Lu,Z.-H. Universal energy-level alignment of molecules on metal oxides. NatureMaterials 11 (2012), 76.

[249] Jasieniak, J., Califano, M., & Watkins, S. E. Size-dependent valence and con-duction band-edge energies of semiconductor nanocrystals. ACS Nano 5 (2011),5888.

[250] Lee, H., Leventis, H. C., Moon, S.-J., Chen, P., Ito, S., Haque, S. A., Torres, T.,Nüesch, F., Geiger, T., Zakeeruddin, S. M., Grätzel, M., & Nazeeruddin, M. K.PbS and CdS Quantum Dot-Sensitized Solid-State Solar Cells: "Old Concepts,New Results". Advanced Functional Materials 19 (2009), 2735.

[251] Park, Y., Choong, V., Gao, Y., Hsieh, B. R., & Tang, C. W. Work function of in-dium tin oxide transparent conductor measured by photoelectron spectroscopy.Applied Physics Letters 68 (1996), 2699.

[252] Sun, Y., Seo, J. H., Takacs, C. J., Seifter, J., & Heeger, A. J. Inverted polymersolar cells integrated with a low-temperature-annealed sol-gel-derived ZnO Filmas an electron transport layer. Advanced Materials 23 (2011), 1679.

[253] Wang, Z. B., Helander, M. G., Qiu, J., Liu, Z. W., Greiner, M. T., & Lu, Z. H.Direct hole injection in to 4,4’-N,N’-dicarbazole-biphenyl: A simple pathway toachieve efficient organic light emitting diodes. Journal of Applied Physics 108(2010), 024510.

[254] Wang, Z. B., Helander, M. G., Qiu, J., Puzzo, D. P., Greiner, M. T., Hudson,Z. M., Wang, S., Liu, Z. W., & Lu, Z. H. Unlocking the full potential of organiclight-emitting diodes on flexible plastic. Nature Photonics 5 (2011), 753.

[255] Hikmet, R. A. M., Talapin, D. V., & Weller, H. Study of conduction mechanismand electroluminescence in CdSe/ZnS quantum dot composites. Journal ofApplied Physics 93 (2003), 3509.

[256] www.roithner-laser.com.

[257] Change, R. R., Prock, A., Silbey, R. Molecular Fluorescence and Energy Trans-fer Near Interfaces. In Advances in Chemical Physics, pp. Vol. 37, 1–65, (JohnWiley & Sons, Inc., New York,1978).

Bibliography 158

[258] Jha, P. P. & Guyot-Sionnest, P. Electrochemical Switching of the Photolumi-nescence of Single Quantum Dots. The Journal of Physical Chemistry C 114(2010), 21138.

[259] Leventis, H. C., O’Mahony, F., Akhtar, J., Afzaal, M., O’Brien, P., & Haque,S. A. Transient optical studies of interfacial charge transfer at nanostructuredmetal oxide/PbS quantum dot/organic hole conductor heterojunctions. Journalof the American Chemical Society 132 (2010), 2743.

[260] Fisher, B. R., Eisler, H.-J., Stott, N. E., & Bawendi, M. G. Emission IntensityDependence and Single-Exponential Behavior In Single Colloidal Quantum DotFluorescence Lifetimes. Journal of Physical Chemistry B 108 (2004), 143.

[261] Buonsanti, R., Llordes, A., Aloni, S., Helms, B. A., & Milliron, D. J. Tunableinfrared absorption and visible transparency of colloidal aluminum-doped zincoxide nanocrystals. Nano Letters 11 (2011), 4706.

[262] Jun Mei, M., Bradley, S., & Bulović, V. Photoluminescence quenching of tris-(8-hydroxyquinoline) aluminum thin films at interfaces with metal oxide filmsof different conductivities. Physical Review B 79 (2009), 235205.

[263] Znaidi, L., Benyahia, S., Marchand, C., Gaston, J. P., Kurdi, J., Guillemoles,J. F. Properties of Pure and Al-Doped ZnO Thin Films Synthesized by SoftChemistry. In 19th European Photovoltaic Solar Energy Conference, volume 49,pp. 301–304, (Paris,2004).

[264] Hines, M. & Scholes, G. Colloidal PbS Nanocrystals with Size-Tunable Near-Infrared Emission: Observation of Post-Synthesis Self-Narrowing of the ParticleSize Distribution. Advanced Materials 15 (2003), 1844.

[265] Kovalenko, M. V., Schaller, R. D., Jarzab, D., Loi, M. A., & Talapin, D. V. Inor-ganically functionalized PbS-CdS colloidal nanocrystals: integration into amor-phous chalcogenide glass and luminescent properties. Journal of the AmericanChemical Society 134 (2012), 2457.

[266] Geyer, S. M., Scherer, J. M., Moloto, N., Jaworski, F. B., & Bawendi, M. G.Efficient luminescent down-shifting detectors based on colloidal quantum dotsfor dual-band detection applications. ACS Nano 5 (2011), 5566.

[267] Znaidi, L. Sol-gel-deposited ZnO thin films: A review. Materials Science andEngineering: B 174 (2010), 18.

[268] Correa, R. E., Dauler, E. A., Nair, G., Pan, S. H., Rosenberg, D., Kerman, A. J.,Molnar, R. J., Hu, X., Marsili, F., Anant, V., Berggren, K. K., & Bawendi,M. G. Single photon counting from individual nanocrystals in the infrared.Nano Letters 12 (2012), 2953.

[269] Thakar, R., Chen, Y., & Snee, P. T. Efficient Emission from Core/(Doped) ShellNanoparticles: Applications for Chemical Sensing. Nano Letters 7 (2007), 3429.

Bibliography 159

[270] Hazarika, A., Layek, A., De, S., Nag, A., Debnath, S., Mahadevan, P., Chowd-hury, A., & Sarma, D. D. Ultranarrow and widely tunable Mn2+-induced pho-toluminescence from single Mn-doped nanocrystals of ZnS-CdS alloys. PhysicalReview Letters 110 (2013), 1.

[271] Hazarika, A., Pandey, A., & Sarma, D. D. Rainbow emission from an atomictransition in doped quantum dots. Journal of Physical Chemistry Letters 5(2014), 2208.

[272] Brown, P. R., Supran, G. J., Grossman, J. C., Bawendi, M. G., & Bulović,V. Transparent luminescent displays enabled by electric-field-induced quench-ing of photoluminescent pixels. US patent application no. PCT/US2016/12971.(2016).

[273] Lee, K. H., Lee, J. H., Song, W. S., Ko, H., Lee, C., Lee, J. H., & Yang,H. Highly efficient, color-pure, color-stable blue quantum dot light-emittingdevices. ACS Nano 7 (2013), 7295.

[274] Chiu, Y.-C., Hung, J.-Y., Chi, Y., Chen, C.-C., Chang, C.-H., Wu, C.-C.,Cheng, Y.-M., Yu, Y.-C., Lee, G.-H., & Chou, P.-T. En Route to High Ex-ternal Quantum Efficiency (12%), Organic True-Blue-Light-Emitting DiodesEmploying Novel Design of Iridium (III) Phosphors. Advanced Materials 21(2009), 2221.

[275] Fleetham, T., Wang, Z., & Li, J. Efficient deep blue electrophosphorescent de-vices based on platinum(II) bis(n-methyl-imidazolyl)benzene chloride. OrganicElectronics: physics, materials, applications 13 (2012), 1430.

[276] Lee, S., Kim, S.-O., Shin, H., Yun, H.-J., Yang, K., Kwon, S.-K., Kim, J.-J.,& Kim, Y.-H. Deep-blue phosphorescence from perfluoro carbonyl-substitutediridium complexes. Journal of the American Chemical Society 135 (2013),14321.

[277] Reineke, S., Lindner, F., Schwartz, G., Seidler, N., Walzer, K., Lüssem, B., &Leo, K. White organic light-emitting diodes with fluorescent tube efficiency.Nature 459 (2009), 234.

[278] Chin, B. D. & Lee, C. Confinement of charge carriers and excitons in elec-trophosphorescent devices: Mechanism of light emission and degradation. Ad-vanced Materials 19 (2007), 2061.

[279] Hwang, E., Smolyaninov, I. I., & Davis, C. C. Surface plasmon polariton en-hanced fluorescence from quantum dots on nanostructured metal surfaces. Nanoletters 10 (2010), 813.

[280] Seth Coe-Sullivanz, Wenhao Liu, P. A. & Steckel, J. S. Quantum Dots for LEDDownconversion in Display Applications. ECS J. Solid State Sci. Technol. 2(2013), R3026.

Bibliography 160

[281] Coe-Sullivan, S. Nanotechnology for Displays: A Potential Breakthrough forOLED Displays and LCDs. In The Society for Information Display, DisplayWeek 2012, 4 June 2012 (Boston, MA).

[282] Universal Display Corporation, PHOLED Features and Performance(http://www.udcoled.com/default.asp?contentID=604) (2015).

[283] Choi, M. K., Yang, J., Kang, K., Kim, D. C., Choi, C., Park, C., Kim, S. J.,Chae, S. I., Kim, T.-H., Kim, J. H., Hyeon, T., & Kim, D.-H. Wearable red-green-blue quantum dot light-emitting diode array using high-resolution intagliotransfer printing. Nature Communications 6 (2015), 7149.

[284] Scholz, S., Kondakov, D., Lüssem, B., & Leo, K. Degradation mechanisms andreactions in organic light-emitting devices. Chemical Reviews 115 (2015), 8449.

[285] Sato, Y., Ichinosawa, S., & Kanai, H. Operation characteristics and degrada-tion of organic electroluminescent devices. IEEE Journal of Selected Topics inQuantum Electronics 4 (1998), 40.

[286] Aziz, H. & Popovic, Z. D. Degradation Phenomena in Small-Molecule OrganicLight-Emitting Devices. Chemistry of Materials 16 (2004), 4522.

Documents

Enhancing quantum-dot luminescence in visible and infrared light emitting devices Geoffrey James