Fabricating designed fullerene nanostructures for functional electronic devices

Christian Larsen

Department of Physics

Umeå 2014

© Christian Larsen

ISBN: 978-91-7601-187-4

Cover art: An almost PCBM transistor

Electronic version available at http://umu.diva-portal.org/

Printed by: Print & Media

Umeå, Sweden 2014

"Shiny"

i

Abstract

A long-term goal within the field of organic electronics has been to develop

flexible and functional devices, which can be processed and patterned with

low-cost and energy-efficient solution-based methods. This thesis presents a

number of functional paths towards the attainment of this goal via the

development and demonstration of novel fabrication and patterning

methods involving the important organic-semiconductor family termed

fullerenes.

Fullerenes are soccer-shaped small molecules, with two often-employed

examples being the symmetric C60 molecule and its more soluble derivative

[6,6]-phenyl-C61-butyric acid methyl ester (PCBM). We show that PCBM can

be photochemically transformed into a dimeric state in a bi-excited reaction

process, and that the exposed material features a significantly reduced

solubility in common solvents as well as an effectively retained electron

mobility. This attractive combination of material properties allows for a

direct and resist-free lithographic patterning of electronic PCBM films down

to a smallest feature size of 1 µm, using a simple and scalable two-step

process constituting light exposure and solution development. In a further

development, it was shown that the two-step method was useful also in the

area-selective transformation of fullerene/conjugated-polymer blend films,

as demonstrated through the realization of a functional complementary logic

circuit comprising a 5-stage ring oscillator.

In another project, we have synthesized highly flexible, single-crystal C60

nanorods with a solution-based self-assembly process termed liquid-liquid

interfacial precipitation. The 1-dimensional nanorods can be deposited from

their synthesis solution and employed as the active material in field-effect

transistor devices. Here, it was revealed that the as-fabricated nanorods can

feature an impressive electron mobility of 1.0 cm2 V-1 s-1, which is on par with

the performance of a work horse in the transistor field, viz. vacuum-

deposited amorphous Si. We further demonstrated that the processability of

the nanorods can be improved by a tuned light-exposure treatment, during

which the nanorod shell is polymerized while the high-mobility interior bulk

is left intact. This has the desired consequence that stabile nanorod

dispersions can be prepared in a wide range of solvents, and we anticipate

that functional electronic devices based on solution-processable nanorods

can be realized in a near future.

ii

iii

Sammanfattning

Ett stående mål för utvecklingen av organisk elektronik är att kunna

realisera flexibla och funktionella elektroniska komponenter genom

kostnads- och energieffektiva tillverkningsmetoder. Denna avhandling

presenterar möjliga vägar för att kunna uppfylla detta mål genom nya

tillverkningsprocesser baserade på en viktig grupp organiska halvledare,

fullerenerna.

Fullerener består av små fotbollsformade och kolbaserade molekyler där de,

inom den organiska elektroniken, mest använda varianterna är C60 och dess

lättlösliga derivat [6,6]-phenyl-C61-butyric acid methyl ester (PCBM). Vi

demonstrerar att PCBM kan transformeras till dimerer genom en fotokemisk

bi-exciterad reaktionsprocess genom enkel ljusexponering. Dimeriserad

PCBM visar en attraktiv kombination av egenskaper genom bibehållna

elektroniska egenskaper samt en minskad löslighet i vanliga lösningsmedel.

Dessa egenskaper möjliggör en direkt och resist-fri mönstringsprocess av det

aktiva lagret i organiska fälteffekttransistorer genom att inkludera en

tvåstegsprocess bestående av exponering samt framkallning av det aktiva

lagret. En mönsterupplösning på 1 µm demonstreras genom mönstring med

UV-ljus. Den enkla tvåstegsprocessen kan även användas för att skapa

funktionella komplementära kretsar vilket demonstreras genom tillverkning

av en 5-stegs ringoscillator.

I ett annat projekt tillverkar vi enkristallina, men flexibla, nanostavar av C60

med hjälp av en uppskalningsbar lösningsmedelsbaserad metod. De 1-

dimensionella nanostavarna har deponerats och använts som det aktiva

materialet i fälteffekttransistorer. Vi visar att dessa nanostavar uppvisar en

imponerande elektronmobilitet på 1.0 cm2 V-1 s-1, vilket är jämförbart med

vakuum-deponerat amorft kisel som är den nuvarande arbetshästen för

kommersiella tunnfilmstransistorer. Vi demonstrerar att processningen av

nanostavar kan förenklas genom en ljusexponering som transformerar C60 i

nanostavens skal till en polymerisk struktur, medan stavens kärna förblir

intakt. Denna transformation resulterar i att nanostavarna kan användas i

ett stort antal lösningsmedel, vilket i sin tur öppnar upp möjligheter för

storskalig och lösningsmedelsbaserad tillverkning av funktionella

elektroniska komponenter baserat på nanostavar av C60.

iv

List of Publications

I. Direct UV Patterning of Electronically Active Fullerene Films

Jia Wang, Christian Larsen, Thomas Wågberg and Ludvig Edman

Advanced Functional Materials, 2011. 21(19): p. 3723-3728.

http://dx.doi.org/10.1002/adfm.201100568

II. Complementary ring oscillator fabricated via direct laser-

exposure and solution-processing of a single-layer organic film

Christian Larsen, Jia Wang and Ludvig Edman

Thin Solid Films, 2012. 520(7): p. 3009-3012.

http://dx.doi.org/10.1016/j.tsf.2011.12.048

III. On the fabrication of crystalline C 60 nanorod transistors

from solution

Christian Larsen, Hamid Reza Barzegar, Florian Nitze, Thomas Wågberg

and Ludvig Edman

Nanotechnology, 2012. 23(34): p. 344015.

http://dx.doi.org/10.1088/0957-4484/23/34/344015

IV. Solution-Based Phototransformation of C60 Nanorods:

Towards Improved Electronic Devices

Hamid Reza Barzegar, Christian Larsen, Ludvig Edman and Thomas

Wågberg

Particle & Particle Systems Characterization, 2013. 30(8): p. 715-720.

http://dx.doi.org/10.1002/ppsc.201300016

V. Palladium nanocrystals supported on photo-transformed C60

nanorods: Effect of crystal morphology and electron mobility on

the electrocatalytic activity towards ethanol oxidation

Hamid Reza Barzegar, Guangzhi Hu, Christian Larsen, Xueen Jia, Ludvig

Edman and Thomas Wågberg

Carbon, 2014. 73(0): p. 34-40.

http://dx.doi.org/10.1016/j.carbon.2014.02.028

VI. An Arylene-Vinylene Based Donor-Acceptor-Donor Small

Molecule for the Donor Compound in High-Voltage Organic Solar

Cells

Javed Iqbal, Jenny Enevold, Christian Larsen, Jia Wang, Thomas

Wågberg, Bertil Eliasson and Ludvig Edman

Submitted

v

Publications not included in this thesis:

Improving the Performance of Light-Emitting Electrochemical

Cells by Optical Design

Nikolai Kaihovirta, Christian Larsen and Ludvig Edman

ACS Applied Materials & Interfaces, 2014. 6(4): p. 2940-2947.

http://dx.doi.org/10.1021/am405530d

vi

List of Acronyms

AFM Atomic force microscopy

BHJ Bulk heterojunction

CV Cyclic voltammetry

DCFL Direct-coupled FET logic

DSC Differential scanning calorimetry

EDP Energy delay product

EQE External quantum efficiency

ETL Electron transport layer

fcc Face-centered cubic

GND Ground

hcp Hexagonal close-packed

HOMO Highest occupied molecular orbital

HRTEM High-resolution transmission electron microscopy

HTL Hole transport layer

IQE Internal quantum efficiency

ITO Indium tin oxide

LLIP Liquid-liquid interfacial precipitation

LUMO Lowest unoccupied molecular orbital

MPP Maximum power point

NR Nanorod

OFET Organic field-effect transistor

OLED Organic light-emitting diode

OPV Organic photovoltaic

OSC Organic semiconductor

PCE Power conversion efficiency

PDN Pull-down network

PDP Power delay product

PNR Polymerized nanorod

PUN Pull-up network

R2R Roll-to-roll

RFID Radio-frequency identification

SAED Selective area electron diffraction

STEM Scanning transmission electron microscopy

TCO Transparent conductive oxide

TEM Transmission electron microscopy

TGA Thermal gravimetric analysis

XRD X-ray diffraction

vii

Table of Contents

1. Introduction 1 2. Organic semiconductors 2

2.1. The fullerenes 4 3. Electronic devices 6

3.1. The field-effect transistor 7 3.1.1. The inverter 11 3.1.2. The ring oscillator 15

3.2. The photovoltaic cell 16 3.2.1. The bilayer structure and operation principles 17 3.2.2. The bulk heterojunction solar cell 20 3.2.3. The OPV performance 21

4. Fullerene photochemical transformation 24 4.1. Raman spectroscopy 24 4.2. Fullerene patterning by resist-free lithography 25 4.3. Patterning efficiency 27 4.4. The photochemical reaction mechanism 29 4.5. PCBM OFETs 31 4.6. Realizing complementary logic circuits with resist-free patterning 33

4.6.1. The inverter 34 4.6.2. The ring oscillator 36

5. Fullerene C60 nanorods 38 5.1. Synthesis 38

5.1.1. Polymerization 40 5.2. Characterization 41

5.2.1. As-grown 41 5.2.2. Annealing 43 5.2.3. Polymerization 44

5.3. The C60 nanorod transistor 46 5.3.1. Nanorod deposition 47 5.3.2. Transistor performance 49

6. Material design for improved photovoltaics 52 6.1. A small-molecule donor compound for OPVs 52 6.2. The C60 nanorod OPV 56

7. Conclusions 58 Acknowledgements 59 References 61

viii

1

1. Introduction

The field of organic electronics includes the design, synthesis,

characterization and application of organic compounds with electronic

properties. These organic compounds are small molecules or polymers with a

conjugated, carbon based, backbone. Since carbon is an abundant material

on earth, these materials promise to be produced and recycled inexpensively

at a very large scale. The chemical tailoring possibilities of conjugated

materials also allow for synthesis of materials with a wide variety of

properties, making them suitable for both electronic applications (such as

transparent, large surface-area electrodes and transistors) and photonic

applications (such as light-emitting devices and solar cells) [1-4]. Conjugated

materials can also be tailored for improved processability in e.g. low-

temperature solution processes. This subsequently allows the use of

unconventional flexible substrates, such as plastic and paper, which are

compatible with roll-to-roll fabrication, allowing for high throughput device

fabrication [5-9]. Organic conjugated materials are already used in

commercial applications, with the most noteworthy being the organic light-

emitting diode (OLED) displays. Possibilities for future applications include

intelligent packaging, flexible displays [10], radio-frequency identification

(RFID) tags [11], transparent or paintable solar cells [12] and light-emitting

wallpaper [5].

This thesis is structured such that the basics of organic semiconductors

(OSCs) are covered in chapter 2. In chapter 3 the device physics of organic

field-effect transistors (OFETs) and organic photovoltaic (OPV) cells are

disclosed. Chapter 4 describes the process of photochemical transformation

of fullerene molecules to create dimers/oligomers/polymers, and how this

transformation can be used to realize efficient electronic devices and

complementary logic circuits. Chapter 5 presents a solution-based method

for the synthesis of crystalline C60 nanorods, and describes how such

nanostructures can be practically implemented as the active material in

OFETs. Finally, chapter 6 presents an innovative solution to the detrimental

issue of fullerene aggregation in OPVs in the form of a morphology-locked

active material based on C60 nanorods.

2

2. Organic semiconductors

Figure 1. (a) The electron orbitals of a sp2 hybridized carbon atom, with the

three sp2 orbitals positioned at a 120° angle with respect to each other, and

the 2pz orbital positioned in a direction perpendicular to the plane spanned

by the sp2 orbitals. (b) The overlap of two 2pz orbitals on neighboring

carbon atoms in a conjugated molecule will give rise to π-orbitals (red), as

shown for ethene.

Organic conjugated materials comprise a molecular backbone of carbon

atoms connected through alternating single and double bonds in a planar

configuration. The planarity originates from the sp2 hybridization, where the

2s, 2px and 2py orbitals of the carbon atom hybridize into three degenerate

in-plane sp2 orbitals, positioned at a 120° angle with respect to each other.

The 2pz-orbital of the carbon atom is positioned in a perpendicular direction

to the plane spanned by the three sp2 orbitals, as shown in figure 1(a). Strong

covalent σ-bonds are formed between two neighboring sp2 hybridized carbon

atoms through the overlap of two neighboring sp2 orbitals. The two

neighboring pz orbitals will also overlap to give rise to a new π-orbital, which

contains delocalized π-electrons. These π-orbitals provide the rigidity of a

conjugated molecule, since they need to be broken in order to rotate the

molecule. The σ- and π-bonding of an ethene molecule (C2H4) is shown in

figure 1(b).

For a molecule in the ground state, the most energetic electron is positioned

in the highest occupied molecular orbital (HOMO). The next available

electron state is termed the lowest unoccupied molecular orbital (LUMO),

and the energetic difference between the LUMO and HOMO levels is termed

the energy gap (Eg). In long-chain conjugated molecules, the most energetic

3

π-electrons can delocalize over several repeat units, and the length of this

delocalization is called the conjugation length. By increasing the conjugation

length, the energy gap between the LUMO and HOMO levels will decrease in

a monotonic fashion, as shown schematically for an increasing number of

connected thiophene monomers in figure 2 [13]. During this process, the

LUMO+1, LUMO+2 levels, etc. can become so close in energy that it, in some

cases, make sense to term them a LUMO band or conduction band; the same

argumentation for the HOMO levels motivates the formation of a HOMO

band or valence band. As the ground-state configuration (at zero Kelvin) for

a conjugated compound is a filled HOMO band and an empty LUMO band,

these materials are appropriately classified as semiconductors.

Figure 2. Electronic structure as a function of number of thiophene rings.

At sufficiently long conjugation length, the discrete energy levels become

degenerate and form a band-like structure.

An organic semiconductor (OSC) can be reduced (oxidized) through the

addition of an electron (hole) into the LUMO (HOMO) band. Once injected,

the electron is free to move within the π-states, and this intra-molecular

transport is commonly facile [14]. An inter-molecular transfer step is

however necessary in order for the electrons to propagate beyond the

conjugation length, and this is a comparatively slow process. The ease with

which charge carriers move is quantified by the mobility (µ, unit: cm2 V-1 s-1),

which can be calculated as the drift velocity of the charge carrier (v, unit: cm

s-1) divided by the driving electric field (E, unit: V cm-1).

4

When OSC molecules are in close proximity of each other, as in a solid film,

the π-orbitals of different molecules can overlap. This overlap can result in a

more efficient inter-molecular charge transport, and molecular ordering,

through crystallization, can increase the molecular π-orbital overlap and the

mobility significantly [15, 16].

The conductivity (σ) of a semiconductor is given by the equation

pn qpqn (1)

where n and µn are the concentration and mobility of the electrons and p and

µp are the concentration and mobility of holes.

2.1. The fullerenes

An exceptionally interesting type of organic semiconductor is the fullerene. A

fullerene comprises at least 20 carbon atoms arranged in hexagonal and

pentagonal rings (the smallest fullerene C20 only comprises pentagonal

rings), and it can form spherical, cylindrical, ellipsoidal and many other

shapes. The most common fullerene is C60, which is built up from 60 carbon

atoms in alternating hexagonal and pentagonal rings to form a shape similar

to a football, as shown in figure 3.

Figure 3. The C60 fullerene (left) with alternating hexagonal and

pentagonal rings of conjugated carbon. The shape can be directly

compared to a football (right).

C60 and several of its derivatives are excellent organic semiconductors, with

some rather unique features, such as high electron mobility and low LUMO

level. They have therefore been used extensively as the active material in

OFETs, and as the electron acceptor material in OPV devices. The symmetric

form of the spherical C60 molecule makes it prone to crystallization.

5

Moreover, the π-orbitals of C60 are positioned around the entire molecule

and inter-molecular π-orbital overlaps are accordingly possible in all

directions. This combination of properties is the origin to why C60 features

one of the highest values for the electron mobility among OSCs [17].

Additionally, different C60 molecules can be made to chemically connect in a

photo-chemical process, so that dimers/oligomers/polymers can be created

[18].

One drawback with C60 is that its solubility in organic solvents is limited,

making thermal evaporation the deposition method of choice. This

deposition method results in high quality films, but suffers from being very

time-consuming and expensive as compared to solution-based processes.

The solubility issue can however be resolved through chemical tailoring. By

adding a side chain to the C60 molecule to produce the derivative [6,6]-

phenyl-C61-butyric acid methyl ester (PCBM), the solubility is drastically

improved.

6

3. Electronic devices

Figure 4. The basic structure for electronic devices consisting of two

electrodes connected by a thin film of an organic semiconductor in a planar

(a) or sandwich (b) configuration.

Since their discovery, OSC materials have been introduced into a wide

variety of applications, notably transistors, solar cells and light-emitting

devices. Although the operational mechanisms differ, these devices share a

common feature in that they comprise a thin OSC film contacted by two

electrodes. The electrode configuration can be either planar (figure 4(a)),

with the current flowing along the substrate, or vertical or sandwich (figure

4(b)), with the current flowing in a direction perpendicular to the substrate.

A schematic energy level diagram for the constituent components of a 2-

electrode structure at open circuit is shown in figure 5(a). As intrinsic

organic semiconductors by definition are undoped, efficient charge injection

can be achieved by minimizing the injection barrier (qΦ) at the

electrode/OSC interface. For electron injection, this injection barrier

corresponds to the energy difference between the cathode work function (EC)

and the OSC LUMO level (ELUMO), and for hole injection it corresponds to the

mismatch between the anode work function and the OSC HOMO level

(EHOMO). In figures 5(a-c) the electrodes were chosen such that the left

electrode, the cathode, has a low work function, hence favoring electron

injection, while the right electrode, the anode, has a high work function that

favors hole injection.

7

Figure 5. Schematic energy level diagram of a 2-electrode device structure

at (a) open circuit, (b) short circuit, (c) during ambipolar hole and electron

injection, and (d) during unipolar electron injection.

The use of electrodes with different work functions introduces an asymmetry

into the device, which results in a built-in-field (Vbi) in the active material

under short-circuit conditions, as indicated in figure 5(b). By applying a

forward bias (VF) larger than the Vbi, the injected charges will be driven by

the electric field towards the opposing electrode, as seen in figure 5(c). If the

device is biased in the reverse direction, the injection of electrons and holes

will be severely limited by the large injection barriers. The described device

is thus a diode, which rectifies the current in the forward direction [19-21]. If

one instead uses the same material for both electrodes, it is possible to

achieve efficient injection of only one carrier type, and the operation of an

electron-only device is depicted in figure 5(d).

3.1. The field-effect transistor

The realization of the transistor and the integrated circuit sparked a

technical revolution, which has resulted in faster, smaller and more efficient

electronic devices. Following the demonstration of the metal oxide

semiconductor field-effect transistor (MOSFET), the efficiency of the

integrated circuits was further increased, eventually leading up to the high-

8

performance electronic tools and toys we have around us today. However,

the processing of the constituent inorganic semiconductor materials, most

commonly Si, is expensive, energy consuming, and executed at high

temperatures not compatible with flexible substrates. OSCs, on the other

hand, can due to their solubility and intrinsic flexibility offer a

complementary category of low-cost and flexible electronic devices, which is

not available for their inorganic counterparts.

OFETs are fabricated in a planar geometry, with the electrodes in contact

with the OSC termed source and drain. The OFET structure also comprises a

third gate electrode separated from the OSC and the source and drain by an

insulating dielectric layer [22-25]. The role of the gate is to capacitively

induce charges in the OSC channel next to the dielectric through the

application of a gate voltage. The OSC conductivity is thus controlled by the

gate voltage, as described by equation (1).

Figure 6. Schematic representation of the four different OFET geometries,

with the definition of the gate length, LG, and the drain width, WD, included

in the bottom-gate, bottom-contact OFET in (a).

There are four OFET geometries, which are distinguished by the position of

the active OSC layer and the electrodes, as schematically depicted in figure 6.

The bottom-gate (BG) geometry has the gate electrode located directly on the

substrate and below the active layer1, whereas the top-gate (TG) geometry

has the active material directly on the substrate with the gate on top.

1 The BG geometry is commonly used in research since doped Si with a thermally grown SiO2 layer is a

substrate which simultaneously fulfills the role as both gate electrode and dielectric layer, making fabrication

procedures simple.

9

Furthermore, the source and drain electrodes can be either in top-contact

(TC) or in bottom-contact (BC) depending on whether they are deposited on

top or below the active material. The schematic figure of the BG-BC

geometry, shown in figure 6(a), also includes the definitions of the gate

length (LG) and the drain width (WD).

Figure 7. Schematic presenting the operation of an OFET (a) in the linear

region, (b) at the pinch-off point and (c) in the saturation region. The dark

red color indicates the charge accumulation in the channel. (d) Typical

output data with the different regions indicated.

The OFET operation can be separated into three regimes: the sub-threshold

region, the linear region and the saturation region. The threshold voltage

(VT) is defined as the gate-source voltage (VGS) at which all immobile trap

states in the OSC channel are filled. Hence, in the sub-threshold region,

when VGS < VT, effectively no mobile charge carriers exist in the OSC channel

and only leakage currents can be measured. When VGS > VT, mobile charge

carriers are induced in the channel and the current through the device is

greatly increased. In the linear regime, when VDS < VGS - VT, the potential in

the entire channel (from source to drain) is above VT, and the drain-source

10

current (IDS) is increasing linearly with increasing VDS, as shown in figure

7(a). When VDS = VGS - VT the channel potential at the drain electrode equals

VT, and this is the pinch-off point, as shown in figure 7(b). When VDS > VGS -

VT, the device operates in the saturation region, as shown in figure 7(c). In

this region the IDS is limited by the amount of charges injected into the

channel from the source electrode and is hence dependent on VGS - VT [26].

The IDS in the linear and saturation regimes can derived by the long-channel

approximation, and the corresponding analytical expressions are [27-29]:

2

2DS

DSTGS

G

DGlinDS,

VVVV

L

WCI

, VGS > VT ; VDS ≤ VGS - VT (2)

2TGS

G

DGsatDS,

2VV

L

WCI

, VGS > VT ; VDS ≥ VGS - VT (3)

Here, CG is the gate capacitance, which depends on the permittivity and

thickness of the dielectric layer, and µ is the charge-carrier mobility. Note

that the mobile charge-carrier in the OFET channel can either be a hole or an

electron, and that the OFET accordingly is termed p-type or n-type. In some

cases, both n-type and p-type operation can be measured on the same OFET

device, and such OFETs are termed ambipolar.

Figure 8. (a) 3D representation of OFET characteristics, with the output

data indicated by black lines and the transfer data indicated by grey lines.

(b) The square root of IDS as a function of VGS, with which it is possible to

extract VT and the mobility, µ, through the relationship in equation (3).

Typical OFET characteristics are shown in the 3D plot in figure 8(a), with the

output data indicated by the black lines and the transfer data indicated by

11

the grey lines. There are a several ways to determine the charge carrier

mobility of the OSC with an OFET, but perhaps the most straightforward

method is to plot the square root of IDS as a function of VGS at saturation

conditions, i.e. at VDS > VGS - VT. By applying a linear fit (solid line in figure

8(b)) it is possible to determine VT as the voltage where the extrapolation of

the linear fit crosses the x-axis (i.e. when IDS = 0). With the aid of equation

(3) and information on the geometrical parameters CG, WD and LG, it is then

possible to also determine the charge carrier mobility [29].

3.1.1. The inverter

The inverter, which is designed to invert an incoming signal, is based on

transistors and is the simplest digital logic element in integrated circuits. It

is therefore used extensively to show that a new transistor technology is

viable for integration into digital integrated circuits. Several inverter designs

exist, and the most suitable for a given transistor technology is set by the

properties of the specific constituent transistors.

Figure 9. Three schematic alternatives to realize an inverter: (a) a resistive

PUN design, (b) a DCFL design and (c) the superior complementary design.

The simplest design is using resistors in a pull-up network (PUN), i.e. the

part of the circuit responsible for pulling the output high (VDD), and an n-

type transistor in a pull-down network (PDN), i.e. the part of the circuit

responsible for pulling the output low (GND, ground), as show in figure 9(a).

A careful selection of RPUN is important in order to achieve fast and robust

operation. It is however notable that this design will always suffer from a

high supply current (IDD) at high input voltage (Vin), and consequently a low

output voltage (Vout), due to the direct current path between VDD and GND

when the n-FETPDN is on. The requirements on RPUN in order to achieve

robust operation will also result in a slow low-to-high Vout switching speed.

12

Another option using only n-type transistors is the direct-coupled FET logic

design (DCFL), where the PUN comprises an FET with the gate and drain

connected2, as shown in figure 9(b). In this design the n-FETPUN will operate

similar to a diode, and a careful design of the WD for both the n-FETPUN and

the n-FETPDN is important to achieve fast and robust operation. Due to the

nonlinear behavior of the PUN in this design it is possible to achieve faster

switching speeds, but the design will still suffer from a direct current path for

IDD at high Vin. Both designs can also, with minor changes, be realized using

p-type FETs [30].

Figure 10. Inverter stability diagram showing Vout (solid line), its inverse

(dotted line), and IDD (dashed line) as a function of Vin. The different

operation regimes for the constituent OFETs are indicated and marked

with the open circles and the letters (a)-(e).

If a transistor technology can integrate both n- and p-type FETs into an

integrated circuit then it is possible to achieve the complementary inverter

design, as shown in the schematic in figure 9(c) [31]. In this design the PUN

comprises a p-type OFET, while the PDN comprises an n-type OFET. This

design is the superior alternative, since only one FET at a time is conducting

2 It is also possible to connect gate and source in the PUN, although this requires a transistor with negative VT

in order to operate at a decent speed.

13

during static operation. This greatly reduces the power consumption of the

inverter circuit, since only leakage currents through FETs operated in the

sub-threshold region will be dissipated during steady-state operation.

The principles of static operation of a balanced complementary inverter (see

figure 9(c)) is presented in the stability diagram in figure 10, which includes

Vout (solid line) and IDD (dashed line) as a function of Vin. The shape of the

curves is dictated by the operation regimes of the constituent OFETs3: (a)

When Vin is low, the p-FET operates in the linear regime, while the n-FET is

in the sub-threshold region, i.e. off. At this point IDD is limited by the low n-

FET sub-threshold current. (b) With increasing Vin, the VGS for the n-FET

will eventually reach and exceed VT. This corresponds to the entering of its

saturation regime, and the point where the channel of the n-FET starts to

become conducting. At this point Vout will begin to decrease and IDD will start

to increase, since both FETs are now entering an on-state operation regime.

(c) By further increasing Vin, the p-FET is also pushed into saturation, thus

further increasing the IDD and allowing a very rapid change in Vout. The slope

of the stability curve, denoted as the gain, reaches a maximum at this point.

This peak value is an interesting metric of the inverter as it describes how

sharp the switching from low to high (and vice versa) is. (d-e) A further

increase in Vin will move the FETs through the reverse behavior, as observed

before the switching threshold.

A functional inverter thus features a high Vout when Vin is low, and vice versa,

as well as a sharp switching threshold which is quantified by the gain, g =

dVout/dVinmax. An indication of the robustness of the inverter is obtained by

plotting the inverse of Vout (dotted line). A large area between Vout and its

inverse is preferable, since this indicates high robustness against noise

caused by e.g. interconnections and device-to-device variations [32].

The description so far have only concerned the steady-state operation of the

inverter. In order to evaluate the dynamic behavior it is important to

consider the charging of capacitances (both intrinsic and parasitic), when

charge carriers move through the device. As a first-order approximation it is

common to sum up all capacitive elements into a fixed single-load

capacitance between the output node and ground (CL), as shown in figure

11(a). The capacitance originating from the overlap between the input and

output electrodes is known as a Miller capacitance and should be counted

twice in CL [32].

3 This description assumes that the n-FET have a positive VT and that the p-FET have a negative VT which is

the most common case for undoped OFETs.

14

Figure 11. (a) Circuit model describing the dynamic behavior of a

complementary inverter and (b) the corresponding dynamic Vout response

following a step change in Vin.

The dynamic operation of the inverter can be pictured as the charging and

discharging of the load capacitance, and the temporal response of Vout to a

step in Vin for a balanced complementary inverter (see figure 10) is plotted in

figure 11(b). First, the inverter input is held at Vin = GND (= low), so that Vout

is high and CL charged. Following a Vin step from GND to VDD (i.e. low-to-

high), the n-FETPDN is forced to open and the p-FETPUN forced to close. This

opens up a path to ground through the n-FETPDN for the charge stored in CL,

and this dissipation will begin to lower Vout. The time it takes for Vout to reach

VDD/2 is defined as the high-to-low propagation delay (tpHL). Eventually all

CL charges will be dissipated, and Vout has now reached the low output. When

Vin is switched back to GND, the n-FETPDN closes and the p-FETPUN opens,

resulting in a charge flow through the open p-FETPUN so that CL becomes

charged and Vout increase. The time for Vout to reach VDD/2 is defined as the

low-to-high propagation delay (tpLH). The total propagation delay (tp) is

defined as the mean of the high-to-low and low-to-high delays, and can be

estimated with the following equation:

PUNDS,

DDL

PDNDS,

DDLpLHpHLp

2

1

2

1

I

VC

I

VCttt (4)

where IDS,PDN and IDS,PUN is measured at |VGS| = |VDS| = VDD for the n-FETPDN

and the p-FETPUN, respectively. It is usually desirable to have inverters with

similar values for tpHL and tpLH, and according to (4) this can be achieved by

balancing the IDS,PDN and IDS,PUN of the constituent OFETs e.g. by varying

their WD [32].

15

The average power consumption (Pav) is another important metric for an

inverter technology and it comprises three contributions:

statdpdynav PPPP (5)

where Pdyn is the dynamic power consumption, Pdp the direct-path power

consumption and Pstat the static power consumption. The Pdyn contribution is

due to the power dissipated through charging and discharging of CL, and is

hence dependent upon the frequency of switching. Pdp originates from the

direct path current which arises if both the n-FETPDN and the p-FETPUN are

on at the same time during a dynamic switching event. This contribution also

depends on the switching frequency. However, this effect can be reduced by

increasing the size of CL [32]. Pstat corresponds to the steady-state leakage of

the inverters. This contribution is significant for organic FETs, due to their

characteristic high sub-threshold currents [33].

3.1.2. The ring oscillator

Figure 12. (a) Schematic showing a 5-stage (N=5) ring oscillator and (b)

typical data based on a 5-stage complementary ring oscillator with Vout

(thick line) and IDD (thin line).

A simple circuit can be realized by connecting an odd number (N) of

inverters in series, with the output of the last inverter connected to the input

of the first, as shown in figure 12(a) [34, 35]. Due to the odd number of

inverters, this circuit will have no stable operation point, and a voltage

transition will propagate through the circuit and looped as fast as the

inverters allow. The voltage at each output node will thus oscillate with a

characteristic frequency. This circuit is hence referred to as a ring oscillator.

Due to its simplicity, the ring oscillator is commonly used to evaluate the

performance and functionality of the inverter technology. A successful ring

16

oscillator operation requires that each inverter is able to switch and drive the

subsequent inverter in the chain. With a successful ring oscillator at hand,

the realization of larger logic circuits is often simply a matter of circuit

design [36-39].

A typical measurement on a complementary ring oscillator is plotted in

figure 12(b), and it presents Vout from one of the output nodes (thick line)

and the supply current (IDD) consumed by the oscillator (thin line) as a

function of time. The time period (T) for a full oscillation of Vout can be

estimated as N × 2tp. The frequency of the oscillation (f) is the inverse of T. It

is also possible to determine Pav for an inverter in the ring oscillator by

integrating IDD over one time period:

T

dttINT

VP

0

DDDD

av (6)

By multiplying Pav with tp the power delay product (PDP) is obtained. The

PDP represents a measure of the average energy consumed per switching

event. By subsequently multiplying the PDP with tp the energy delay product

(EDP) is obtained, and the EDP thus represents a combined energy and

performance metric. These metrics allow for a comparison of different

inverter technologies, as they describe the performance and efficiency of

each technology in a quantitative manner [32, 40].

3.2. The photovoltaic cell

Our modern lifestyle consumes an increasing amount of energy each year.

Historically, this energy has been almost exclusively supplied through fossil

fuels (coal, oil and gas), a non-renewable energy resource. Along with the

problem of exhausting earth's fossil energy resources comes the release of

the greenhouse gases (CO2, CH4, etc.), the inevitable product from fossil fuel

combustion which, once in the atmosphere, contributes to global warming.

Renewable energy alternatives are thus necessary in order to alleviate these

problems. One such alternative is the photovoltaic (PV) cell in which solar

energy is converted to electrical energy. The current generation of

commercial PV cells comprises inorganic Si, and such cells can feature a high

efficiency thanks to the maturity of the Si technology. The refining and

processing of the Si is however an expensive process, both in terms of price

and energy.

In 1992, an ultra-fast photo-induced electron transfer from a conjugated

polymer to C60 fullerene molecule was demonstrated [2]. This discovery of

17

efficient charge separation in organic materials started the development of a

new generation of light-energy-harvesting devices, organic photovoltaics

(OPVs), which offer a low-cost and easy-to-fabricate alternative to the

existing Si-based PVs [3].

3.2.1. The bilayer structure and operation principles

Figure 13. The (a) device structure and (b) energy diagram for a bi-layer

organic solar cell.

In order to understand the operation principles of an OPV cell it is

educational to start with the simple bi-layer structure. This structure is

depicted in figure 13(a) and comprises two organic semiconductor layers

sandwiched between two electrodes, of which at least one needs to be

transparent. Transparent conductive oxides (TCO), such as indium tin oxide

(ITO), is usually used as the transparent electrode, although significant

efforts are put into finding replacements based on more abundant, low-cost

and flexible materials [41-43].

A thin film of a donor (D) material is first deposited onto the TCO. The role

of the donor is to absorb the incoming photons to excite electrons from the

HOMO into the LUMO to create excitons, viz. bound electron and hole pairs.

An effective donor material should thus feature a high absorption coefficient

(~105 cm-1) to allow for complete light absorption for a film with a thickness

on the order of 100 nm. Since the exciton is a neutral particle it will diffuse

randomly in the donor, independent on the electric field, until it either

recombines or dissociates into a free electron and hole. Due to the low

permittivity of organic materials, the binding energy of the exciton (Eexciton >

18

100 meV) is much larger than the average phonon energy at room

temperature (~25 meV), implying that the exciton will recombine unless a

driving force will cause it to dissociate [44].

Figure 14. Schematic highlighting the desired (green solid arrows) and

non-desired (black dashed arrows) processes related to (a) photon

absorption, (b) exciton dissociation, (c) charge transport and (d) charge

collection in an OPV at short-circuit condition.

19

A driving force for exciton dissociation can however be an interface with an

acceptor (A) material with lower ELUMO and EHOMO than the donor, so that

the electron, but not the hole, finds a more favorable state in the acceptor.

The bound exciton can thus dissociate into a free electron located in the

acceptor LUMO and a free hole located in the donor HOMO at the D/A

interface. Once separated, the charges are free to move in an external field.

By selecting a top electrode material with a lower work function than the

TCO, it is possible to create an internal electric field which will act as a

driving force on the electrons and holes towards the top cathode and the

bottom TCO anode, respectively. The successful collection of the charge

carriers at the respective electrode results in a PV current. A schematic

energy diagram for a functional bi-layer solar cell is shown in figure 13(b).

In order to maximize the performance of an OPV it is important to

understand and minimize the loss factors. The efficiency with which the OPV

converts incoming photons into current exiting the device is given by the

equation for the external quantum efficiency (EQE):

ccctedabsabs IQEEQE (7)

The internal quantum efficiency (IQE) refers to the efficiency with which the

absorbed photons, i.e. those photons that are not reflected or transmitted out

of the cell, can generate current. ηabs describes the efficiency of absorption,

and this process is presented in figure 14(a), where the green solid arrow

indicates the desired events (in this case photon absorption to create an

exciton) and the black dashed arrow indicates the non-desired event (a non-

absorbed photon reflected from or transmitted through the device). ηed

represents the exciton dissociation efficiency, as schematically shown in

figure 14(b). Once created through absorption, the exciton will diffuse in the

organic semiconductor until it is either separated into free charge carriers at

an interface (green solid arrow) or recombined under the generation of heat

or light (black dashed arrow).

ηct describes the charge transport efficiency, as depicted schematically in

figure 14(c). Once separated, the free charge carriers need to be transported

to the correct electrodes in order to be collected (green solid arrows). This

process requires a driving force from an electric field or a charge

concentration gradient, as well as uninterrupted pathways for the charge

carriers. The free electrons and holes can, however, also recombine at a D/A

interface before reaching the electrodes (black dashed arrow). ηcc is the

charge collection efficiency at the electrodes, as described in figure 14(d),

where the non-preferred process is charge collection at the wrong electrode.

20

For bi-layer devices, the good news is that ηabs is efficient at a film thickness

of ~100 nm, ηct can be high due to the good mobility in the pure donor and

acceptor materials, and ηcc is close to unity since the electrodes are only

contacting materials carrying a single charge-carrier type. ηed is, however,

greatly limited by the short diffusion length (~10-20 nm) of excitons in OSCs

[45]. As a result, only photons absorbed within an exciton diffusion length

away from a D/A interface will contribute to the total EQE, thus greatly

limiting the total current produced in a bi-layer device.

3.2.2. The bulk heterojunction solar cell

Figure 15. (a) A schematic of the BHJ-OPV device structure, with ETL and

HTL layers, and (b) the corresponding energy diagram.

The bulk heterojunction (BHJ) OPV cell is designed to alleviate the problem

of inefficient exciton dissociation. In a BHJ device the donor and acceptor

are intimately mixed in a single active-layer film. In such a film, the donor

and acceptor materials are able to form interfaces on length scales of only a

few nm, i.e. well below the exciton diffusion length, which allows for a high

ηed. However, with a randomly intermixed material network come charge

transport and collection problems. Since both donor and acceptor materials

are in contact with the electrodes in a BHJ device, it is important to ensure

that only the correct type of charge carrier is collected at the electrodes.

Therefore, a hole-transport layer (HTL), designed to block electrons while

efficiently transporting holes, and an electron-transport layer (ETL),

designed to block holes while transporting electrons, can be deposited in

21

between the active layer and the anode and the cathode, respectively. The

inclusion of a HTL and an ETL minimizes charge collection losses, i.e. keeps

ηcc close to unity. A typical device structure for a BHJ-OPV cell is shown in

figure 15(a) and the accompanying energy diagram is shown in figure 15(b).

Maximizing the charge transport efficiency ηct in a BHJ cell is a delicate

matter. In an intimately mixed network the interface surface area is greatly

increased, resulting in increased exciton dissociation but also higher

recombination probability [46, 47]. The D/A random network (as shown in

figure 15(a)) formed during thin-film deposition can also comprise one-

phase material islands, resulting in interrupted pathways for charge

transport. This random D/A network is however not permanent and it can be

adjusted after deposition in order to optimize the film morphology. Methods

for optimizing the film morphology include thermal annealing [48-50],

solvent treatment [51] and light exposure [52, 53]. Still, the overall aim is the

same, i.e. controlling the phase separation to reduce charge recombination

while maintaining efficient exciton dissociation.

3.2.3. The OPV performance

Figure 16. (a) Energy diagram highlighting parameters vital to the OPV

cell performance. (b) The AM1.5G irradiance spectrum showing the

spectral irradiance as a function of wavelength and energy. The total

irradiance is 1 kW m-2.

Various material and device parameters determine the maximum

performance of an OPV cell, with the energy levels of the D/A materials

playing an important role. The maximum voltage is determined by the

difference in energy between the acceptor LUMO and donor HOMO, Emax =

22

ELUMO,A - EHOMO,D, as this represents the energy difference between the

separated electrons and holes, as seen in figure 16(a). To maximize the

voltage, the EHOMO,D is increased and/or the ELUMO,A is decreased. In order to

allow for efficient exciton dissociation, and therefore a high current density,

the driving force for exciton dissociation, ΔE = ELUMO,D - ELUMO,A, must be

larger than Eexciton. Moreover, a lower donor energy gap (Eg = ELUMO,D -

EHOMO,D) results in the absorption of a larger portion of the solar spectrum,

as shown in figure 16(b). It is thus clear that a tradeoff exists between the

maximum voltage and the maximum current density from an OPV [54].

Figure 17. Typical data obtained from a J-V measurement on an OPV cell

with the dark current indicated by the black line and the AM1.5G

illuminated indicated in red. The Jsc, Voc and MPP are marked as a

reference.

In order to generate power, the device needs to be operated in the forward

bias regime. In this regime, the solar-induced current will compete with the

dark current (as measured on a non-illuminated dark OPV and indicated by

the black line in figure 17), which is due to charge injection in the diode

structure created by the asymmetric electrodes. A typical current density –

voltage (J-V) measurement performed on an AM1.5G illuminated organic

solar cell is shown in figure 17 (red curve). This current density represents

the net of the generated and the dark currents.

The point where the generated and dark currents balance, and no net current

is produced by the device, is termed the open-circuit voltage (Voc).

Consequently, when the voltage measured over the device is zero, no net

current is injected from the electrodes. The current density measured at this

23

point is termed the short-circuit current density (Jsc). The total generated

power is the product of the current and voltage; hence no power is generated

by the device at Jsc or Voc. Instead, the maximum power point (MPP) defines

the point where the product between current and voltage is maximized, as

visualized by the green rectangle with dashed borders in figure 17. In an ideal

device, the MPP would equal the product of Jsc×Voc (red square with solid

borders in figure 17) but because of, e.g., the dark current the measured MPP

is reduced. The ratio between the MPP and the ideal (Jsc×Voc) is defined as

the fill factor (FF).

In the end, the most important metric for a PV cell is its total power

conversion efficiency (PCE), as calculated by

in

ocsc

P

FFVJPCE

(8)

where Pin is the total power density of the of the light impinging upon the

surface of the PV cell, which commonly is 1 kW m-2 in experiments [45]. A lot

of effort has been invested on optimizing materials and processes in order to

maximize the PCE of BHJ-OPV cells [55]. Record values for the PCE are

however almost exclusively measured during a short time period following

device fabrication, but a practical device should maintain this level of

performance for an extended period of light illumination. This requirement

is not simple for a BHJ-OPV cell, since the temperature increase during

continuous operation can drive the phase separation beyond its optimum,

and in the process reduces the device efficiency. A fabrication process aimed

at locking the film morphology at its optimum is thus essential for long-term

continuous operation. Several methods have been suggested with varying

success and chapter 6.2 will present a fully solution-processable alternative

for solving this problem.

24

4. Fullerene photochemical transformation

Figure 18. The photochemical dimerization of PCBM.

The fullerene C60 has been demonstrated to photochemically transform into

dimers, oligomers and polymers during exposure to light [18, 56]. It has also

been shown that the C60 derivative PCBM can be dimerized during exposure

to green laser [57] and UV [58] light. The two fullerene monomers are

connected to a dimer in a [2+2] cycloaddition dimerization reaction, as

depicted in figure 18.

4.1. Raman spectroscopy

Figure 19. Raman spectra showing the Ag(2) vibrational mode for PCBM

films. While the pristine film (a) indicates a fully monomeric film, both the

green laser exposed (b) and UV exposed (c) films shows a clear downshift,

indicating the presence of PCBM dimers.

25

The presence of dimerized PCBM in a film can be established with Raman

vibrational spectroscopy. In this technique, the sample is exposed to a

monochromatic excitation light, which is scattered by the molecules in the

sample. A small portion of the excitation light will interact with the

vibrational modes (phonons) of the molecule, resulting in an inelastic

scattering process where the scattered photon is shifted either up or down in

energy by an amount corresponding to the phonon energy. So by measuring

the scattered light spectrum it is possible to obtain information about the

vibrational modes of a molecule.

The Ag(2) pentagonal pinch mode gives valuable information about the

degree of polymerization of fullerenes. It is located at 1469 cm-1 for monomer

C60, but is down-shifted by ~5 cm-1 for each chemical connection to the C60

cage. For the C60 molecule, one downshift thus indicates the presence of

dimers, two downshifts indicate linear oligomers/polymers, and three

downshifts indicate a branched polymeric structure [18, 59-62].

Since a solubilizing side-group is chemically attached to the C60 cage in

PCBM, the Ag(2) mode of the PCBM monomer is located at 1465 cm-1, i.e.

downshifted by 4 cm-1 in comparison to the C60 monomer [57]. The Raman

spectrum for a 100 nm thick pristine PCBM film is shown in figure 19(a), as

obtained using a Raman excitation wavelength of λRaman = 780 nm at a power

density of PRaman < 10 mW cm-2.

The photochemical transformation of the PCBM film was performed by

exposure to either a green laser (λ = 532 nm) or a hand-held UV lamp (λ =

365 nm), and the corresponding Raman spectra are presented in figure 19(b)

and figure 19(c). Both spectra show a clear downshift of the Ag(2) mode by 5

cm-1 as highlighted by the two dotted vertical lines, which indicate the

formation of PCBM dimers in both films. Interestingly, no further degree of

polymerization is observed in the exposed films. The lack of a higher degree

of polymerization of PCBM is motivated by the steric hindrance of the PCBM

side-group [57].

4.2. Fullerene patterning by resist-free lithography

An exciting and useful property of the photochemically transformed PCBM

dimers is that their solubility in organic solvents, as compared to the PCBM

monomers, is drastically reduced. Thus, by selectively exposing certain

regions of a pristine PCBM film, it is possible to create lateral regions of

lowered solubility which can be developed by immersion into a tuned

solution [57, 58, 63-65]. This process is known as photolithography.

26

Figure 20. Schematic representation of the key steps in the resist-free

photolithographic patterning of PCBM: (a) film deposition, (b) light

exposure through a shadow mask, (c) development in a tuned developer

solution, and (d) the patterned PCBM dimer film. The optical micrographs

in (e-g) display (dark brown) patterned PCBM, in the form of lines with

different line widths and a square 2D structure.

The resist-free photolithographic patterning of a PCBM film is schematically

depicted in figures 20(a-d). The process begins with the deposition of a

PCBM film through spin-coating, as shown in figure 20(a). The as-cast film

is then exposed to light through a shadow-mask, featuring the pattern to be

transferred as the open areas, as shown in figure 20(b). The exposure will

transfer the pattern of the shadow mask into PCBM dimer regions with

reduced solubility. By immersing the exposed film into a carefully tuned

developer solution (e.g. chloroform:acetone in a 1:7 volume ratio), as shown

in figure 20(c), the monomer PCBM is washed away, leaving only the

patterned dimers on the substrate, as shown in figure 20(d).

Optical micrographs of patterned (dark brown) PCBM films on SiO2

substrates are shown in figures 20(e-g). The line width of the patterned

PCBM lines decreases from 100 µm to 10 µm, in steps of 10 µm, in figure

20(e), and from 10 µm to 1 µm, in steps of 1 µm, in figure 20(f). Figure 20(g)

shows a 22.5 µm square pattern. The successful patterning of the smallest 1

µm sized lines in figure 20(f), and the sharp corners in figure 20(g),

demonstrate the method’s capacity for high resolution. Moreover, the

27

measured line-edge roughness of 56 nm, in combination with that the

smallest feature on the shadow-mask was 1 µm, indicates that an even

smaller feature size of <1 µm should be achievable.

4.3. Patterning efficiency

Figure 21. (a) Optical micrograph of a patterned PCBM film, with only the

left part of the film being exposed. (b) An AFM height scan was performed

in a direction perpendicular to the patterned interface, as indicated by the

dotted line in (b). (c) The UV-vis absorption spectrum of a pristine PCBM

film. (d) The normalized retained PCBM film thickness as a function of

exposure dose, as extracted from AFM measurements. A mercury lamp

(solid squares) and a green laser (open circles) were used for the exposure

in (d).

In order to evaluate the photochemical dimerization efficiency for PCBM, a

set of pristine PCBM films were exposed to light in varying doses. Half of the

sample was masked during the exposure, resulting in a step profile following

the development step. The normalized retained thickness of the developed

28

film was measured by atomic force microscopy (AFM), as depicted in figure

21(a) and figure 21(b). Since the developer solution is assumed to selectively

remove the PCBM monomers, the normalized retained thickness can be

viewed as an estimate of the dimerization fraction in the exposed and

developed film.

Due to the wide absorption of PCBM, as shown in the UV-vis absorption

spectrum in figure 21(c), a wide range of wavelengths are able to

photochemically transform PCBM monomers. From a resolution

perspective, it is preferable with a short wavelength since the minimum

feature size in photolithography is dependent on the wavelength of the

exposure light. Additionally, PCBM shows a significantly higher absorption

coefficient in the UV region than in the visible region, with the anticipated

consequence being that the dimerization process should be more efficient

with UV-light exposure. A hand held mercury UV-light source (λ = 365 nm, P

= 8.4 mW cm-2) and a longer wavelength green laser (λ = 532 nm, P = 61.3

mW cm-2) were chosen to investigate the effect of the wavelength on the

process efficiency.

The normalized retained PCBM film thickness as a function of exposure dose

and wavelength is shown in figure 21(d), with the exposure dose being

controlled by the exposure time. We quantify the process efficiency with the

dose required for 50% retained thickness. It is clear that the UV exposure

(solid squares) requires less energy to fully dimerize the PCBM film than the

green laser (open circles), as quantified by that the 50% dose for the former

process is 3.57 J cm-2 and 26.1 J cm-2 for the latter process.

Some assumptions are necessary in order to transform the measured dose

into a more conceptually accessible “dimerization events per absorbed

photon” as represented by the symbol Φ. First, the intensity of the exposure

light is uniform throughout the film thickness, and no reflections at the film

or substrate surface are present, implying that the incoming light passes

through the film once. Second, the absorption coefficient remains constant

(equal to that of pristine PCBM) throughout the entire exposure step. Third,

a successful dimerization event requires that two neighboring PCBM

monomers are simultaneously excited by one photon each. Based on these

assumptions it is possible to formulate an expression for Φ based on a

modified version of Beer-Lambert's law:

29

dehc

E

M

VN

N

NΦ

1tot

A

abs

dim (9)

where Ndim is the minimum number of photons necessary to fully dimerize

the PCBM film, i.e. assuming one absorbed photon per PCBM monomer. It is

given by the density of PCBM (ρ = 1.5 g cm-3), the volume of the dimerized

PCBM film (V), Avogadro's constant (NA) and the molar weight of PCBM

(M). Nabs represents the measured number of absorbed photons for full

dimerization. It is given by the total energy of the incoming light (Etot), the

wavelength of the exposure light (λ), Planck's constant (h), the speed of light

in vacuum (c), the absorption coefficient of PCBM at the exposure

wavelength (α), and the PCBM film thickness (d).

By including the relevant numbers into equation (9), notably Nabs,365 =

2.4×1019 cm-2 and Nabs,532 = 1.3×1020 cm-2, we find that the photon-to-

dimerization efficiency is: Φ365 = 4.2×10-4 and Φ532 = 7.4×10-5. The fact that

Φ365 > Φ532 highlights the advantage of UV light over visible light as it both

results in a more efficient dimerization process and a higher resolution.

4.4. The photochemical reaction mechanism

Figure 22. (a) Jablonski diagram showing the PCBM energy levels and the

proposed reaction pathways for PCBM dimerization. (b) Normalized

residual film thickness of exposed and developed PCBM films, as a function

of the exposure dose and intensity. The exposure was effectuated by a high-

30

power UV-diode (λpeak = 365 nm) using different intensities and exposure

times.

Understanding the underlying mechanism of the PCBM photo-dimerization

reaction is crucial in order to optimize the exposure conditions. The energy

levels of PCBM and the possible reaction pathways during electronic

excitation, relaxation and dimer formation are presented in the Jablonski

diagram in figure 22(a). The energy levels are the singlet ground state (M),

the excited singlet state (1M*), the excited triplet state (3M*), and the PCBM

dimer state (D) [18, 66], and the transitions between the different states are

marked with arrows and associated rate constants, k.

The PCBM fluorescence (1M* → M) is in comparison to the intersystem

crossing to the triplet state (1M* → 3M*) a very slow process. It is thus

reasonable to state that k2 is negligible [18, 66]. This assumption has the

consequence that all excited electrons end up in the 3M* state, and this entire

transition can be described by the combined rate constant k13. C60 dimers

begin to dissociate into monomers at a high temperature of 400 K [18, 62],

and we therefore assume that the PCBM dimer dissociation rate k6 can be

neglected during an exposure step executed at room temperature. Once a

PCBM monomer is excited to the 3M* state, it can either relax to the ground

state (3M* → M) with a rate of k4, or form a dimer. The latter dimer-

formation process is described by two alternative pathways, the uni-excited

reaction and the bi-excited reaction, as described by the rate constants k5

and k’5, respectively. In the uni-excited reaction, a dimer is formed by the

combination of one fullerene monomer in the triplet excited state and

another neighboring fullerene monomer in the singlet ground state, i.e. 3M*

+ M → D. Whereas in the bi-excited reaction, two neighboring monomers

need to be simultaneously excited in order to form a dimer: 3M* + 3M* → D.

A Raman spectroscopy study by Eklund et al. came to the conclusion that the

uni-excited reaction described the dimerization rate of C60 fullerenes [18]. It

is however here contradicted by the data in figure 22(b), which shows that

higher-intensity exposure light results in a lower exposure dose for the

dimerization of a PCBM film. The uni-excited reaction pathway, which only

requires that one monomer is excited in order to form a dimer, will show an

intensity-independent dose, and it can therefore not be the accurate

description of the behavior seen in figure 22(b). The bi-excited reaction

pathway, which requires two neighboring monomers to be simultaneously

excited, will feature the observed intensity-dependence of the dose. It is

therefore concluded that the bi-excited reaction pathway describes the

photochemical dimerization process of PCBM.

31

The detailed understanding of the reaction processes for PCBM dimerization

results in the practical conclusion that high-intensity UV-light is an ideal

light source for the exposure step. The short wavelength will ensure high

resolution, strong absorption and low dose. For example, short wavelength

UV-light, with a wavelength of 365 nm and an intensity of 8000 mW cm-2,

will result in an exposure time of 1 mere second. This short exposure time

fits well with large-scale processes, such as a roll-to-roll (R2R) fabrication.

4.5. PCBM OFETs

Figure 23. Transfer (left) and output (right) data for typical PCBM OFETs,

as measured at different instants during the patterning process. The

transfer data were measured at VDS = 60 V. The output data were

measured in VGS steps of 20 V, starting from 0 V (lower trace) and ending

at 80 V (upper trace).

32

The impact of UV-patterning on the electronic transport properties of PCBM

was investigated with BG-TC OFETs, comprising a PCBM film as the active

material, Au as the source and drain electrodes, and SiO2/p:Si as the gate-

oxide/gate; the experimental details are presented in Paper I. The transfer

and output characteristics are presented in the left and right part of figure

23, respectively, as measured on: (i) pristine PCBM, (ii) UV-exposed PCBM

(λ = 365 nm, P = 8.4 mW cm-2, t = 15 min), (iii) UV-exposed and developed

PCBM, and (iv) unexposed and developed PCBM.

Treatment µn

(cm2 V-1 s-1) [a]

VT

(V)

Ion/Ioff

(a.u.)

Pristine 1.8(±0.6)×10-2 2 2.0×104

Exposed 1.6(±0.5)×10-2 7 1.2×104

Exposed + Developed 1.2(±0.2)×10-2 10 1.2×104

Solely Developed --- --- ---

[a] Mean value with standard deviation in parenthesis

Table 1. Performance metrics of PCBM transistors as measured during the

UV-patterning procedure.

The transistor data demonstrate well-behaved n-type OFETs for (i)-(iii),

with monotonously increasing IDS with increasing VGS in the transfer

characteristics, and clear saturation in the output characteristics. As

expected, the solely developed sample shows no OFET characteristics,

simply because the unexposed PCBM were dissolved and removed by the

developer solution. Both µn and VT were extracted from the transfer data, as

described in section 3.1, and the Ion/Ioff ratio was calculated as the ratio of IDS

at (VDS = VGS = 60 V) and (VDS = 60 V, VGS = 0 V). The average values from

>60 characterized OFETs are summarized in table 1. To our satisfaction,

PCBM features a respectable mobility following patterning, although it is

slightly reduced in comparison to the pristine state. We tentatively attribute

the decrease to a reduction of intermolecular π-orbital overlap following the

transformation of some sp2 orbitals to sp3 orbitals; see figure 18.

Interestingly, the dimerization also makes PCBM less sensitive to ambient

air. By exposing the OFETs to ambient air for 24 h prior to the FET

characterization, we found that the mobility of pristine PCBM dropped to a

mere 1 % of its initial value, whereas the UV-patterned PCBM retains 39 % of

its initial mobility. We speculate that this improvement is due to a more

33

dense and possibly more crystalline film, which could act as an improved

barrier towards O2 permeation into the active layer [58, 67].

Most importantly, the electronic mobility of PCBM (and C60) is only slightly

reduced by the photochemical transformation [57, 63]. This fact indicates

that high-resolution and resist-free patterning of the fullerene-based active

material in organic electronic devices could be possible. In OFETs, this

advantage could be used to reduce inter-device and fringe-field leakage

currents, and to realize advanced circuits such as memory devices.

4.6. Realizing complementary logic circuits with resist-free

patterning

The resist-free patterning technique can also be utilized to realize

complementary circuits with a single-layer active material, comprising a

blend of n-type PCBM and p-type poly(3-hexylthiophene-2,5-diyl) (P3HT).

By selecting an active material with high PCBM content, a predominantly n-

type device is obtained. Interestingly, it is possible to transform this device

into featuring p-type behavior, by immersing the device into a developer

solution tuned to selectively dissolve (non-exposed) PCBM monomers, but

not (exposed) PCBM dimers and P3HT. By using this single-layer blend film

as the common active material for an array of transistors, exposing some of

the transistors to high-intensity light, and then developing the entire array

an interesting and attractive result is obtained. The exposed and developed

transistors remain n-type, as the active material is left effectively intact by

the developer solution, while the solely developed transistors have

transformed into being p-type, following the removal of the (non-exposed)

PCBM monomers [64].

34

4.6.1. The inverter

Figure 24. (a) Photograph and (b) schematic diagram of the PCBM:P3HT

complementary inverter fabricated with the resist-free patterning

technique. The output data of (c) the p-type FET and (d) the n-type FET,

with VGS changed in steps of 10 V starting from 0 V and ending at 50 V

and VDS swept at a rate of 2 V s-1, as indicated by the arrows. (e) The static

and (f) the dynamic inverter characteristics, as recorded with VDD = 50 V

and VDD = 44 V, respectively.

35

A schematic of the complementary inverter circuit, comprising a p-type and

an n-type FET, is depicted in figure 24(b). The FETs were fabricated in a BG-

TC geometry, comprising a common active material of PCBM:P3HT in a

mass ratio = 5:1, Au as the source and drain electrodes, and SiO2/p:Si as the

gate-oxide/gate. The active material in the upper and lower FETs was then

tuned to be p-type and n-type, respectively, following the procedure outlined

in the previous section. More specifically, a green laser (λ = 532 nm, P = 60

mW cm-2, t = 30 min) selectively exposed the active material in the FET

designed to be n-type, where after the entire active material film was

developed in the developer solution. Figure 24(a) shows a photograph of two

inverters, with the exposed dimer-PCBM:P3HT and the unexposed P3HT

regions distinguished by their dark blue-green and red-brown appearances,

respectively.

A typical output device performance of the p-type and n-type OFETs are

shown in figure 24(c) and figure 24(d), respectively. Both devices show well-

behaved operation with clear current saturation and low hysteresis. A slight

ambipolarity of the n-type FET is observed as a slight increase in IDS at high

VDS as VGS approaches 0 V. The µ and VT extracted from the transfer

measurements on 5 pair of transistors are shown in table 2.

Device µ

(cm2 V-1 s-1) [a]

VT

(V) [a]

p-FETPUN 1.3(±0.4)×10-4 -14.8(±2.5)

a-FETPDN 1.7(±0.8)×10-4 19.4(±1.7)

[a] Mean value with standard deviation in parenthesis

Table 2. Performance metrics of the two OFETs forming the inverter.

Figure 24(e) shows the static inverter characteristics, as recorded at VDD = 50

V. The inverter features a sharp switch of Vout (blue solid line) at Vin ≈ 30 V,

as quantified by a high gain of g = 11 ±1. The IDD (red dotted line) shows a

peak at the switching threshold, as expected, a low leakage current of <1 nA

at Vin > 35 V, and a larger leakage current for Vin < 25 V. The latter is an

effect of the slightly ambipolar character of the n-type FET, which results in

a direct current path from VDD to GND.

Figure 24(f) shows the dynamic inverter behavior, as measured at VDD = 44

V. The high-to-low transition (tpHL = 2.9 s) is delayed by an overshoot of Vout,

as it is pulled towards GND by the n-type FET. The low-to-high transition

(tpLH = 0.6 s), in contrast, shows an ideal behavior as Vout is pulled towards

36

VDD by the p-type FET. A total propagation delay of 1.75 s is measured at a

VDD of 44 V.

4.6.2. The ring oscillator

Figure 25. (a) Schematic of a 5-stage ring oscillator realized with the

complementary PCBM:P3HT inverters, and (b) the corresponding

temporal behavior of Vout and IDD at VDD = 50 V. (c) the delay time and the

average power consumption as a function of supply voltage, and (d) the

PDP and EDP as a function of the supply voltage.

A 5-stage ring oscillator was fabricated by connecting 5 inverters in series,

with the output of the last inverter connected to input of the first, as depicted

schematically in figure 25(a). When the supply voltage was connected, the

circuit began to oscillate, and temporal data on Vout and IDD at VDD = 50 V are

shown in figure 25(b). The frequency of operation is 10 mHz, and the circuit

consumes an average IDD ≈ 15 nA. Interestingly, the measured Vout only

oscillates between 25-50 V and thus never operates in the ambipolar region.

37

Figure 25(c) shows that the delay time, tp, decreases and the average power

consumption, Pav, increases with increasing VDD, as expected, for the 5-stage

ring oscillator. The superlinear increase of Pav with increasing VDD indicates

that the power consumption can become an issue at high VDD.

The dependence of the power delay product, PDP, and energy delay product,

EDP, on VDD are shown in figure 25(d). A PDP minimum of 2.0 µJ is found at

VDD = 46 V, indicating that this is the most efficient VDD for operation.

However, no minimum is found for EDP in the measured interval, since the

reduction of tp is greater than the increase of Pav as VDD increases. This

indicates that the supply voltage for optimum operation will be found at VDD

> 50 V.

In summary, it has been shown that functional complementary inverters and

ring oscillators can be fabricated with a facile photolithographic procedure

developed in-house, without the use of a sacrificial photo-resist material. By

considering the high values for the intrinsic capacitance (Ci = 14.7 pF) and

the extrinsic capacitance (CL = 1.44 ±0.15 nF) for the inverters in this

particular circuit, it is clear that further improvements in terms of speed and

power consumption by several orders of magnitude can be achieved by

reducing these parasitic capacitances.

38

5. Fullerene C60 nanorods

For facile electron transport, a crystalline material is often preferred over its

amorphous counterpart [14-16, 68]. Self-assembly is a common method for

the fabrication of crystalline organic materials with controlled size and

shape, and crystalline C60 structures ranging from 0-D dots, through 1-D

rods, to 2-D discs have been realized using such methods [17, 69-73]. A

notably straightforward solution-based self-assembly fabrication method is

the liquid-liquid interfacial precipitation (LLIP) method, which allows

synthesis of crystalline 1-D C60 nanorods (NRs) with sub-µm diameter [74-

80].

5.1. Synthesis

The LLIP synthesis of C60 NRs is a 3-step process: i) interface formation, ii)

nucleation, and iii) crystal growth. An example of a fabrication process is

detailed below: C60 is dissolved in a good solvent (1 mg ml-1 in m-

dichlorobenzene, m-DCB) and 1 ml of the solution is added to a glass vial. 10

ml of a poor C60 solvent (ethanol) is gently added to the vial, so that an

interface is formed between the good and poor solvents; see left vial in figure

26(a). The two-component solution is ultrasonicated to induce partial

mixing of the two solvents at the interface. The mixing results in local

oversaturation of C60, so that nucleation and precipitation of C60 seed

particles can take place, as manifested in the greenish interfacial region in

the middle vial in figure 26(a). The hermetically closed vial is then left at

room temperature, and after 2 days C60 NRs are observed to begin to deposit

at the bottom of the vial; see right vial in figure 26(a). After 3 days the crystal

growth is terminated by vigorously shaking the vial and the resulting as-

grown NRs are ready for use.

An optical micrograph and a transmission electron microscope (TEM) image

of a batch of as-grown NRs are shown in figure 26(b) and figure 26(c),

respectively. The NRs feature a hexagonal cross section, lengths of ~1 mm,

an aspect ratio of >103, and they are highly flexible [81]. The size of the NRs

is controlled by the number of crystal seeds formed during the nucleation

step. This number can be controlled by the solvent intermixing during the

ultrasonication, and an increased ultrasonication power results in enlarged

intermixing and an increased number of crystal seeds, which in turn results

in smaller NRs [82]. We have utilized two different ultrasonication powers

for the fabrication of two types of NRs with: (i) a mean diameter of Ø1 = 250

nm, within a range of 100-400 nm (Paper III); (ii) Ø2 = 172 nm, within a

39

range of 50-360 nm (Paper IV). The size distribution of the latter is

presented in figure 26(d).

Figure 26. (a) Photograph of vials comprising solutions in the different

stages of the LLIP process: (left) interface formation, (middle) nucleation,

and (right) crystal growth. (b) Optical micrograph and (c) TEM image of

the as-grown NRs. (d) Diameter distribution of NRs with an average

diameter (Ø2) of 172 nm.

40

5.1.1. Polymerization

Figure 27. Schematic representation of the photochemical transformation

of NRs into PNRs. The photograph in the inset shows vials containing

dispersions of as-grown NRs (left) and PNRs (right), depicting the color

shift from green-brown to red-brown following the exposure.

Exposing the as-grown NRs to high-intensity light transforms the

monomeric C60 in the as-grown NRs into polymeric C60. The exposure was

performed by directing a laser beam (λ = 532 nm, P = 15 mW mm-2, t = 20 h)

into the NR dispersion through the bottom of the vial, as depicted in figure

27. The width of the laser beam was adjusted to match the width of the vial

(~1 cm), and the sides of the vial were covered in Al foil to trap the light

inside the vial. The exposed NRs featured a color shift from green-brown to

red-brown, as seen in the photograph inset in figure 27, while the NR

diameter distribution was found to be unaffected. The photochemically

transformed NRs will hereafter be referred to as polymerized nanorods

(PNRs). It should be noted that the exposure was performed on NRs in

dispersion, making it a scalable process.

41

5.2. Characterization

The NR structure was characterized by TEM, high-resolution TEM

(HRTEM), scanning TEM (STEM), Raman spectroscopy, x-ray diffraction

(XRD), and selective area electron diffraction (SAED), while the thermal

characterization was performed using thermal gravimetric analysis (TGA).

5.2.1. As-grown

Figure 28. Characterization of as-grown NRs with (a) XRD and (b)

HRTEM, with the (top left) inset showing a SAED image.

Figure 28(a) shows the XRD of as-grown NRs, with the peak pattern

indicating a hexagonal close-packed (hcp) structure, with a unit cell size of a

= b = 23.71 Å and c = 10.18 Å. The sharp distinct peaks indicate a single-

crystal structure, which is supported by both the HRTEM image in figure

28(b), in which both crystalline faults and dislocations are absent, and the

SAED pattern in the inset in figure 28(b). It is thus concluded that the as-

grown NRs feature a single-crystalline hcp crystal structure.

TGA of as-grown NRs reveals a linear weight loss of 10.7 % between 85-171

°C, as shown in figure 29(a). This weight loss is assigned to the evaporation

of guest m-DCB molecules from within the NRs. The weight loss with an

onset of ~400 °C is due to the sublimation of C60, which features a

sublimation temperature of 390 °C in the solid state.

The Raman spectra of as-grown NRs and pristine C60 are displayed in the

upper red curve and lower blue curve in figure 29(b), respectively. The as-

grown spectrum features 10 characteristic C60 peaks (8 Hg and 2 Ag) and 3

additional peaks, as marked with asterisks, and these additional peaks can be

42

removed by annealing at 200 °C. Furthermore, the Hg(1) and Ag(1) modes of

the as-grown NRs are downshifted in comparison to pristine C60 (see inset

in figure 29(b)). The new and shifted Raman features of as-grown NRs are

assigned to the existence of solvated m-DCB guest molecules in the NR

lattice [72, 83, 84].

Figure 29. (a) TGA of as-grown NRs. (b) Raman spectra of as-grown NRs

(i, red upper curve) and pristine C60 (ii, blue lower curve). The inset in (b)

highlights the downshift of the Hg(1) and Ag(1) modes for as-grown NRs (i)

in comparison to pristine C60 (ii).

43

5.2.2. Annealing

Figure 30. (a) XRD patterns of annealed NRs (i, red upper curve) and

pristine C60 (ii, blue lower curve). (b) STEM, (c) HRTEM, and (d) SAED of

the annealed NRs, which reveal the existence of a porous and

polycrystalline structure.

A vacuum annealing step (T = 200 °C, p <1 kPa, t = 24 h) was introduced in

order to remove the solvated m-DCB molecules, and the upper red curve in

Figure 30(a) shows the XRD of annealed NRs. Interestingly, the crystal

structure has transformed from hcp to face-centered cubic (fcc), with a unit

cell size of a = b = c = 14.13 Å; i.e. the same crystal structure as pristine C60

(blue lower curve) [70]. Further inspection of the annealed NRs by STEM

(figure 30(b)), HRTEM (figure 30(c)) and SAED (figure 30(d)) reveals the

existence of a porous polycrystalline structure, with clearly distinguishable

crystal grain boundaries.

44

5.2.3. Polymerization

Figure 31. The Ag(2) Raman mode of PNRs (upper panels, solid line) and

pristine C60 (lower panels). The deconvoluted peaks of the PNRs (dotted

lines) are indicated in the upper panels. The excitation laser wavelength

was 514 nm in (a) and 785 nm in (b), with the longer wavelength probing

deeper into the sample.

Figure 31 shows the Ag(2) Raman mode of PNRs (upper panels, solid line)

and pristine C60 (lower panels), with the deconvoluted PNR peaks indicated

by dotted lines in the upper panels. The excitation wavelength was λRaman =

514 nm and λRaman = 785 nm in figure 31(a) and figure 31(b), respectively.

Since the C60 absorption is significantly higher at 514 nm than 785 nm, the

former will probe the surface and the latter the bulk. A comparison of the

deconvoluted peaks of the Ag(2) mode shows a relative larger downshift at

514 nm, which is assigned to a higher degree of polymerization on the

surface than in the bulk. In other words, the PNRs can be described as

comprising a highly polymerized shell surrounding a mainly monomeric

core, as schematically indicated in the cartoon above the figure.

In order to investigate the effects of polymerization on the solubility, both

as-grown and PNRs were filtered out of the synthesis dispersion and added

to pure m-DCB in a concentration of 1 mg ml-1. The as-grown NRs were

observed to readily dissolve in the good C60 solvent, as evident by the

formation of a characteristic clear purple solution; see left vial in the

photograph in figure 32(a). Interestingly, the PNRs remained intact and well

dispersed in the m-DCB, as evident by the opaque brown dispersion in the

45

right vial in figure 32(a). In fact, subsequently recorded TEM images of the

PNRs reveal that these feature a completely smooth surface and are

completely unharmed by the good C60 solvent (see figure 32(b)). Similar tests

using common hydrophobic solvents showed similar results, indicating that

the PNRs can be processed in a wide range of organic solvents without

damage.

Figure 32. (a) Vials containing dissolved as-grown NRs (left vial with

purple solution) and dispersed PNRs (right vial with brown dispersion) at a

concentration of 1 mg ml-1 in m-DCB. (b) TEM image of PNRs, with the

inset showing a single PNR at higher magnification.

46

5.3. The C60 nanorod transistor

Figure 33. Schematic showing the NR FET structure, with the NRs aligned

in the shortest sourced-drain direction, i.e. perpendicular to the source-

drain interfaces.

OFETs with NRs as the active material were realized in order to evaluate the

electronic transport properties of the NRs in the pristine state and following

the structural changes induced by annealing and photo-polymerization. A

schematic of the employed BG-TC device structure is shown in figure 33,

with p:Si/SiO2 (200 nm) as the gate/gate-oxide, and Cr/Au (15/15 nm) as the

source-drain electrodes. In order to accurately measure µn with equation (3),

it is important to accurately estimate the effective WD and LG of the NRs

involved in the transistor operation i.e. the NRs bridging the gated channel

between the source and drain electrodes. The WD can be readily calculated as

the product of the mean NR diameter (Ø) and the observed number of NRs

crossing the channel, while the estimation of LG benefits from the alignment

of the NRs in the shortest source-drain direction; see figure 33. This

geometry will also produce the largest source-drain current.

47

5.3.1. Nanorod deposition

Figure 34. Schematic representation of the drop-casting of as-grown NRs

from an ethanol/m-DCB dispersion onto a transistor structure,

highlighting (a) the initial ethanol evaporation and (b) the subsequent m-

DCB evaporation and the formation of a random network of NRs and

needle-like C60 crystals.

The NRs can conveniently be deposited by drop-casting, whereby a (~25 µl)

drop of NR dispersion was applied onto the transistor structure. The

dispersion comprises a solvent mixture of m-DCB and ethanol, with the two

solvents differing greatly in vapor pressure (0.252 kPa versus 7.87 kPa at 25

°C) [85]. As a result, the high-vapor pressure ethanol will evaporate much

faster during drying, leaving the slow-drying m-DCB on the substrate, as

depicted in figure 34. As a consequence, some of the NRs will dissolve in the

remaining m-DCB solvent, while the remaining undissolved NRs will deposit

randomly on the substrate. The dissolved C60 is observed to exclusively

deposit on the Au electrodes in the form of needle-like structures,

presumably due to the large surface energy difference between the bare SiO2

channel and the Au electrodes. The random positioning of the NRs have the

unfortunate consequence that the effective LG of drop-cast deposited NRs is

difficult to determine with good accuracy.

To circumvent these issues, the method of controlled dip-coating was

utilized. The transistor substrate was immersed into the NR dispersion with

the long side of the transistor channel parallel to the solution interface, and

then slowly withdrawn at a fixed angle of 70° and a constant rate of 3 cm

min-1; see schematic in figure 35(a). A photograph of an aligned NR in the

transistor channel is shown in the photograph in figure 36(a). The

corresponding angular distribution of the dip-coated NRs is shown in figure

35(b), where the measured distribution of the drop-cast NRs is included for

comparison as the dashed blue line. It is clear that dip-coating results in a

48

good alignment of the NRs in the transistor channel and that the issue of C60

needle formation is effectively resolved. It is thus reasonable to approximate

LG with the shortest source/drain electrode distance in the calculation of the

mobility in transistor measurements on dip-coated NRs.

Figure 35. (a) Schematic of the dip-coating procedure and an optical

micrograph of dip-coated NR on a transistor structure (the light and dark

regions correspond to the Au electrodes and the SiO2 channel, respectively).

(b) Angular distributions of the dip-coated NRs and drop-cast NRs, relative

to the channel.

49

5.3.2. Transistor performance

Figure 36(a) presents typical transfer characteristics of as-grown NRs (Ø1 =

250 nm) deposited by drop-casting, before and after annealing. The

undesired large hysteresis is suppressed after the annealing, and this

improvement in performance is assigned to the removal of the charge-

trapping solvent molecules from the NRs. We have chosen to not include

data on the absolute mobility of the drop-cast NRs, due to the difficulty in

establishing the correct value for the gate length, LG, in drop-cast NR-based

OFETs, as discussed in detail in section 5.3.1.

Figure 36. Transfer characteristics of (a) drop-cast and (b) dip-coated NR

OFETs in different states. (c) Mobility distribution of as-grown NRs and

PNRs. (d) Output characteristics of a dip-coated PNR OFET.

Figure 36(b) shows the transfer characteristics of OFETs based on dip-

coated NRs (Ø2 = 172 nm) as the active material, with the NRs either being

characterized as-grown or after photo-induced polymerization. Quantitative

performance data for dip-coated NRs as a function of nanorod diameter,

annealing and photo-polymerizaton are summarized in table 3. We find that

the highest mobility is recorded on as-grown NRs with the lower average

diameter of 172 nm. Both the annealing and photo-polymerization

50

procedures result in a significant drop for both the mobility and the on/off

ratio. We assign this reduction to the introduction of grain boundaries

during annealing and to the formation of lower-mobility C60 polymers during

photo-polymerization.

Treatment Ø

(nm)

µn

(cm2 V-1 s-1) [a]

VT

(V) [a]

Ion/Ioff

(a.u.)

As-grown 250 4.4(±2.7)×10-2 17.4(±3.5) 103

Annealed 250 8.8(±2.6)×10-3 18.9(±1.7) 102

As-grown 172 3.0(±2.9)×10-1 21.5(±3.8) 104

Polymerized 172 4.7(±3.9)×10-3 20.4(±5.5) 103

[a] Mean value with standard deviation in parenthesis

Table 3. Transistor data on dip-coated NR OFETs.

We also point out that the data in table 3 are average values (and standard

deviations), and that the variation in performance can be rather significant,

as demonstrated by the mobility distributions for as-grown and polymerized

NRs in figure 36(c). In fact, the best-performing NR-OFET, comprising as-

grown NRs with an average diameter of 172 nm, did break the mobility

"unity" barrier at 1.0 cm2 V-1 s-1. This proves that solution-processable NRs

can compete with state-of-the-art vacuum-deposited C60 and a-Si [6, 86-88].

We further note that the drop in mobility following photo-polymerization is

in agreement with previous studies on the effects of chemical transformation

of C60 [63, 79]. In this context, it is, however, relevant to remind ourselves

that the charge accumulation and charge transport in an OFET take place in

the first few monolayers of the OSC deposited on the gate insulator, which

for PNRs correspond to the highly polymerized shell [19, 28, 89]. It is thus

reasonable to assume that the highly monomeric PNR core could exhibit a

higher mobility, similar to what was measured in as-grown NRs.

Typical output characteristics of a dip-coated PNR OFETs are shown in

figure 36(d). The devices exhibits a good IDS saturation at high VDS, but the

exponential increase at low VDS indicates a Schottky-type injection barrier at

the source electrode interface [90]. This observation, in combination with

the utilized simplification that all NRs are perfectly aligned in the transistor

channel, implies that the documented values for the mobility are

underestimating the true value [91, 92].

51

In summary, we have shown that single-crystal NRs can be synthesized in

solution using the LLIP technique, and that such NRs can be applied as the

active material in OFETs with a high electron mobility. A light-exposure step

can transform the NR surface to become polymeric, so that the resulting

PNRs can form a stabile dispersion in common hydrophobic solvents. With

these properties demonstrated, the NRs can be exploited in several

applications, and we mention the decoration of PNRs with Pd catalyst

particles, which resulted in an improvement in the oxidation of ethanol

(Paper V).

52

6. Material design for improved photovoltaics

The efficiency and practical usefulness of BHJ-OPVs are influenced by a

wide range of factors, and dedicated work in the field has resulted in the

demonstration of small-area devices with a PCE >10 % [55, 93, 94]. These

improvements are stemming from a reduction of the Eg of the constituent

compounds, so that the active material can absorb more of the incoming

solar radiation, and an improved active-material phase morphology as

manifested in an increased Jsc [55, 95-98]. It is also possible to attain a high

output voltage from BHJ-OPVs, but such reports are comparatively sparse

[55, 99]. Moreover, a well-documented issue with current BHJ-OPVs is the

poor morphological stability of the active material [52, 53, 100], and the

development of functional methods for the resolution of this setback is

highly needed.

6.1. A small-molecule donor compound for OPVs

Figure 37. (a) Molecular structure of ZOPTAN-TPA. (b) UV-vis absorption

spectra of ZOPTAN-TPA in solution (open circles) and as a thin film (solid

squares).

The BHJ-OPV commonly comprises an active-material blend of conjugated

polymer as the donor and a small-molecule fullerene as the acceptor. An

important advantage with small molecules is that they are comparatively

easy to synthesize and purify, and it is therefore of interest to employ small-

molecules also for the donor compound. For the attainment of a small

molecule donor compound with appropriate energy levels for high OPV

performance, we choose to synthesize a compound with an internal donor-

acceptor-donor (D-A-D) structure. By chemically connecting a (2Z,2Z)-2,2´-

(2,5-bis(octyloxy)-1,4-phenylene)bis(3-(5-bromothiophen-2-yl)acrylonitrile)

53

acceptor unit with two peripheral 4-(diphenylamino)phenylboronic acid

donor units, we attained the small-molecule donor compound ZOPTAN-

TPA, with its chemical structure presented in figure 37(a); for a description

of the synthesis procedure, see Paper VI.

ZOPTAN-TPA was thermally characterized with thermogravimetric analysis

(TGA) and differential scanning calorimetry (DSC). The TGA showed that

the material is stable (under inert atmosphere) up to 410 °C, whereas the

DSC revealed that ZOPTAN-TPA can crystallize and that its melting point is

145 °C. Figure 37(b) shows the UV-vis absorption spectra of ZOPTAN-TPA in

solution (open circles) and as a thin film (solid squares). The film features a

red-shifted absorption onset in comparison to the solution, which is

positioned at ~600 nm; the latter corresponds to an optical energy gap of

Eg,OPT = 2.1 eV. The HOMO and LUMO levels of ZOPTAN-TPA were

established with cyclic voltammetry (CV), and these data and the

electrochemical energy gap (Eg,EC) and Eg,OPT are presented in table 4.

Compound HOMO

(eV) LUMO

(eV) Eg,EC

(eV) Eg,OPT

(eV)

ZOPTAN-TPA –5.2 –3.0 2.2 2.1

Table 4. Measured energy levels and energy gaps for ZOPTAN-TPA.

By comparing the LUMO level of ZOPTAN-TPA with the LUMO level of

PCBM (-3.7 eV), we find that ΔELUMO = 0.7 eV in a PCBM:ZOPTAN-TPA

blend. Thus, it is expected that excitons, generated in the ZOPTAN-TPA

phase, will be efficiently split up at the ZOPTAN-TPA/PCBM interfaces in a

BHJ-OPV device. Moreover, as the difference in energy between the LUMO

of the PCBM acceptor and the HOMO of the ZOPTAN-TPA donor is notably

large at 1.5 eV, and this value dictates the maximum for the open-circuit

voltage, Voc, (see section 3.2.3), we also anticipate the opportunity for a high

output voltage.

BHJ-OPV cells were fabricated with an ITO/PEDOT:PSS anode and an Al

cathode sandwiching a PCBM:ZOPTAN-TPA active layer (see Paper VI for

the experimental procedure). The average photovoltaic performance of these

7×3 mm2 devices as a function of active material composition is presented in

figure 38(a)-(e). As anticipated, a high Voc was attained, with the maximum

of 1.03 V obtained for a PCBM:ZOPTAN-TPA mass ratio of 1:1. The Jsc, and

accordingly the EQE, featured a peak in performance for the 2:1 mass ratio.

The FF is relatively modest, with a maximum of 32 % for the 1:1 mass ratio.

In order to investigate whether recombination at the cathode could be the

54

culprit to the low FF, we also fabricated and tested devices with a 2nm LiF

electron-transport layer inserted between the cathode and the active

materials. As this procedure did not result in a significant improvement in

performance, we exclude cathodic recombination as the culprit. With the

information of Voc, Jsc and FF at hand, it is straightforward to calculate the

PCE, which features a peak value of 1.17 % for the 2:1 mass ratio.

Figure 38. (a) The open-circuit voltage, (b) the short-circuit current

density, (c) the fill factor, (d) the power conversion efficiency, and (e) the

external quantum efficiency for ITO/PEDOT:PSS/PCBM:ZOPTAN-TPA/Al

solar cells as a function of the PCBM:ZOPTAN-TPA mass ratio in the active

material.

We also wish to emphasize that the data in figure 38 represented the average

values (and standard deviations) derived from characterization of 14

independent devices, and that it is common that the performance of BHJ-

OPV can vary significantly between different devices. We have therefore

chosen to include results from our “champion” device, being a non-annealed

ITO/PEDOT-PSS/PCBM:ZOPTAN-TPA/Al OPV cell, with a

PCBM:ZOPTAN-TPA mass ratio of 2:1. The champion device featured the

following performance metrics: Jsc = 6.01 mA cm-2, Voc = 0.98 V, FF = 32 %,

and PCE = 1.86 %.

The OPV performance as a function of annealing is presented in figure 39(a).

The 2:1 mass ratio was selected for this study, since it featured the best

performance (see figure 38). While only minor changes in Voc and FF were

found (data not shown), the Jsc, and consequently the PCE, showed major

drops in performance following annealing. For instance, a 60 s annealing at

70 °C under inert N2 atmosphere resulted in a 15 % and 19 % drop in Jsc and

PCE, respectively, and an increase in temperature to 90 °C resulted in a drop

of 26 % and 33 %, respectively.

55

Figure 39. Relative change of (a) Jsc and (b) PCE as measured on

PCBM:ZOPTAN-TPA (2:1) devices following annealing at 70 °C (solid line)

and 90 °C (dotted line) for varying time. AFM micrographs of similar (c)

as-prepared and (d) annealed (90 °C for 5 min) films.

AFM micrographs of the same type of PCBM:ZOPTAN-TPA blend film

directly after spin-coating and after a 5 min thermal annealing at 90 °C

under N2 atmosphere are displayed in figure 39(c) and figure 39(d),

respectively. The as-prepared film comprises small depressed features of

unknown origin, with a typical vertical size of <20 nm, on an otherwise

smooth film surface. With thermal annealing, the depressed features vanish

and are replaced by protruding features. The size and density of these

protruding features are observed to increase with both temperature and

annealing time. Based on the observation that ZOPTAN-TPA can crystallize

(with a melting temperature of Tm = 145 C, as established with DSC), we

attribute the observed growth of the protruding features to the enrichment

and subsequent crystallization of ZOPTAN-TPA within the blend film. As the

lateral size of these features of ~100 nm is much larger than the typical

exciton diffusion of ~10 nm in organic semiconductors, it is straightforward

to assign the observed decrease of Jsc with thermal annealing to a decrease in

the area of the critical charge-generating donor/acceptor interfaces in the

active-material film when (too) large ZOPTAN-TPA crystallites form.

56

6.2. The C60 nanorod OPV

Figure 40. Schematic device structure of an OPV based on C60 polymerized-

nanorods.

One severe stability issue with BHJ-OPVs originates in the propensity of the

active material constituents to phase separate over time, with a reduction of

the exciton dissociation efficiency, and a lowering of the Jsc and PCE being a

common consequence [101, 102]. An example of this detrimental phase-

separation process was presented in the last paragraph in the previous

section, where an elevated temperature, as often encountered during OPV

operation, was shown to drive the undesired phase-separation process. Here,

we propose that the issue with an unstable phase morphology in BHJ-OPVs

could be alleviated by employing the polymerized nanorods as the acceptor

material for the realization of the device schematically depicted in figure 40.

We anticipate that not only will the charge-separation process be stabilized

in such a device, but also the subsequent charge-transport processes be

improved if the nanorods can be aligned with their long axes in the

interelectrode direction.

In order to investigate the feasibility of the outlined concept, we have

fabricated OPVs with a blend of PNR:P3HT. The as-grown NRs were first

ultrasonicated for 1 h to decrease their size to a length of 1-3 µm and a

diameter of 170 nm, thereafter photo-polymerized, and finally dispersed in a

P3HT-in-m-DCB solution, at a PNR:P3HT mass ratio of 1:2. This blend

solution/dispersion was spin-coated onto a PEDOT:PSS/ITO/glass

57

substrate, whereafter an Al cathode was thermally evaporated on top. The

performance of the OPV device was modest, but it is straightforward to

realize that a significant improvement should be attainable via a decrease in

NR size and through an alignment of the NRs, so that exciton split-up and

charge transport can be improved. In this context, it is interesting to note

that NRs with a small diameter of 50 nm can be synthesized; see figure

26(d).

58

7. Conclusions

We demonstrate that the fullerene PCBM can be photochemically

transformed into a dimeric state in a bi-excited reaction process, and that the

exposed material features a significantly reduced solubility in common

solvents as well as an effectively retained electron mobility. This attractive

combination of material properties allows for a direct and resist-free

lithographic patterning of electronic PCBM films down to a smallest feature

size of 1 µm, using a simple and scalable two-step process constituting light

exposure and solution development. In a further development, it was shown

that the two-step method was useful also in the area-selective transformation

of fullerene/conjugated-polymer blend films, as demonstrated through the

realization of a functional complementary logic circuit in the form of a 5-

stage ring oscillator.

In another project, we have synthesized highly flexible, single-crystal C60

nanorods with a solution-based self-assembly process termed liquid-liquid

interfacial precipitation. The 1-dimensional nanorods can be deposited from

their synthesis solution and employed as the active material in field-effect

transistor devices. Here, it was revealed that the as-fabricated nanorods can

feature an impressive electron mobility of 1.0 cm2 V-1 s-1, which is on par with

the performance of a work horse in the transistor field, viz. vacuum-

deposited amorphous Si. We further demonstrated that the processability of

the nanorods can be improved by a tuned light-exposure treatment, during

which the nanorod shell is polymerized while the high-mobility interior bulk

is left intact. This has the desired consequence that stabile nanorod

dispersions can be prepared in a wide range of solvents, and we anticipate

that functional electronic devices, including organic solar cells, based on

solution-processable nanorods can be realized in a near future.

59

Acknowledgements

First and foremost I would like to thank my brilliant supervisor Ludvig

Edman for giving me the chance to come to Umeå and be a part of the

Organic Photonics and Electronics Group. I have truly learned a lot during

the last 5 years and much of it is thanks to you.

Next up I would like to thank the incredible people in the OPEG group,

starting with the office squad Amir Asadpoordarvish and Andreas

Sandström; may the science you are throwing at the wall be vitorious.

Nikolai Kaihovirta, good luck in Turku and may your frolf discs fly far and

true. A special gratitude is aimed towards my permanent collaborators Jia

Wang and Jenny Enevold for countless discussions about solar cells,

fullerenes, and other things. Special thanks also to the ink-maestro Shi Tang,

Mattias Lindh, Petter Lundberg and Ning Sun. I would also like to

acknowledge the previous group members Nahid Morshed, Jimmy

Granström, Antoni Munar and Piotr Matyba. Our lab-technician Roushdey

Sahl also deserves a big thanks for making all of our lives easier by keeping

the labs running.

I also owe gratitude towards my collaborators outside of the group, with

special mentions of the brilliant Hamid Barzegar and his equally brilliant

supervisor Thomas Wågberg. It was a true pleasure to work together with

you. I am also grateful for the science done and discussed together with

Javed Iqbal, Bertil Eliasson, Florian Nitze, Guangzhi Hu, Xueen Jia, Tiva

Sharifi, Srikanth Revoju, Eduardo Gracia, Nathaniel Robinson, Deyu Tu and

Robert Forchheimer.

I would also like to specially thank Patric Stafshede for offering me the

opportunity to join the LunaLEC crew. Together we will do great things.

I would also like to thank the people I got into contact with during my

teaching endeavors: Jens Zamanian, Leif Hassmyr, Hans Forsman, Gabriella

Allansson, Sune Pettersson, Ludvig Lizana, Lars-Erik Svensson, Patrik

Norqvist, Jonas Larsson, Isak Silander, Mats Forsberg and Michael Bradley.

A big thanks also to all students.

A big thanks to Jörgen Eriksson, Katarina Hassler, Lilian Andersson, Lena

Burström, Karin Rinnefeldt and Nils Blix for all the help with my various

silly questions and problems. Special thanks to the "bluecoats" of the

workshop Lena Åström, Tomas Gustafsson and Peter Wikström for very

professional help with complex demands. I would also like to acknowledge

60

the remaining staff at the department of physics for making it a great place

to work.

I am also grateful for the support from my friends from the south: Anil Dey

(although I suppose it is a bit more west than south now), Fredrik and Veddy

Hägerström and Tim Larsson. I moved 1000 km north but still you did not

get rid of me. I would also like to thank all my relatives for their

encouragements.

Finally, I am infinitely grateful for my brothers and sister Mattias, Felicia,

Kenny and Sebastian. And last, but certainly not least, my parents Marianne

and Lennart. Thank you for everything.

61

References

1. J.H. Burroughes, D.D.C. Bradley, A.R. Brown, R.N. Marks, K.

Mackay, R.H. Friend, P.L. Burns, and A.B. Holmes, Light-emitting diodes based on conjugated polymers. Nature, 1990. 347(6293): p. 539-541.

2. N.S. Sariciftci, L. Smilowitz, A.J. Heeger, and F. Wudl, Photoinduced Electron Transfer from a Conducting Polymer to Buckminsterfullerene. Science, 1992. 258(5087): p. 1474-1476.

3. N.S. Sariciftci, D. Braun, C. Zhang, V.I. Srdanov, A.J. Heeger, G. Stucky, and F. Wudl, Semiconducting Polymer-Buckminsterfullerene Heterojunctions - Diodes, Photodiodes, and Photovoltaic Cells. Applied Physics Letters, 1993. 62(6): p. 585-587.

4. Q. Pei, G. Yu, C. Zhang, Y. Yang, and A.J. Heeger, Polymer light-

emitting electrochemical cells. Science, 1995. 269(5227): p. 1086-8.

5. A. Sandstrom, A. Asadpoordarvish, J. Enevold, and L. Edman, Spraying Light: Ambient-Air Fabrication of Large-Area Emissive Devices on Complex-Shaped Surfaces. Adv Mater, 2014.

6. C.D. Dimitrakopoulos and P.R.L. Malenfant, Organic Thin Film Transistors for Large Area Electronics. Advanced Materials, 2002. 14(2): p. 99-117.

7. S. Bae, H. Kim, Y. Lee, X. Xu, J.S. Park, Y. Zheng, J. Balakrishnan, T.

Lei, H.R. Kim, Y.I. Song, Y.J. Kim, K.S. Kim, B. Ozyilmaz, J.H. Ahn, B.H. Hong, and S. Iijima, Roll-to-roll production of 30-inch graphene films for transparent electrodes. Nat Nanotechnol, 2010. 5(8): p. 574-8.

8. A. Sandstrom, H.F. Dam, F.C. Krebs, and L. Edman, Ambient fabrication of flexible and large-area organic light-emitting devices using slot-die coating. Nat Commun, 2012. 3: p. 1002.

9. M. Hösel, R.R. Søndergaard, D. Angmo, and F.C. Krebs, Comparison of Fast Roll-to-Roll Flexographic, Inkjet, Flatbed, and Rotary Screen Printing of Metal Back Electrodes for Polymer Solar Cells. Advanced Engineering Materials, 2013. 15(10): p. 995-1001.

10. S.R. Forrest, The path to ubiquitous and low-cost organic electronic

appliances on plastic. Nature, 2004. 428(6986): p. 911-8.

62

11. P.F. Baude, D.A. Ender, M.A. Haase, T.W. Kelley, D.V. Muyres, and S.D. Theiss, Pentacene-based radio-frequency identification circuitry. Applied Physics Letters, 2003. 82(22): p. 3964.

12. C.C. Chen, L. Dou, R. Zhu, C.H. Chung, T.B. Song, Y.B. Zheng, S.

Hawks, G. Li, P.S. Weiss, and Y. Yang, Visibly transparent polymer solar cells produced by solution processing. ACS Nano, 2012. 6(8): p. 7185-90.

13. U. Salzner, J.B. Lagowski, P.G. Pickup, and R.A. Poirier, Comparison of geometries and electronic structures of polyacetylene, polyborole, polycyclopentadiene, polypyrrole, polyfuran, polysilole, polyphosphole, polythiophene, polyselenophene and polytellurophene. Synthetic Metals, 1998. 96(3): p. 177-189.

14. S.S. Lee, J.M. Mativetsky, M.A. Loth, J.E. Anthony, and Y.L. Loo, Quantifying resistances across nanoscale low- and high-angle interspherulite boundaries in solution-processed organic semiconductor thin films. ACS Nano, 2012. 6(11): p. 9879-86.

15. H. Sirringhaus, P.J. Brown, R.H. Friend, M.M. Nielsen, K.

Bechgaard, B.M.W. Langeveld-Voss, A.J.H. Spiering, R.A.J. Janssen, E.W. Meijer, P. Herwig, and D.M. de Leeuw, Two-dimensional charge transport in self-organized, high-mobility conjugated polymers. Nature, 1999. 401(6754): p. 685-688.

16. J.-F. Chang, B. Sun, D.W. Breiby, M.M. Nielsen, T.I. Sölling, M. Giles, I. McCulloch, and H. Sirringhaus, Enhanced Mobility of Poly(3-hexylthiophene) Transistors by Spin-Coating from High-Boiling-Point Solvents. Chemistry of Materials, 2004. 16(23): p. 4772-4776.

17. H. Li, B.C.K. Tee, J.J. Cha, Y. Cui, J.W. Chung, S.Y. Lee, and Z. Bao, High-Mobility Field-Effect Transistors from Large-Area Solution-Grown Aligned C60 Single Crystals. Journal of the American Chemical Society, 2012. 134(5): p. 2760-2765.

18. P.C. Eklund, A.M. Rao, P. Zhou, and Y. Wang, Photochemical

Transformation of C-60 and C-70 Films. Thin Solid Films, 1995. 257(2): p. 185-203.

19. J.H. Burroughes, C.A. Jones, and R.H. Friend, New semiconductor device physics in polymer diodes and transistors. Nature, 1988. 335(6186): p. 137-141.

63

20. S. Steudel, K. Myny, V. Arkhipov, C. Deibel, S. De Vusser, J. Genoe, and P. Heremans, 50 MHz rectifier based on an organic diode. Nat Mater, 2005. 4(8): p. 597-600.

21. J. Cornil, J.L. Brédas, J. Zaumseil, and H. Sirringhaus, Ambipolar

Transport in Organic Conjugated Materials. Advanced Materials, 2007. 19(14): p. 1791-1799.

22. J. Veres, S. Ogier, G. Lloyd, and D. de Leeuw, Gate Insulators in Organic Field-Effect Transistors. Chemistry of Materials, 2004. 16(23): p. 4543-4555.

23. J. Veres, S.D. Ogier, S.W. Leeming, D.C. Cupertino, and S. Mohialdin Khaffaf, Low-k Insulators as the Choice of Dielectrics in Organic Field-Effect Transistors. Advanced Functional Materials, 2003. 13(3): p. 199-204.

24. L.-L. Chua, J. Zaumseil, J.-F. Chang, E.C.W. Ou, P.K.H. Ho, H.

Sirringhaus, and R.H. Friend, General observation of n-type field-effect behaviour in organic semiconductors. Nature, 2005. 434(7030): p. 194-199.

25. A. Facchetti, M.H. Yoon, and T.J. Marks, Gate Dielectrics for Organic Field-Effect Transistors: New Opportunities for Organic Electronics. Advanced Materials, 2005. 17(14): p. 1705-1725.

26. L. Burgi, H. Sirringhaus, and R.H. Friend, Noncontact potentiometry of polymer field-effect transistors. Applied Physics Letters, 2002. 80(16): p. 2913.

27. S.M. Sze, Semiconductor Devices, Physics and Technology. 2nd ed.

2002: John Wiley & Sons.

28. C.R. Newman, C.D. Frisbie, D.A. da Silva Filho, J.-L. Brédas, P.C. Ewbank, and K.R. Mann, Introduction to Organic Thin Film Transistors and Design of n-Channel Organic Semiconductors. Chemistry of Materials, 2004. 16(23): p. 4436-4451.

29. J. Zaumseil and H. Sirringhaus, Electron and ambipolar transport in organic field-effect transistors. Chem Rev, 2007. 107(4): p. 1296-323.

30. L. Herlogsson, M. Colle, S. Tierney, X. Crispin, and M. Berggren,

Low-voltage ring oscillators based on polyelectrolyte-gated polymer thin-film transistors. Adv Mater, 2010. 22(1): p. 72-6.

64

31. A. Dodabalapur, J. Laquindanum, H.E. Katz, and Z. Bao, Complementary circuits with organic transistors. Applied Physics Letters, 1996. 69(27): p. 4227.

32. J.M. Rabaey, A. Chandrakasan, and B. Nikolic, Digital Integrated

Circuits, A Design Perspective. Second ed. 2003: Prentice Hall.

33. A. Dodabalapur, Organic and polymer transistors for electronics. Materials Today, 2006. 9(4): p. 24-30.

34. Y.Y. Lin, A. Dodabalapur, R. Sarpeshkar, Z. Bao, W. Li, K. Baldwin, V.R. Raju, and H.E. Katz, Organic complementary ring oscillators. Applied Physics Letters, 1999. 74(18): p. 2714.

35. M. Kitamura, Y. Kuzumoto, S. Aomori, and Y. Arakawa, High-

Frequency Organic Complementary Ring Oscillator Operating up to 200 kHz. Applied Physics Express, 2011. 4(5): p. 051601.

36. B. Crone, A. Dodabalapur, Y.Y. Lin, R.W. Filas, Z. Bao, A. LaDuca, R. Sarpeshkar, H.E. Katz, and W. Li, Large-scale complementary integrated circuits based on organic transistors. Nature, 2000. 403(6769): p. 521-3.

37. B.K. Crone, A. Dodabalapur, R. Sarpeshkar, R.W. Filas, Y.Y. Lin, Z. Bao, J.H. O’Neill, W. Li, and H.E. Katz, Design and fabrication of organic complementary circuits. Journal of Applied Physics, 2001. 89(9): p. 5125.

38. D. Bode, K. Myny, B. Verreet, B. van der Putten, P. Bakalov, S.

Steudel, S. Smout, P. Vicca, J. Genoe, and P. Heremans, Organic complementary oscillators with stage-delays below 1 μs. Applied Physics Letters, 2010. 96(13): p. 133307.

39. K.-J. Baeg, D. Khim, D.-Y. Kim, S.-W. Jung, J.B. Koo, I.-K. You, H. Yan, A. Facchetti, and Y.-Y. Noh, High speeds complementary integrated circuits fabricated with all-printed polymeric semiconductors. Journal of Polymer Science Part B: Polymer Physics, 2011. 49(1): p. 62-67.

40. R. Chau, S. Datta, M. Doczy, B. Doyle, B. Jin, J. Kavalieros, A. Majumdar, M. Metz, and M. Radosavljevic, Benchmarking Nanotechnology for High-Performance and Low-Power Logic Transistor Applications. IEEE Transactions On Nanotechnology, 2005. 4(2): p. 153-158.

65

41. D.S. Hecht, L. Hu, and G. Irvin, Emerging transparent electrodes based on thin films of carbon nanotubes, graphene, and metallic nanostructures. Adv Mater, 2011. 23(13): p. 1482-513.

42. S. De, T.M. Higgins, P.E. Lyons, E.M. Doherty, P.N. Nirmalraj, W.J.

Blau, J.J. Boland, and J.N. Coleman, Silver Nanowire Networks as Flexible, Transparent, Conducting Films: Extremely High DC to Optical Conductivity Ratios. ACS Nano, 2009. 3(7): p. 1767-74.

43. H.A. Becerril, J. Mao, Z. Liu, R.M. Stoltenberg, Z. Bao, and Y. Chen, Evaluation of solution-processed reduced graphene oxide films as transparent conductors. ACS Nano, 2008. 2(3): p. 463-70.

44. J.L. Bredas, J.E. Norton, J. Cornil, and V. Coropceanu, Molecular understanding of organic solar cells: the challenges. Acc Chem Res, 2009. 42(11): p. 1691-9.

45. S. Gunes, H. Neugebauer, and N.S. Sariciftci, Conjugated polymer-

based organic solar cells. Chem Rev, 2007. 107(4): p. 1324-38.

46. J.K.J. van Duren, X. Yang, J. Loos, C.W.T. Bulle-Lieuwma, A.B.

Sieval, J.C. Hummelen, and R.A.J. Janssen, Relating the Morphology of Poly(p-phenylene vinylene)/Methanofullerene Blends to Solar-Cell Performance. Advanced Functional Materials, 2004. 14(5): p. 425-434.

47. L.J.A. Koster, M. Kemerink, M.M. Wienk, K. Maturová, and R.A.J. Janssen, Quantifying Bimolecular Recombination Losses in Organic Bulk Heterojunction Solar Cells. Advanced Materials, 2011. 23(14): p. 1670-1674.

48. M. Campoy-Quiles, T. Ferenczi, T. Agostinelli, P.G. Etchegoin, Y.

Kim, T.D. Anthopoulos, P.N. Stavrinou, D.D. Bradley, and J. Nelson, Morphology evolution via self-organization and lateral and vertical diffusion in polymer:fullerene solar cell blends. Nat Mater, 2008. 7(2): p. 158-64.

49. B. Watts, W.J. Belcher, L. Thomsen, H. Ade, and P.C. Dastoor, A Quantitative Study of PCBM Diffusion during Annealing of P3HT:PCBM Blend Films. Macromolecules, 2009. 42(21): p. 8392-8397.

50. N.D. Treat, M.A. Brady, G. Smith, M.F. Toney, E.J. Kramer, C.J. Hawker, and M.L. Chabinyc, Interdiffusion of PCBM and P3HT Reveals Miscibility in a Photovoltaically Active Blend. Advanced Energy Materials, 2011. 1(1): p. 82-89.

66

51. G. Lu, L. Li, and X. Yang, Creating a uniform distribution of fullerene C60 nanorods in a polymer matrix and its photovoltaic applications. Small, 2008. 4(5): p. 601-6.

52. Z. Li, H.C. Wong, Z.G. Huang, H.L. Zhong, C.H. Tan, W.C. Tsoi, J.S.

Kim, J.R. Durrant, and J.T. Cabral, Performance enhancement of fullerene-based solar cells by light processing. Nature Communications, 2013. 4.

53. H.C. Wong, Z. Li, C.H. Tan, H. Zhong, Z. Huang, H. Bronstein, I. McCulloch, J.T. Cabral, and J.R. Durrant, Morphological stability and performance of polymer-fullerene solar cells under thermal stress: the impact of photoinduced PC60BM oligomerization. ACS Nano, 2014. 8(2): p. 1297-308.

54. M.C. Scharber, D. Mühlbacher, M. Koppe, P. Denk, C. Waldauf, A.J. Heeger, and C.J. Brabec, Design Rules for Donors in Bulk-Heterojunction Solar Cells—Towards 10 % Energy-Conversion

Efficiency. Advanced Materials, 2006. 18(6): p. 789-794.

55. M. Jørgensen, J.E. Carlé, R.R. Søndergaard, M. Lauritzen, N.A.

Dagnæs-Hansen, S.L. Byskov, T.R. Andersen, T.T. Larsen-Olsen, A.P.L. Böttiger, B. Andreasen, L. Fu, L. Zuo, Y. Liu, E. Bundgaard, X. Zhan, H. Chen, and F.C. Krebs, The state of organic solar cells—A meta analysis. Solar Energy Materials and Solar Cells, 2013. 119: p. 84-93.

56. A.M. Rao, P. Zhou, K.A. Wang, G.T. Hager, J.M. Holden, Y. Wang, W.T. Lee, X.X. Bi, P.C. Eklund, D.S. Cornett, M.A. Duncan, and I.J. Amster, Photoinduced polymerization of solid C-60 films. Science, 1993. 259(5097): p. 955-957.

57. A. Dzwilewski, T. Wagberg, and L. Edman, Photo-induced and resist-free imprint patterning of fullerene materials for use in functional electronics. J Am Chem Soc, 2009. 131(11): p. 4006-11.

58. J. Wang, C. Larsen, T. Wågberg, and L. Edman, Direct UV

Patterning of Electronically Active Fullerene Films. Advanced Functional Materials, 2011. 21(19): p. 3723-3728.

59. T. Wagberg, P.A. Persson, and B. Sundqvist, Structural evolution of low-pressure polymerised C-60 with polymerisation conditions. Journal of Physics and Chemistry of Solids, 1999. 60(12): p. 1989-1994.

67

60. H.R. Barzegar, C. Larsen, L. Edman, and T. Wågberg, Solution-Based Phototransformation of C60 Nanorods: Towards Improved Electronic Devices. Particle & Particle Systems Characterization, 2013. 30(8): p. 715-720.

61. P.C. Eklund, P. Zhou, K.A. Wang, G. Dresselhaus, and M.S.

Dresselhaus, Optical Phonon Modes in Solid and Doped C-60. Journal of Physics and Chemistry of Solids, 1992. 53(11): p. 1391-1413.

62. B. Burger, J. Winter, and H. Kuzmany, Dimer and cluster formation in C-60 photoreaction. Zeitschrift Fur Physik B-Condensed Matter, 1996. 101(2): p. 227-233.

63. A. Dzwilewski, T. Wagberg, and L. Edman, C-60 field-effect transistors: Effects of polymerization on electronic properties and device performance. Physical Review B, 2007. 75(7): p. 9.

64. A. Dzwilewski, P. Matyba, and L. Edman, Facile fabrication of

efficient organic CMOS circuits. J Phys Chem B, 2010. 114(1): p. 135-40.

65. C. Larsen, J. Wang, and L. Edman, Complementary ring oscillator fabricated via direct laser-exposure and solution-processing of a single-layer organic film. Thin Solid Films, 2012. 520(7): p. 3009-3012.

66. J. Wang, J. Enevold, and L. Edman, Photochemical Transformation of Fullerenes. Advanced Functional Materials, 2013. 23(25): p. 3220-3225.

67. A. Tapponnier, I. Biaggio, and P. Gunter, Ultrapure C[sub 60] field-

effect transistors and the effects of oxygen exposure. Applied Physics Letters, 2005. 86(11): p. 112114.

68. A.L. Briseno, S.C.B. Mannsfeld, S.A. Jenekhe, Z. Bao, and Y.N. Xia, Introducing organic nanowire transistors. Materials Today, 2008. 11(4): p. 38-47.

69. K. Miyazawa, Y. Kuwasaki, K. Hamamoto, S. Nagata, A. Obayashi, and M. Kuwabara, Structural characterization of C60 nanowhiskers formed by the liquid/liquid interfacial precipitation method. Surface and Interface Analysis, 2003. 35(1): p. 117-120.

68

70. K. Miyazawa, Y. Kuwasaki, A. Obayashi, and M. Kuwabara, C(60) nanowhiskers formed by the liquid-liquid interfacial precipitation method. Journal of Materials Research, 2002. 17(1): p. 83-88.

71. C. Park, H.J. Song, and H.C. Choi, The critical effect of solvent

geometry on the determination of fullerene (C(60)) self-assembly into dot, wire and disk structures. Chemical Communications, 2009(32): p. 4803-4805.

72. M.G. Yao, B.M. Andersson, P. Stenmark, B. Sundqvist, B.B. Liu, and T. Wagberg, Synthesis and growth mechanism of differently shaped C(60) nano/microcrystals produced by evaporation of various aromatic C(60) solutions. Carbon, 2009. 47(4): p. 1181-1188.

73. A.L. Briseno, S.C.B. Mannsfeld, M.M. Ling, S.H. Liu, R.J. Tseng, C. Reese, M.E. Roberts, Y. Yang, F. Wudl, and Z.N. Bao, Patterning organic single-crystal transistor arrays. Nature, 2006. 444(7121): p. 913-917.

74. K. Ogawa, T. Kato, A. Ikegami, H. Tsuji, N. Aoki, Y. Ochiai, and J.P.

Bird, Electrical properties of field-effect transistors based on C-60 nanowhiskers. Applied Physics Letters, 2006. 88(11).

75. M. Larsson, J. Kjelstrup-Hansen, and S. Lucyszyn, DC Characterisation of C60 Whiskers and Nanowhiskers. ECS Transactions, 2007. 2(12): p. 27-38.

76. K.I. Ogawa, N. Aoki, K. Miyazawa, S. Nakamura, T. Mashino, J.P. Bird, and Y. Ochiai, C(60) nanowhisker field-effect-transistor application for nano-electronics. Japanese Journal of Applied Physics, 2008. 47(1): p. 501-504.

77. M.S. Xu, Y. Pathak, D.S. Fujita, C. Ringor, and K. Miyazawa,

Covered conduction of individual C(60) nanowhiskers. Nanotechnology, 2008. 19(7).

78. Y. Ochiai, K. Ogawa, N. Aoki, and J.P. Bird, Field-effect-transistor characteristics of solvate C60fullerene nanowhiskers. Journal of Physics: Conference Series, 2009. 159: p. 012004.

79. T. Doi, K. Koyama, Y. Chiba, H. Tsuji, M. Ueno, S.-R. Chen, N. Aoki, J.P. Bird, and Y. Ochiai, Electron Transport Properties in Photo and Supersonic Wave Irradiated C60Fullerene Nano-Whisker Field-Effect Transistors. Japanese Journal of Applied Physics, 2010. 49(4): p. 04DN12.

69

80. K. Miyazawa, Y. Kuwasaki, A. Obayashi, and M. Kuwabara, C60 Nanowhiskers Formed by the Liquid–liquid Interfacial Precipitation Method. Journal of Materials Research, 2011. 17(01): p. 83-88.

81. K. Miyazawa, K. Hamamoto, S. Nagata, and T. Suga, Structural

investigation of the C-60/C-70 whiskers fabricated by forming liquid-liquid interfaces of toluene with dissolved C-60/C-70 and isopropyl alcohol. Journal of Materials Research, 2003. 18(5): p. 1096-1103.

82. H.R. Barzegar, F. Nitze, A. Malolepszy, L. Stobinski, C.W. Tai, and T. Wagberg, Water Assisted Growth of C-60 Rods and Tubes by Liquid-Liquid Interfacial Precipitation Method. Molecules, 2012. 17(6): p. 6840-6853.

83. A. Talyzin and U. Jansson, C-60 and C-70 solvates studied by Raman spectroscopy. Journal of Physical Chemistry B, 2000. 104(21): p. 5064-5071.

84. L. Wang, B.B. Liu, S.D. Yu, M.G. Yao, D.D. Liu, Y.Y. Hou, T. Cui,

G.T. Zou, B. Sundqvist, H. You, D.K. Zhang, and D.G. Ma, Highly enhanced luminescence from single-crystalline C-60 center dot 1m-xylene nanorods. Chemistry of Materials, 2006. 18(17): p. 4190-4194.

85. D.R. Lide, CRC handbook of chemistry and physics. 2005, Boca Raton: CRC Press.

86. K. Itaka, M. Yamashiro, J. Yamaguchi, M. Haemori, S. Yaginuma, Y. Matsumoto, M. Kondo, and H. Koinuma, High-Mobility C60 Field-Effect Transistors Fabricated on Molecular- Wetting Controlled Substrates. Advanced Materials, 2006. 18(13): p. 1713-1716.

87. S. Kobayashi, T. Takenobu, S. Mori, A. Fujiwara, and Y. Iwasa,

Fabrication and characterization of C[sub 60] thin-film transistors with high field-effect mobility. Applied Physics Letters, 2003. 82(25): p. 4581.

88. T.D. Anthopoulos, B. Singh, N. Marjanovic, N.S. Sariciftci, A.M. Ramil, H. Sitter, M. Colle, and D.M. de Leeuw, High performance n-channel organic field-effect transistors and ring oscillators based on C-60 fullerene films. Applied Physics Letters, 2006. 89(21): p. 3.

89. H. Sirringhaus, Device Physics of Solution-Processed Organic Field-Effect Transistors. Advanced Materials, 2005. 17(20): p. 2411-2425.

70

90. L. Edman, J. Swensen, D. Moses, and A.J. Heeger, Toward improved and tunable polymer field-effect transistors. Applied Physics Letters, 2004. 84(19): p. 3744-3746.

91. D.J. Gundlach, L. Zhou, J.A. Nichols, T.N. Jackson, P.V. Necliudov,

and M.S. Shur, An experimental study of contact effects in organic thin film transistors. Journal of Applied Physics, 2006. 100(2).

92. L. Burgi, T.J. Richards, R.H. Friend, and H. Sirringhaus, Close look at charge carrier injection in polymer field-effect transistors. Journal of Applied Physics, 2003. 94(9): p. 6129-6137.

93. Y. Liu, C.C. Chen, Z. Hong, J. Gao, Y.M. Yang, H. Zhou, L. Dou, G. Li, and Y. Yang, Solution-processed small-molecule solar cells: breaking the 10% power conversion efficiency. Sci Rep, 2013. 3: p. 3356.

94. J. You, L. Dou, K. Yoshimura, T. Kato, K. Ohya, T. Moriarty, K.

Emery, C.C. Chen, J. Gao, G. Li, and Y. Yang, A polymer tandem solar cell with 10.6% power conversion efficiency. Nat Commun, 2013. 4: p. 1446.

95. J. Zhou, Y. Zuo, X. Wan, G. Long, Q. Zhang, W. Ni, Y. Liu, Z. Li, G. He, C. Li, B. Kan, M. Li, and Y. Chen, Solution-processed and high-performance organic solar cells using small molecules with a benzodithiophene unit. J Am Chem Soc, 2013. 135(23): p. 8484-7.

96. J. Peet, J.Y. Kim, N.E. Coates, W.L. Ma, D. Moses, A.J. Heeger, and G.C. Bazan, Efficiency enhancement in low-bandgap polymer solar cells by processing with alkane dithiols. Nat Mater, 2007. 6(7): p. 497-500.

97. W. Li, K.H. Hendriks, W.S.C. Roelofs, Y. Kim, M.M. Wienk, and

R.A.J. Janssen, Efficient Small Bandgap Polymer Solar Cells with High Fill Factors for 300 nm Thick Films. Advanced Materials, 2013. 25(23): p. 3182-3186.

98. Z. Li, G.R. He, X.J. Wan, Y.S. Liu, J.Y. Zhou, G.K. Long, Y. Zuo, M.T. Zhang, and Y.S. Chen, Solution Processable Rhodanine-Based Small Molecule Organic Photovoltaic Cells with a Power Conversion Efficiency of 6.1%. Advanced Energy Materials, 2012. 2(1): p. 74-77.

99. E. Ripaud, T. Rousseau, P. Leriche, and J. Roncali, Unsymmetrical Triphenylamine-Oligothiophene Hybrid Conjugated Systems as Donor Materials for High-Voltage Solution-Processed Organic Solar Cells. Advanced Energy Materials, 2011. 1(4): p. 540-545.

71

100. Y. Zhang, Y. Diao, H. Lee, T.J. Mirabito, R.W. Johnson, E. Puodziukynaite, J. John, K.R. Carter, T. Emrick, S.C.B. Mannsfeld, and A.L. Briseno, Intrinsic and Extrinsic Parameters for Controlling the Growth of Organic Single-Crystalline Nanopillars in Photovoltaics. Nano Letters, 2014. 14(10): p. 5547-5554.

101. M. Jorgensen, K. Norrman, S.A. Gevorgyan, T. Tromholt, B.

Andreasen, and F.C. Krebs, Stability of polymer solar cells. Adv Mater, 2012. 24(5): p. 580-612.

102. D. Angmo, P.M. Sommeling, R. Gupta, M. Hösel, S.A. Gevorgyan, J.M. Kroon, G.U. Kulkarni, and F.C. Krebs, Outdoor Operational Stability of Indium-Free Flexible Polymer Solar Modules Over 1 Year Studied in India, Holland, and Denmark. Advanced Engineering Materials, 2014.