inkjet-printed quantum dot

8/9/2019 Inkjet-printed Quantum Dot

1/84

U N I V E R S I TAT I S O U L U E N S I S

ACTAC

TECHNICA

U N I V E R S I TAT I S O U L U E N S I S

ACTAC

TECHNICA

OULU 2010

C 351

Hanna Haverinen

INKJET-PRINTED QUANTUM

DOT HYBRID LIGHT-EMITTING

DEVICESTOWARDS DISPLAY

APPLICATIONS

FACULTY OF TECHNOLOGY,

DEPARTMENT OF ELECTRICAL AND INFORMATION ENGINEERING,

OPTOELECTRONICS AND MEASUREMENT TECHNIQUES LABORATORY,UNIVERSITY OF OULU


2/84


3/84

A CTA UNIVE RS ITA T I S O UL

C Techn i c a 35 1

HANNA HAVERINEN

INKJET-PRINTED QUANTUM

HYBRID LIGHT-EMITTING

DEVICESTOWARDS DISPLA

APPLICATIONS

Academic dissertation to be presented with

the Faculty of Technology of the Universitypublic defence in OP-sali (Auditorium L10), L

19 March 2010, at 12 noon


4/84

Copyright 2010

Acta Univ. Oul. C 351, 2010

Supervised byProfessor Risto Myllyl

Professor Ghassan E. Jabbour

Reviewed by

Doctor Arto MaaninenDoctor Patrick J. Smith

ISBN 978-951-42-6126-8 (Paperback)ISBN 978-951-42-6127-5 (PDF)

http://herkules.oulu.fi/isbn9789514261275/ISSN 0355-3213 (Printed)

ISSN 1796-2226 (Online)http://herkules.oulu.fi/issn03553213/


5/84

Haverinen, Hanna, Inkjet-printed quantum dot hybrid lig

devicestowards display applicationsFaculty of Technology, Department of Electrical and Information Engineering, Opand Measurement Techniques Laboratory, University of Oulu, P.O.Box 450

University of Oulu, Finland

Acta Univ. Oul. C 351, 2010

Oulu, Finland

Abstract

This thesis presents a novel method for fabricating quantum dot light-emitting devic

based on colloidal inorganic light-emitting nanoparticles incorporated into

semiconductor matrix. CdSe core/ZnS shell nanoparticles were inkjet-printe

sandwiched between organic hole and electron transport layers to produce effi

emissive media. The light-emitting devices fabricated here were tested as individu

integrated into a display setting, thus endorsing the capability of this method as a m

approach for full-colour high-definition displays.

By choosing inkjet printing as a deposition method for quantum dots, seve

currently inevitable with alternative methods are addressed. First, inkjet printing pr

patterning due to its drop-on-demand concept, thus overruling a need for com

laborious patterning methods. Secondly, manufacturing costs can be reduced sig

introducing this prudent fabrication step for very expensive nanoparticles.

Since there are no prior demonstrations of inkjet printing of electroluminescen

devices in the literature, this work dives into the basics of inkjet printing of l

relatively highly volatile quantum dot inks: piezo driver requirements, jetting par

dynamics in the cartridge and on the surface, nanoparticle assembly in a wet drople

of dots on the surface are main concerns in the experimental part. Device performan

discussed and plays an important role in this thesis. Several compositional QDLED

described. In addition, different pixel geometries are discussed. The last part of th

deals with the principles of QDLED displays and their basic components: RGB pixe

thin-film transistor (OTFT) drivers. Work related to transistors is intertwined with Q

ideas for surface treatments that enhance nanoparticle packing are carried ov

assembled monolayer (SAM) studies in the OTFT field. Moreover, all the work don

project was consolidated by one method, atomic force microscopy (AFM), which

throughout the entire thesis.

Keywords: atomic force microscopy (AFM), inkjet printing, inorganic n

organic thin-film transistors, quantum dots, RGB displays, self-assembled

surface treatment


6/84


7/84


8/84


9/84

Acknowledgements

This study is based on research carried out in the Flexible Displa

Arizona State University, USA, and the Optoelectronics and M

Techniques Laboratory at the University of Oulu during the years 200

I thank the Graduate School of Modern Optics and Photon

(Finnish Funding Agency for Technology and Innovation),

Synjkangas Foundation and the Nokia Science Foundation for research.

I want to acknowledge and thank all my colleagues at Ar

University for their help, guidance, discussion and support during m

research program. My understanding in the field was strongly in

several people. Dr. Brigitte Wex introduced me to organic chemist

dots and several processing methods and encouraged me. Dr. Mike

Evan Williams and Dr. Parul Dhagat have given me technical su

grateful to Dr. Madhusudan Singh, Mrs. Tricia Youngbull and Dr. In

their fruitful discussions and friendship.

Not to forget my colleagues in the Optoelectronics and

Techniques Laboratory, I direct my thanks to Dr. Erkki Alarousu fo

ideas, experiences and dreams and Dr. Tapio Fabritius, Dr. Juha

Lic.Tech. Eija Vieri-Gashi for their discussions and friendship.I have been fortunate to have two supervisors during my years

Professor Risto Myllyl has my greatest gratitude for always believin

pushing me forward. Professor Ghassan E. Jabbour got me excit

topic, gave me a great deal of knowledge and encouraged me wh

doubt. He is acknowledged for being a major part of my success.

My family, my parents and my brother, have been my source without their support, love and guidance, I wouldnt have accom

dreams. I forever carry my gratitude for them with me.

My deepest gratitude goes to my fiance, partner in life and

Without him I would have never believed in myself and would nev


10/84


11/84

Abbreviations and symbols

AC alternating current

AFM atomic force microscopy

AM active matrix

CIE commission internationale de l'clairage

CRT cathode ray tube

DOD drop-on-demandEQE external quantum efficiency

ETL electron transporting layer

FWHM full width at half maximum

HOMO highest occupied molecular orbital

HTL hole transporting layer

LCD liquid crystal display

LED light-emitting device/diode

LUMO lowest unoccupied molecular orbital

OLED organic light-emitting device/diode

OTFT organic thin-film transistor

PLED polymer light-emitting device/diode

PM passive matrix

QD quantum dotQDLED quantum dot light-emitting device/diode

QVGA quarter video graphics array

QY quantum yield

RGB red, green, blue

UV ultraviolet

Alq3 Tris(8-hydroxyquinolinato)aluminium

CdS Cadmium Sulfide

ITO Indium Tin Oxide

MeO-TPD N,N,N,N-tetrakis(4-methoxyphenyl)-benzidine


12/84

PEDOT:PSS Poly(3,4-ethylenedioxythiophene) poly(styrenes

poly-TPD poly-(N, N-bis(4-butylphenyl-N,N-bis(phenyl)bePPV p-paraphenylene vinylene

PVK poly(N-vinylcarbazole)

SiO Silicon Monoxide

TOPO Trioctylohosphine oxide

TPBI 1,3,5-tri(phenyl-2-benzimidazole)-benzene

TPD N,N bis(3-methylphenyl) N,N bis(phenyl)ZnS Zinc Sulfide

II-VI refers to the periodic table classification; materi

group two and six materials

a.u. arbitrary unit

c speed of light, 2.99792458108 m/s

cd/m2 candela per square meter

cP centipoises (1 cP = 0.01 Pas)

eV electron volt

g/ml grams per milliliter

h Plancks constant, 4.135667331015 eVs

reduced Plancks constant, 6.582118991016 eVk wave vector

k wave vector, scalarm mass of an electron, 9.10938188 10-31 kg

mM millimole

mg/ml milligrams per milliliter

m/s meters per secondn nano, 10-9

nm nanometer

p momentum

rms root mean square


13/84

I current

V voltI-V current-voltage

ngstrom, 10-10

wavelength

angular frequency

micro, 10-6

s microsecond

theta, contact angle

liquid free surface energy of a liquid

solid free surface energy of a solid


14/84


15/84

List of original papers

This thesis is a summary of the following papers published in internat

journals:

I Haverinen HM, Myllyla RA & Jabbour GE (2009) Inkjet printing of

quantum dots. Applied Physics Letters94: 073108.

II Haverinen HM, Myllyla RA & Jabbour GE (2010) Inkjet printed RGB

hybrid LED. IEEE Journal of Display Technology 6(3).

III Singh M, Haverinen HM, Dhagat P & Jabbour GE (2009) Progress

printing process and applications. Advanced Materials 21: 113.

IV Dhagat P, Haverinen HM, Kline RJ, Jung Y, Fischer DA, DeLongc

Jabbour GE (2009) Influence of dielectric surface chemistry on the micr

carrier mobility of an n-type organic semiconductor. Advanced Functio

19: 23652372.

V Li FM, Dhagat P, Haverinen HM, McCulloch I, Heeney M, Jabbour GE

(2008) Polymer thin-film transistor without surface pretreatment on a gate dielectric. Applied Physics Letters 93: 073305.

Authors contribution to publications I-V:

I II The author defined the research plan and conducted the

with measurements. She analysed the research results

the literature review, wrote the manuscript and producpaper.

III The author discussed the topic with her co-authors and

structure of the review paper. She was responsible for t

review of all the nanoparticle applications, wrote the r

them and helped to organise the final version adjunct

authors.IVV The author helped define the structural analysis and ca

the atomic force microscopy measurements. She

measurement part of the publication and helped revie

paper.


16/84


17/84

Contents

Abstract

Acknowledgements

Abbreviations and symbols

List of original papers

Contents

1 Introduction 1.1 Related work on quantum dot light-emitting devices ..............1.2 Organisation of the thesis ........................................................

2 Quantum dots 2.1 Propagation of photons and electrons .....................................

2.1.1 Free space ......................................................................2.1.2 Confinement effects ......................................................

2.2 Quantum-confined materials ...................................................2.2.1 Core-shell structure of fluorescent quantum dots .........2.2.2 Optical properties of quantum dots ..............................2.2.3 Colloidal processing method ........................................

3 Light-emitting devices 3.1 Organic light-emitting devices ................................................3.2 Quantum dot light-emitting devices .........................................3.3 Fundamentals of display technology ........................................

3.3.1 Engineering points to look at when designing a display3.3.2 Active vs. passive displays ............................................3.3.3 Active matrix OLED displays .......................................3.3.4 Why a QDLED display? ...............................................

4

Inkjet printing 4.1 Background of printing technology ........................................4.2 Piezo inkjet printing technology .............................................4.3 Fluid dynamics ........................................................................

4.3.1 Printhead dynamics ......................................................


18/84

6 Results 6.1 Early steps of device fabrication ................................6.1.1 Spin-coated devices ..........................................

6.1.2 Printed devices .................................................6.2 Poly-TPD-based monochromatic devices ..................6.3 RGB devices ..............................................................

6.3.1 Monochromatic devices ...................................6.3.2

Patterned RGB devices ....................................6.4 Debate on the wetting properties of nanoparticle ink ..

7 Discussion 7.1 Accomplished results ..................................................7.2 Limitations .................................................................7.3 Future work ................................................................

8 Conclusion References

Original Papers


19/84

1 Introduction

Based on a display market survey conducted by the DisplaySearch

2008, over three-billion-dollar markets were estimated for flat panel

the end of 2009 (DisplaySearch Company 2008). Moreover, another

the Freedonia Group indicated that the annual increase in display ma

be 13.2% until 2008 (Freedonia Group 2004). These numbers clearl

important a business display technology is worldwide. The mo

application areas include, in addition to television screens and comput

cellular phones and portable IT devices. Currently, liquid crystal dis

and plasma flat screen technologies lead the market, overshadowing

display technology based on cathode ray tubes (CRTs). However, ne

trying to step into the worldwide markets; one of the first displ

organic light-emitting diodes (OLEDs) were lately debuted by LG a

(www.oled-displays.net 2009, Samsung Electronics 2009) in additiTrends indicate favourability for very thin, fast-switching, high col

displays with affordable price tags.

Ever since the discovery of OLEDs in 1987 (Tang et al. 198

leading towards thin, flexible displays has been rapid. OLED displ

several benefits over their competitors, including a wide viewing a

excellent colour contrast and fast response. Furthermore, organic manaturally emissive, there is no need for backlights or colour filters,

very thin devices (Overwijk et al. 2004). Polymer-based OLED

attraction due to their suitability for solution processing, thu

manufacturing cost drastically. Now, inkjet printing technology beca

thus revolutionising the display manufacturing business. Regard

superior features of OLED TVs, they are still facing some challengemitters are used for red, green and blue pixels, and they are al

different compounds with several different properties, i.e. each of

different lifetime and has to be optimised separately. The bottle

lifetime properties of OLEDs has been the short lifetime (less than 2


20/84

structure). By choosing inorganic lumophores, colour purity

the very narrow emission spectra (FWHM ~ 20 nm) the(Dabbousi et al. 1997). Quantum dot hybrid LEDs utilise th

and thermal properties of inorganic crystals: high temperature

defect toleration as well as virtually no optical quenching

2004). In addition, their applicability in creating flexible,

ensured by organic layers around the quantum dots. Thu

address the problems that both entirely organic and entirelwould individually have.

This study shed light on the following questions: can we

a deposition method to fabricate QDLEDs? What thin-film ch

be achieved in order to produce efficient and bright devices?

properties and ink properties need to be modified? What are t

approach? How do the devices behave? What are their e

properties? Can we use inkjet-printed QDLEDs in display ap

be commercialised? In a nutshell, a conclusion about the

printing technology as a fabrication method for QDLEDs is

this thesis.

This thesis is a compendium of five separate publications

with inkjet printing of nanoparticles. The latter part of this th

publications related to OTFTs, thus introducing the display thon RGB displays and integration of the proposed LED str

backplanes. The results are presented and discussed, taking in

main ambition of the thesis project.

1.1 Related work on quantum dot light-emitting de

One of the earliest studies of light emission from colloidal n

et al. 1982) goes back to the early 1980s, although the qu

effects of IIVI semiconductors were under investigation a

earlier (Bergstresser et al. 1967, Lawson et al. 1960). Howev


21/84

luminance of 100cd/m2. Regardless of the inefficient device stru

results started a new era in the research on hybrid QDLEDs. (Colvin eSimilar to the work by Colvin et al., another report using t

structure was published, this time by Kumaret al. (Kumaret al. 1997

sulfide (CdS) quantum dots were spin coated onto the PPV laye

aluminium was used as the cathode as opposed to the magnesium use

In this particular device, quantum dot layer functioned as charge tra

emission was dominantly from the PPV layer, not from the nLuminance reported was 150 cd/m2 and maximum efficiency reached

A true breakthrough was made in 2002, when a 25-fold

luminescent power efficiency in QDLEDs was reported. S. Coe et al.

new approach, where quantum dots were dispersed in a N,N

methylphenyl) (1,1 biphenyl) -4,4 diamine (TPD) solution an

resulting in a monolayer formation through phase separation (Coe et a

electron transporter, Tris(8-hydroxyquinolinato)aluminium, (Alq3),

sandwich structure, which became the most common structure ov

Luminance was still close to a few hundred candelas per square

efficiency reached the highest value reported thus far, 0.52%.

improvement was explained by a double exciton-forming mechanis

energy transfer from (HTL) - (ETL) excitons to quantum dots and a d

injection into the dots.The same group mastered this approach later on. The need for

study played a big role, and by adjusting the quantum dot layer thickn

quantum efficiency was doubled; 1.1% was reported (Coe-Sullivan

Laudable device performance reported in 2005 set the bar for eff

brightness really high; 2% EQE and 7000 cd/m2 was by far a record i

of QDLED development (Coe-Sullivan et al. 2005). Interestingly, used the same phase separation approach, yet did not achieve sup

performance; currents of less than 1 mA were measured (Zhao et al. 2

Spin casting has been the most popular approach to fabricatin

mainly due to the simple method for achieving a nearly m


22/84

polar solvents as the HTL and ETL and they spun QDs from

thus avoiding dissolving the first hole transport layenanoparticles, and later on protecting the QD layers when th

layer was deposited (Chaudhary et al. 2004). This polymer

0.22% EQE and 500 cd/m2 brightness. On the other han

Gingers groups approached this problem by introducing

linkable hole transport layer. The initial device, with nearly

cd/m2 brightness, was improved in a later study by optimisin

EQE was doubled and luminance peaked at 6000 cd/m2.(Zha

al. 2007.)

Char et al. recently showed how by attaching an anc

quantum dots, a hybrid material can be produced when

polymer surrounds the thiol-anchored QDs. Thus, an ea

resulted in devices where EQE increased threefold compare

unmodified dots (Zorn et al. 2009). Interestingly, highly ef

reported without a cross-linked HTL; CdSe/ZnS quantum d

spun directly onto a polymer TPD layer, resulting in green-e

maximum brightness above 10 000 cd/m2 and 1.4% EQE (B

improvement in efficiency was explained as being caused by

chemical composition gradient that increased the photolumine

above 80%. The same group also reported highly efficient bsimilar device structure, however emphasising QD processin

2009b). In addition to the studies reported here, various grou

casting as a primary deposition method for colloidal quant

(Son et al. 2004, Cheng et al. 2009), red (Stouwdam et al. 20

al. 2007) QDLEDs have been fabricated by spin casting, to

The common denominator of these is the similar device strucomposition structure altered by doping or material selection.

In attempting to permit patterning, the spin coating met

avoided by microcontact printing. Typically, polydimethyl

moulded to form a desired pattern, followed by spin casting


23/84

varied between 0.52%, depending on the position of the QD layer in

i.e. in the HTL, the ETL or between both. Moreover, both groups same method for fabricating white QDLEDs: mixing red, green and

solution in the correct ratio and then transferring the spin-coated lay

PDMS stamp to the organic HTL surface. Furthermore, the dev

current densities in the same order of magnitude (Anikeeva et al. 20

al. 2008a), thus being fully comparable.

Other approaches to depositing QDs include mist fabrication

2008), dip coating (Lee et al. 2009) and inkjet printing (Haverinen

Haverinen et al. 2010, Tayloret al. 2007, Wood et al. 2009), the latte

suitable for low-cost fabrication with exquisite patterning capabiliti

demonstration of electroluminescent inkjet-printed quantum dots was

Haverinen et al. Lately, AC-driven hybrid QDLEDs in which the do

inkjet printed were reported. However, the basis of this thesis repres

attempts to fabricate monochromatic as well as RGB QDLEDs

printing technology.

1.2 Organisation of the thesis

This thesis is organised as follows:

Chapter 2 focuses on the theory of quantum dots: quantum

effects, optical properties and processing methods where th

structure is discussed and the properties of organic ligands are exp

Chapter 3 dives into the properties of organic LEDs and quantu

LEDs, thus explaining in principle how these devices work. Furt

fundamentals of display technology are introduced in this chapter

Chapter 4 concentrates on inkjet technology: the principles of in jetting properties and ink requirements. The second part di

dynamics in the cartridge and on the surface.

Chapter 5 introduces the materials and methods used and briefly d

QDLED f b i i


24/84

Chapter 7 contains a discussion on the obtained results

contribution of this thesis on inkjet-printed QDLEDs. Alimitations of the method used, with some suggestions for

Chapter 8 provides a summary of the results.


25/84

2 Quantum dots

This chapter briefly discusses electron and photon propagation in fre

in a confined system, thus shedding light on the definition of q

confinement. The manifestations of quantum confinement incl

properties; the effect of quantum dot size and structure on optical

described. Furthermore, processing approaches are introduced by ex

simple steps of core-shell fabrication.

2.1 Propagation of photons and electrons

The physical behaviour of photons and electrons is rather similar.

described as a particle and as a wave, known as dualism (Feynman

However, classical physics emphasises that a photon is mainly

transporting electromagnetic wave, while an electron is treated as a charged particle with the lowest mass of matter. However, quantum p

photons and electrons analogously (Prasad 2004). For example, w

described with de Broglies postulate, = h/p, where h is Plancks co

is particle momentum. Electrons possess greater momentum, thus ha

wavelengths, which also has another consequence: confinement

photons occur at larger scales (Atkins & dePaula et al. 2002)Photon propagation is typically described by the electromagnetic

of a medium, where the refractive index of matter resists propagati

propagation, on the other hand, is described by time-independent

equations, where kinetic energy and potential energy are taken into a

equation called the Hamiltonian operator postulates that electron pr

dependent on the kinetic energy of a charged particle and is retarded interactions, described as the potential energy of the particle. Sol

Schrdinger equation gives the allowed energy states of electrons.

The biggest difference between electrons and photons relates t

description: photons have a vector field, while electrons have a sca


26/84

no spatial variance in their refractive index. In this situa

electromagnetic field behaves as a plane wave with a complexvectork implicates the direction of the wave; positive value

direction when a Cartesian plane is the reference plane. Th

direct relationship to momentum and an inverse relation to w

= /, wherep is reduced Plancks constant. Photon proportional to angular frequency; E = (Born & Wolf 1998

The description of free-space propagation of electrons in

the Schrdinger equations resulting in an oscillating sinu

similar to photon propagation. Moreover, wave vector k c

movement. The energy of an electron is postulated in the fol

2k2/2m, indicating all the energy values of electrons allowe

The same reflects to photons; all the frequency values allowed

Therefore, continuous values of E and form a band, thu

electrons and photons within the band. (Born & Wolf 1998.)

2.1.2 Confinement effects

Confinement simply indicates that propagation is dimension

interaction potential that causes reflection, thus inhibiting th

For photons, this can be introduced by trapping a light wavehigher refractive index (Saleh & Teich 1991).

For electrons, confinement is created by an internal ene

from increased potential energy. This traps electrons betw

creating a box where electrons can travel freely inside but w

potential energy of the particle is smaller than the barrier e

Merzbacher 1998). This barrier can be thought of as a barriatoms in a material. For quantum dots, the diameter of a

determines the length of the box. The most important

quantisation of allowed energies, which is visualised in

between atom orbitals forces band splitting to occur due to


27/84

Fig. 1. Molecular orbitals create band splitting when more atoms are close proximity. Bulk crystal has smaller spacing between adjacent e

compared with e.g. quantum dots.

Typical semiconducting quantum dots have a few hundred to seve

atoms, thus their actual size ranges between 210 nm (Borovitsk

2002). Now, wave functions cannot extend fully inside the dots, whic

energy state. Intuitively, smaller dots have a higher energy state than

Therefore, smaller dots have a larger energy gap, thus they emit at a h

(Fig. 2). Size-dependent emission is one of the most intriguing featu

by changing only quantum dot size, emission colour can be tuned th

entire visible spectrum (Houtepen 2007, Borovitskaya & Shur 2002).


28/84

Fig. 2. Quantum dot size relationship to emission wavelength.

2.2 Quantum-confined materials

2.2.1 Core-shell structure of fluorescent quantum d

The original quantum dots had only a core without any p

However, the optical properties of quantum dots can be dra

coating the light-emitting core with a monolayer or a few mo

band gap material shell (Wilson et al. 1993). By passivatingthe core surface with shell molecules, the charge-carrier prop

efficiency and band gap profile can be modified (Prasad200

material is chosen to allow emission from the core through t

absorption losses. Even though the most popular structure inv

layer, multiple shell layers have been demonstrated. A so-ca

has double or triple shell layers with alternating low and high(Schooss et al. 1994). This structure further modifies emissio

to red wavelengths, thus widening the possibilities of tuning e

quantum dots used in this thesis project were CdSe/ZnS dots

Dabboussi et al.


29/84

drawback in this approach is that ligands function as a dielectric, th

charge transport efficiency.

2.2.2 Optical properties of quantum dots

Due to three-dimensional confinement, quantum dots are known to

narrow emission band. FWHM is dictated by the uniformity

distribution; 20 40 nm FWHM is typical nowadays. Size distrib

reason for spectrum broadening; quantum physical properties such a

spectral line width of quantum dots and temperature dependence o

structure also widen the spectra of quantum dots.

Interestingly, the onset of quantum dot photoluminescence is ind

the wavelength of excitation, as long as it is lower than the first abso

as seen in Fig. 3. It is worth mentioning that the peak of lowest energ

and the photoluminescence emission peak have a small shift. ThisStokes shift and is a typical phenomenon with quantum dots. Fur

decreases when dot size increases, reaching a point when a Stokes s

occur (Bagga et al. 2001)


30/84


31/84

3 Light-emitting devices

By nature, organic compounds, like polymers for example, are in

should not be good conductors. However, a surprising discovery

Shirakawa and Heeger opened doors for organic electronics. In t

polyacetylene polymer was doped with iodine, thus rendering an ins

conductor (Chiang et al. 1977). The conduction mechanism

compounds is mainly explained by two mechanisms: 1) overlappin

orbitals with adjacent molecules create a band structure for charg

propagate and/or 2) hopping of carriers over a small barrier to

molecules, enabled by perturbation from surrounding molecules (

Brenner 1964). This is the basis for OLED and QDLED operation.

3.1 Organic light-emitting devices

An over-simplistic model of an OLED consists of two organic ma

transporting and electron-transporting, sandwiched between low an

function electrodes. In some cases, like in polymer light-emitt

(PLEDs), one layer functions as the hole and electron transporter

emitter (Burroughes et al. 1990). In Fig. 4, the operation of a two-lay

used as an example. A high work function material, Indium Tin Oxused as a transparent anode. Holes are injected into the hole transpo

highest occupied molecular orbital (HOMO) and transferred towards

transport layer. At the same time, electrons are injected from th

function cathode into the electron transport materials lowest

molecular orbital (LUMO) and transferred towards the ETL/HTL in

to Coulomb attractive interaction, the holes and electrons form a p

exciton. An exciton site can be formed between two different molecu

a single material, depending on material choices. After a recombina

photon is emitted. An exciton can also relax non-radiatively when ma

energy is released into the system.


32/84

electron injection material (typically a fluoride-based com

between the cathode and the ETL to assist in the injection some devices an additional layer, an emissive layer, can be a

and HTL inject charge carriers into the emissive layer, which

them. In addition, the hole-blocking layer at the cathode site i

exciton-forming site away from the contacts to prevent n

quenching. Regardless of the device configuration, the final

(in the order of a hundred nanometers), and in the case of

virtually immune to bending and twisting.

Fig. 4. Energy level diagram of a typical two-layer OLED system

formed between the ETL and HTL.

3.2 Quantum dot light-emitting devices

A quantum dot light-emitting device has a similar structure

OLED. Here the emitter is a semiconductor nanoparticle

between the charge carrier layers. However, several s


33/84

The reasoning behind the popularity of the sandwich structure

optimal exciton formation in the quantum dots. Fig. 5 a) depicts how dots transport electrons. In transferring an electron from the conduc

one dot to another, the electron faces a barrier due to a wider ban

which is heightened further by a barrier composed of ligands. N

quantum dot layer is very thin, in the order of a monolayer, surroun

layers can easily transport charge carriers into the dot, as seen in Fi

direct injection of charge carriers is assumed to be the most common p

for creating an exciton in the device.

An exciton can also be formed via an energy transfer route. An e

pair is formed in the organic molecule and transferred as a bound e

dots, where it relaxes radiatively. This Frster energy transfer can b

congenial material choice to assist in exciton formation, thus resultin

and more efficient devices (Coe-Sullivan 2005).


34/84

3.3 Fundamentals of display technology

3.3.1 Engineering points to look at when designing

Regardless of display type - LCD, plasma or OLED

specifications that make a display exceptional. Brightness is

consumer estimates first. When RGB pixels have 23% diff

the human eye can sense it. It is a bit more tolerant of vid

change in brightness is acceptable due to the quickly switch

2001). The human eye is also most sensitive to waveleng

which has to be taken into account when designing the optim

of emitting pixels. Good contrast, meaning the boundary bet

regions, resolution, the size of a pixel and the viewing angle

viewers experience of the visualised data. From the engine

due to limitations of the driving circuitry, the driving power30 V if driven with direct current. Furthermore, the switching

be fast; microsecond speeds are achievable, especially w

However, the human eye cannot resolve anything faster th

known that high speed consumes more power, which d

Addressability and degradation mechanisms are both engine

need to be solved. Ultimately, the consumer is looking for ayears to come, which emphasises the importance of improv

display (Pankove 1980).

3.3.2 Active vs. passive displays

Displays can be divided into categories, not only by the emisalso by the pixel driving method. Typically, displays are e

(PM) or active matrix (AM) displays. The pixels of both disp

transistor circuitry. Moreover, a display is formed by rows an

When a passive matrix display is driven, a bias is given to a


35/84

3.3.3 Active matrix OLED displays

Since an OLED display is closest to a QDLED display, it is used

example. An OLED display incorporates five main parts. An integ

with RGB pixels is positioned on a display component carrier. This sy

have a hermetically sealed transparent cover and a set of wire

encapsulated at the end (Pankove 1980). The integrated circuitry inclu

transistor that is responsible for charging and discharging a storage

drive transistor addresses the LED and is bigger in physical dimensiohigher current requirement. However, the bigger size decreases the a

and is the main limiting factor when designing the emissive area of

simplest AM LED display has two resistors and a capacitor to drive o

It is evident that transistors create a bottleneck for opti

performance; the higher the driving current provided by the transisto

the brightness. Most drivers of OLED displays, e.g. Sanyo and Kutilised polysilicon transistors. While the efficiencies of OLEDs hav

lower-mobility transistors have become possible; a-Si:H or organ

transistors (OTFTs) can be used. OTFTs would enable flexi

manufacturing, thus revolutionising display markets. However,

concern in using OTFTs is whether they can provide a steady curren

enough time. Dielectric integrity is also an important factor, which is

Papers 4 and 5 in this thesis. Furthermore, device passivation and TFT

are properties that need to be optimised (Klauk 2006).

3.3.4 Why a QDLED display?

To fabricate full-colour displays from OLEDs, several processin

required, which complicate the manufacturing process. One approac

white OLED as an emitter and colour filters to produce RGB images

pixels of each colour with shadow masks. OLEDs can also be stacke

conversion of blue light can be used to create red and green (Burrows


36/84

QDLEDs one can combine the benefits of inorganic lumoph

the flexibility and simple fabrication routes provided by organ


37/84

4 Inkjet printing

4.1 Background of printing technology

The history of printing technology is long, and most credit bel

Chinese, who first demonstrated printed text and images via woodbl

in the 2nd century (Hind 1935). It took 1700 years to come up with

non-contact printing. Prior to that, all printing methods used phys

parts to make an image. Moreover, most methods invented to date

physical phenomena to deposit material on a substrate. Flexography

use a relief method, offset lithography relies on surface energy, scr

require masking, embossing uses high force and pressure to replace m

ablation removes material, and laser transfer uses excessive force an

release material (Khan 2006). Inkjet technology is the only method

dispenses material, without contact, to the receiving material. Furthertechnology is the only method that uses remotely amount of material

potential to be a high-throughput, low-cost process.

Inkjet printing is also called a digital printing method, since

controlled by electronic devices. The dispensing method div

technologies into two groups: continuous and drop-on-demand (

bubble jet technique is a DOD technique, which differs from piezoein terms of droplet production (Gamota et al. 2004). Fig. 6 demo

principles of each method. The continuous method electrically defle

of droplets by using high-voltage deflection plates. Droplets that

deposited are collected by a gutter and re-circulated. Piezoelectric m

piezoelectric crystal chamber that expands or contracts when a bia

This material deformation creates a pressure wave that forces theexpand and exit through an orifice. Bubble jet is also called a ther

method. As the name describes, a vapour bubble is heated, creating

bubble that results in the formation of an acoustic wave, thus forcing

of the chamber (Le 1998)


38/84

thermal competitors, and they can be used with a wider range

(Clymer & Asaba 2004). Piezoelectric inkjet is used in this wo

Fig. 6. Three main categories of inkjet technology. a) continuous

c) bubble jet.

Inkjet printing is a simple method in theory, but is rather invo

Some of the main things that need to be controlled during the

drop volume, drop shape and formation, printing distance an

quality, evaporation kinetics of the droplet, surface energy

surface tension of the droplet, penetration or spreading p

thickness. Overall, material printing differs considerably

printing; uniformity and excellent film quality are required in

while graphic arts printing accept discontinuous films,

resolution of the human eye (55 m or more). Furthermore,


39/84

The active ingredients in the ink can be polymers, metal n

biological cells or carbon nanotubes, to name only a few. It can

material printing has created new challenges in the design of ink

Therefore, printing should be divided into conventional graphic art

functional material printing techniques.

4.2 Piezo inkjet printing technology

Regardless of the actual implementation of a piezoelectric inkjet syste

thermal inkjet system, the following features can be found in each o

print driver is the heart of the manipulation process. This digital imag

converts data into a location map of dots and also drives the electr

head, creating motion in the piezo crystal.

The print cartridge contains a reservoir for the ink and is atta

actual printhead. The printhead has a small chamber surrounded by a pand an electronic subsystem that is responsible for ejecting co

optimally shaped droplets. This small module is a technologica

system that has to control various functions during and after jetti

maintain thermal and fluid stability. The printhead also has a mainten

for cleaning between print cycles by jetting excessive material, thus

orifice and forcing dried ink out from the opening. Typical prinnozzles in one column; the number of nozzles varies from a few

(Clymer & Asaba 2004).

In addition to the main parts mentioned here, some printer

designed for functional materials have sophisticated piezo control fea

as camera systems in order to have ultimate control of the jetting prop

4.3 Fluid dynamics

4.3.1 Printhead dynamics


40/84

Piezo actuation has several steps, as seen in Fig. 7. Pie

starts the formation of the pressure wave; negative pressure

the change in the shape of the channel. Shortly after the ch

shape of the channel returns to its original state. Meanwh

negative pressure wave is reflected from the bottom of th

become positive pressure. The most efficient way to

amplitude is to amplify the pressure wave after reflection. Jet

propagating pressure wave has overcome surface tension and

entire process typically takes 1020 s, resulting in droplet s

m/s inside the chamber (Wijshoff 2008).

Fig. 7. Piezo actuation and acoustic wave behaviour in the cha

crystal is actuated.


41/84

Channel geometry has other effects, as well. A small noz

accelerates ink movement in the channel, thus decreasing the driv

Furthermore, a small opening together with short channel lengths

overfill effect due to faster pressure wave damping. Moreover, chan

not the only characteristic affecting damping time; higher-viscosity in

damping more efficiently. As a conclusion, optimal damping of the pr

for refilling is not optimal for avoiding overfill, thus piezo actuation

trade-off.

4.3.2 Drop dynamics

Drop shape is an important issue for excellent quality print. Ideally

needs to be fully spherical without following drops, called satellites, b

hits the substrate. However, if small droplets are created near the m

they tend to recombine prior landing onto the substrate thus causinFurthermore, in reality drop shape tends to be more tear drop shape th

yet resulting excellent quality films. In this chapter theoretical exp

drop formation is discussed.

Various parameters affect drop formation: nozzle area and shape,

waveform shape and duration and ink properties, such as viscosity

tension, which are explained here. Nozzle diameter and voltage bomost impact on the maximum volume of the droplet, even th

oscillating parameters modifies the drop volume. Nozzle quality

chips, grooves or dirt can alter the trajectory of the droplets, thus

registration.

Fig. 8 depicts how a drop is formed in an ideal case. Drop speed

as the meniscus speed before ejection from the nozzle. Right after a d

pinched off from the main meniscus, drop speed is reduced typically

initial speed. The tail of the liquid cone slows down the falling ink dr

tail is short enough, the effect of decreasing velocity is less (improved

accuracy due to higher speed), satellite drops cannot be formed and


42/84

Fig. 8. Ideal jetting results in a perfect spherical drop prior t

substrate. In addition to jetting parameters such as piezo acproperties play an important role in drop formation.

Viscosity actually has the biggest impact on the length of th

estimated if in addition to viscosity; also density and surface

Tail length, also called viscous length, is directly proportio

inversely proportional to the product of density and surface

highly viscous inks have a longer tail, which is also symviscosity induces a smaller tail. In addition, higher concentrat

tails. Drop formation is complete when the tail merges w

droplet, creating a perfect spherical drop (Wijshoff 2008).

If the tail is too long, it breaks into smaller drops, calle

break-off in surface tension. These small drops may have diff

can end up around the main drop on the substrate, thus dectremendously. A typical satellite-forming mechanism is high

drop, which on the other hand is created by a high pie

Moreover, satellites can be produced if the pulse width of the

too short or too long Interestingly the tail is not affected b


43/84

power are the main determinants of drop size formation. In dro

inkjet printing, drop diameter is similar to the nozzle diameter. (Wijsh

Drop formation can be easily disturbed by several factors, w

biggest challenges to overcome in order to increase printhead life. N

easily, especially when highly volatile solvents are involved in the in

The clogging problem can be partially solved by introducing clea

between runs and while the printer is idle. Furthermore, air entrap

channel disrupts ejection of droplets. However, the main challenge, a

the research presented in this thesis, is overfill, which means exceplate wetting with ink. The reasons for this include too high a repe

drops, an overfilled channel, wetting properties of the nozzle plate

tension, which allows merging of the ink blob on the nozzle pla

meniscus. Low-viscosity inks tend to overfill easier than higher-vi

Once overfill occurs, it cannot be reversed and the nozzle in that

(Wijshoff 2008, Shin & Smith 2008).

4.3.3 Fluid dynamics at the surface

Fluid dynamics at the surface are a rather complex process. Ink chemi

perfected to meet required properties, such as viscosity, printability, r

surface energy. Moreover, dry ink properties must also be taken iwhat is the desired structure of the film (in terms of physical s

toughness), the adhesion and cohesion attributes and the surface p

printed features (Gamota et al. 2004). Printability of an ink is domina

by surface tension and rheology. Rheology describes the deformation

materials and influences the processability of the ink; thus, film

uniformity and the final properties of the ink are important param

process. Moreover, by definition, surface energy/tension describes

between ink and substrate. Fig. 9 depicts the definition of Yangs equa

equation describes the relationship between surface energies and

which can be described as a physical measure called the contact angle


44/84

Fig. 9. The contact angle is dependent on the free surface energ

and solid/liquid interface.

The contact angle can be derived if the surface energies of th

substrate/ink are known. The contact angle can be though

energy difference between the ink and the substrate. Yangs eq

solidsolid/liquid = liquid cos,

where gamma is the free surface energy at a given point an

between the substrate and the tangent of the ink/substrate

theoretical calculation is rather difficult in reality; it is easie

using a camera and a protractor. However, theoretical c

difficult in reality; it is easier to be measured by using a camer

A large contact angle refers to a pinned, round droplet, wlow surface energy of the substrate and high surface energy

contact angle prohibits spreading of the droplet, thus wetting

Inadequate wetting, however, decreases adhesion (Tekin et

high surface energy of the substrate and low surface energy o

small contact angle, thus enhancing wetting.

The final footprint of the ink is a result of the surfac

described in three steps. The droplets impact with the substr

on drop shape. In addition, gas pressure plays a big role in d

surface. After hitting the surface, the droplet starts expand

order of magnitude comparable with the drop diameter while


45/84

Fig. 10. Marangoni effect on a pinned drop. At t1 the evaporation rate is h

circumference. Marangoni flow (red arrows) takes place at t2 and so

towards the edge. The result is highest particulate concentration at the bo

The solvent evaporating rate is higher at the edge of the droplet, cre

from the centre of the droplet towards the circumference. Since the

cannot retract, the solute will migrate towards the boundary until al

has evaporated (Crasteret al. 2009). This phenomenon can be slow

even avoided if the substrate is heated, thus allowing less time for th

flow to take place, or if the solute/solvent ratio is increased or if low

boiling-point solvents are mixed. The benefit of mixing solvents com

fact that a high-boiling-point solvent at the circumference will evap

which in turn increases the concentration of low-boiling-point solven

location, thus balancing evaporation rates throughout the whole drople

However, in the case of nanoparticles, a high contact angle mig

Ko, et al. recently showed that in the case of spherical nanoparticontact angle provides uniform gas pressure throughout the drople

well as longer drying time, which will assist in close packing of nano

et al. 2004). Furthermore, Perelaeret al. dove deeper into this topic

light into packing of nanoparticles in inkjet printed droplet (Perelaer

Similar results were seen during the research presented in this thesis

the conventional rules of surface properties may not be valid in nanoparticles.


46/84


47/84

5 Materials and methods

5.1 Fabrication steps for QDLEDs

All the devices fabricated during this research had the same fabr

regardless of the material choices, e.g. in the charge carrier layers. F

the materials used were taken from the same batch to make a compari

each device set easier. Device preparation involved pre-cleaning of th

spin coating of the hole injection and hole transporting layers,

quantum dots and thermal evaporation of the electron transport

cathode. Testing occurred typically a few hours after device fabricatio

5.1.1 Device preparation

Substrate

Transparent Indium Tin Oxide (ITO) coated on glass was used a

contact of the device. When pixilated ITO was used, UV -cured pho

patterned to define the pixel area; no ITO was etched out. The sub

cleaned with a three solvent method in which the substrates were s

minutes in methanol, acetone and isopropyl alcohol and dried wnitrogen. Prior to any deposition step, the ITO/glass was also plasma

2 minutes with oxygen plasma. Based on Atomic Force Microscop

root mean square (rms) roughness was typically around 0.50.9 n

highest peaks around 4060 nm.

Hole injection and hole transport layers

Due to the roughness of the ITO surface, which causes high local e

that in the worst case could short the device, the hole injection layer, c

d i l PEDOT PSS (B P4083) i d i


48/84

and less than 10 nm maximum peaks were measured. T

PEDOT:PSS film was 30 nm, measured with a Dektak 3030 s

The hole injection material in the device was mostly

poly-TPD [(poly-(N,N-bis(4-butylphenyl-N,N-bis(phenyl)b

thermally cross-linkable poly(N-vinylcarbazole) (PVK) de

Corning was also used. In the case of poly-TPD, a 6 mg/ml co

utilising chloroform as a solvent was prepared. PVK, howev

solution from the manufacturer, with 3 wt% PVK content

ketone (MIBK) solvent, and it was used as received, but wwhen dopants were added. The poly-TPD layer was cross-link

minutes, and the PVK, at 190 C for 5 minutes. After cross-li

was 40 nm thick (0.65 nm rms roughness) and the PVK (0.39

was 50 nm thick. Polymerisation strength was tested for bo

coating bare solvent on the cross-linked layer and checked

removal of un-polymerised material. Both materials seemedusing the given method.

Quantum dot layer

The quantum dots used in the first set of monochromatic dev

CdSe/ZnS with an octadecylamine ligand. These dots were Nanotech, LLC from Arkansas. However, the RGB devic

commercial quantum dots from Evident Technologies, als

core/shell structure. The organic ligand in this case was n

provider, but was most likely based on oleic acid or oleyl a

soluble in non-polar solvents. The quantum dot concentratio

to study between 2 mg/ml and 10 mg/ml. The dots were di

cyclohexylbenzene or a toluene:dichlorobenzene mixture.

The inkjet printer used in this study was a Fuji Dimat

(DMP2800 series), which has one row of 16 piezo driven no

nozzle opening and 254 m spacing between adjacent noz


49/84

adjacent drops was set at 510 m. The printed films were annealed

after deposition at 180 C for 10 minutes to remove the organic lig

remaining solvent. The substrates were annealed again in a glove

thermal evaporation in order to remove adsorbed water.

Electron transport layer and cathode

The electron transport material, TPBi (1,3,5-tri(phenyl-2-ben

benzene), was thermally evaporated in a 10-6 Torr vacuum at the ratLayer thickness was 40 nm. The cathode consisted of a 1.5 nm laye

fluoride (LiF), which served as the electron injection layer, and 150

aluminium. The LiF was deposited at a rate of 0.04 /s and the alu

higher rate, typically a bit below 2 /s. The substrates were transfe

nitrogen-filled tester glove box immediately after deposition.

5.1.2 Testing and measurements

Various measurement tools were utilised throughout the experimen

absorption properties of the quantum dots were measured with a Car

Vis spectrophotometer at a range of 200800 nm. Photolumine

measured with a Jobin Yvon FL4-1057 Fluorolog setup. The surface and topography of the various films were analysed by Veeco DI3

Force Microscopy. The same system was also used for thickness mea

addition to a Dektak 3030 profilometer.

Electroluminescence spectra were measured with an Ocean Opt

spectrometer and I-V measurements were carried out using a Ke

source meter. Optical power was measured using a Newport 2832 C p


50/84


51/84

6 Results

The results of this thesis project involve device performanceengineered structures along with a structural analysis of the deposited

AFM. The work presented in Papers IV and V is strongly bound to

of quantum dot layers, therefore making them an important part o

obtained in this thesis project. All of the major topics - device ch

AFM work and surface modification - are intertwined into the result

thesis and are not separated into individual topics.

6.1 Early steps of device fabrication

6.1.1 Spin-coated devices

Optimisation of the device architecture started with spin coating qonto various QDLED structures in order to find the most promising c

to be used in printed devices. Fig. 11 reveals the first set of devices te

energy band diagram associated with them. A device with only PED

the anode side was investigated to shed light on the exciton-forming

and to see how important a role the hole transport layer has in the ac

only injection of holes is possible in the proposed device. Moreoverlower HOMO level of PVK than of PEDOT:PSS, PVK was introduce

transport layer in order to achieve more efficient hole injection. The

PVK 3 wt% (roughly 50 nm) was thought to be enough to smooth

surface, therefore no PEDOT:PSS layer was spun onto the ITO

arrangement, the device was kept as simple as possible in order to ke

voltages at reasonable levels, but also to decrease the number of fabri

Now, by doping PVK with PBD [2-(4-tert-butylphenyl)-5-(4-biphe

oxadiazole], which has traditionally been used as an emitter in OLE

and possesses electron-transporting properties (Lee et al. 2000), the en

of the holes was further decreased (HOMO level at -6 2 eV) Howeve


52/84

Fig. 11. First device structures and their corresponding energy in energy levels are electron volts [eV].

For these devices the quantum dots were dispersed in tolue

concentration, spin coated at 1000 rpm for 60 seconds and he

minutes. The quantum dots used in devices 1, 2 and 4 had

600 nm ((QY > 40% and FWHM 25 nm), emitting in the dee

visible spectrum. The quantum dots used in device 3 were demission peak at 620 nm ((QY > 40% and FWHM 23 nm). It

here that for dots with a 600 nm emission peak, their energ

Eg= hc/, was 2.06 eV, and for dots emitting at 620nm Eg, it

this small gap difference of 0.07 eV, comparisons between de

feasible; the PVK:PBD device still has the smallest energy ba

quantum dot boundary.Fig. 12 depicts the emission spectrum of each device. Em

the quantum dots, indicating good surface coverage by dots a

4. Pinholes in the quantum dot layers would show some e

l l th N b d th b d di


53/84

nm. This, in turn, would suggest there is no physical hindrance

forming a closely packed layer.

Fig. 12. Normalised emission spectrum of each device. Insert: phoilluminated devices; top picture for 600 nm emission peak dots and low

620 nm emission peak dots.

As expected, the device with only PEDOT:PSS on the p-type side of

the lowest brightness and external quantum efficiency (EQE). Bright

a maximum of 125 cd/m2 at 10 V, while maximum EQE was 0.012%

to the small overall thickness of the device, the turn-on voltag

Regardless of the largest energy barrier of the holes at the QD sit

based device reached maximum brightness of 298 cd/m2 at 15.8V. T

of this structure was low EQE due to an unbalanced hole-electron co

only 0 058 % was measured By doping PVK with PBD thus l


54/84

Fig. 13. I-V curves of each device. TPD-doped PVK (black

performance.

6.1.2 Printed devices

The same structures were used when the quantum dot layer w

this point, the dots were dispersed in toluene and th

concentration was chosen. Due to the really poor performa

based device, it was not used in the study. Therefore, only de

11 were fabricated by printing.All the devices had 10-micron drop spacing, which was su

surface completely. However, it turned out that the drop spaci

overlapping of adjacent rows created thick lines at the edg


55/84

Fig. 14. Example of a printed device with dark non-emitting regions due

concentration and drop spacing (a). Optical interferometry measuremen

structure in more detail (b), scan area roughly 600 m x 400 m.

Furthermore, the devices had low brightness and low EQE. Device

PVK on the anode side, showed maximum brightness of 33 cd/m2

afor the same device was measured to be 0.03% at 18 V. The PBD-

device had the highest EQE in this batch of devices; 0.04% at 30 V w

Luminance, however, was only 20 cd/m2 at 35 V. The TPD-doped

had really low EQE, 0.024%, and luminance of 21 cd/m2 at 25 V.

It was evident, based on these first printed devices that much effo

be placed on optimising the QD ink. Concentration adjustment w

parameter to be changed. In addition, the solvent choices were recon

on. Fig. 15 shows the emission spectra of TPD-doped PVK device

printed QD ink had concentrations of 10 mg/ml, 7 mg/ml an

Interestingly, the highest-concentration QD ink resulted in the po

devices; exciplex emission was more prominent in these devices.

indicate either larger pinholes or a higher density of smaller uncovere

reason for this phenomenon is most likely explained by the solvent/so5 mg/ml ink droplet has more solvent per drop, therefore its drying

longer. Thus, quantum dots would get more time to reassemble on the

form a more uniform layer.


56/84

Fig. 15. Concentration dependence between QD ink and the spec

doped PVK devices. Lower-concentration ink results in better QDoptimal solvent/solute ratio.

When the concentration was decreased further to 2 mg/ml, fu

achieved with the given drop spacing. Fig. 16 shows the A

quantum dots on a TPD-doped PVK surface. Fig. 16 a) is a he

nm scan area. Brighter colour indicates a higher surface. T

section of one scan line and clearly depicts monolayer islanFig. 16 b) is an amplitude image of the same scan showing i

vividly. The rms roughness of the measured area was 2.05 n

was ~5 nm. The dots had a total diameter of 46 nm

measurements. Therefore, the observed islands are monolayer


57/84

Fig. 16. AFM data of QD (2 mg/ml) on a PVK:TPD surface. Full cover

obtained with the given concentration and drop spacing. Coverage is

Scan size is 500 nm in both figures.

6.2 Poly-TPD-based monochromatic devices

As mentioned earlier, solvent choice needed to be re-evaluated. Tolue

viscosity (0.590 cP), low-density (0.867 g/ml) solvent which also h

pressure of 2.9 kPa. A solvent with better jetting capability as well as

drying was needed. Chlorobenzene was the choice due to its higher v

cP) and lower vapour pressure (1.6 kPa), which both assist in unformation. Furthermore, chlorobenzene has a density of 1.1 g/ml, thu

favourably with toluene in this aspect, as well.

Fig. 17 shows how solvent choice impacted jetting parameters

shows jetting of toluene-based ink. As can be seen, satellite droplet

regardless of optimisation of piezo actuation. However, when the

switched to chlorobenzene, jetting was more controllable and individwere formed, as seen in Fig. 17 b). Therefore, small changes in visco

pressure and density affected jetting characteristics positively. N

nozzles are actuated simultaneously in Fig. 17 a), but only one nozz

b) Th i t k ith th i t


58/84

Fig. 17. Drop watcher images of the jetting nozzles. Toluene (a) a

had different properties, where chlorobenzene had better con

higher viscosity and lower vapour pressure.

When the solvent was changed to chlorobenzene, a new

Contact angle measurement indicated that the QD ink drop

behaved hydrophobically, as seen in Fig. 18. The contact a

was 78.4. Quick oxygen plasma treatment of PVK:TPD

sufficiently, and the contact angle was close to zero a

However, regardless of the short ashing time (10 s), t

morphologically and ashing also introduced an extra step iwas not ideal when processing was designed to be kept as

Therefore, a new hole transport layer was investigated. Based

with spin-coated devices, TPD-based devices seemed

performance. Now, UV cross-linkable poly-TPD was desig

rigid HTL layer that also tolerated the solution process after

showed good compatibility with QD ink as far as wettabilityseen in Fig. 18 b). The contact angle was measured to be close


59/84

Fig. 18. Contact angle measurement of a QD/chlorobenzene drop on PVK

a QD/chlorobenzene drop on poly-TPD (b).

Now, using poly-TPD as the HTL, the concentration study was partia

Since the previous results suggested that the most optimal concentr

be between 2 mg/ml and 5 mg/ml, two devices were fabricated with 4 mg/ml QD ink. In order to verify the best surface coverage, AFM m

was executed. The AFM scan was performed in air after printing th

tapping mode with an aluminium-coated tip was used for phase ima

the small size of the dots and the surrounding material, only ph

important information about the dots. Phase imaging is optimal f

smooth samples due to its excellent sensitivity to morphological chasurface. The exceptional properties of phase imaging are also describ

paper I in the supplements. Two phase images are shown in Fig. 1

shows the QD surface when the concentration was 3 mg/ml and

displays the printed QD surface with 4 mg/ml ink. Hard quantum d

brighter in the scan as opposed to soft organic material, thus

individual dots on the film. It is evident on the basis of the AFM

mg/ml shows better surface coverage than 3 mg/ml, therefore the d

prepared using 4 mg/ml ink.


60/84

Fig. 19. AFM phase images of 3 mg/ml ink printed on poly-TPD (a

Scan size is 500 nm in both images.

However, printed QDs do not show a hexagonal closely

reported by other groups when dots were spin coated [19].

even though small areas of closely packed structure can be

this was sought from the underlying poly-TPD layer. A cros

test was conducted by introducing a solvent layer on a cured p

measurement was conducted prior to and after the solvent trea

a solvent layer was deposited on a poly-TPD layer via a spin

a printing process.

The poly-TPD layer was spin coated at 2000 rpm for 6cured at 254 nm for 10 minutes. AFM measurement was us

imaging and roughness measurement. The poly-TPD film

smooth after deposition and curing; the rms value was 0.248

valley value was 8.34 nm. The solvent used for this study was

in the case of spin coating, it was spun onto the cross-linked H

60 seconds and dried prior to the AFM scans. Furthermore, was used as the solvent deposition method, chlorobenzene w

TPD and dried in air post-deposition. Fig. 20 points out th

prior deposition (a), after spin coating (b) and after inkjet prin


61/84

Fig. 20. AFM height images of cross-linked poly-TPD (a), the same lay

coating chlorobenzene (b) and after inkjet printing chlorobenzene (c). Sc

micrometers.

The rms roughness of the deposited poly-TPD was 0.248 nm an

coating, 0.514 nm. The AFM scan indicated that solvent treatmen

coating had virtually no effect on the topography of the HTL laye

when the solvent was jetted, apparent changes occurred. The rms roug

layer was 0.972 nm. The explanation for this difference between the t

solvent treatments arises from the solvent interaction time with th

layer. When the solvent is spin cast it has a relatively short time to

the underlying layer; within the first few seconds a majority of the

been removed and only a really thin layer remains throughout

process. Nevertheless, inkjet-printed solvent stays on the surfaccompletely dried, thus having more time to interact with the polym

linking is not complete, non-cured material can start to move around

dynamic laws and rearrange itself in lumps, since they are

favourable structures. The small material islands in Fig. 20 c) depict d

packing. Island height is roughly 5 nm.

Regardless of the less than optimal QD layer, the devices were obtain valuable information on QD performance in a device setting. T

in these devices was ITO/ poly-TPD/QDs/TPBi/LiF/Al. The HTL wa

thick, whereas the QD layer was in the order of a few monolayer

increase in device brightness only Furthermore EQE was 1


62/84

increase in device brightness only. Furthermore, EQE was 1

compared with the TPD-doped PVK devices.

Fig. 21. EL properties of a poly-TPD-based device. The sp

dominant emission from the dots. The IV characteristics (b) s

cd/m2

brightness and ~ 0.2 maximum EQE (Paper I American In

At this point, attention focused on patterned devices. The fo

towards display applications, which meant printing onto pixi

first attempt to show the capability of inkjet-printed QD devi

substrates was demonstrated; refer to Paper I. Fig. 22 summ

amount of jetted material was optimised by analysing the spedevice (a). Fifteen drops per pixel (ca. 150 pL per pixel

exciplex emission significantly, indicating better film coverag

also had a coherent response to layer thickness (b); 15 dr

roughly 2.5 V higher turn-on bias compared with the 10 dro

pL per pixel).


63/84

Fig. 22. Optimising ink volume per pixel. Fifteen drops per pixel had b

response than 10 drops/pixel (a). However, driving bias was increase

thicker quantum dot layer (b). (Paper I American Institute of Physics).

6.3 RGB devices

Studying monochromatic red, green and blue quantum dot LED str

naturally one step closer to RGB displays based on QDL

demonstration of these structures, integrated devices would follow

changed in these structures compared with the previous ones was usi

ITO as a substrate, adding a PEDOT:PSS layer under the HTL layer

solvent mixture in the QD ink. Now, a 1:1 ratio of toluene:dichfunctioned as the solvent and 2.5 mg/ml of quantum dots were dispe

Solvent mixture was essential when jetting was optimized since th

dispersed into toluene when purchased. Jetting properties were imp

dichlorobenzene was added into toluene. Due to the lower concentrat

of drops per pixel was increased. The typical number of jetted drops p

18 for red and green ink and 39 for blue ink. The volume of blue ink p

to be increased because it seemed to be harder to obtain dominant em

the dots only. The reason for this is unknown, as the company provid

did not reveal details about the ligand. If the ligand in blue dots were

i ht h h d i t tti ti

shoulder can be seen in red and green emission The spectra


64/84

shoulder can be seen in red and green emission. The spectra

the red and blue devices and at 20 V for the green device. As

concerned, the red device had the greatest brightness; 352 cd/17.5 V and EQE at the same bias was 0.23%. Energetically

case since red dots have a smaller energy barrier from the H

which assists in more efficient hole injection. When the dot s

valence band also goes deeper, thus increasing the energy ba

layer is kept the same in these devices, the red devices have t

due to the most favourable structure of the device.Green devices had the best brightness, 270 cd/m2 at 18 V

at the same voltage. Furthermore, blue devices exhibited 122

corresponding EQE of 0.13%. More details about device

analysis can be found in Paper II in the supplements.

Fig. 23. Device characteristics of R, G and B devices. The EL s

emission from the dots (a). However, the red device was

energetically most favourable structure (b). (Paper II IEEE

Technology).

The purity of emission colour can be estimated by calculating

of the emission spectrum. The CIE (Commission Internatio

coordinates for the red device were (0.57, 0.37), for green

blue (0 12 0 13) as seen in Fig 24 The red and green dev

that emit in deep blue wavelengths. Regardless of quantum dot choi


65/84

p g g q

the process remains unaltered.

Fig. 24. Photographs of monochromatic devices. 232 pixels are driven sim

The red device has CIE coordinates of (0.57, 0.37), the green device (band the blue device (c), (0.12, 0.56). (Paper II IEEE Journal of Display Te

6.3.2 Patterned RGB devices

Integrating RGB into one device was a rather simple task, since th

parameters of each QD ink were established. The biggest challenge w

a routine for position accuracy when the ink cartridges were changedone ink can be deposited at a time, each ink needs to be in an individu

Furthermore, the cartridges are manually changed between print

increases the position error of the following layer However a succes

balance. After all, each pixel is driven simultaneously with a


66/84

, p y

increases the red emission ratio in the entire spectrum. E

driven individually for optimal colour balance.

Fig. 25. Integrated RGB device. The EL is dominated by emissioInsert: Photograph of 232 sub-pixels illuminated simultaneously

high brightness and EQE comparable with monochromatic de

IEEE Journal of Display Technology).

The integrated RGB device turned on at 5.2 V, which is optim

lower than in devices fabricated earlier. Maximum brightnes

at 17.5 V and EQE 0.14 at the same bias. Typical video brig

which was obtained at 9.3 V with 0.24% efficiency; see Fig. 2

6.4 Debate on the wetting properties of nanopartic

The conventional approach to printing functional materials is

energy of the substrate to allow complete wetting. Ideally, t

will be diminished, thus resulting in a more uniform layesupplements). However, some studies have indicated th

nanoparticles, hydrophobicity would be beneficial in nanopa

to the longer drying time and more uniformly distributed

the surface hydrophilic. Contact angle measurement gave a ~0 angl


67/84

droplet on glass. Now, an octadecyltrichlorosilane (OTS-18) -treated

was altered to be hydrophobic, and had a contact angle of 95-100. Fan octanedecanethiol (ODT) -treated gold surface with a 110 c

functioned as a superhydrophobic surface. The OTS treatment

explained in Paper IV. Thiol treatment, however, included soaking o

substrates for 10 minutes in ethanol and then transferring them int

ODT solution for 18 hours. The impact of surface treatment, or on th

a non-treated surface, on OTFT performance was studied in Papers Ished light on packing of molecules onto the surface as well as the nec

self-assembled monolayers when the properties of the deposited lay

optimised. The results from those works were reflected and exploited

this study.

Fig. 26 shows three AFM scans of each prepared substrate. T

contains amplitude images of the scans and the second rocorresponding phase images. Two images per scan are included to cla

what was concluded. The first column (a) and d) images) shows how

dots occurred on bare glass, i.e. on a hydrophilic surface. The rms rou

2.02 nm and the peak-to-valley value was 13.6 nm. The amplitud

reveals this surface to be most disorganised. Moreover, the phase ima

same scan depicts voids in packing. In the second column, OTS trea

to enhance packing of dots. The rms roughness of this surface was

the peak-to-valley value 6.73 nm. However, the best packing visuall

the thiol-treated surface, seen in the third column. The rms value was

be 0.612 nm and peak-to-valley, 5.21 nm. All the scans are 500 nm.


68/84

Fig. 26. AFM scans of printed dots on three different surfaces

either hydrophobic (middle and right column) or hydrophilic (l

row shows amplitude images of each surface (a, b, c) and th

corresponding phase images (d, e, f). Scan size in each image is

Based on these results, it seems that a hydrophobic surfac

favourably. However, it is hard to achieve a hydrophobic su

device setting, thus the question is how important this is for

This will be discussed further in Chapter 7.

7 Discussion


69/84

Since the most important aim of this thesis project was to demapplicability of inkjet printing technology in the fabrication of QD

perspective dominates the following discussion on the results. T

providing plausible explanations for the obtained results and di

limitations of the applied methods.

7.1 Accomplished results

The results of the work introduced in this thesis clearly show the rele

method used to fabricate LEDs based on colloidal nanoparticle

Moreover, the described method has major benefits over other proc

make these results remarkable. The strongest argument supporting in

as a deposition method arises from the capability to pattern in innum

with virtually no material loss during the process. Colloidal q

chemistry seems rather simple; however, obtaining infinitely

distribution is a tremendous challenge. To achieve this is laborious

which also is the main cost factor of QDLED manufacturing. There

fabrication should be as simple as possible, where inkjet printing is

solution.

Even though the studied device structures were not fully ablesuper-brightness, it is justifiable to claim that inkjet-printed QDLED

potential. Since spin-coated devices showed results comparable

printed devices, there is not much difference in the devices as far as

is concerned. Thus, the limiting factor of devices with high brightne

EQE is not the inkjet process; lower brightness compared with ot

results is a consequence of something else.Plenty of effort has been put into this research field by other grou

hexagonally closely packed QD monolayers. This issue was also c

during this research, but interestingly, it seemed not to have much effe

similar performance, which questions the requirement of a p


70/84

layer.

The most significant result of this thesis project is the dedevices. These results substantiate the great potential that in

have for display manufacturing, as well. Moreover, the devic

a level where the process could be easily transferred to a

fabrication. In addition, integration with OTFTs, studied in

feasible and could be achieved with reasonable effort.

The results achieved in this thesis project will also benefields as well, and are not limited to serving only display ma

sensor applications using nanoparticles in their process woul

applicable. More applications are mentioned in Paper III.

7.2 Limitations

A major limitation of using inkjet printing as a QD depositio

from ink/organic layer compatibility. The hole transport lay

cross-linked in order to withstand the harsh organic solvent

seen in chapter 6, spin coating tolerates solvents better due to

time compared with jetted solvents. Therefore, device c

dependent on HTL quality, which limits configurations sign

cross-linkable HTL materials other than the ones used in thi

allow successful device design.

Moreover, modification of QD ink is limited to solvent

conventional methods that alter, e.g. viscosity, with additives

be utilised. Thus, the material printers limitations in viscosi

an important role when designing QD inks. For example, the

in this study has a viscosity range of 120 cP, whereas threality is typically below 1 cP. Overcoming this dilemma

challenging part of the work, and much effort was needed

actuators successfully.

7.3 Future work


71/84

Future work includes integrating RGB QDLEDs into an active matrdemonstrate a QVGA AM-QDLED display. In addition, more atten

placed on defining the best solvent choice or mixture that would be

with the cross-linked HTL. Furthermore, there is room for device opt

possibly emphasising charge balance by utilising hole blocking

blocking layers. Some studies also show that QDs should not be

between the HTL and ETL, but should be embedded into organic laye

White light QDLEDs produced by inkjet printing is also a pros

step and would be studied by printing mixtures of dots or by print

R,G and B inks onto pixels in an optimal ratio. Moreover, QDLED

sensor applications is also an interesting research area.


72/84

8 Conclusion


73/84

The main ambition of this thesis project was to investigate an inmethod used with low-viscosity nanoparticle inks to fabricate quantu

emitting devices. First, spin coating was used as a reference method

device configurations were made and tested. Based on these resu

attempts were made to print QDs and investigate device performance.

In order to improve device performance, fine tuning of

configuration became essential along with ink modification, whemphasis was on solvent mixtures. Atomic force microscopy measur

employed to assist in determining the optimal parameters of the

process itself.

Monochromatic printed QDLEDs were fabricated successfully, a

fully integrated RGB device was presented. The film quality requ

assessed on the basis of a study consisting of printing nanopartic

energetically altered surface. The results showed no significant impa

performance, regardless of whether dot packing was optimised or no

further studies are needed to fully confirm this statement.

The results of this thesis project clarify that inkjet printing is an a

method in QDLED fabrication, and this technique avoids various ob

with other methods. The most significant achievements would be eas

of RGB pixels in display fabrication, lowering the cost of the mprocess and making it an environmentally friendly process where to

are not wasted. Furthermore, up-scaling of the process is feasible an

by other technologies associated with display manufacturing, not th

manufacturing process. For example, backplane technology is a lim

when designing larger-scale displays.


74/84

References


75/84

Anikeeva PO, Halpert JE, Bawendi MG & Bulovi V (2007) Electroluminemixed red-green-blue colloidal quantum dot monolayer. Nanoletters 7(8)Anikeeva PO, Madigan CF, Halpert JE, Bawendi MG & Bulovi V (2008) E

excitonic processes in light-emitting devices based on organic materials

quantum dots. Physical Review B 78: 085434.

Atkins P & dePaula J (2002) Physical Chemistry. 7th Edition. WH Freeman.

Bae WK, Kwak J, Park JW, Char K, Lee C & Lee S (2009a) Highly efficien

emitting diodes based on CdSe/ZnS quantum dots with a chemica

gradient. Advanced Materials 21: 15.

Bae WK, Kwak J, Lim J, Lee D, Nam MK, Char K, Lee C & Lee S (2009

light emitting diodes based on Cd1-xZnxS/ZnS quantum dots. Nanote

075202.

Bagga A, Chattopadhyay PK & Ghosh S (2006) Origin of stokes shift in In

quantum dots: exchange splitting of excitonic states. Physical Review B

Bergstresser TK & Cohen ML (1967) Electronic structure and optical

hexagonal CdSe, CdS, and ZnS. Physical Review.Bhattacharya P, Ghosh S & Stiff-Roberts AD (2004) Quantum dot opto-elect

Annual Review of Material Research 43: 140.

Born M & Wolf E (1998) Principles of optics. 7th Edition. Oxford, Pergamon

Borovitskaya E & Shur MS (2002) Quantum dots. World Scientific.

Burroughes JH, Bradley DDC, Brown AR, Marks RN, Mackay K, Friend RH

Holmes AB (1990) Light-emitting diodes based on conjugated polymer

539541.Burrows PE, Gu G, Bulovic V, Shen Z, Forrest SR & Thompson ME (199

full-color organic light-emitting devices for lightweight, flat-panel d

Transactions on Electron Devices 44(8): 11881203.

Chaudhary S, Oxkan M & Chan WC (2004) Trilayer hybrid polymer-quan

emitting diodes. Applied Physics Letters 84(15).

Cheng G, Mazzeo M, Rizzo A, Li Y, Duan Y & Gigli G (2009) White

devices based on the combined emission from red CdSe/ZnS quantum

phosphorescent and blue fluorescent organic molecules. Applied Physi

243506.

Chiang CK, Fincher CR, Heeger AJ, Shirakawa H, Louis EJ, Gau SC & Ma

(1977) Electrical conductivity in doped polyacetylene. Physical Revie

Coe-Sullivan S, Woo WK, Steckel J, Bawendi M & Buloviperformance of hybrid organic/inorganic quantum dot light-em


76/84

Electronics 4: 123130.

Coe-Sullivan S, Steckel JS, Woo WK, Bawendi MG & Buloviordered quantum-dot monolayers via phase separation during s

Functional Materials 15: 11171124.

Coe-Sullivan S (2005) Hybrid organic-quantum dot thin film s

Dissertation, Department of Electrical Engineering and

Massachusetts Institute of Technology.

Colvin VL, Schlamp MC & Alivisatos AP (1994) Light-emittin

cadmium selenide nanocrystals and a semiconducting polymer. Craster RV, Matar OK & Sefiane K (2009) Pinning, retraction and t

droplets containing nanoparticles. Langmuir 25(6).

Dabbousi BO, Rodrigues-Viejo J, Mikulec FV, Heine JR, Mattoussi

& Bawendi MG (1997) (CdSe)ZnS core-shell quantum

characterization of a size series of highly luminescent nano

Physical Chemistry B 101: 9463.

Dabbousi BO, Radzilowski LH, Thomas EL, Bawendi MG &Electroluminescence from heterostructures of Poly(Pheny

inorganic CdSe nanocrystals. Journal of Applied Physics 83: 79

inkjet-printed quantum dot

Documents