Rights / License: Research Collection In Copyright - Non ...41808/... · Trap states in organic...
Transcript of Rights / License: Research Collection In Copyright - Non ...41808/... · Trap states in organic...
Research Collection
Doctoral Thesis
Trap states in organic field-effect transistors: quantification,identification and elimination
Author(s): Kalb, Wolfgang L.
Publication Date: 2009
Permanent Link: https://doi.org/10.3929/ethz-a-005813523
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
Diss. ETH No. 18324
Trap states in organic field-effect transistors:Quantification, identification and elimination
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY
ZURICH
for the degree of
Doctor of Natural Sciences
presented by
WOLFGANG L. KALB
Dipl.-Phys., RWTH Aachen
born on the 5th of December 1977
citizen of Germany
accepted on the recommendation of
Prof. Dr. B. Batlogg, examiner
Prof. Dr. G. Horowitz, co-examiner
2009
to Ludovica
We are all, at heart, gradualists,our expectations set by the steady passage of time.But the world of the Tipping Point is a placewhere the unexpected becomes expected,where radical change is more than possibility.It is - contrary to all our expectations -a certainty.
Malcolm Gladwell, The Tipping Point
Contents
Frequently used symbols vii
Frequently used acronyms ix
Abstract 1
Kurzfassung 3
1 Introduction 7
2 Oligomeric semiconductors: Electronic structure, charge transport and tran-
sistor operation 13
2.1 Intrinsic electronic states . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Transport mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1 Band transport . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.2 Hopping transport . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.3 Diffusive transport limited by thermal disorder . . . . . . . . . . 22
2.3 Trap-controlled transport . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.1 Trap-controlled band transport . . . . . . . . . . . . . . . . . . . 24
2.3.2 Trap-controlled hopping transport . . . . . . . . . . . . . . . . . 27
2.4 Causes of trap states in oligomeric semiconductors . . . . . . . . . . . . 29
2.4.1 Structural defects . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.4.2 Chemical impurities . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4.3 Trap states due to the gate dielectric . . . . . . . . . . . . . . . . 37
2.5 Transistor operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5.1 Qualitative description . . . . . . . . . . . . . . . . . . . . . . . 40
ii Contents
2.5.2 Analytical description of an ideal field-effect transistor . . . . . . 42
2.5.3 Deviations from the ideal transistor behaviour . . . . . . . . . . . 47
2.5.4 Electrical stability of organic field-effect transistors . . . . . . . . 49
3 Experimental details 51
3.1 Organic semiconductors investigated in this study . . . . . . . . . . . . . 51
3.2 Purification of pentacene and rubrene . . . . . . . . . . . . . . . . . . . 52
3.3 Preparation of the gate dielectric . . . . . . . . . . . . . . . . . . . . . . 52
3.3.1 Cleaning of Si/SiO2 substrates . . . . . . . . . . . . . . . . . . . 54
3.3.2 Surface modification with self-assembled monolayers of octade-
cyltrichlorosilane (OTS) . . . . . . . . . . . . . . . . . . . . . . 55
3.3.3 Polymeric gate insulators and polymeric buffer layers . . . . . . . 56
3.4 Growth of the semiconductor and electrical characterization . . . . . . . 58
3.4.1 Evaporation of organic films . . . . . . . . . . . . . . . . . . . . 58
3.4.2 Single crystal growth . . . . . . . . . . . . . . . . . . . . . . . . 59
3.4.3 Electrode deposition . . . . . . . . . . . . . . . . . . . . . . . . 59
3.4.4 Electrical characterization of transistors . . . . . . . . . . . . . . 60
3.5 Additional thin-film characterization . . . . . . . . . . . . . . . . . . . . 61
3.5.1 Static water contact angles . . . . . . . . . . . . . . . . . . . . . 61
3.5.2 Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . . 63
3.5.3 X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.5.4 Surface step profiling . . . . . . . . . . . . . . . . . . . . . . . . 64
3.5.5 Leakage current and capacitance measurements . . . . . . . . . . 64
3.6 Advanced fabrication and characterization of thin-film transistors . . . . . 64
3.6.1 Device fabrication and characterization system . . . . . . . . . . 65
3.6.2 Electrical characterization by gated four-terminal measurements . 67
3.6.3 Parameter extraction . . . . . . . . . . . . . . . . . . . . . . . . 69
4 Quinoid heteropentacenes as promising organic semiconductors for field-effect
transistor applications 73
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2 Experimental details . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2.1 Synthesis of 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX) . . 75
4.2.2 Device fabrication . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.3 Electrical characterization . . . . . . . . . . . . . . . . . . . . . 77
Contents iii
4.2.4 X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.3.1 Electrical characterization . . . . . . . . . . . . . . . . . . . . . 78
4.3.2 Crystal structure . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5 Organic small molecule field-effect transistors with fluoropolymer gate di-
electric: Eliminating gate bias stress effects 87
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.3.1 Properties of the gate dielectric . . . . . . . . . . . . . . . . . . . 90
5.3.2 Comparison of different devices . . . . . . . . . . . . . . . . . . 91
5.3.3 Reproducibility . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.3.4 Gate bias stress experiments . . . . . . . . . . . . . . . . . . . . 93
5.3.5 Contact effects in SC-FET’s . . . . . . . . . . . . . . . . . . . . 95
5.3.6 Temperature-dependent measurements . . . . . . . . . . . . . . . 96
5.3.7 Trap-controlled transport in TFT’s . . . . . . . . . . . . . . . . . 100
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6 Defect healing at room temperature in pentacene thin films and improved
transistor performance 105
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6.2 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.2.1 Device fabrication . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.2.2 Electrical characterization . . . . . . . . . . . . . . . . . . . . . 108
6.3 Parameter extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.3.1 Basic parameter extraction . . . . . . . . . . . . . . . . . . . . . 109
6.3.2 Advanced parameter extraction . . . . . . . . . . . . . . . . . . 112
6.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.4.1 Improvement of the device performance with time . . . . . . . . 116
6.4.2 Influence on the density of states function . . . . . . . . . . . . . 118
6.4.3 Comparison of several experiments . . . . . . . . . . . . . . . . 121
6.4.4 Influence of oxygen and nitrogen . . . . . . . . . . . . . . . . . 122
iv Contents
6.4.5 Annealing at slightly elevated temperatures . . . . . . . . . . . . 126
6.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.5.1 Defect healing at room temperature . . . . . . . . . . . . . . . . 127
6.5.2 Defects and contact resistance . . . . . . . . . . . . . . . . . . . 128
6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7 Oxygen-related traps in pentacene thin films: Energetic position and impli-
cations for transistor performance 131
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
7.3 Charge transport parameters . . . . . . . . . . . . . . . . . . . . . . . . 134
7.3.1 Field-effect conductivity, field-effect mobility and contact resistance134
7.3.2 Spectral density of trap states and free hole density . . . . . . . . 135
7.3.3 Fraction of free holes and band mobility . . . . . . . . . . . . . . 138
7.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.4.1 Extraction method and the influence of the contact resistance . . . 139
7.4.2 Oxygen-related device degradation . . . . . . . . . . . . . . . . 141
7.4.3 Oxygen-related traps . . . . . . . . . . . . . . . . . . . . . . . . 144
7.4.4 Trap induced changes in the free hole density . . . . . . . . . . . 147
7.4.5 Stability of the oxygen-related defects . . . . . . . . . . . . . . . 148
7.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.5.1 Effect of oxygen on the trap DOS . . . . . . . . . . . . . . . . . 148
7.5.2 Influence of oxygen-related traps on the field-effect mobility . . . 150
7.5.3 Consistency check: trapped holes vs. traps . . . . . . . . . . . . 151
7.5.4 Deep traps and device performance . . . . . . . . . . . . . . . . 151
7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8 Summary, conclusions and outlook 153
Bibliography 159
Acknowledgments 173
Curriculum Vitae 177
Publication list 179
Contents v
Contributions at conferences 181
Frequently used symbols
Ci Capacitance per gate unit area
d Thickness of the organic semiconducting layer
EV Valence band edge / mobility edge
Ea Activation energy of the field-effect conductivity
E0 = kT0 Characteristic slope of the trap density of states
εi Dielectric constant of the gate insulator
εs Dielectric constant of the organic semiconductor
Id Drain current
L Channel length
L′ Distance between the voltage sensing electrodes
l Thickness of the gate insulator
µ0 Intrinsic mobility
µe f f Effective mobility
NV Effective density of valence band states
N(E) Density of states (DOS)
P Holes per gate unit area (Ptotal , P f ree, Ptrapped)
p Volume hole density (p, p f ree, ptrapped)
Rcontact Contact resistance
Rchannel Channel resistance
σ Field-effect conductivity
S Subthreshold swing
Ug = |Vg−VFB| Effective gate voltage
Vg Gate voltage
viii Frequently used symbols
Vd Drain voltage
V1,V2 In-channel potentials
Vt Threshold voltage
Von Onset voltage
VFB Flatband voltage
V0 Interface potential
W Channel width
Frequently used acronyms
AFM Atomic force microscopy
CB Conduction band
Cytop Cyclic transparent optical polymer
DOS Density of states
DPCX 7,14-Diphenyl-chromeno[2,3-b]xanthene
HOMO Highest occupied molecular orbital
ITO Indium tin oxide
LUMO Lowest occupied molecular orbital
MOSFET Metal-oxide-semiconductor field-effect transistor
MIS Metal-insulator-semiconductor
RMS Root mean square
SC-FET Single crystal field-effect transistor
TFT Thin-film transistor
VB Valence band
XRD X-ray diffraction
Abstract
This study contributes to the broad topic of quantification, identification and elimination
of electronic trap states in organic field-effect transistors based on oligomeric semicon-
ductors and to an understanding of the charge transport mechanism in organic semicon-
ductors.
Field-effect transistors with new oligomeric semiconductors were fabricated and
characterized in order to identify possible relations between the chemical structure and
the crystal structure of the oligomeric material on one hand, and the transistor perfor-
mance on the other hand. The transistor performance reflects the efficiency of the intrinsic
transport mechanism and the extent of charge carrier trapping. Field-effect mobilities of
µ = 0.16 cm2/Vs and µ = 0.01 cm2/Vs were achieved, respectively, in single crystal and
thin-film transistors with the new quinoid heteropentacene 7,14-Diphenyl-chromeno[2,3-
b]xanthene (DPCX). The transistors show favourable properties, such as a near zero on-
set/threshold voltage and a small current hysteresis. A comparison of DPCX thin films
on octadecyltrichlorosilane (OTS)-treated and bare SiO2 gate dielectrics provides clear
evidence that the frequently used OTS surface treatment leads to organic thin films with a
better structural order. The experiments confirm the crucial importance of structural order
of the organic semiconductor for charge transport and thus for the transistor performance.
Furthermore, we studied the influence of different polymeric gate dielectrics and
polymeric buffer layers on the performance of pentacene- and rubrene-based field-effect
transistors. The experiments identify two causes of electrical instability in organic field-
effect transistors: traps due to the surface of the gate dielectric and due to structural dis-
order of the semiconductor. Organic single crystal transistors with a highly hydrophobic
and low permittivity fluoropolymer gate dielectric are essentially unaffected by long-term
gate bias stress. This highlights the high intrinsic electrical stability of oligomeric semi-
conductors. Due to the low density of interface traps and structural defects in rubrene
2 Abstract
single crystal field-effect transistors with fluoropolymer gate dielectric, we have achieved
a subthreshold swing of 0.75 nFV/(dec cm2). This is only ≈ 4 times larger than the in-
trinsic MOSFET limit of 60 mV/dec at room temperature. In the context of this study
we also address the issue of parasitic contact resistances since they may limit current in-
jection/extraction in high quality organic single crystal transistors for their low channel
resistance.
Moreover, we report on measurements of pentacene-based thin-film transistors with
a new experimental setup. This setup allows for measurements under highly controlled
conditions, e.g. without exposing the samples to ambient air between the transistor fabri-
cation steps and the electrical characterization. Great care was taken in order to separate
parasitic contact effects from properties of the semiconducting layer. The transistors were
characterized by gated four-terminal measurements. A key finding is that, under high
vacuum conditions (base pressure of order 10−8 mbar), the device performance improves
with time. The effective field-effect mobility increases by as much as a factor of two
and we obtained mobilities up to 0.45 cm2/Vs. In addition, the contact resistance de-
creases by more than an order of magnitude. These effects are attributed to a healing
of structural defects within the pentacene film at room temperature. This peculiar effect
is a direct consequence of the weak intermolecular interaction which is characteristic of
organic semiconductors. We show that the relevant structural defects that anneal at room
temperature are associated with shallow traps≤ 0.15 eV from the mobility edge/transport
level.
Finally, we studied the influence of oxygen on the spectral density of trap states of
a pentacene thin film. This was done by carrying out gated four-terminal measurements
on pentacene-based thin-film transistors as a function of temperature and without ever ex-
posing the samples to ambient air. We developed a scheme which allows for a calculation
of the spectral density of trap states from the dependence of the drain current on gate volt-
age and on temperature in a straightforward and unambiguous fashion. Photo-oxidation
of pentacene is shown to lead to a broad peak of trap states centered at 0.28 eV from the
mobility edge, with trap densities of the order of 1018 cm−3. The experiments support the
assumption of a mobility edge for charge transport, and contribute to a detailed under-
standing of an important degradation mechanism of organic field-effect transistors. We
can conclude that deep traps in an organic field-effect transistor, in general, reduce the
effective field-effect mobility by reducing the number of free carriers and their mobility
above the mobility edge.
Kurzfassung
Diese Dissertation trägt zu dem breiten Themengebiet der Quantifizierung, Identifizierung
und Ellimination von elektronischen Defektzuständen in organischen Feldeffekttransis-
toren auf der Basis von Oligomeren bei, sowie zum Verständnis des Ladungstransportes
in organischen Halbleitern.
Es wurden Feldeffekt-Transistoren mit neuen Oligomeren hergestellt und charak-
erisiert mit dem Ziel, mögliche Zusammenhänge zu identifizieren zwischen der chemis-
chen Struktur und der Kristallstruktur der Oligomere auf der einen und der Leis-
tungsfähigkeit der Transistoren auf der anderen Seite. Die elektrische Qualität
der Transistoren spiegelt die Effizienz des intrinsischen Ladungstransportes und die
Dichte von Defektzuständen wieder. Mit dem neuen chinoiden Pentazenderivat 7,14-
Diphenyl-chromeno[2,3-b]xanthene (DPCX) wurden Ladungsträger-Beweglichkeiten
von µ = 0.16 cm2/Vs und µ = 0.01 cm2/Vs entsprechend in Einkristall-Transistoren und
Dünnfilm-Transistoren erreicht. Die Transistoren mit DPCX haben günstige Eigen-
schaften wie etwa eine nahezu bei Null liegende Schwellenspannung und eine kleine Hys-
terese in den Transistor-Kennlinien. Ein Vergleich von Dünnfilmen aus DPCX auf mit
Octadecyltrichlorosilan (OTS)-behandeltem und unbehandeltem Gate-Isolator aus SiO2
stellt einen klaren Beweis dafür dar, dass die häufig verwendete Oberflächenbehandlung
mit OTS zu organischen Dünnfilmen mit besserer struktureller Ordnung führt. Die Exper-
imente bestätigen die grosse Bedeutung von struktureller Ordnung im organischen Hal-
bleitermaterial für den Ladungstransport und somit für die Leistungsfähigkeit der Tran-
sistioren.
Desweiteren wurde der Einfluss verschiedener löslicher Polymere als Gate-Isolator
in Feldeffekt-Transistoren mit Pentazen und Rubren im Hinblick auf die Leistungs-
fähigkeit der Transistoren untersucht. Durch die Experimente konnten wir zwei Ursachen
4 Kurzfassung
von elektrischer Instabilität organischer Feldeffekt-Transistoren identifizieren: Defek-
tzustände aufgrund der Oberflächen des Gate-Dielektrikums und strukturelle Unordnung
des Halbleiters. Organische Einkristall-Transistoren mit einem stark wasserabweisenden
Fluorpolymer sehr niedriger Permeabilität sind im wesentlichen von langzeitigem Gate
Bias-Stress unbeeinflusst. Dieses Ergebnis unterstreicht die hohe intrinsische elektrische
Stabilität von Oligomeren. Aufgrund der niedrigen Dichte von Grenzflächenzustände
und strukturellen Defekten in Einkristall-Transistoren auf der Basis von Rubren mit dem
Fluorpolymer als Gate-Isolator haben wir einen sehr schmalen Einschaltbereich unter-
halb der Schwellenspannung von nur 0.75 nFV/(dec cm2) erreichen können. Dies ist
nur ≈ 4 mal grösser als das intrinsische Limit in MOSFETs (60 mV/dec bei Raumtem-
peratur). Im Zusammenhang mit dieser Studie wird auch auf parasitäre Kontaktwider-
stände hingewiesen, denn diese können die Injektion/Extrahierung der Ladungsträger
in hochqualitativen organischen Einkristall-Transistoren aufgrund des niedrigen Kanal-
widerstandes limitieren.
Ferner beschreiben wir Messungen an Pentazen Dünnfilm-Transistoren mit einem
neuen experimentellen Aufbau. Dieser Aufbau ermöglicht Messungen unter höchst kon-
trollierten Bedingungen, also z. B. ohne die Proben zwischen der Herstellung der Transis-
toren und der elektrischen Charakterisierung der Umgebungsluft auszusetzen. Mit grosser
Sorgfalt wurden zwischen parasitäre Kontakteffekten und Eigenschaften der Halbleiter-
schicht unterschieden. Die Transistoren wurden durch Vier-Punkt-Messungen mit zusät-
zlicher Gate-Elektrode charakterisiert. Ein zentrales Ergebnis ist, dass sich die Qualität
der Transistoren im Hochvakuum (Basisdruck in der Grössenordnung 10−8 mbar) mit der
Zeit verbessert. Die effektive Ladungsträgerbeweglichkeit nimmt um bis zu einem Faktor
zwei zu und es wurden Beweglichkeiten bis zu 0.45 cm2/Vs erreicht. Zusätzlich nimmt
der Kontaktwiderstand um mehr als eine Grössenordnung ab. Diese Effekte werden einem
Ausheilen von strukturellen Defekten in den Pentazen-Dünnfilmen bei Raumtemperatur
zugeschrieben. Dieser eigentümliche Effekt ist eine direkte Konsequenz der für organ-
ische Halbleiter charakteristischen schwachen intermolekularen Wechselwirkung. Wir
zeigen, dass die relevanten strukturellen Defekte, welche bei Raumtemperatur ausheilen,
flachen Defektzuständen mit Energien ≤ 0.15 eV von der Mobilitätskante (bzw. von dem
Transportlevel) zuzuordnen sind.
Schliesslich wurde der Einfluss von Sauerstoff auf die energieaufgelösten De-
fektzustandsdichten von Pentazen-Dünnfilmen untersucht. Dazu wurden Vier-Punkt-
Messungen mit Gate an Pentazen-Dünnfilmtransistoren als Funktion der Temperatur
Kurzfassung 5
durchgeführt, und zwar ohne die Proben jemals der Umgebungsluft auszusetzen. Wir
haben ein Verfahren entwickelt mit dem man die energieaufgelösten Defektzustands-
dichten aus der gemessenen Abhängigkeit des Drain-Stroms von der Gate-Spannung und
von der Temperatur und in direkter und eindeutiger Art und Weise ausrechnen kann.
Es wird gezeigt das die Fotooxidation von Pentazen zu einem breiten Peak von Defek-
tzuständen führt, der bei 0.28 eV zentriert ist. Die Dichte der Fallenzustände ist in der
Grössenordnung von 1018 cm−3. Die Experimente bestätigen die Richtigkeit der An-
nahme einer Mobilitätskante für den Ladungstransport und tragen zum detaillierten Ver-
ständnis eines wichtigen Degradationsmechanismuses von organischen Feldeffekttransi-
storen bei. Im Allgemeinen reduzieren tiefe Defektzustände in organischen Feldeffekt-
transistoren die effektive Ladungsträgerbeweglichkeit durch die sich verringernde Anzahl
freier Ladungsträger und durch die Reduzierung der Beweglichkeit oberhalb der Mobil-
itätskante.
1 Introduction
Organic semiconductors have been studied for about 100 years now. The discovery of
photoconduction in anthracene in 1906 can be referred to as the beginning of organic
semiconductor research. [1] Although the principle of the metal-oxide-semiconductor
field-effect transistor (MOSFET) was already proposed in 1930 [2] and although the first
silicon-based MOSFET’s were fabricated in 1960 [3], it took another two decades until
the first organic field-effect transistors were realized. [4]
Organic semiconductors can be classified as either oligomeric materials (also called
small molecules) or conjugated polymers. Fig. 1.1 and 1.2 show some frequently used
materials. Representatives from both classes have loosely bound π-electrons which are
the source of charge conduction. Oligomers, on the one hand, tend to be crystalline and
can be obtained in high purity. Polymers, on the other hand, tend to be amorphous and
are difficult to purify. Within the long polymeric chains, there is very good orbital overlap
that leads to efficient intrachain transport.
One of the first reports on organic field-effect transistors dates back to 1983, when
Ebisawa et al. reported on field-effect transistors with (insoluble) polyacetylene as the
active semiconductor (see Fig. 1.2 for the chemical structure of this material). [4] In the
following years, several reports on field-effect transistors with conjugated polymers were
published. For example, polyacetylene was used as the active semiconductor and the
films were prepared by a precursor route. [5, 6] Field-effect mobilities in these devices
ranged between 10−6 and 10−4 cm2/Vs. Other studies with (insoluble) polythiophene
were published in the 1980’s and the polymer films were electrochemically polymerized.
[7–9] In this way, field-effect mobilities up to 10−3 cm2/Vs were achieved. [9] One of the
first reports with a soluble organic semiconductor was published in 1988. [10] Poly(3-
hexylthiophene) (P3HT, Fig. 1.2) served as the semiconductor and was deposited by spin
coating. Field-effect mobilities up to 10−4 cm2/Vs were measured. [10] The first organic
8 Introduction
Figure 1.1: Common oligomeric semiconductors, also called small molecules. Sexithiophene(6T) was one of the first oligomers that was used in organic field-effect transistors. Pentacene andits soluble derivatives (e.g. TIPS pentacene) are very promising for transistor applications; field-effect mobilities in excess of 1 cm2/Vs have been achieved. Rubrene is often used in fundamentalresearch due to the outstanding field-effect mobilities > 20 cm2/Vs in the best devices. Most ofthe experimental work for this thesis was done with pentacene and rubrene.
field-effect transistors with (thermally evaporated) oligomers as the active semiconductor
emerged at the end of the 1980’s as well. [11, 12] Horowitz et al. reported on transistors
based on vacuum evaporated sexithiophene (6T, Fig. 1.1) and obtained mobilities up to
10−3 cm2/Vs. [12]
Research on organic field-effect transistors has intensified substantially in recent
years. The research efforts are fueled by the fact that organic semiconductors can be de-
posited by thermal evaporation or from solution at low cost on large areas. Virtually any
substrate (including flexible plastic substrates) can be used since all fabrication steps can
be carried out by keeping the substrate at room temperature or at relatively low elevated
temperatures. Consequently, organic semiconductors are promising candidates for future
9
Figure 1.2: Common polymeric semiconductors. The first transistors with a polymeric semicon-ductor employed the insoluble polymers polyacetylene or polythiophene. Nowadays, the solublematerials P3HT and PQT-12 are frequently used and mobilities in excess of 0.1 cm2/Vs have beenobtained with both materials.
flexible and/or low-cost electronics. Apart from the easy deposition of organic semicon-
ductors, an important advantage is that the properties of the organic semiconductor can
be easily adjusted by means of synthetic organic chemistry.
The performance of organic field-effect transistors has improved dramatically since
the 1980’s. Progress was often due to the discovery of a new material and the subse-
quent optimization of the device fabrication. [13] Currently, field-effect mobilities up
to 5 cm2/Vs have been achieved both with vacuum-evaporated pentacene films (“p-type
semiconductor”, Fig. 1.1) or C60 films (“n-type semiconductor”). [14,15] Other examples
of the substantial progress are field-effect mobilities ≥ 20 cm2/Vs measured with rubrene
single crystals field-effect transistors [16] and mobilities in excess of 1 cm2/Vs with the
solution processable pentacene derivative TIPS pentacene [17].1 Structural order is of
importance for oligomeric semiconductors and seems to be relevant also in the case of
conjugated polymers. Mobilities of 0.1 cm2/Vs have now been achieved with polycrys-
talline P3HT films. [19] Moreover, it has been demonstrated that polymers exhibiting
1 In an earlier report transistors with TIPS pentacene showed mobilities up to 0.4 cm2/Vs but the TIPSpentacene was deposited by vacuum evaporation. [18]
10 Introduction
liquid-crystalline behaviour reach mobilities of 0.6-0.7 cm2/Vs. [20] Consequently, or-
ganic semiconductors are outperforming hydrogenated amorphous silicon (a-Si:H, typi-
cal field-effect mobility: 1 cm2/Vs) as the active semiconductor in thin-film transistors in
terms of field-effect mobility.
Although field-effect mobilities in organic transistors are already adequate for many
applications, other requirements need to be met for a successful commercialization of
organic field-effect transistors. It is clear that, for low-cost electronics the organic tran-
sistors should be inexpensive. Apart from low-cost fabrication steps, low-cost synthesis
and purification of all involved materials are required. In addition to a high field-effect
mobility, useful organic transistors must have a near zero threshold voltage, a steep sub-
threshold swing and a low current at zero applied gate bias. Moreover, a high electrical
and environmental stability of the devices is mandatory. Any changes of the transistor
characteristics during the lifetime of a device are to be avoided.
The device parameters and stability of organic field-effect transistors are intimately
related to the efficiency of the charge transport mechanism and the extend of charge car-
rier trapping in extrinsic traps. The main scientific challenge thus is to clarify the nature of
the charge transport in organic semiconductors and the microscopic origin of charge car-
rier traps. Pioneering work in this matter was done by Karl and coworkers. [21] Organic
semiconductors are distinct in that they generally consist of neutral molecules which in-
teract by rather weak forces (predominantly Van der Waals forces). The energy associated
with a Van der Waals bond is 10−3-10−2 eV, i.e. orders of magnitude smaller than the en-
ergy of a covalent bond. [22] The weak interaction strength leads to narrow energy bands
and may render band transport an inconsistent description of charge transport in organic
semiconductors, at least at room temperature. [23, 24]
All of the experimental work we performed was done with oligomeric semiconduc-
tors grown from the vapour phase. In Chap. 4 we describe a new oligomeric semicon-
ductor that was synthesized by Ciba Speciality Chemicals Inc. Most of the experiments
(Chap. 5-7) were, however, conducted on the common organic semiconductors pentacene
and rubrene since high field-effect mobilities can readily be obtained with these materials.
The thesis is organized as follows: in Chap. 2 we describe concepts to rationalize
the charge transport in organic semiconductors and particularly in crystalline oligomers.
Moreover, we summarize our knowledge on the microscopic origin of traps in oligomeric
semiconductors. Particular attention is paid to structural defects in pentacene thin films,
as much of the experimental work was done with pentacene-based thin-film transistors.
11
In Chap. 2 we also describe possible causes of additional traps in the vicinity of the inter-
face between the organic semiconductor and the gate dielectric. These traps may become
important since charge transport in a field-effect transistor takes place in the first few
nanometers at the insulator-semiconductor interface. Chap. 2 concludes by describing the
operation of an ideal field-effect transistor followed by a discussion of deviations from the
ideal transistor behaviour. Chap. 3 is intended to give an overview over the materials and
methods that were used for the experiments. More specific experimental details can be
found in Chap. 4-7, where the respective experimental results are described. The order of
the results chapters 4-7 is chronological. This means that the results in Chap. 4 were ob-
tained in the beginning of the experimental work for this thesis and the results in Chap. 7
towards the end. In Chap. 4 we describe the study of a new quinoid heteropentacene which
is promising as organic semiconductor for field-effect transistor applications. Chap. 5 ad-
dresses the issue of electrical stability. We show that, if a suitable gate dielectric is used
and if the organic semiconductor has a high degree of structural order, organic field-effect
transistors can have a very high electrical stability and outperform a-Si:H also in that re-
spect. The high degree of structural order is reached by employing organic single crystals
as the active semiconductor. Chap. 6 and 7 contribute to an understanding of traps in
organic thin films. The experiments in Chap. 6 and 7 were done with a vacuum prober
station which was attached to an evaporation system. These studies were thus performed
under highly controlled conditions. We developed and used a scheme which allows for
the extraction of the spectral density of trap states from the measured transfer character-
istics with high accuracy in an unambiguous and straightforward fashion. The results in
Chap. 4-7 are published in scientific journals (see publication list on p. 179).
2 Oligomeric semiconductors:Electronic structure, charge transportand transistor operation
Here we introduce concepts to rationalize the charge transport in oligomeric semicon-
ductors. We begin by describing the nature of the intrinsic electronic states followed by
a discussion of possible transport mechanisms in organic crystals of high quality. Ex-
trinsic trap states, however, can dominate the charge conduction; in Sec. 2.3 we describe
trap-controlled band and trap-controlled hopping transport. Then, we give a summary of
knowledge on the microscopic origin of trap states in oligomeric semiconductors. Finally,
we describe the operation of an ideal organic field-effect transistor and discuss deviations
from the ideal transistor behaviour.
2.1 Intrinsic electronic states in oligomericsemiconductors
Estimating the energy levels of the isolated molecules that make up organic crystals can
already be a very complex task. An approach that has long been applied to aromatic
hydrocarbons is the separation of σ- and π-electronic systems. [25] The 2s-, 2px and
2py-orbitals of each carbon atom can be combined to form a sp2-hybrid orbital. The
sp2-orbitals are orthogonal to the remaining 2pz-orbitals and their interaction can thus be
neglected. [25] The sp2-orbitals form molecular σ-orbitals which are localized between
two adjacent carbon atoms. The σ-electrons are strongly bound in the molecule. The
π-orbitals of the molecule, on the contrary, result from the overlap of the remaining 2pz-
orbitals at each carbon atom. This is illustrated in Fig. 2.1 for benzene. The π-electrons
are delocalized over the whole molecule and are more loosely bound compared to the
14 Oligomeric semiconductors
ð2pz
X
Z
Y
Figure 2.1: The π-electronic system in an aromatic hydrocarbon (e.g. benzene) results from theoverlap of atomic 2pz-orbitals. The σ-electrons are localized between two adjacent carbon atomsand are more strongly bound. Adapted from [26, 27].
b
a
a
c
c´
Figure 2.2: Herringbone structure of pentacene. In a simplistic view, the structure can be seenas a result of the interplay of weak intermolecular forces, i.e. dispersion forces and weak hydro-gen bonds. The pentacene crystal structure has a space group P1, two molecules per unit cell,a = 7.93 Å, b = 6.14 Å, c = 16.03 Å, α = 101.9, β = 112.6, γ = 85.8 as determined by Camp-bell et al. [30] This structure is often called bulk phase. Several other polymorphs exist due to theweak intermolecular interaction. [31, 32] Adapted from [33].
σ-electrons. The highest occupied molecular orbital (HOMO) and the lowest unoccupied
molecular orbital (LUMO) are π-type orbitals.
In organic crystals, the molecules are closely packed and adopt different structures,
such as a herringbone structure in the case of pentacene (Fig. 2.2). The atoms in the
molecules are bound by strong covalent bonds, but the intermolecular forces are relatively
weak. We may distinguish dispersion forces and weak hydrogen bonds. Dispersion forces
are due to fluctuations in the π-electronic clouds of the molecules and contribute to the
attraction of the molecules. Polarizabilities are often anisotropic with different values
along the different crystallographic directions. In addition, we have an attraction caused
by the weak hydrogen bonds between hydrogen and carbon atoms of adjacent molecules.
This latter type of interaction promotes edge to plane orientation of the molecules as in
the case of the herringbone structure. [28, 29]
2.1 Intrinsic electronic states 15
When forming the condensed phase, the weak intermolecular forces produce much
smaller changes in the electronic structure of the isolated molecules compared to co-
valently bound inorganic materials. The molecules in the crystal essentially keep their
identity. [22] In the tight binding-model, the energy levels in the organic crystal are seen
to arise from the molecular energy levels. The interaction among the HOMO’s and the
LUMO’s of a large number of molecules leads to the formation of the valence and conduc-
tion band. The intermolecular transfer integral of a certain electronic state of the isolated
molecule determines the width of the resulting band in the crystal. For example, the elec-
tronic bandwidth of a one-dimensional infinite stack of molecules is simply four times
the transfer integral. [34] X-ray analysis of the electron density in aromatic hydrocarbons
yields a maximum electron density around the carbon atoms which practically drops to
zero between the molecules. [22] Consequently, all transfer integrals have a rather small
value and the valence and conduction bands are narrow.
Brédas et al. calculated the influence of the intermolecular distance on the electronic
splitting of the HOMO and LUMO levels. [34] Fig. 2.3 shows the splitting of the HOMO
and LUMO of a cofacial sexithienyl dimer as a function of the intermolecular distance.
For a larger number of molecules, this splitting translates into the width of respectively
the valence and conduction band. Clearly, the intermolecular transfer integral and the
transfer of charge are closely related. Consequently, the larger the splitting of the HOMO
and the LUMO in the organic crystal is, the higher the mobility of holes and electrons
results to be. Fig. 2.3 shows a larger splitting of the HOMO which predicts a higher
intrinsic mobility for holes in sexithienyl. The calculations confirm the crucial importance
of electronic overlap on the electronic splitting and thus on the intrinsic mobility. Since
organic crystals are often highly anisotropic, the degree of electronic overlap can be very
different along the different crystallographic directions. For pentacene, the overlap is best
in the ab-plane (Fig. 2.2), leading to a superior charge carrier mobility within this plane
than in the c-direction.
Puschnig et al. studied the electronic band structure of the oligomers of poly(para-
phenylene) by first-principles calculations within the framework of density functional the-
ory. [35] The calculations confirm that the main character of the molecular π-orbitals is
preserved in the crystal structure. The band splittings are found to be 0.1− 0.2 eV for
the highest valence bands and lowest conduction bands. [35] Similar calculations for pen-
tacene combined with experimental evidence indicate significant differences for the dif-
ferent polymorphs under investigation. For the pentacene polymorph reported in [30] by
16 Oligomeric semiconductors
Figure 2.3: Calculated electronic splittings of the HOMO and LUMO levels in a cofacial dimermade of two sexithienyl molecules as a function of the distance between the two molecules. From[34].
Figure 2.4: Characteristic energies in an isolated molecule and in the condensed phase. Theinteraction of the HOMO’s and the LUMO’s of a large number of molecules leads to the formationof a narrow valence band (VB) and conduction band (CB) in the condensed phase. The ionizationenergy I and electron affinity A are changed in the condensed phase since the charge carrier in thecrystal is stabilized by polarization. Adapted from [1], p. 203.
Campbell et al., which is also called bulk phase, the width of the valence and conduction
band respectively are 0.34 eV and 0.53 eV. [36] In evaporated films of pentacene one often
finds a particular thin-film phase (see also Sec. 2.4.1, Fig. 2.9). In the thin-film phase the
molecules only have a small tilting angle and are standing almost upright. This leads to
a better overlap of the π-orbitals in the thin-film phase and the width of the valence and
conduction band are determined to be 0.64 eV and 0.62 eV. [36] The width of the valence
band is almost twice as large as for the bulk phase. The dispersion of the top valence
band is a measure for the hole mobility [35] and the increased valence band width in the
2.2 Transport mechanisms 17
Table 2.1: Ionization energies and electron affinities of an isolated pentacene molecule (Ig andAg) and of a pentacene crystal (Ic and Ac). The bandgap of pentacene is Eg = 2.2 eV. Adaptedfrom [1], p. 204.
Ig (eV) Ag (eV) Ic (eV) Eg (eV) Ac = (Ic−Eg) (eV)
6.7 1.2 5.1 2.2 2.9
thin-film phase may thus contribute to an understanding of the outstanding mobilities in
pentacene thin-film transistors.
Fig. 2.4 shows the formation of the valence and conduction band when going from
an isolated molecule (gas phase) to the condensed phase. The electron affinity Ag is the
energy related to binding an electron to a neutral molecule in the gas phase. The ioniza-
tion energy Ig is the energy which is necessary to remove an electron from the neutral
molecule. As illustrated in Fig. 2.4, the electron affinity and the ionization energy in the
crystal are changed. This is mainly attributable to the polarizing effect of the positively
or negatively charged molecule in the crystal which leads to stabilization. [1] Table 2.1
summarizes measured electron affinities and ionization energies for pentacene. The po-
larization energies which can be estimated with Pe ≈ (Ac −Ag) and Ph ≈ (Ig − Ic) are
significant, i.e. Pe ≈ 1.7 eV and Ph ≈ 1.6 eV.
2.2 Transport mechanisms
Only the loosely bound π-electrons (and possibly loosely bound electron pairs at het-
eroatoms) are transferred from molecule to molecule and, therefore, are the source of
charge transport in an organic crystal. The nature of the charge transport is still contro-
versial. It has been suggested that the charge carrier moves as a delocalized wave, as in the
case of many inorganic materials. Scattering, e.g. at phonons, would be seen as a small
perturbation. In the case of band transport, the charge carrier mobility µ0 is expected to
increase as the temperature is decreased. This is because the scattering of electrons at
phonons becomes less pronounced at a reduced temperature. For scattering at acoustic
phonons (one-phonon process), for example, the mobility is expected to vary as T−3/2 for
a wide band (W > kT ) and as T−2 for a narrow band (W ≤ kT , Ref. [1], p. 347).
As we will see, translational symmetry is not sufficient to make band transport a
self-consistent description of charge transport. In the hopping model the charge carrier
is highly localized. The polarization of adjacent molecules and the relaxation of inter-
18 Oligomeric semiconductors
Figure 2.5: Effect of phonons on the charge transport in (a) the band model and (b) a hoppingmodel. A phonon is illustrated as a peak in the energy surface. The peak cannot be surmounted bythe moving electron. The mobility is expected to increase with decreasing temperature in the bandmodel and to decrease with decreasing temperature in the hopping model. Adapted from [1, 27].
and intramolecular vibrations contribute to the localization. The localized charge carrier
hops from one molecule to the next. This process is expected to be assisted by phonons.
Consequently, one would expect a charge carrier mobility µ0 that decreases as the tem-
perature is decreased. The effect of phonons on charge transport in the band model and in
the hopping model is illustrated in Fig. 2.5.
2.2.1 Band transport
Band transport in the one electron approximation considers a perfect crystal with fixed
molecules, causing a periodic potential in which the electron moves. Scattering, e.g.
by phonons, is seen as a small perturbation of the zeroth order Hamiltonian. The small
perturbation induces transitions between Bloch states. The macroscopic mobility can be
approximated as follows:
2.2 Transport mechanisms 19
µ0 ≈ eτ0
kTW 2a2
h2 . (2.1)
W is a measure of the bandwidth, a is a length of the order of the lattice constant and
τ0 is the mean lifetime of the Bloch states. [1] We have already mentioned that, in the
frame of the band model, it is assumed that scattering is weak. Consequently, the energy
broadening related to the finite lifetime τ0 of the Bloch states through the uncertainty
principle has to be significantly smaller than the bandwidth W . By introducing W À h/τ0
into Eq. 2.1, we have
µ0 À ea2
hWkT
. (2.2)
This criterion has to be fulfilled in order for band transport to be a self-consistent de-
scription. By inserting typical values for W and a, we can conclude that the band the-
ory is self-consistent only if the mobility at room temperature is significantly larger than
1 cm2/Vs. [1] The mobilities in organic crystals at room temperature are around 1 cm2/Vs,
however. With the definition of a mean free path λ0 = τ0v (v≈Wa/h is the carrier veloc-
ity) we can deduce from Eq. 2.2 that the mean free path λ0 should be at least 3−4 times
the lattice constant. [1]
The temperature dependence of the charge carrier mobility was measured e.g. by
Karl et al. in ultrapure naphthalene crystals using the time-of-flight method. [23] Impor-
tantly, charge carriers are created within the crystal by illumination and are capacitively
measured. This means that the time-of-flight method does not depend so much on high
quality contacts to the organic crystals which are difficult to achieve. [37] In Fig. 2.6 we
show the temperature dependence of the electron and hole mobilities parallel to the crys-
tallographic a-direction. [23] The electron mobilities are lower than the hole mobilities.
The temperature-dependent data were fitted to µ0 ∝ T n, and exponents of n = −1.4 and
n =−2.9 were obtained respectively for electrons and holes. Karl et al. argue that charge
transport cannot proceed via hopping in these ultrapure crystals. This is because hop-
ping transport is not consistent with the increase in mobility with decreasing temperature.
Moreover, especially the mobilities at low temperatures are too high to be compatible with
hopping transport. [23] The conclusion of the authors is that the standard band model is
consistent for the low-temperature mobilities. Around room temperature, however, the
mean free path λ0 was estimated to be smaller than the lattice constant. This is inconsis-
tent with band transport. A complete description of charge transport in organic crystals
20 Oligomeric semiconductors
Figure 2.6: Measured temperature dependence of the electron and hole mobility in the a-directionof an ultrapure naphtalene crystal. The mobility increases as temperature decreases. Hole mobil-ities are in excess of 100 cm2/Vs at low temperatures. Hole mobilities are always higher thanelectron mobilities. From [23].
would have to explain both the low-temperature and the high-temperature regime and
should link the regimes in a unified way. [23]
Since the molecules in the crystal have highly polarizable π-orbitals, polarization ef-
fects may not be negligible in a suitable description of charge transport. Holstein’s polaron
band model considers electron-electron interactions. [38,39] With increasing temperature,
the polaron mass increases. This effect is accompanied by a bandwidth narrowing and in-
evitably results in a localization of the charge carrier. Consequently, this model predicts
a transition from band transport at low temperature to phonon-assisted hopping transport
at higher temperatures (e.g. room temperature). Recently, Holstein’s model was extended
by Hannewald et al. [40] The calculations predict a large increase of the bandwidth with
decreasing temperature in crystals of naphtalene, anthracene and tetracene, as shown in
Fig. 2.7.
2.2 Transport mechanisms 21
Figure 2.7: Effective polaron bandwidths as a function of temperature for tetracene (Tc), an-thracene (Ac) and naphthalene (Nph) as calculated with the extended Holstein model. From [40].
2.2.2 Hopping transport
If the intermolecular transfer integral is small compared to energies associated with
electron-phonon coupling or to energies associated with the polarization of adjacent
molecules due to the presence of a charge carrier, charge transport by polaron hopping
has to be considered. The time the charge carrier stays at a given site is relatively large,
since the electron is stabilized by the polarization of adjacent molecules and the relaxation
of intermolecular vibrations. There is a large number of hopping models, which generally
lead to a temperature dependence of the form
µ0 ∝1
T n exp(− E
kT
)(2.3)
with a value of n not exceeding 1.5. [23]
From diffusion theory and from Einstein’s relation between diffusivity D and mo-
bility,
µ0 =DekT
(2.4)
a hopping mobility
µ0 =ea2
kTP (2.5)
22 Oligomeric semiconductors
follows. P is the probability for charge transfer between two hopping sites and a is the
spacing between the sites. [41, 42]
Holstein et al. [39] and Yamashita et al. [43] propose different approaches to the
hopping probability P. In the high temperature limit both approaches lead to the same
thermally activated hopping probability. The expression for the hopping mobility in this
limit is
µ0 =ea2
kT1h
(π
2EbkT
)1/2
J2 exp(− Eb
2kT
)
︸ ︷︷ ︸P
. (2.6)
J is the intermolecular transfer integral and Eb is the polaron binding energy. [42]
Another important hopping model is the semiclassical Marcus electron transfer
theory. [44, 45] The transport of polarons is thought to be governed by the physics of
electron-transfer processes, as established by Marcus for chemical reactions and biologi-
cal electron-transfer processes. [44] The charge transfer rate kmarcus can be written as
kmarcus =t2
h
√π
λkTexp
(− λ
4kT
), (2.7)
where t is the electronic coupling and λ is the reorganization energy. [46] The reorga-
nization energy is the energy that is required for the structural adjustments which are
necessary in order for the electron transfer to occur between molecule A and D, i.e.
A+D A·−+D·+. (2.8)
The system has to reach a transition state in which the charge transfer occurs. [46] This
hopping model also leads to a thermally activated mobility which is not in agreement with
the results from measurements on high quality crystals (e.g. Fig. 2.6). Very recently it has
however been shown that a full quantum mechanical golden rule treatment would affect
the reorganization energy λ and would lead to a bandlike mobility behaviour. [46]
2.2.3 Diffusive transport limited by thermal disorder
The measured mobility in pure organic crystals decreases as a function of temperature up
to room temperature according to a power law µ0 ∝ T n (e.g. Fig. 2.6). This might be
consistent with band transport. However, the mobilities at room temperature are around
2.2 Transport mechanisms 23
1 cm2/Vs and the estimated mean free path λ0 is comparable to the lattice constants. It has
often been noticed that this is not consistent with band transport (see Sec. 2.2.1). Troisi
et al. suggested a transport model for organic crystals that resolves this contradiction. As
described in Sec. 2.1, the intermolecular transfer integral is very sensitive to small dis-
placements of the molecules. Consequently, it is reasonable to expect that the fluctuations
of the intermolecular transfer integrals caused by the thermal motions of the molecules
are significant. Troisi et al. have shown that, at least for temperatures above 100 K, the
fluctuation of the transfer integral is of the same order of magnitude as the transfer inte-
gral itself in organic semiconductors such as pentacene, anthracene or rubrene. [47, 48]
As a consequence, the fluctuations do not introduce a small correction, but determine the
transport mechanism and limit the charge carrier mobility. [49] The translational symme-
try of the Hamiltonian is broken, the band transport breaks down and the charge carriers
become localized. [24, 47]
In this scenario, the temperature dependent charge carrier mobility can be de-
termined with Einstein’s relation (Eq. 2.4). The diffusion coefficient D is computed
with a semi-classical model Hamiltonian and a one-dimensional stack of planar conju-
gated molecules. [24] The calculations predict a mobility that decreases with increas-
ing temperature up to room temperature, according to a power law. This is in excellent
agreement with the measured temperature-dependence in pure organic crystals (e.g. in
Fig. 2.6). [24,48] The model predicts mobilities at room temperature in the range between
0.1 cm2/Vs and 50 cm2/Vs, which also is in good agreement with the experiments. [24,48]
Interestingly, the importance of thermal disorder is supported by recent tetrahertz transient
conductivity measurements on pentacene crystals. [50]
This transport model may however not be consistent with other experimental facts.
Due to the thermal motions of the molecules, the electronic structure of organic crys-
tals is expected to resemble the electronic structure of amorphous materials even in the
absence of defects. [47, 48] However, SCLC measurements in the temperature range of
100−200 K clearly show a sharp peak of trap states in rubrene crystals caused by chemi-
cal impurities that were deliberately introduced. [51] The question whether this is consis-
tent with the model by Troisi et al. arises, or should the thermal motions of the molecules
lead to a significant broadening of the peak?
24 Oligomeric semiconductors
2.3 Trap-controlled transport
In Sec. 2.2, we have described charge transport in ultrapure organic crystals which is lim-
ited by electron-phonon coupling, by polarization effects and possibly by thermal disor-
der. Chemical impurities or structural defects were assumed to have a very limited effect
on charge transport, e.g. by scattering the charge carriers. Organic thin films are grown
rather rapidly and generally are polycrystalline and sometimes even amorphous. The sit-
uation can arrive in which the large number of extrinsic defects completely dominate the
charge transport. The charge carriers are multiply trapped by and thermally released from
trap states associated with the defects. [52, 53] Even at relatively high charge carrier den-
sities, most of the charge may be trapped on a time average. This leads to an effective
mobility
µe f f =n f
n f +ntµ0 (2.9)
where n f and nt respectively are the density of the free and trapped electrons and µ0 is
the intrinsic mobility. In this scenario, the Fermi level would be below the transport level.
The main effect of decreasing the temperature would be a reduction in the number of
free electrons in the transport level due to the sensitive dependence of the Fermi function
on temperature. The temperature dependence of the “intrinsic” mobility µ0 can therefore
be a secondary effect. The measured effective mobility µe f f may decrease with decreas-
ing temperature even if µ0 increases upon cooling. Sec. 2.3.1 describes trap-controlled
band transport and the mobility edge picture. In Sec. 2.3.2 we rationalize trap-controlled
hopping transport. We will see that even in a completely disordered solid, the hopping
transport can be described by a transport level and a distribution of localized states below
the transport level.
2.3.1 Trap-controlled band transport
The simplest case is to consider transport in extended states and localized states with
a fixed trap depth, e.g. due to a specific chemical impurity. The extended states have
an effective density NC at an energy EC and the localized states have a specific energy
Et and concentration Nt . If EC and Et are well separated from the the Fermi level EF ,
Boltzmann’s approximation can be applied. In that case, the density of free electrons is
n f ≈ NC exp[−(EC−EF)
kT
](2.10)
2.3 Trap-controlled transport 25
and the density of trapped electrons is
nt ≈ Nt exp[−(Et −EF)
kT
]. (2.11)
By inserting Eq. 2.10 and Eq. 2.11 into Eq. 2.9 we obtain [54]
µe f f ≈ 1
1+ NtNC
exp(
EC−EtkT
)µ0, (2.12)
where µ0 is the mobility in the extended states. Provided that the trapped charge exceeds
the free charge (nt À n f , low-temperature approximation), we have [54]
µe f f ≈n f
ntµ0 =
NC
Ntexp
[−(EC−Et)
kT
]µ0. (2.13)
The temperature dependence of the exponential factor in Eq. 2.13 may dominate the tem-
perature dependence of the effective mobility µe f f . In this case, measuring the mobility
as a function of temperature reveals the trap depth EC−Et . [21, 54, 55]
Especially in the case of organic thin films, it is more realistic to assume trap states
with a continuous distribution of the trap depth. According to extensive studies in our
group, this description is also appropriate for organic single crystals. [51] The situation in
organic thin films may be described in analogy to charge transport in amorphous inorganic
semiconductors, where the mobility edge picture has been developed. [54, 56] Fig. 2.8(a)
shows the influence on the spectral density of trap states of an increasing degree of dis-
order in a solid: the discrete trap levels which may exist in a crystal are smeared out as
disorder increases. The mobility edge (line between hatched and white area in Fig. 2.8(a))
separates extended from localized states. As shown in Fig. 2.8(a), its position depends on
the degree of disorder. At high levels of disorder, localization extends well into the energy
bands. At the mobility edge, the mobility as a function of energy increases abruptly, as
shown in Fig. 2.8(b). This can be rationalized as follows. We have seen that the mobility
in extended states is at least 1 cm2/Vs. Charge carriers may also be transported by tunnel-
ing between localized states. Mott et al. have however demonstrated that, for tunneling
26 Oligomeric semiconductors
(a)
(b)
Figure 2.8: (a) Influence of increasing disorder on the band structure of a solid. The hatched areasmark localized states, i.e. the spectral density of trap states. As disorder increases, the width ofdiscrete trap levels (left) increases and, eventually, the discrete trap levels are completely smearedout (right). In addition, more and more states become localized. (b) Schematic representationof the mobility edge. The mobility edge separates localized states (hatched area) from extendedstates. At the mobility edge, the mobility as a function of energy abruptly rises. From [54].
between sites of the same energy, the mobility is
µ∼= eν0R2
6kTexp
(−2R
R0
)exp
(− E
kT
). (2.14)
2.3 Trap-controlled transport 27
ν0 is the frequency of the hopping attempts (1012− 1013 Hz), R is the mean separation
between the sites, R0 is the localization radius of the charge carrier1 and E is the activa-
tion energy associated with the tunneling process. [54, 57] Even in the case of very high
densities of localized states (small R) and even if we assume E = 0, we can conclude that
µ ≤ 10−2 cm2/Vs. Hopping in localized states is thus expected to be negligible if trans-
port in extended states exists and we therefore have an abrupt increase in mobility at the
mobility edge. Only the charge carriers that are thermally activated to states above the
mobility edge contribute to the transport of charge.
In the low-temperature approximation (nt À n f ), the mobility can be expressed as
µe f f ≈n f
ntµ0 =
(Z ∞
EC
N(E) f (E)dE)(Z EC
EF
N(E) f (E)dE)−1
µ0 (2.15)
in analogy to Eq. 2.13. EC and N(E) are respectively the energetic position of the mobility
edge and the density of states, and f (E) is the Fermi function. [54] Again, the tempera-
ture dependence of the Fermi function can dominate the temperature-dependence of the
measured mobility µe f f when cooling down a sample.
2.3.2 Trap-controlled hopping transport
Trap-controlled hopping transport is very similar to trap-controlled band transport. In
both scenarios, the essence is thermal activation to a transport level. For the simple case
of traps with a discrete energy Et and a density Nt , the low-temperature approximation of
the mobility is essentially given by Eq. 2.13. Clearly, the effective density of extended
states NC needs to be replaced by the density of hopping sites NH at the energy EH . We
assume that the intrinsic hopping mobility µ0 can be described by Eq. 2.6. The traps
reduce this mobility to an effective mobility which can be written as
µe f f ∝NH
Ntexp
[−(EH −Et)+Eb/2
kT
]. (2.16)
If the activation energy of the hopping process Eb/2 is known, Eq. 2.16 can be used to
estimate the trap depth EH −Et and vice versa. For example, this approach has been used
1 For a localized charge carrier, the localization radius may be approximated with the Bohr radius.
28 Oligomeric semiconductors
to estimate the depth of discrete traps in (inorganic) As4S4. [41] The authors estimate
a trap depth of 0.26 eV with an activation energy of the intrinsic hopping transport of
0.14 eV. [41] Thermal activation to a hopping level may also be suitable for crystals of
small molecule organic semiconductors. [21]
In amorphous inorganic semiconductors such as amorphous Si, the existence of ex-
tended electronic states is attributed to the similarity of the short-range configuration of
the atoms in the amorphous phase compared to the configuration of the atoms in the corre-
sponding crystalline phase. [54] In the case of a completely disordered material, however,
one would expect a complete disappearance of extended states. The density of states is
not high enough to allow for the occurrence of extended states and all electronic states
are hopping sites. This description may not be suitable for crystalline or polycrystalline
organic semiconductors, but should be applicable to highly disordered conjugated poly-
mers. Bässler and Arkhipov et al. suggest a model which is based on variable-range
hopping and on the Miller-Abrahams equation for the hopping rate. [58] They show that
the complicated situation can be substantially simplified by introducing an effective trans-
port level and a broad distribution of localized states below this transport level. [59] The
density of states of the completely disordered system is a Gaussian function of energy.
Carriers in localized states jump to a shallower state with a universal energy (the transport
energy) and are transported away from the starting site. It is important to note that the po-
sition of the transport level remains unchanged, if the charge density is increased. There
is, however, an essential difference to trap-controlled band transport, where the proba-
bility of the charge carrier to be captured by the same localized state after reaching the
transport level is negligibly small. In the present situation, the site in the transport level is
still a localized state and adjacent to the starting site. Therefore, the carrier might return
to the initial site (backward jump). Consequently, we have to distinguish between the
energy level onto which most carriers jump from deeper states and the “genuine transport
level”. [59] Jumps onto the genuine transport level lead to a transport of the carrier away
from the initial site. It can be shown that, for a Gaussian distribution of states
N(E) =Nt√2πσ
exp(− E2
2σ2
)(2.17)
and with some simplifying assumptions (high temperatures and low carrier concentra-
tions), the charge mobility is
2.4 Causes of trap states in oligomeric semiconductors 29
µ =eν0
kT N3/2t
exp
[−1.2
(6γ3
πNt
)1/3]
exp[− σ2
2(kT )2
]. (2.18)
γ is the inverse localization radius and ν0 is the attempt-to-jump frequency. [59] Eq. 2.18
is valid for relatively low carrier concentrations only and is not applicable to operating
field-effect transistors. However, the concept of the effective transport level has also been
applied to a situation with a high charge density. [60]
2.4 Causes of trap states in oligomeric semiconductors
In this section we deal with microscopic causes of trap states in organic field effect tran-
sistors. Charge carrier traps within the organic semiconductor are caused by structural
defects or chemical impurities. Chemical impurities may also cause a surrounding of
structural defects by distorting the host lattice. [21] On the other hand, chemical impuri-
ties tend to accumulate in regions with increased structural disorder. [1] Apart from traps
within the semiconductor, trap states caused by the gate dielectric can become important
in organic field-effect transistors. Trap states may be caused by certain chemical groups
on the surface of the gate dielectric, so that trapping is expected to depend on the specific
surface chemistry of the gate dielectric. The ability of chemical groups to cause traps will,
however, depend on the nature of the organic semiconductor. In addition, water adsorbed
on the gate dielectric may cause traps. Moreover, polar gate dielectrics may broaden
the density of states within the semiconductor at the interface, since randomly oriented
dipoles within the gate dielectric locally modify the polarization energy. [61, 62]
2.4.1 Structural defects
Structural defects are classified as point defects or extended defects. An example of a
point defect is a vacancy and examples of extended defects are edge dislocations, screw
dislocations, or (low-angle) grain boundaries. Structural defects modify the energy levels
in their vicinity and often lead to energy levels in the band gap, i.e. to charge carrier
traps. Such traps are electrically neutral when empty and become charged upon trapping
a charge carrier. Certain structural defects can result in scattering centers, but do not
cause traps (so called antitraps). A vacancy, for example, is a point defect that leads to
a decreased polarization energy, if a charge carrier occupies a molecule adjacent to the
vacancy. Consequently, the energy levels around the vacancy are not as low as in the rest
30 Oligomeric semiconductors
of the crystal and we have antitraps around the vacancy. Typical densities of vacancies in
anthracene or naphthalene crystals are of the order of 1014−1015 cm−3 (Ref. [1], p. 222).
Vacancies as an example of point defects are expected to be concentrated close to other
structural defects due to a reduced formation energy. [1] Consequently, most structural
defects in organic crystals are located close to each other in extended lines or areas. Ex-
tended structural defects can result in significant defect densities in organic crystals, e.g.
1019 cm−3 (Ref. [1], p.226). Therefore, extended structural defects are thought to be the
main source of traps in ultrapure organic crystals. [63] Structural defects are often thermo-
dynamically unstable. Annealing e.g. naphthalene or anthracene crystals for 100−300 h
reduces the density of structural defects by typically one order of magnitude (Ref. [1],
p.228). The density of structural defects in an organic crystal sensitively depends on the
preparation method and also on the handling of the crystal after growth. [64, 65]
Clearly, organic thin films have a higher density of structural defects than organic
crystals. In the following, we discuss structural defects in vacuum evaporated pentacene
films, which are of particular relevance for our work. Pentacene films are grown rather
rapidly. For example, 50 nm thick films are evaporated in 30 min or less. The films are
often polycrystalline and have a layered structure within the grains. The molecules in the
layers are almost perpendicular to the substrate. The structure of vacuum evaporated thin
films is the result of the thin-film growth which depends on properties of the substrate’s
surface, including surface roughness [66, 67], surface free energy [68, 69] and the pres-
ence or absence of sites for heterogenous nucleation [27]. A low surface free energy, for
example, is thought to favour 3D-growth (Volmer-Weber growth). [70] In the case of a
high surface free energy, the material tends to wet the surface and layer-by-layer growth
is more likely. [70] This means that we have the completion of a molecular layer prior to
the nucleation of the next layer. On the other hand, the growth kinetics is substantially
influenced by the choice of the deposition rate and the substrate temperature during the
deposition. [71, 72] In the case of homogenous nucleation, the increase of the substrate
temperature and the decrease of the deposition rate are both expected to lead to films with
an increased grain size. [70]
Since pentacene films are often polycrystalline, large angle grain boundaries pro-
duce additional structural defects. Another important cause of structural disorder in pen-
tacene films is polymorphism. Pentacene can crystallize in at least four different struc-
tures (phases). This is due to the weak interaction among the molecules. [31, 32] It is
quite common that at least two of these phases coexist in pentacene thin films. [71, 73]
2.4 Causes of trap states in oligomeric semiconductors 31
Figure 2.9: Structure of a pentacene film evaporated on a SiO2 surface as suggested by [73],based on measurements with X-ray diffraction (XRD) and Raman spectroscopy. The films have alayered structure and contain several crystalline phases. The tilt of the molecules increases as thefilm thickness is increased. From [73].
Fig. 2.9 shows a pentacene thin film as a mixture of different pentacene polymorphs. [73]
According to this study by Cheng et al., an orthorhomic phase grows on a SiO2 substrate
which transforms to a thin-film phase and eventually to the triclinic bulk phase as the film
thickness increases. [73] The tilt angle of the molecules with respect to the surface normal
depends on the crystallographic phase. Close to the insulator-semiconductor interface, the
pentacene molecules are oriented almost perpendicular to the substrate.
Verlaak et al. theoretically studied structural defects in vacuum evaporated pen-
tacene films. [74] The study is restricted to defects, where the pentacene molecules are
translated within the molecular layers parallel to the substrate, or rotated along the long
molecular axis. It is argued that translation or rotation out of a molecular layer is not
favourable. [74] Moreover, any interaction with the substrate is neglected. The theoretical
study identifies an interesting effect which is shown in Fig. 2.10. Structural defects are
formed during the film growth. Upon addition of more and more “defective” molecules at
a given site, the ideal crystal structure becomes energetically more and more favourable.
The system eventually relaxes into the ideal crystal structure during the continuation of
the film growth. [74] The relaxation happens, provided that the evaporation rate is low
enough and that there is enough time for relaxation. Otherwise, the defective molecules
would become “locked” upon addition of more molecules. [74] It is suggested that struc-
tural defects within the grains of a pentacene film that resist relaxation cannot exceed
densities of 1016 cm−3, at typical growth conditions. A structural defect can, however,
influence the electronic levels of 10 surrounding molecules even if these molecules are in
32 Oligomeric semiconductors
Figure 2.10: Formation of in-grain structural defects during the growth of a pentacene film. Uponadding more “defective” molecules, relaxation into the ideal crystal structure becomes energeti-cally more favourable. However, the increased number of defective molecules not only reducesthe energy of the respective ideal crystal but also increases the energy barrier which needs tobe overcome. The height of the barrier is related to the time needed for the relaxation process.Most structural defects which exist during the growth of a film are expected to have relaxed intothe ideal crystal structure at the end of the film growth, as long as typical growth conditions arechosen. From [74].
the perfect crystal configuration. The authors conclude that grain boundaries are the most
prominent cause of structural defects in pentacene thin films. [74]
Although Verlaak et al. argue that translation out of a molecular layer is not
favourable, an experimental study identifies pentacene molecules that are displaced
slightly out of the molecular layers. [75] By means of high impedance scanning tunneling
microscopy (STM), specific defect islands in pentacene films with monolayer coverage
are detected. Within the defect islands, the pentacene molecules are displaced up to 2.5 Å
along the long molecular axis out of the pentacene layer with a broad distribution in the
magnitude of the displacements. However, the two-dimensional packing within the layer
is not disturbed. Electronic structural calculations show that the displaced molecules lead
to traps for both electrons and holes. These traps are located very close to the valence
or conduction band edge. The maximum displacement of the pentacene molecules as
seen by STM is 2.5 Å and this corresponds to a maximum trap depth of 0.1 eV. [75] The
situation is illustrated in Fig. 2.11.
The influence of annealing on the structure of pentacene films for field-effect tran-
sistor applications has been studied rather extensively. [76–81] Annealing after device
2.4 Causes of trap states in oligomeric semiconductors 33
Figure 2.11: Pentacene molecules that are displaced out of the pentacene layers along the longmolecular axis cause trap states very close to the valence and conduction band edge. Calculationspredict trap depths in the range of 0−0.1 eV for the experimentally observed displacements up to2.5 Å. From [75].
fabrication has also been done with pentacene/C60-based photovoltaic cells. [82] Several
studies agree that annealing pentacene thin films at moderate temperatures (e.g. 50 C)
results in an improved crystallinity of the films as measured by XRD. [76,77,79] In some
cases, the annealing is found to result in an increased field-effect mobility. [79] The mo-
bility can, however, also remain unchanged, although structural changes were identified
with XRD. [76] The latter results (improved crystallinity and unchanged mobility) are
rationalized by an improved ordering within the grains in most of the molecular layers,
but an unchanged molecular packing in the first layer close to the interface where the
charge is transported in a field-effect transistor. The effect of annealing in this scenario is
illustrated in Fig. 2.12.
2.4.2 Chemical impurities
The best method to produce organic crystals includes a zone refinement step in the purifi-
cation procedure (Ref. [1], p. 224). Even such crystals still have a considerable content of
impurities. Anthracene, for example, still has an impurity content of 0.1 ppm in the best
34 Oligomeric semiconductors
Figure 2.12: Thermal annealing at 50 C for 1 h leads to a significantly improved crystallinity ofpentacene films as seen by XRD. The film morphology, as examined with atomic force microscopy(AFM), remains unaffected. Consequently, the annealing results in an improved ordering withinthe grains. From [76].
crystals, which corresponds to a volume density of ≈ 1014 cm−3 (Ref. [1], p. 224). Zone
refinement produces organic materials of much higher purity as compared to purification
by sublimation. [83] However, zone refinement can only be applied if the material can
be molten without a chemical reaction or a decomposition to occur. This is not possible
for many small molecule semiconductors. For the acenes, there is a crossover between
materials which can be purified by zone refinement and those which can only be purified
by sublimation. This crossover occurs between anthracene and tetracene, i.e. tetracene
or pentacene cannot be purified by zone refinement. Thus, much higher impurity concen-
trations are expected in tetracene or pentacene. [83] An experimental study indicates that
in tetracene single crystals the charge carrier mobility is limited by chemical impurities
rather than by structural defects. [83]
Clearly, chemical impurities have different energy levels than the host material. This
may lead to empty states in the bandgap, i.e. traps located at the impurity molecules. For
a given impurity and a given host crystal, a simplistic approach can be used to estimate
the trap depths ∆Eet and ∆Eh
t , respectively for electron and hole traps. The situation is
illustrated in Fig. 2.13. The trap depth for electrons is approximated by the difference in
the electron affinity of the guest crystal (Ac)guest and of the host crystal (Ac)host .2 The
difference in the crystal electron affinities can be approximated by the difference in the
gas phase electron affinities, and the trap depth for electrons can thus be written as [1]
∆Eet ≈ (Ac)guest − (Ac)host ≈ (Ag)guest − (Ag)host . (2.19)
2 The term guest crystal describes a crystal made of the impurity material.
2.4 Causes of trap states in oligomeric semiconductors 35
Figure 2.13: Energy levels of a host material with chemical impurities. The impurities lead toelectron traps at Ee
t and hole traps at Eht . The trap depths ∆Ee
t for electrons can be estimated withthe difference in electron affinity of the host and guest material. The trap depth for holes ∆Eh
t canbe derived with the respective difference in ionization energy. [27]
In analogy, the depth of hole traps due to a specific impurity can be estimated with the
ionization energies according to
∆Eht ≈ (Ic)host − (Ic)guest ≈ (Ig)host − (Ig)guest . (2.20)
Fig. 2.14 compares measured trap depths in anthracene with the corresponding estimates.
The simplistic approach predicts the trap depths surprisingly well in several cases. [84]
If the electron affinity of the host is larger than the electron affinity of the guest, the
impurity itself cannot act as a trap for electrons. However, due to difference in size, the
impurity may deform the host lattice which causes a local change in the polarizability of
the lattice. If the impurity is larger than the host molecule, we have a compression of
the lattice, an increase in polarization energy and thus traps located on host molecules
adjacent to the chemical impurity. On the other hand, a smaller impurity results in a
reduced polarization energy. This cannot lead to an increase in stabilization in the vicinity
of the impurity but can generate a scattering of charge carriers in more polarizable regions
of the crystal, thus impeding the charge transport. [1]
Much of our experimental work was done with pentacene as the organic semicon-
ductor. Therefore, we now discuss chemical impurities in pentacene. The center ring of
the pentacene molecule is expected to be the most reactive, as illustrated in Fig. 2.15(a).
[85, 86] An important impurity in pentacene is thus thought to be the oxidized pentacene
species 6,13-pentacenequinone, where two oxygen atoms form double bonds with the
36 Oligomeric semiconductors
Figure 2.14: Measured trap levels in anthracene due to various impurities (solid lines). Thecorresponding trap levels, as estimated with the electron affinities and ionization energies of thehost and guest material (Eq. 2.19 and 2.20), are shown as dashed lines. Anthracenequinone (e), forexample, leads to electron traps, at approximately 0.6 eV from the conduction band. From [84].
(a) (b) (c)
Figure 2.15: The center ring of the pentacene molecule is expected to be the most reactive (a).Common impurities in pentacene are the oxidized species 6,13-pentacenequinone (b) and 6,13-dihydropentacene (c).
carbon atoms at the 6,13-positions (Fig. 2.15(b)). This disrupts the π-electron system
of the molecule. Pentacenequinone is expected to lead to trap states in the band gap of
pentacene. [86] Repeated purification of pentacene by vacuum sublimation can result in
2.4 Causes of trap states in oligomeric semiconductors 37
very high mobilities in pentacene single crystals. [87] This effect is attributed to reduc-
ing the concentration of pentacenequinone which degrades the transport properties by
scattering the charge carriers. [87] Another common impurity in pentacene may be 6,13-
dihydropentacene (Fig. 2.15(c)), where additional hydrogen atoms are bound both at the
6- and at the 13-position. [85]
2.4.3 Trap states due to the gate dielectric
Properties of the gate dielectric’s surface such as surface roughness, surface free energy
and density of heterogenous nucleation sites are expected to play a key role for the growth
of organic films from the vapour phase and thus influence the quality of the films. Apart
from growth-related effects, the sole presence of the gate dielectric may influence the
charge transport in a field-effect transistor especially because the charge is transported
in the first few molecular layers at the insulator-semiconductor interface. The chemical
nature of the gate dielectric and, in particular, water adsorbed on its surface appear to
be aspects of particular importance. Moreover, the polarity of the gate dielectric may
influence the charge transport.
Chemical nature of the gate dielectric
The surface of the gate dielectric contains chemical groups that may act as charge carrier
traps. The trapping mechanism may be as simple as the one discussed above for chemical
impurities. It may also involve a reversible or irreversible electrochemical reaction driven
by the application of a gate voltage. Chemical groups on the surface of the gate dielectric
are certainly important for the transport of electrons in a field-effect transistor. [88–90]
Silanol groups (-Si-OH) and carbonyl groups (-CO) on the surface of the gate dielectric
are identified as electron traps. [89,90] Silanol groups are present on the surface of a SiO2
gate dielectric particularly if the substrate cleaning is carried out with piranha solution
(30 % hydrogen peroxide in 70 % sulfuric acid). The efficiency of electron trapping by
these groups is expected to depend on the choice of the organic semiconductor. [90] The
trapping of electrons by various chemical groups is illustrated in Fig. 2.16. The presence
of chemical groups at the insulator-semiconductor interface with a high potential to trap
electrons explains why most organic field-effect transistors show p-type, but no n-type op-
eration. [88, 89] The chemical species are thought to completely suppress the conduction
of electrons, while good hole transport is still possible. Pentacene-based transistors, for
38 Oligomeric semiconductors
Figure 2.16: Surface chemistry of polystyrene (PSn) and poly(vinyl alcohol) (PVA) gate di-electrics, of a hexamethyldisilazan (HMDS)-treated and a bare SiO2 gate dielectric and of a PSdielectric which is exposed to an oxygen plasma (PS-Ox). The efficiency of electron trapping atthese surfaces increases from left to right. From [90].
Figure 2.17: A high dielectric constant of the gate insulator leads to a polar surface. The randomlyoriented dipoles at the insulator-semiconductor interface modify the local polarization energy. Thenet effect is a broadening of the bulk density of states at the interface and thus a reduced mobilityof the charge carriers. The density of states in the bulk of the organic semiconductor is a Gaussiandensity of states which results from static disorder. The carrier population is also shown both forthe bulk density of states and the broader interface density of states. From [62].
example, generally show p-type but no n-type operation, although the intrinsic mobilities
for electrons and holes in pentacene are not expected to be very different.
2.4 Causes of trap states in oligomeric semiconductors 39
Adsorbed water
Water adsorbed on the gate dielectric may dissociate and react with pentacene. One pos-
sible reaction product is 6,13-dihydropentacene (Fig. 2.15(c)). The number of the impuri-
ties that are formed may depend on the electrochemical potential and would thus increase
as the gate voltage is ramped up in a field-effect transistor. [86]
It has also been suggested that water causes traps by reacting with the surface of the
gate dielectric. Water on a SiO2 gate dielectric treated with piranha solution (i.e. with a
large number of silanol groups (-Si-OH)) causes the formation of SiO−-groups according
the the reaction [91]
SiOH+H2O SiO−+H3O+. (2.21)
In addition to chemical reactions involving water, this species may impede the
charge transport in a different way. Polar impurities such as water can act as traps them-
selves if either Eq. 2.19 or Eq. 2.20 result in a positive trap depth. In that case, the polar
character of the impurity results in an electric field dependent trap depth. [92] Even if
the polar impurity does not lead to positive trap depths, its dipole moment modifies the
local value of the polarization energy. This results in traps in the vicinity of the water
molecules. [92, 93] We always have the polarization energy caused by the interaction be-
tween the charge carrier and the induced dipoles on the neighboring molecules Pc. Now,
we also find contributions due to the interaction of the charge carrier with the permanent
dipoles of the water molecules Pcp and between the permanent dipole moments and the
induced dipole moments Ppi. The local polarization energy can thus be written as
Ploc = Pc +∆Ploc, (2.22)
∆Ploc = Pcp +Ppi. (2.23)
The net effect is a broadening of the bulk density of states function at the insulator-
semiconductor interface. [92]
Dielectric constant of the gate dielectric
We have seen that polar molecules such as water can lead to additional traps due to a
local change of the polarization energy. Veres et al. have suggested that the polarity of
40 Oligomeric semiconductors
the gate dielectric surface impedes the charge transport in the same way. [61, 62] A more
polar surface has randomly oriented dipoles which lead to a modification of the local
polarization energy within the semiconductor and thus to a change of the site energies. As
in the case of water, this brings a broadening of the density of states function. Fig. 2.17
illustrates the effect of a polar gate dielectric on a Gaussian distribution of hopping sites.
Therefore, gate dielectrics with a low polarity result in organic field-effect transistors with
the best performance. If the gate dielectric has a low dielectric constant, we expect the
formation of a nonpolar surface. In this context, it is important to realize that surfaces
with a low polarity have a low surface free energy and are expected to have a high water
repellency as well. The high water repellency would also lead to a a reduced amount of
water at the critical insulator-semiconductor interface. [62] A high static water contact
angle on the insulator surface thus indicates that the surface has a low surface free energy,
a high water repellency and a low polarity.
In this context, it is interesting to notice that a clear correlation between the hole
mobility in rubrene single crystals field-effect transistors and the dielectric constant of the
gate dielectric has been observed. [94] This study is particularly meaningful, because the
single crystals are grown separately and growth-related effects can be excluded. A recent
explanation is somewhat different from the interpretation by Veres et al.: the dependence
of the mobility on the dielectric constant of the gate dielectric may be accounted for by a
two-dimensional Fröhlich polaron model. [95]
2.5 Transistor operation
2.5.1 Qualitative description
The operation principle of an organic field effect transistor is illustrated in Fig. 2.18 for
p-type operation. If no voltages are applied to the device, the density of mobile holes in
the semiconductor is low and may depend on the concentration of impurities that act as
dopants (Fig. 2.18(a)). [96] If a drain voltage Vd is now applied between the grounded
source and the drain electrode, the resulting current is very low. In principle, it would be
given by
Io f f =WL
dσ0Vd (2.24)
for sufficiently low drain voltages (ohmic regime). L is the channel length, W is the chan-
nel width and d is the thickness of the semiconductor. However, in organic field-effect
2.5 Transistor operation 41
transistors with an oligomeric semiconductor of high purity the (off-state) conductivity σ0
is generally so low that the current Io f f as described by Eq. 2.24 is unmeasurably small.
We have an off-current of typically 10−12 A which is due to leakage currents/experimental
limitations. In any case, the off-state of an organic field-effect transistor is guaranteed by
the low conductivity of the organic semiconductor.
If we apply a gate voltage Vg between the source and the gate but no drain volt-
age, holes are injected from the source electrode and are accumulated at the insulator-
semiconductor interface as shown in Fig. 2.18(b). This is analogous to charging a plane
plate capacitor, but the screening length La in a semiconductor is somewhat larger than in
a metal.
The application of a small drain voltage would now produce a large current due to
the highly conducting interface region (Fig. 2.18(c)). At a fixed drain voltage Vd , the drain
current Id can thus be modulated by several orders of magnitude by simply ramping up
the gate voltage. This is shown in Fig. 2.19(b) and the transistor characteristic is called
transfer characteristic. For a low drain voltage, the term “linear regime” is used.
In Fig. 2.18(c), the drain voltage is much lower than the gate voltage and the charge
density in the active channel is, to a good approximation, uniform all along the transistor
channel. The situation is more complicated if the drain voltage is increased. The charge
density now depends on the distance from the source electrode. At the grounded source
electrode, the voltage drop is Vg but at the drain electrode, the voltage drop is Vg−Vd .
Consequently, the density of accumulated holes at the interface is lower at the drain than
in the region at the source (Fig. 2.18(d)). [96]
When Vg = Vd , the channel is said to pinch of. If the drain voltage is further in-
creased, a depletion zone grows from the drain electrode (Fig. 2.18(e)). Due to the ex-
panding depletion zone, the drain current no longer increases if the drain voltage is further
increased (at a constant gate voltage). The drain current saturates. This behaviour of the
transistor can be observed in the output characteristic at sufficiently high source-drain
voltages (Fig. 2.19(a)). The transfer characteristic corresponding to the saturation regime
is measured by applying a high drain voltage which is kept constant and by sweeping the
gate voltage. We call this transistor characteristic the saturation regime transfer charac-
teristic (Fig. 2.19(c)).
42 Oligomeric semiconductors
Figure 2.18: Operation principle of an organic field-effect transistor. The scheme describes a p-type device. In the off-state, the density of mobile holes in the organic semiconductor is very lowand can be caused by chemical impurities that act as dopants (a). By applying a gate voltage Vg
between the gate electrode and the grounded source electrode, holes are injected from the sourceand are accumulated at the insulator-semiconductor interface (b). The application of a low drainvoltage Vd results in a high drain current Id (c). The situation is more complicated when the drainvoltage is increased. Now, the interfacial charge density depends on the position in the channeland is higher at the source than at the drain (d). If the drain voltage is increased even further sothat |Vd| > |Vg|, the drain region is depleted. This leads to a saturation of the drain current; at afixed gate voltage, the drain current is constant even if the drain voltage is increased. Adaptedfrom [96].
2.5.2 Analytical description of an ideal field-effect transistor
In this section we present an analytical description of an organic field-effect transistor.
This description is expected to be valid for samples with a low trap density and negligible
contact resistances. In addition, all traps are assumed to be “fast” traps. This means that
2.5 Transistor operation 43
Figure 2.19: Transistor characteristics of an ideal organic field-effect transistor: (a) output char-acteristic, (b) linear regime transfer characteristic and (c) saturation regime transfer characteristic.In the linear regime (low drain voltage), the drain current linearly depends on the gate voltage (b).In the saturation regime (high drain voltage), the square root of the drain current linearly dependson the gate voltage (c). From [97].
Figure 2.20: Definition of coordinates and variables to describe the organic field-effect transistor.The schematic shows the channel length L and the channel width W . The gate insulator has athickness l and a dielectric constant εi, while the semiconductor has a thickness d and a dielectricconstant εs.
the trapping and release times are much shorter than the time necessary to measure a
transistor characteristic (e.g. 1 min).
Transistor equations, mobility and threshold voltage
We begin with the derivation of two simple equations that describe the dependence of
the drain current on gate voltage both in the linear and in the saturation regime. These
equations can be used to extract the “trap-free” mobility µ0 and a threshold voltage Vt
from the measured transfer characteristics.
We use V for the electrical potential in the semiconductor and V (x = 0) = V0 is the
electrical potential at the insulator-semiconductor interface. Since the dielectric strength
D at the insulator-semiconductor interface must be continuous, we have
Dx = ε0εiFx(x = 0) = ε0εiVg−V0
l= Ci(Vg−V0) =−ε0εs
(dVdx
)
x=0, (2.25)
44 Oligomeric semiconductors
where F is the electric field and Ci = ε0εi/l is the capacitance of the gate dielectric per
unit area (see Fig. 2.20 for the definition of the variables). On the other hand, Gauss’s law
yields
(dVdx
)
x=0−
(dVdx
)
x=d=
(dVdx
)
x=0=
1ε0εs
Qtotal (2.26)
as long as the electric field on the backside of the semiconductor (x = d) vanishes and the
channel length L is much longer than the thickness of the gate insulator l (L À l). Qtotal
is the total charge per unit area (trapped and free). Combining Eq. 2.25 and Eq. 2.26, we
immediately find that
Qtotal =−Ci(Vg−V0). (2.27)
Note that V0 =V0(y) in Eq. 2.27 depends on the distance from the source if a drain voltage
is applied. The threshold voltage Vt is defined as the gate voltage above which essentially
all of the incrementally added gate-induced charge is free. The threshold voltage depends
on the density of charge carrier traps in the device and on the value of the flatband volt-
age VFB. The flatband voltage is the gate voltage which needs to be applied in order to
enforce flat bands at the insulator-semiconductor interface. A non-zero flatband voltage
can result from a difference of the Fermi level in the semiconductor and in the gate. More
importantly, the flatband voltage is influenced by charge that is permanently trapped at
the interface or within the gate dielectric. We replace the gate voltage Vg in Eq. 2.27 by
the effective gate voltage Vg−Vt and obtain an expression for the free charge per gate unit
area:
Q f ree =−Ci(Vg−Vt −V0). (2.28)
Eq. 2.28 will be inserted into an equation that is derived in the following.
The drain current Id is given by
Id =Z
jd(x,y)dxdz = WZ d
0jd(x,y)dx =
= WZ d
0µ0ep f ree(x,y)Fy(y)dx, (2.29)
2.5 Transistor operation 45
where µ0 and p f ree are the mobility and density of free holes. Fy is the component of the
electric field in the direction of the current flow. We can write
ID =−WdV0(y)
dyµ0
Z d
0ep f ree(x,y)dx
︸ ︷︷ ︸Q f ree(y)
=
=−WdV0(y)
dyµ0Q f ree(y). (2.30)
Combining Eq. 2.30 with Eq. 2.28, we have the differential equation
dV0
dy=
Id
Wµ0
[1
Ci(Vg−Vt −V0)
]. (2.31)
Variable separation (variables V0 and y) and integration gives
Z Vd
0Ci(Vg−Vt −V0)dV0 =
Id
Wµ0
Z L
0dy (2.32)
and eventually results in
Id =WL
µ0Ci
(Vg−Vt − Vd
2
)Vd (2.33)
as long as |Vd| ≤ |Vg−Vt |. The term which is quadratic in Vd is often neglected and so, in
the linear regime, we have
Id =WL
µ0Ci(Vg−Vt)Vd |Vd| ¿ |Vg−Vt |. (2.34)
In the linear regime, Eq. 2.34 predicts a linear dependence of the drain current on the
effective gate voltage Id ∝ (Vg−Vt). This behaviour was already mentioned above and
is represented in Fig. 2.19(b). A linear regression of the measured transfer characteristic
would yield the trap-free mobility µ0 and the threshold voltage Vt .
At Vg−Vt = Vd the depletion zone at the drain electrode is about to form and the
drain current saturates. By introducing Vg−Vt = Vd in Eq. 2.33, we obtain
Id =WL
µ0Ci
2(Vg−Vt)2 |Vd| ≥ |Vg−Vt |. (2.35)
46 Oligomeric semiconductors
Eq. 2.33 was derived by assuming |Vd| ≤ |Vg−Vt |. Due to the saturation of the drain cur-
rent, Eq. 2.35 is valid not only for |Vd|= |Vg−Vt | but also for |Vd|> |Vg−Vt |. According
to Eq. 2.35, the drain current in the saturation regime quadratically depends on gate volt-
age, i.e. Id ∝ (Vg−Vt)2. Fitting the square root of the measured saturation regime drain
current to a straight line yields the mobility and the threshold voltage. This is illustrated
in Fig. 2.19(c).
Onset voltage, subthreshold swing and correlation with trap densities
In addition to the threshold voltage Vt , the onset voltage Von is a useful parameter. It
is defined as the gate voltage where the drain current exceeds the noise level which is
typically at 10−12 A. The onset voltage is indicated in Fig. 2.19(b). The onset voltage
Von and the flatband voltage VFB are approximately equal in many cases. Deviations may
however exist in certain situations. Consequently, the trapped charge per unit area is
approximately Ci|Vt −Von|. The total density of traps per unit area N2 (unit: cm−2) can
thus be estimated with
N2 ≈ Ci|Vt −Von|e
. (2.36)
In addition to the onset voltage, the subthreshold swing S is an important device
parameter. It is a measure of how easily a transistor can be switched from the off-state to
the on-state. The subthreshold swing3 is defined as [3]
S =dVg
d(logId). (2.37)
With the simplistic assumption that both the density of deep bulk traps Nbulk (in
cm−3eV−1) and the density of interface traps Nint (in cm−2eV−1) are independent of
energy, the subthreshold swing may be written as [98]
S =kT ln10
e
[1+
eCi
(√
εsNbulk + eNint)]. (2.38)
3 The subthreshold swing S is the inverse of the subthreshold slope.
2.5 Transistor operation 47
This may be simplified as follows:
S =kT ln10
e
[1+
e2
CiN2
]. (2.39)
Both the deep bulk traps and the interface traps contribute to N2 (unit: cm−2eV−1).
[90, 99] Consequently, the subthreshold swing is a simple measure of the deep trap den-
sity. The lower limit of the subthreshold swing is obtained with N2 = 0 in Eq. 2.39:
S =kT ln10
e. (2.40)
At T = 300 K this lower limit is≈ 60 mV/dec. In other words, it is impossible to measure
a subthreshold swing steeper than 60 mV/dec in a field-effect transistor with the operation
principle described above.
2.5.3 Deviations from the ideal transistor behaviour
It is common practice in the field of organic transistors to use Eq. 2.34 and Eq. 2.35 in
order to estimate a mobility and a threshold voltage. If the experimental transfer charac-
teristics are linear in the linear regime and quadratic in the saturation regime, the approach
is self-consistent and the extracted mobility and threshold voltage have physical meaning.
The extensive experimental studies for this thesis have shown that organic field-
effect transistors exhibit ideal characteristics only in rare cases. For samples with a very
low trap density (steep subthreshold swing, high on-currents), the drain current often
increases sub-linearly in the linear regime. A typical example of such a transfer charac-
teristic is shown in Fig. 2.21(a). For this type of device, the transconductance (∂Id/∂Vg)Vd
does not saturate, but decreases even at high gate voltages.4 On the other hand, in sam-
ples with an increased trap density (thin-film transistors), the drain current in the linear
regime increases faster than linearly (Fig. 2.21(b)). The transconductance is a function
that monotonically increases with gate voltage. These deviations from the ideal transis-
tor behaviour can easily be understood. Eq. 2.34 and Eq. 2.35 rest on two simplifying
assumptions:
4 For an ideal transistor, the transconductance is expected to saturate above the threshold voltage at a valueof (∂Id/∂Vg)Vd = (W/L)µ0CiVd (Eq. 2.34).
48 Oligomeric semiconductors
(a) (b)
Contacts
Gate Voltage (V)
Tran
scon
duct
ance
Contacts
Dra
in C
urre
nt (A
)
Gate Voltage (V)
Traps
Tran
scon
duct
ance
Gate Voltage (V)
Traps
Dra
in C
urre
nt (A
)
Gate Voltage (V)
Figure 2.21: Deviations from the ideal transistor behaviour. The device in (a) is severely af-fected by parasitic contact resistances and the drain current increases less than lineraly in thelinear regime. An increased trap density is attributed to the device in (b). The traps lead to asuperlinear dependence of the drain current on gate voltage. The insets show the linear-regimetransconductance (∂Id/∂Vg)Vd .
1. The effect of parasitic contact resistances at the source and at the drain are negligi-
ble. All the drain voltage Vd drops across the transistor channel.
2. The trap density is very low and the threshold voltage is a rather small voltage.
Upon reaching the threshold voltage, essentially all the traps are filled. All the
charge which is accumulated above the threshold voltage is free and contributes to
the drain current.
We suggest that, for samples with a low trap density, the few traps are readily filled as the
gate voltage is ramped up, resulting in a high current density. The contacts cannot supply
and/or extract a sufficient amount of charge and the current is thus severely affected by the
contacts. This interpretation is supported by measurements of single crystal transistors in
Chap. 5. On the contrary, in samples with an increased trap density, the majority of the
charge carriers induced by the gate is trapped even at relatively high gate voltages. The
percentage of the gate-induced charge that is free increases with gate voltage and this leads
to a “superlinear” transfer characteristic (Fig. 2.21(b)). Strictly speaking, the threshold
voltage is not reached even if large gate voltages are applied. Eq. 2.34 and Eq. 2.35
are not suitable for these devices. [100, 101] This is as in the case of amorphous silicon
field-effect transistors, where the trap densities are substancial. [102] Pentacene thin-film
transistors with a fluoropolymer gate dielectric (Chap. 5) or with a SiO2 gate dielectric
2.5 Transistor operation 49
(Chap. 6 and Chap. 7) exhibit transfer characteristics with a superlinear dependence on
gate voltage. Due to the reduced current densities, contact effects are less severe. We
have developed and used an analytical approach for organic thin-film transistors with an
increased trap density. The analytical approach is outlined in Chap. 3 and is described in
detail in Chap. 6 and Chap. 7.
2.5.4 Electrical stability of organic field-effect transistors
In Sec. 2.5.2 we have also assumed that all traps are “fast” traps. However, organic field-
effect transistors often exhibit a current hysteresis (difference between the forward and the
reverse sweeps). Moreover, we often have a persistent shift of the transfer characteristic
when a gate bias is applied for a prolonged time. These phenomena are known as elec-
trical instability or gate bias stress effects. Electrical instability is most likely caused by
the trapping of charge in long-lived trap states. The term “long-lived” refers to a trapping
and release time which is long compared to the time needed to measure a transfer charac-
teristic (e.g. 1 min.). The shift of the transistor characteristic (threshold voltage shift) ∆Vt
can be used to estimate the surface density of long-lived traps according to [103]
N2 =Ci∆Vt
e. (2.41)
The long-lived states may be extrinsic in nature (e.g. certain chemical groups on the sur-
face of the gate dielectric). [103] The long-lived trap states may also be located within the
semiconductor in regions with increased structural disorder, i.e. may result from structural
defects. [67,80,104] Furthermore, a reversible electrochemical reaction of adsorbed water
with the organic semiconductor or with chemical groups on the surface of the gate dielec-
tric may play a role as well. [86, 105–107] It has also been suggested that gate bias stress
effects are an intrinsic phenomenon and are due to the formation of hole bipolarons, i.e. a
tightly bound state with a very low mobility. [108] Impurity states might even catalyze the
formation of bipolarons. [109] Other more marginal causes of electrical instability of an
organic field-effect transistor include the transfer of charge from the semiconductor to the
gate dielectric [110], or the movement of charged ions within the gate dielectric [111].
3 Experimental details
We now present an overview of the materials and methods that were employed for our
experimental work. More specific experimental details can be found in Chap. 4-7, where
we describe the respective experimental results. Here we begin by introducing the or-
ganic semiconductors that were employed. We then describe the transistor fabrication
and characterization steps with the equipment that had been established before this thesis
was started. This equipment was used for the studies in Chap. 4 and Chap. 5. In Chap. 6
and Chap. 7 we present measurements that were carried out with a new “device fabrica-
tion and characterization system” which we set up as a part of this thesis. In the present
chapter we introduce this system.
3.1 Organic semiconductors investigated in this study
All the experimental work was done with oligomeric semiconductors. The semiconductor
was always grown from the vapour phase by vacuum evaporation or by physical vapour
transport. Most of the experimental work was performed using the common materials
pentacene and rubrene (Fig. 1.1). The first project, however, was to fabricate and char-
acterize field-effect transistors with new oligomeric semiconductors in order to identify
relations between the chemical structure and the crystal structure of the oligomeric ma-
terial, on the one hand, and the transistor performance, on the other hand. This project
was carried out in collaboration with Ciba Speciality Chemicals Inc., Basel, Switzerland.
Fig. 3.1 shows most of the new organic semiconductors that we have investigated. The
materials that were obtained from Ciba were used as received, without any further pu-
rification. We made both thin-film transistors (TFT’s) and single crystal field-effect tran-
sistors (SC-FET’s) with the new oligomeric semiconductors (see Sec. 3.3-3.4 for details
on the transistor fabrication and characterization). Fig. 3.2 reveals the best field-effect
52 Experimental details
mobilities that were measured respectively in SC-FET’s and TFT’s. General relations,
e.g. between the chemical structure of the molecules and the device performance, could
not be identified. However, we see that most of the new oligomers which were stud-
ied led to operating transistors with mobilities ranging from 10−7 cm2/Vs to 0.1 cm2/Vs.
Field-effect mobilities from single crystal transistors (SC-FET’s) were found to be always
higher than mobilities from thin-film transistors (TFT’s), as long as operating devices of
both types could be made (factor of 3− 670 depending on the material). This is likely
caused by a better structural order of the semiconductor in the SC-FET’s. The electrical
performance of the new organic semiconductor 7,14-Diphenyl-chromeno[2,3-b]xanthene
(DPCX, Fig. 3.1) was found to be the most promising, and this material was thus studied
extensively. The study of DPCX highlights the crucial importance of structural order of
the semiconductor and is described in Chap. 4.
3.2 Purification of pentacene and rubrene
The experimental work in Chap. 5-7 was done with pentacene and rubrene. Both materials
were purchased from Sigma-Aldrich (pentacene: purum, Prod. No.: 76440; rubrene:
Prod. No.: R2206). The materials were purified by recrystallization in vacuum. This
was done by placing the starting material at the closed end of a horizontal glass tube.
The glass tube was introduced into a home-made holder and was evacuated with a turbo
pump. The side with the starting material was heated to a temperature slightly above
the sublimation temperature at the given pressure (e.g. ≈ 290 C for pentacene). While
heating the starting material, the other side of the tube was water-cooled, thus establishing
a temperature gradient along the tube. The sublimed material condensed at the walls of the
glass tube, in a specific region of the tube corresponding to a specific temperature. Lighter
impurities condensed further along the tube and heavier impurities did not sublime. The
recrystallization process was repeated with the recrystallized material at least once. A
schematic drawing of the purification system can be found in [112].
3.3 Preparation of the gate dielectric
An organic field-effect transistor consists of a gate electrode, a gate insulator, the active
organic semiconductor and the source and drain electrodes. The most convenient way
to fabricate organic field-effect transistors is to purchase highly-doped Si/SiO2 wafers.
3.3 Preparation of the gate dielectric 53
Figure 3.1: Most of the new oligomeric semiconductors that were studied in organic field-effecttransistors. The figure contains materials that were synthesized by Ciba Speciality Chemicals Inc.7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX) was found to be a promising organic semicon-ductor and was extensively studied.
54 Experimental details
10-710-610-510-410-310-210-1100101
No SC-FET operation
Amorp.films
AGH2SOL3
SOL1SOL2
NBR1
5,12-B
TBRAGH1
DPCX
XPEN2
XPEN1
Fiel
d-ef
fect
mob
ility
(cm
2 /Vs) SC-FET
TFT
Figure 3.2: Field-effect mobilities measured in single-crystal field-effect transistors (SC-FET’s)and in thin-film transistors (TFT’s) with new oligomeric semiconductors. The mobilities in SC-FET’s are always higher than in TFT’s. In several cases, no suitable crystals could be grown.Operating TFT’s were obtained with most of the materials.
The highly doped Si acts as gate electrode and the amorphous SiO2 layer (typical thick-
ness: 300 nm) acts as gate insulator. We purchased Si/SiO2 wafers (n/phosphorus) from
SilChem, Germany. The wafers were cut into appropriate pieces with an automatic dic-
ing saw. The transistors were fabricated by cleaning these substrates and by depositing
the semiconductor and the source/drain electrodes. A thin-film transistor of this type is
represented in Fig. 3.3(a). In Chap. 2 we have already seen that the surface of the gate
dielectric might influence the device performance in various ways. Consequently, other
device structures with an ultrathin organic buffer layer on the SiO2 gate dielectric were
also studied (Fig. 3.3(b)). Finally, we made transistors where the gate dielectric entirely
consisted of a spin-coated polymeric layer (Fig. 3.3(c)).
3.3.1 Cleaning of Si/SiO2 substrates
If we use a bare SiO2 gate dielectric, the device performance critically depends on the de-
tails of the cleaning procedure. Initially, the Si/SiO2 samples were washed with ultrapure
water (18.2 MΩcm) and were then cleaned with hot acetone and hot isopropanol in an
ultrasonic bath. Subsequently, the samples were immersed in piranha solution (30 % hy-
drogen peroxide in 70 % sulfuric acid). Finally, the samples were thoroughly washed with
ultrapure water and dried with nitrogen. The piranha treatment is important if the SiO2
is to be modified with a self-assembled monolayer of octadecyltrichlorosilane (OTS, see
3.3 Preparation of the gate dielectric 55
Figure 3.3: Organic field-effect transistors with three types of gate insulators were used, asschematically shown for the TFT’s. We used a bare SiO2 gate dielectric (a), a SiO2 gate dielectricwith an additional ultrathin organic layer as surface modification (b) or a spin-coated polymericgate dielectric on a glass/ITO substrate. The highly doped Si in (a) and (b) and the ITO layer in(c) act as the gate electrode. For the TFT’s, the contacts were formed after the semiconductor inall cases (top-contact devices).
Sec. 3.3.2), since it creates reactive silanol groups (-Si-OH). In the course of the experi-
mental work, we realized that for TFT’s with a bare SiO2 gate dielectric the device per-
formance is much better if the cleaning with piranha solution is omitted (see also [112]).
Consequently, for the projects described in Chap. 6 and 7, the SiO2 was cleaned solely
with hot acetone and hot isopropanol in an ultrasonic bath. This led to a rather passive
surface of the gate dielectric.
3.3.2 Surface modification with self-assembled monolayers of oc-tadecyltrichlorosilane (OTS)
For the study of the new organic semiconductor DPCX in Chap. 4, we have modified the
SiO2 of some substrates with a self-assembled monolayer of OTS prior to the semicon-
ductor deposition, as illustrated in Fig. 3.3(b). The OTS-treatment was initially applied to
organic field-effect transistors by Lin et al. in the year 1997. [113, 114]
56 Experimental details
More generally speaking, self-assembled monolayers (SAM’s) are used as a means
of surface modification for various applications. The molecules have a rather long alkyl
chain and a head and tail group at the respective ends of the alkyl chain. The head group
is chosen so that it can bond to the substrate’s surface. The head groups of the SAM-
molecules attach to the surface to be modified and the van der Waals interaction between
the long alkyl chains pulls the molecular chains up. This process results in highly ordered
monolayers. [115, 116] The alkyl chains are almost perpendicular to the surface of the
substrate and have a thickness of typically 2−3 nm, depending on the length of the alkyl
chain. Once formed, the monolayer is very stable. [117–119] The collectivity of the tail
groups forms the new surface. Consequently, the tail group can be chosen according to
the desired properties of the new surface. [116]
In the case of OTS, the chemical structure is CH3(CH2)17SiCl3. We have a silane
group which covalently binds to the substrate, and the methyl groups (-CH3) at the other
end of the alkyl chains form a surface with a low surface free energy. The surface mod-
ification with OTS can be done from solution or by a vapour process. We have used the
latter approach. OTS was purchased from Sigma-Aldrich (≥ 90 %, Prod. No.: 104817).
0.1-0.2 ml of OTS were placed at the closed end of a horizontal glass tube. The cleaned
Si/SiO2 substrates were placed into the glass tube, ≈ 10 cm from the OTS, and then
the glass tube was evacuated. Subsequently, the OTS and the samples were heated to
120−125 C for 1−3 h.
3.3.3 Polymeric gate insulators and polymeric buffer layers
Several polymeric insulators were deposited from solution by spin-coating (Fig. 3.4). The
solution of the polymer was spin-coated either on ITO-coated glass substrates to form a
polymeric gate insulator (Fig. 3.3(c)), or onto Si/SiO2 samples to form a polymeric buffer
layer (Fig. 3.3(b)). The concentration of the solution and the spinning speed were adjusted
to yield the appropriate film thickness. By decreasing the concentration of the solution
or by increasing the rotation speed, thinner films were obtained. Typical rotation speeds
were 500−3000 rpm and a typical spinning time was 30 s. The polymeric films were then
dried on a hotplate. Both the spin-coating and the drying were carried out in air. A typical
thickness was 500 nm for a polymeric gate dielectric and 10− 30 nm for a polymeric
buffer layer.
The nonfluorinated materials polystyrene (PS), poly(α-methylstyrene) (AMS) and
poly(vinyl alcohol) (PVA) are all commercially available from Sigma-Aldrich (PS: Prod.
3.3 Preparation of the gate dielectric 57
Figure 3.4: Several solution-processable polymers were used as polymeric gate dielectrics orpolymeric buffer layers in organic field-effect transistors. We worked with polymers such as thenonfluorinated materials polystyrene (PS), poly(α-methylstyrene) (AMS) and poly(vinyl alcohol)(PVA) as well as the fluoropolymer CytopT M. All materials are commercially available.
No.: 182427, AMS: 81520 (mol wt. ∼ 100000), PVA: 324590). These materials had
been used to some extend as gate dielectrics or polymeric buffer layers in pentacene-
based thin-film transistors, before we investigated the materials (PS: [90], AMS: [14,66],
PVA: [111]). We were able to reproduce some results including high field-effect mobil-
ities in pentacene thin-film transistors with AMS buffer layers1 and electron transport in
pentacene films on PVA gate dielectrics. However, the devices with AMS buffer layers
or PVA gate dielectrics were unsatisfactory in terms of electrical stability. The devices
had a rather large current hysteresis. Field-effect transistors with a PS gate dielectric
only showed a small current hysteresis in accordance with [90]. The best results were
achieved with the amorphous fluoropolymer CytopT M (Cytop is a registered trademark
of Asahi Glass and is a short hand for Cyclic Transparent Optical Polymer). CytopT M
was purchased in the dissolved state as a liquid (Cytop CTL-809M, concentration: 9 %)
along with a suitable solvent (CT-Solv.180) for further dilution which is very convenient.
Initially, we purchased the CytopT M from Asahi Glass, Japan and later from Bellex In-
ternational Cooperation, USA. CytopT M had previously been utilized by Veres et al. in
1 Record mobilities around 5 cm2/Vs have been achieved with pentacene thin films on AMS buffer layersby Kelley et al. [14]
58 Experimental details
Figure 3.5: Structure of the fluoropolymer CytopT M (Cyclic Transparent Optical Polymer). [120]This material was extensively studied as polymeric gate insulator in organic field-effect transistors.
combination with solution-processed semiconductors. [61, 62] The structure of the mate-
rial is illustrated in Fig. 3.5. We used CytopT M as gate insulator with vacuum-evaporated
pentacene films and with pentacene and rubrene single crystals. Transistors fabricated
with CytopT M showed a very high electrical stability, combined with high field-effect
mobilities and an excellent subthreshold performance. Therefore, this material was stud-
ied most extensively in field-effect transistors and the results are described in Chap. 5.
3.4 Growth of the semiconductor, electrode depositionand electrical characterization
After having detailed the preparation of the gate dielectric, we now proceed by describing
the vacuum evaporation of organic films and the growth of organic crystals by physi-
cal vapour transport. We then introduce the electrode deposition. Finally, the electrical
characterization of the completed transistors is described.
3.4.1 Evaporation of organic films
The established approach to evaporate organic thin films was to use a home built evapora-
tion chamber. The system was turbo-pumped with a base pressure of 2−3×10−6 mbar.
The samples with the gate insulators were fixed on a sample holder and shadow masks
were fastened on top of the samples. The sample holder was attached to the top lid of
the evaporation chamber. The temperature of the samples during the thin-film deposition
could be adjusted (between≈ 0 C and≈ 140 C) by simply heating or cooling the top lid
with the sample holder from the outside. The organic semiconductor in a molybdenum
boat within the evaporation chamber was resistively heated to temperatures of typically
200− 300 C (pentacene: 230− 260 C) in order to sublime the semiconductor. The
deposition rate was measured with a water-cooled quartz crystal microbalance and was
typically held at 0.3−0.5 Å/s. The nominal film thickness was 50 nm in most cases.
3.4 Growth of the semiconductor and electrical characterization 59
3.4.2 Single crystal growth
Organic single crystals were grown by physical vapour transport. [121] The apparatus
essentially consisted of an inner and an outer horizontal glass tube. The inner glass tube
had a gas inlet on one side and a gas outlet on the other side. A heating wire was wound
around the outer glass tube with an increasing spacing between the windings and thus
produced a temperature gradient. High purity argon was used as the transport gas and was
constantly flowing at a rate of 7−20 ml/min through the inner glass tube during the crystal
growth. Applying the temperature gradient with the heating wire led to the sublimation
of the source material in the inner tube and to the growth of single crystals in a region of
lower temperature. Rubrene powder, for example, was typically heated up to ≈ 290 C,
in order to have a sufficient sublimation rate. The single crystal growth method results in
a spatial separation of the organic semiconductor from lighter and heavier impurities and
is thus expected to result in a further purification.
3.4.3 Electrode deposition
All transistors incorporated vacuum-evaporated gold electrodes. In the established
approach to device fabrication, a turbo-pumped Tectra evaporator (base pressure:
2× 10−6 mbar) was used for the gold deposition. Again, the deposition rate was mon-
itored with a water-cooled quartz crystal microbalance.
All thin-film transistors were fabricated as top-contact devices, i.e. the electrodes
were deposited after the deposition of the organic semiconductor. This is known to lead to
better transistors, compared to devices where the electrodes are evaporated first (bottom-
contact devices). In the latter case, the thin-film growth encounters a step at the gold
electrode resulting in structural disorder in the vicinity of the electrode. [122] For the
deposition of gold electrodes, the samples remained attached to the sample holder after
the deposition of the organic film. The shadow masks for the semiconductor deposition
were replaced by shadow masks for the gold evaporation. The mask exchange was carried
out in air. The deposition rate of the gold was 0.5−1 Å/s and the film thickness typically
was 50 nm for the TFT’s. Fig. 3.6 is a photograph of completed thin-film transistor test
structures.
For the SC-FET’s, the gold electrodes were directly evaporated onto the gate di-
electric through shadow masks. The thickness of the electrodes was 15− 30 nm for the
SC-FET’s and the deposition rate was 0.5− 1 Å/s. The single crystals were then placed
60 Experimental details
Figure 3.6: Top view of thin-film transistor test structures. The use of a shadow mask for thedeposition of the organic semiconductor results in a “masked” organic film. The photograph showsvarious test structures with a channel width of W = 500 µm and a channel lengths of L = 200 µm,L = 150 µm, L = 100 µm and L = 50 µm. Typically, TFT’s with a channel length of L = 100 µmwere measured.
Figure 3.7: Device structure of a SC-FET with a polymeric gate dielectric. First, the contactswere formed (bottom-contact devices) and, afterwards, the separately grown single crystals wereplaced on the electrodes.
onto the electrodes in air and were held in place by electrostatic forces. The bottom
contacts are thought to lead to a superior device performance in the case of SC-FET’s,
because the charge is injected close to the active channel. [123] In the case of top-contact
SC-FET’s, the charge would have to pass through the crystals (e.g. 500 nm thick) and this
might result in a large access resistance. [123] The structure of a SC-FET with a glass/ITO
substrate and an organic dielectric is shown in Fig. 3.7.
3.4.4 Electrical characterization of transistors
The completed samples were quickly transported through air to a prober station in a
glovebox with a dry He atmosphere. We used a HP 4155A semiconductor parameter
analyzer for all electrical measurements. The parameter analyzer is equipped with four
source/monitor units (SMU, ±100 V) and with a ground connection. For convenience,
3.5 Additional thin-film characterization 61
the HP 4155A was controlled by means of a computer program which also allows for the
automatic measurement of test sequences. [112]
The transistors were generally characterized by measuring transfer characteristics
both in the linear regime (with a low drain voltage Vd) and in the saturation regime (with
a high drain voltage). Additionally, output characteristics were measured. These gated
two-terminal measurements were done by connecting three of the SMU’s to the source,
drain and to the gate of the transistor. The SMU’s were operated in the “force voltage -
measure current” mode.
The transfer characteristic in the saturation regime was generally used to estimate
the mobility µ0 and the threshold voltage Vt with Eq. 2.35. Other device parameters
including the onset voltage Von and the subthreshold swing S were extracted from the
transfer characteristic as well (see Sec. 2.5.2). Output characteristics were measured in
order to confirm the transistor operation and to quantify to which extend the device is
influenced by parasitic contact resistances.
Typically, both the forward and the reverse sweeps were measured. The difference
between the forward and the reverse sweeps gives a first indication of the electrical sta-
bility of the transistor. In some cases, gate bias stress studies were carried out to further
elucidate the electrical stability. A gate bias was applied for an extended period of time
while the source was grounded and the drain potential was held at 0 V. The transfer char-
acteristic measured after the gate bias stress period is often shifted, compared to a trans-
fer characteristic measured prior to the stress period. The magnitude of the shift of the
transfer characteristic in gate bias stress experiments is a measure of the device stability
(Sec. 2.5.4).
3.5 Additional thin-film characterization
Occasionally, we used additional techniques to characterize the gate insulator and the
evaporated organic semiconductor film and these methods are described in the following.
3.5.1 Static water contact angles
Static water contact angles on the surface of the gate dielectric were measured by carefully
placing a small droplet of ultrapure water on the surface of the sample. A photograph of
the droplet was taken immediately after placing the droplet. As an example, Fig. 3.8
shows a photograph of a water droplet on a CytopT M gate insulator. The water contact
62 Experimental details
Figure 3.8: Photograph of a water droplet on a highly hydrophobic amorphous fluoropolymerlayer (CytopT M). The static water contact angle is α = 112. In general, the water contact angleα is a function of the interfacial tensions/energies of all involved interfaces, i.e. substrate-air (sa),substrate-water (sw) and water-air (wa).
angle α was estimated with the photograph. The contact angle results from the interplay of
three different interfacial tensions/energies, i.e. σsa (substrate-air), σsw (substrate-water)
and σwa (water-air). [124] From Fig. 3.8 we see that
cos(α) =σsa−σsw
σwa≈ σs−σsw
σw. (3.1)
The interfacial tension σsa (substrate-air) can be approximated with the surface tension of
the substrate σs and similarly σwa ≈ σw, where σw is the surface tension of water.2
On the one hand, we measured contact angles in order to check the quality of the
SAM’s of OTS. The quality of a SAM critically depends on many factors, including the
cleaning procedure of the substrate, and is reflected in the static water contact angle: the
larger the contact angle is, the better the SAM. Contact angles of α = 90−95 were gen-
erally achieved on OTS-treated SiO2 with the vapour process. Sometimes, significantly
lower contact angles were measured and such samples were discarded. On the other
hand, the method was used to characterize the surface of polymeric gate dielectrics and
polymeric buffer layers. The contact angles on polymeric gate dielectrics and polymeric
buffer layers were more reproducible than on OTS and showed only small deviations from
2 It should be kept in mind that, in practice, the measured contact angle does not only depend on interfacialtensions. It is also affected by the roughness of the surface. A decrease in roughness should cause adecrease in contact angle.
3.5 Additional thin-film characterization 63
Table 3.1: Typical static water contact angles, as measured on OTS and on polymeric layers thatwere used as gate insulators or polymeric buffer layers.
Surface Static water contact angle
OTS 90-95
AMS 95
PVA water-soluble
PS 95
CytopT M 112
a mean value. Typical water contact angles from our measurements are summarized in
Table 3.1. In Chap. 2 we have seen that a high water contact angle is indicative of a non-
polar and water repellent surface which reduces the density of traps in the interface region
of an organic field-effect transistor.
3.5.2 Atomic force microscopy
In Chap. 2 we also mentioned that the roughness of the gate dielectric influences the
thin-film growth and thus the structure and quality of evaporated films. The roughness
of the polymeric gate insulators was occasionally determined by tapping mode atomic
force microscopy (AFM). Tapping mode AFM is suitable for soft polymeric surfaces that
are easily damaged, because the tip is not dragged over the surface as in contact mode. In
order to measure the surface roughness, AFM images with a scan size of 4 µm×4 µm were
recorded and the root mean square (RMS) roughness was calculated from the images. An
Asylum Research AFM located in the cleanroom of the FIRST Center for Micro- and
Nanoscience was used in order to obtain high quality AFM images.
3.5.3 X-ray diffraction
The structure of evaporated organic films was sometimes investigated with X-ray diffrac-
tion in the Bragg-Brentano geometry (Θ− 2Θ-mode). This was done with an in-house
Stoe Stadi P diffractometer. In the Θ−2Θ-mode, the angle of the incident X-rays is equal
to the angle of the scattered and detected radiation. As a consequence, the scatter vec-
tor K is perpendicular to the surface of the sample and the Laue condition K = G can
only be fulfilled with a reciprocal lattice vector G that is perpendicular to the surface of
the sample. Therefore, measurements in the Bragg-Brentano geometry can only reveal
64 Experimental details
information about lattice planes that are parallel to the surface of the sample. We recall
that evaporated organic films, such as pentacene films, often have a polycrystalline and
layered structure. The molecules in the layers are almost perpendicular to the substrate
and the layers are parallel to the substrate. It should be kept in mind, however, that X-ray
diffraction measurements give information about the whole film (typically 50 nm thick).
On the contrary, the charge transport in a field-effect transistor takes place in the first few
molecular layers. For pentacene, the thickness of one molecular layer is ≈ 1.5 nm.
3.5.4 Surface step profiling
For samples with a polymeric gate insulator (Fig. 3.3), the thickness of the gate insulator
was measured with a Tencor Alpha-Step 500 at the FIRST Center. This was done by
scratching the soft polymer film with plastic tweezers. A profile of the scratch was mea-
sured and the measured thickness l was used to calculate the gate capacitance per unit area
Ci = ε0εi/l. This value of Ci was used for the calculation of the field-effect mobilities.
3.5.5 Leakage current and capacitance measurements
Leakage current and capacitance measurements were sometimes carried out to further
characterize the gate insulator. These measurements were done with an open-air prober
station and respectively an Agilent 4339B high resistance meter and an Agilent 4192A
impedance analyzer. For these measurements, circular gold electrodes were evaporated
onto the gate insulator and the area of the circular gold electrodes was 0.985 mm2.
3.6 Advanced fabrication and characterization of thin-film transistors
Previously, we have described the established approach to fabricate and characterize or-
ganic field-effect transistors. These experimental details are relevant for the studies in
Chap. 4 and Chap. 5. The results in Chap. 6 and Chap. 7 were obtained with a “device fab-
rication and characterization system” which allows for measurements of organic thin-film
transistors under highly controlled conditions, e.g. without exposing the samples to am-
bient air between the transistor fabrication and the electrical characterization. Instead of
the commonly employed gated two-terminal measurements, we used gated four-terminal
measurements in these studies to rule out the effect of parasitic contact resistances. In
3.6 Advanced fabrication and characterization of thin-film transistors 65
addition, the system contains a cryostat and allows for temperature-dependent measure-
ments of the transistors. Finally, we developed and used a scheme for the parameter
extraction that is suitable for samples with an increased trap density and sufficiently good
contacts.
3.6.1 Device fabrication and characterization system
At the beginning of our work for this thesis, the laboratory had been equipped with a
new evaporation chamber purchased from Kurt J. Lesker company, Great Britain. The
chamber had a base pressure of 10−8 mbar. It allowed for the thermal evaporation of or-
ganic semiconductors and metals. A mask-positioning mechanism had been incorporated
in the system, in order to deposit pentacene films and gold electrodes without breaking
the high vacuum in between the two deposition steps. We designed a cryogenic vacuum
prober station with the aim to attach this prober station to the existing evaporation cham-
ber and to eventually characterize organic thin-film transistors without any exposure to
air between the deposition of the semiconductor, the deposition of the contacts and the
electrical characterization. This project was very promising, because organic semicon-
ductors were thought to be sensitive to ambient gases including water vapour [105, 125]
and oxygen [126–130]. Organic field-effect transistors, in general, had most often been
characterized after a transfer of the samples to the measurement chamber through air.
Electrical characterization without air exposure is still very rare. [131, 132] More specifi-
cally, the goal of the experiments with the device fabrication and characterization system
was to learn about the microscopic origin of trap states in organic field-effect transistors
by exploiting the high degree of control this setup has to offer. The challenges in the
design of the prober station were:
1) the realization of the sample transfer from the evaporation chamber to the cryostat in
the prober station without breaking the high vacuum,
2) the development of a mechanism which assures that the samples reach sufficiently low
temperatures when cooling down the cryostat in the prober station and
3) the reliable measurement of the temperature of the samples without breaking the high
vacuum.
The prober station was built by CryoVac, Germany. Drawings of the completed system
are shown in Fig. 3.9. Fig. 3.10 is a photograph of the system.
66 Experimental details
(a)
Evaporation chamber
Gate valve 1
Transfer rod 2
Load lock
Transfer rod 1
Gate valve 2
Prober station
(b)
Load lock
Cryostat
Sample holder / samples
Micro-Prober
Microscope
Evaporation chamber
Prober station
Figure 3.9: Schematic drawings of the device fabrication and characterization system: top view(a) and side view (b). The system essentially consists of an evaporation chamber, a cryogenicprober station and a load lock. It allows for the fabrication and electrical characterization oforganic thin-film transistors without breaking the high vacuum of the order of 10−8 mbar.
3.6 Advanced fabrication and characterization of thin-film transistors 67
Figure 3.10: Photograph of the device fabrication and characterization system.
For this thesis, the device fabrication and characterization system was exclusively
used with pentacene and gold. The substrates (e.g. highly doped Si/SiO2 samples) were
mounted on a sample holder (Fig. 3.11) and were introduced into the device fabrication
and characterization system via the load lock (Fig. 3.9(a)). The base pressure of the sys-
tem was of the order of 10−8 mbar. The substrates were introduced into the evaporation
chamber with transfer rod 1 (Fig. 3.9(a)) in order to evaporate pentacene and gold. After
the completion of the device fabrication, the samples were transported to the prober sta-
tion with transfer rod 1 and 2 (Fig. 3.9(a)). Transfer rod 2 was equipped with a threaded
bar and was screwed into the sample holder. The sample holder was then pushed under
the clamps on the cryostat in the prober station with transfer rod 2 (Fig. 3.12(a)). In the
prober station we carried out the characterization of the transistors.
3.6.2 Electrical characterization by gated four-terminal measure-ments
The prober station is equipped with five micro-probers for the electrical characteriza-
tion. By means of an electrical feedthrough to the cryostat, a gate bias could be applied
to the transistor test structures. In order to measure the temperature on the surface of
the samples, a thermocouple was attached to one of the five micro-probers as shown in
68 Experimental details
(a) (b)
Figure 3.11: Top view (a) and side view (b) of the sample holder with four samples.
Fig. 3.12(b). The thermocouple was carefully pressed against the surface of the sample
with the micro-prober at each temperature.
Gated four-terminal measurements have the advantage that the measurements can
be corrected for the influence of parasitic contact resistances at the source and at the
drain. The contact-corrected channel conductivity as a function of gate voltage σ(Vg) can
be used to calculate the contact-corrected field-effect mobility. The gated four-terminal
measurement also allows for an estimation of the device contact resistance. Four-terminal
measurements have previously been used to estimate the field-effect mobility and the
contact resistance in TFT’s [133, 134] and SC-FET’s [135–137].
The transistor test structure for the gated four-terminal measurements is schemati-
cally shown in Fig. 3.13. The transistors consisted of a well-defined stripe of pentacene (a
“masked” pentacene film) and had two voltage sensing electrodes with little overlap to the
pentacene film. For the electrical measurements we used a HP 4155A semiconductor pa-
rameter analyzer. In principle, the gated four-terminal measurement is the measurement
of a transfer characteristic and, in addition, the potentials V1(Vg) and V2(Vg) are measured.
For each applied gate voltage Vg, the drain current Id and the potentials V1 and V2 between
the grounded source electrode and the respective voltage sensing electrode were measured
while keeping the source-drain voltage constant. This was done by connecting the source
of the transistor to the ground connector of the HP 4155A and by measuring the channel
potentials V1 and V2 with two SMU’s in the “source current - measure voltage” mode with
a sourced current of 0 A. The other two SMU’s were connected to the gate and the drain,
and were used as for the gated two-terminal measurements. The drain voltage was gener-
ally held at Vd = −2 V. We should point out that the gated four-terminal measurement is
most useful if measured with a low drain voltage Vd . In that case we can assume a linear
voltage drop all along the transistor channel. [138]
3.6 Advanced fabrication and characterization of thin-film transistors 69
(a)
(b)
Figure 3.12: (a) Top view of the opened prober station. The sample holder (left) was pushedunder the clamps on the cryostat with transfer rod 2. The prober station is equipped with fivemicro-probers. The prober arms were connected to the cryostat with thick copper braids and werethus cooled when the cryostat and the sample holder with the samples were cooled down. (b)Four micro-probers were used for the four-terminal measurements. A thermocouple was attachedto one of the micro-probers and could thus be pressed against the surface of the samples. Thisallowed for the reliable measurement of the temperature in temperature-dependent measurements.
3.6.3 Parameter extraction
Since transistor characteristics are expected to critically depend on trap states, the field-
effect transistor can be used as a tool to extract the underlying spectral density of “fast”
traps. This has been extensively done with thin-film transistors employing amorphous
semiconductors (e.g. [139–147]) or polycrystalline silicon as the semiconductor (e.g.
70 Experimental details
Figure 3.13: Transistor test structure for the gated four-terminal measurements. The alignmentof the electrodes with respect to the pentacene was achieved by means of a high precision maskpositioning mechanism. The channel length and width were L = 450 µm and W = 1000 µm. Thedistance between the voltage sensing electrodes was L′ = 300 µm. With the voltage sensing elec-trodes, the potentials V1 and V2 were measured.
[148–151]). The method is emerging in the field of organic transistors as well. On the
one hand, a density of states function can be postulated a priori and the corresponding
transistor characteristic can be calculated by means of a suitable device simulation pro-
gram. The density of states function is then iteratively refined until, after a number of
predictor-corrector loops, good agreement between the measured characteristic and the
simulated curve is achieved. [19, 152–157] On the other hand, the density of states func-
tion can be calculated from the linear regime transfer characteristics in a straightforward
fashion. [52, 53, 128, 158–162] The straightforward approach has the advantage of giv-
ing an unambiguous result but, depending on the complexity of the extraction scheme,
spurious errors may result from simplifying assumptions. We underline that, although
contact effects can be severe in organic field-effect transistors, the influence of the contact
resistance is almost always neglected when extracting the trap DOS.
In Chap. 6 we used an extraction scheme for the interpretation of experimen-
tal results which was originally developed for amorphous inorganic semiconductors by
Grünewald et al. [140, 142, 145] We adapted this scheme to gated four-terminal mea-
surements in order to correct for the contact resistance. The original scheme neglects
the influence of parasitic contact resistances and had been used with gated two-terminal
measurements only. The scheme allows for the calculation of the trap DOS from a trans-
fer characteristic/gated four-terminal measurement obtained at a single temperature (e.g.
room temperature). It is therefore particularly suitable for the experimental study in
Chap. 6, since metastable trap states are under investigation. The scheme rests on the
assumption of trap-controlled transport above a mobility edge or in a transport level (see
3.6 Advanced fabrication and characterization of thin-film transistors 71
Sec. 2.3). For Chap. 7 we developed an improved scheme to extract the trap DOS from
temperature-dependent gated four-terminal measurements. This new scheme
1) rests on a minimal set of simplifying assumptions,
2) gives the spectral density of trap states relative to the mobility edge / transport level in
an unambiguous and straightforward fashion,
3) confirms the assumption of trap-controlled transport above a mobility edge or in a
transport level and
4) provides information about other important transport parameters including the “intrin-
sic mobility”.
This scheme was used to study the influence of oxygen on the trap DOS with the device
fabrication and characterization system (Chap. 7).
4 Quinoid heteropentacenes aspromising organic semiconductors forfield-effect transistor applications
In this chapter we describe experiments with a quinoid heteropentacene as p-type semi-
conductor in organic field-effect transistors. Both single crystal and thin-film transistors
were fabricated with 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX). In this new small
molecule organic semiconductor the field-effect mobility is as high as 0.16 cm2/Vs in
single-crystal devices and 0.01 cm2/Vs in thin-film devices. In addition, the transistors
show favourable properties such as near zero onset/threshold voltages and a small current
hysteresis. X-ray diffraction experiments show the molecules to be arranged in slipped
stacks and to have a flat backbone in the crystals. For thin films of DPCX the situation
is complicated by the co-existence of a thin-film phase with the bulk phase. However, a
comparison of DPCX thin films on octadecyltrichlorosilane (OTS)-treated and bare SiO2
gate dielectrics provides clear evidence that the OTS surface treatment leads to organic
thin films with a better structural order. The low-cost synthesis and purification of DPCX
along with the improved processability and the good electrical characteristics suggest that
quinoid heteropentacenes are promising materials for organic field-effect transistors.1
1 The studies described in this chapter contributed to patent application WO/2007/118779and are published inW. L. Kalb, A. F. Stassen, B. Batlogg, U. Berens, B. Schmidhalter, F. Bienewald, A. Hafner, T. Wagner,J. Appl. Phys. 105, 043705 (2009).
74 Quinoid heteropentacenes as promising organic semiconductors
4.1 Introduction
Apart from a high field-effect mobility, other requirements such as a near zero threshold
voltage, a steep subthreshold swing, a low current at zero applied gate bias and a high
electrical and environmental stability are important for practical applications. Moreover,
the organic transistors should be inexpensive, i.e. low-cost synthesis, purification and
processing of all involved materials are required.
The semiconductor material plays an important role for the performance of an or-
ganic field-effect transistor. An advantage of organic semiconductors is that their prop-
erties can be adjusted by means of synthetic organic chemistry. Consequently, intense
research efforts are currently being undertaken to synthesize new organic semiconduc-
tors. [37, 163–165] Pentacene derivatives and heteropentacenes are promising classes of
materials and are synthesized with the aim to mimic the excellent transport properties of
the benchmark material pentacene and to also have improved properties, such as a better
stability or solubility. [18]
We have seen that the subthreshold swing and the threshold voltage of a field-effect
transistor are dominated by charge carrier traps close to the interface between the gate
dielectric and the organic semiconductor (Sec. 2.5.2). The effective field-effect mobil-
ity can be affected by many factors such as traps, thin-film morphology or polymor-
phism. [166,167] The overlap of the π-orbitals of adjacent molecules in the crystal struc-
ture certainly plays an important, intrinsic role as well (e.g. Fig. 2.3 or [168, 169]). In
Chap. 2 we have pointed out that reversible and irreversible oxidation of the organic semi-
conductor by water and oxygen radicals are thought to be major causes of electrical and
environmental instability. [106,107,170] The susceptibility of the organic semiconductor
to oxidation can thus lead to low quality device characteristics with a poor subthreshold
performance and a non-zero threshold voltage due to the creation of traps.
Most heteropentacenes synthesized up to now use thiophene units instead of some
of the benzene rings in the pentacene molecule. [164] Other heteropentacenes containing
nitrogen have recently been reported. [171,172] Here we report on the synthesis, transistor
performance and crystal structure of the new oxygen-containing heteropentacene 7,14-
Diphenyl-chromeno[2,3-b]xanthene (DPCX). [173] The material was investigated in both
single crystal and thin-film transistors. It is remarkable that the synthesis of DPCX is
described in a report that was published in the year 1934. [174] In that work, however, a
somewhat different synthetic approach was used.
4.2 Experimental details 75
OH
OH
OH
OH
O
O
(a)
(b)
DPCX
Figure 4.1: Chemical structure of the intermediate 2,5-Bis-(hydroxy-diphenyl-methyl)-benzene-1,4-diol (a) and the new organic semiconductor 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX)(b).
4.2 Experimental details
4.2.1 Synthesis of 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX)
The synthesis of the new organic semiconductor was done by Dr. Ulrich Berens together
with Dr. Beat Schmidhalter, Dr. Frank Bienewald and Dr. Andreas Hafner at Ciba Spe-
ciality Chemicals Inc. It rests on the synthesis of the intermediate 2,5-Bis-(hydroxy-
diphenyl-methyl)-benzene-1,4-diol (Fig. 4.1(a)). In order to synthesize this intermediate,
a Schlenk flask was charged with 1,4-bis-(1-ethoxy-ethoxy)-benzene (10.16 g, 40 mmol)
and was flushed with nitrogen. Then, diethyl ether (50 ml), n-BuLi (29.6 ml of a 2.7 N so-
lution in heptane, 80 mmol) and N,N,N’,N’-tetramethyl ethylenediamine (9.3 g, 80 mmol)
were added. After stirring over night, a solution of benzophenone (14.58 g, 80 mmol) in
diethyl ether (in total 50 ml) was added dropwise. A greenish blue solution formed and
was stirred for another 45 min. Solvents were removed and the residue was treated with
a mixture of ether (≈ 150 ml) and an ammonium chloride solution (ca. 100 ml of a 10 %
solution). The clear organic layer was separated and the solvent was removed, leaving
26.3 g of oil. After being re-dissolved in methanol (ca. 200 ml) and, after an addition of
≈ 0.5 ml of 36 % HCl, 10.55 g of 2,5-bis-(hydroxy-diphenyl-methyl)-benzene-1,4-diol, a
colourless solid, formed over night.
2,5-bis-(hydroxy-diphenyl-methyl)-benzene-1,4-diol (9.5 g) was added to 50 g of ni-
trobenzene. The obtained suspension was heated at reflux for ≈ 45 min, and was then
allowed to cool over night. The formed crystals were filtered off, washed several times
76 Quinoid heteropentacenes as promising organic semiconductors
with ethanol and dried at 10−3 mbar/190 C for two hours, giving 4.47 g of greenish red
crystals. Elementary analysis found C: 87.62 % and 87.78 % (calculated: 88.05 %); H:
4.29 % and 4.65 % (calculated: 4.62 %). The chemical structure of the resulting 7,14-
Diphenyl-chromeno[2,3-b]xanthene is shown in Fig. 4.1(b).
4.2.2 Device fabrication
Single crystal field-effect transistors (SC-FET’s)
Single crystals of DPCX were grown by physical vapour transport in a horizontal oven
with argon as the inert carrier gas (see Sec. 3.4.2). A temperature gradient was applied,
resulting in the evaporation of DPCX at 295 C and crystallization between 270 C and
240 C.
Heavily-doped Si wafers with 300 nm thick thermally grown SiO2 were used as
gate electrode and gate insulator. The samples were cleaned with hot solvents and pi-
ranha solution (30 % hydrogen peroxide in 70 % sulfuric acid). Afterwards, the substrates
were thoroughly washed with ultra pure water (18.2 MΩcm) and, subsequently, the SiO2
was treated with octadecyltrichlorosilane (OTS). [113] This was done by exposing the
substrates to OTS vapour in vacuum at 120 C for 1 h (Sec. 3.3.2). Static water contact
angles on OTS-treated substrates were 90-95 (see Sec. 3.5.1). 18 nm thick gold source
and drain electrodes were evaporated through shadow masks in high vacuum and, finally,
the single crystal field-effect transistors (SC-FET’s) were completed by placing the DPCX
single crystals on the gold electrodes in air. [135, 136]
Thin-film transistors (TFT’s)
As for the SC-FET’s, heavily doped Si-wafers with a 300 nm thick SiO2 layer were used
for the thin-film transistors (TFT’s). The substrates were cleaned with solvents and pi-
ranha then and were eventually washed with ultra pure water. Subsequently, half of the
substrates were treated with OTS with the vapour process. DPCX was vacuum-evaporated
on both types of surfaces (bare SiO2 and OTS-treated SiO2) in the same deposition run.
Shadow masks were used and the base pressure was ≈ 2× 10−6 mbar. The temperature
of the substrates Tsub during the deposition was kept constant at various temperatures be-
tween 0 C and 135 C. The deposition rate and nominal film thickness were 0.5 Å/s and
50 nm, as measured with a water-cooled quartz crystal in the deposition chamber. Gold
4.2 Experimental details 77
source and drain electrodes were vacuum-evaporated onto the DPCX thin films in a sep-
arate high vacuum chamber resulting in multiple transistor test structures with a channel
length of L = 100 µm and a channel width of W = 500 µm. More details on the evapora-
tion of organic films and gold electrodes can be found in Sec. 3.4.1 and Sec. 3.4.3.
4.2.3 Electrical characterization
All electrical measurements were done with a HP 4155A semiconductor parameter ana-
lyzer in a dry He atmosphere. The SC-FET’s and TFT’s were characterized by measuring
both transfer and output characteristics.
According to common practice, the mobility µ0 was estimated with Eq. 2.35, i.e. by
a linear regression of the square root of the drain current Id measured in the saturation
regime (Vd = −50 V). Eq. 2.35 implicitly assumes that, above a threshold voltage Vt , all
of the incrementally added, gate-induced charge is free and moves with the “trap-free”
mobility µ0 (see Sec. 2.5.2). We have pointed out in Sec. 2.5.3 that this assumption is
often not valid. In this case, µ0 may be interpreted as an average effective mobility.
In addition to the threshold voltage Vt , we extracted the onset voltage Von, which
is defined as the gate voltage where the drain current exceeds the noise level (typically
10−12 A). Moreover, the on-off current ratio Ion/Io f f was extracted and is given as an
order of magnitude. Finally, the subthreshold swing was determined as a measure of how
easily a device can be turned from the off-state to the on-state. The subthreshold swing is
defined by Eq. 2.37. S depends on the gate capacitance Ci and the figure of merit is the
normalized subthreshold swing CiS.
4.2.4 X-ray diffraction
Single crystal diffraction data were collected at 100 K with a Bruker AXS SMART 6000
CCD detector on a three-circle platform goniometer with graphite-monochromatized
Cu(Kα) radiation (λ = 1.54178 Å) from a sealed-tube generator.
X-ray diffraction patterns of DPCX thin films were obtained in the Θ−2Θ geometry
with a Stoe Stadi P diffractometer and Cu(Kα) radiation.
78 Quinoid heteropentacenes as promising organic semiconductors
Table 4.1: Mobility µ0, subthreshold swing S, normalized subthreshold swing CiS, onset voltageVon, threshold voltage Vt and on-off current ratio Ion/Io f f from the devices in Fig. 4.2. For theTFT’s, the OTS-treatment leads to a drastic increase in mobility (factor of ≈ 500). This largeeffect is due to a different growth of the vacuum-evaporated DPCX thin films on the differentsurfaces.
Device µ0 S CiS Von Vt Ion/Io f f
(cm2/Vs) (V/dec) (nF V/(dec cm2)) (V) (V)
SC-FET (OTS) 0.16 1.3 15.0 +1 +1 106
TFT (OTS) 0.01 1.9 21.9 -1 -4 105
TFT (bare SiO2) 2×10−5 20.7 238.3 -13 −12 103
4.3 Results and discussion
4.3.1 Electrical characterization
In Fig. 4.2 we show transfer characteristics from a DPCX SC-FET and TFT both with
OTS-treated SiO2 gate dielectric and from a TFT with a bare SiO2 gate dielectric. The de-
vice parameters, as extracted from these measurements, are summarized in Table 4.1. The
DPCX thin films in Fig. 4.2 were evaporated at a substrate temperature of Tsub = 0 C for
which the best mobilities were obtained. The field-effect mobilities are µ = 0.16 cm2/Vs
for the SC-FET and 0.01 cm2/Vs for the TFT with OTS. Moreover, the on-off ratios are
respectively 106 and 105 for the SC-FET and TFT. Fig. 4.2 and Table 4.1 reveal a signif-
icant advantage of DPCX over some other organic semiconductors: the material leads to
devices with near zero onset/threshold voltages and a small current hysteresis. The sub-
threshold swing S normalized by the gate capacitance Ci = 11.51 nF/cm2 is CiS = 15.0 nF
V/(dec cm2) and CiS = 21.9 nF V/(dec cm2) respectively for the SC-FET and TFT with
OTS. As a consequence of the near zero onset/threshold voltage and the good subthresh-
old swing, the devices are turned off when no gate bias is applied and can be easily turned
on. This enables a low power consumption if the devices are to be employed in com-
plementary circuits. The near zero onset voltages of DPCX-based devices are in contrast
to some other organic semiconductors which lead to shifted transfer characteristics, even
when organic buffer layers such as OTS [130] or poly(α-methylstyrene) (AMS, [175]) are
used.
The OTS-treatment has a drastic influence on the performance of the thin-
film transistors. The field-effect mobility increases from µ0 = 2× 10−5 cm2/Vs to
µ0 = 0.01 cm2/Vs, i.e. by a factor of ≈ 500. X-ray diffraction experiments show a better
4.3 Results and discussion 79
0 -20 -40 -60 -8010-12
10-11
10-10
10-9
10-8
10-7
10-6
10-5
Vd = -50 V
TFT (bare SiO2)
TFT (OTS)
SC-FET (OTS)
| Id |
(A)
Vg (V)
Figure 4.2: Transfer characteristic from a DPCX single-crystal transistor (SC-FET) with OTS-treated SiO2 (µ0 = 0.16 cm2/Vs). Also shown are the transfer characteristics from a DPCXthin-film transistor (TFT) with OTS (µ0 = 0.01 cm2/Vs) and with a bare SiO2 gate dielectric(µ0 = 2× 10−5 cm2/Vs). Remarkable is the near zero onset voltage Von in the case of the OTS-treated substrates and the small current hysteresis. Depending on the type of device, the field-effectmobility spans five orders of magnitude. This highlights the importance of structural order to ob-tain high field-effect mobilities.
structural order of the films grown on OTS (see Sec. 4.3.2 of this chapter), compared to
the films grown on bare SiO2. This information and the following discussion demonstrate
that the large effect of the OTS-treatment on the field-effect mobility is due to a better
structural order of the molecules in the films on OTS.
The OTS-treatment reduces the surface free energy of the SiO2 gate dielectric, which
is indicated by the increased water contact angle on the OTS surface (90-95). A low
surface free energy is equivalent with a non-polar surface. This may be favourable, since
it decreases the extend of dipolar disorder at the insulator-semiconductor interface. [61]
In addition, a low-energy surface is water repellent. Water may impede the charge trans-
port because of the dipolar character of the water molecules and/or by an electrochemical
reaction with the organic semiconductor. [61, 86] Moreover, the OTS-treatment might
passivate certain chemical groups on the SiO2 surface and this may play a role even in
the case of hole transport. We suggest that the low surface energy results in a reduced in-
teraction between the small molecules and the substrate during thin-film growth and that
this interaction is a cause of structural disorder. Consequently, a smooth surface with a
low surface energy leads to evaporated organic films with improved structural order. This
80 Quinoid heteropentacenes as promising organic semiconductors
effect may be more or less important during the growth of an organic thin film, depending
on the strength of the interaction between the molecules, i.e. depending on the orbital
overlap/charge carrier mobility: for vacuum-deposited pentacene films (typical mobil-
ity 0.1− 1 cm2/Vs), the OTS treatment has a significant influence on the subthreshold
performance but typically leads to an improvement of less than 2 in mobility. [114] For
tetracene thin films (mobilities of ≈ 0.1 cm2/Vs with OTS), the effect of the OTS treat-
ment on the field-effect mobility is larger than for pentacene, and a factor of ≈ 10 has
been reported. [166] The present study (mobilities of ≈ 0.01 cm2/Vs with OTS) reveals a
∼ 500 times better mobility on OTS-treated surfaces than on the bare SiO2.
Single crystal transistors are expected to reveal the ultimate performance of a ma-
terial since they should have the lowest degree of structural disorder. Indeed, the DPCX
SC-FET has an even better subthreshold swing and a higher mobility than the TFT on the
same type of substrate.
Fig. 4.3 shows the output characteristic of the same devices with OTS as in Fig. 4.2.
The output characteristics show clear p-type operation, and deviations from the ideal field-
effect transistor behaviour are minor. Apparently, gold is a suitable material for the source
and drain contacts of a DPCX field-effect transistor.
The substrate temperature Tsub during the deposition of the DPCX thin films has
also a pronounced effect on the field-effect mobility. For a given substrate temperature,
the DPCX was deposited onto a sample with bare SiO2 and with OTS-treated SiO2. Typ-
ically, three devices were measured on each sample and the values for the mobility in
Fig. 4.4 are averages for each sample. The drastic increase in field-effect mobility caused
by the OTS-treatment was observed in all experiments. We measure the highest mo-
bilities on films deposited at the lowest temperatures, which is rather unusual. Heating
the substrates during the vacuum-deposition of small molecule organic semiconductors
often leads to improved mobilities due to a better growth/structural order of the result-
ing films. [71, 72, 167, 176] The unusual temperature dependence in the case of DPCX
is likely attributable to a better connectivity between grains in films deposited at lower
substrate temperatures. This is supported by the fact that, for a substrate temperature of
Tsub = 135 C, there is no film on either type of substrate. The surface is not covered
because all the molecules have re-evaporated from the substrate at Tsub = 135 C. How-
ever, high-quality characteristics can be obtained with DPCX deposited at low substrate
temperatures. This may turn out as a major advantage when flexible plastic substrates are
used.
4.3 Results and discussion 81
0 -20 -40 -600.0
-0.2
-0.4
-0.6
0 -20 -40 -60 -800
-1
-2
-3
-4
-5
-30 V
-50 V
-50 V
I d (A
)
Vd (V)
(b)
(a)
TFT (OTS)
-40 V
Vg = -60 V
-30 V
-40 V
Vg = -60 V
SC-FET (OTS)
I d (A
)
Vd (V)
Figure 4.3: Output characteristics of a DPCX SC-FET (a) and a TFT (b) with OTS-treated gatedielectric (SC-FET: µ0 = 0.16 cm2/Vs, TFT: µ0 = 0.01 cm2/Vs). There are only minor deviationsfrom the ideal transistor behaviour. This indicates gold to be a suitable contact material for thesource and drain electrodes of DPCX devices.
4.3.2 Crystal structure
Crystallographic data (excluding structure factors) have been deposited with the Cam-
bridge Crystallographic Data Centre as supplementary publication number CCDC
696211.2 Fig. 4.5 shows the crystal structure of DPCX. In contrast to pentacene, which
is known to crystallize in a herringbone structure (e.g. Fig. 2.2 in Chap. 2), the molecules
form slipped stacks along the b-axis. The space group is C2/c with a = 18.961(5) Å,
b = 6.058(2) Å, c = 20.229(5) Å, β = 116.351(9) and Z = 4. The backbone of DPCX
2 Copies of the data can be obtained free of charge on application to CCDC, 12 Union Road, CambridgeCB2 1EZ, UK [fax (+44) 1223 336033, email: [email protected]].
82 Quinoid heteropentacenes as promising organic semiconductors
0 20 40 60 80 100 120 1400
5x10-3
1x10-2
0 20 40 60 80 100 120 140
0
1x10-5
2x10-5
3x10-5
4x10-5
with OTS
bare SiO2
(cm
2 /Vs)
no filmno film
TSub
(°C) (cm
2 /Vs)
TSub (°C)
bare SiO2
no film
Figure 4.4: Influence of the substrate temperature Tsub during the deposition of DPCX on thefield-effect mobility. Red squares: TFT’s with OTS-treatment; blue circles: TFT’s with bare SiO2.The OTS treatment reproducibly leads to a drastic increase in field-effect mobility. Interestingly,the highest mobilities are obtained for the lowest substrate temperatures.
in the crystal is flat, and the dihedral angle between the phenyl ring and the backbone is
70.22.
The oxygen atoms in the DPCX molecules disrupt the π-electronic system; it is sur-
prising that mobilities in excess of 0.1 cm2/Vs can be achieved with DPCX. A theoretical
approach to the electronic structure of the material may help to understand whether the
π-electrons from the phenyl rings and possibly the lone electron pairs at the oxygen atoms
contribute to the charge conduction.
In Fig. 4.6 we show X-ray diffraction patterns from 50 nm thick DPCX films grown
in the same deposition run at a substrate temperature of Tsub = 0 C, both on OTS-treated
and bare SiO2. For the OTS-treated sample, two peaks can clearly be observed, indicating
a polycrystalline nature of the DPCX. For the untreated SiO2, no peaks can be discerned.
The film is amorphous.
Importantly, the X-rays are scattered from the whole film, while in field-effect tran-
sistors the current flow is located close to the insulator-semiconductor interface. However,
if no peaks are present in a diffraction pattern, it is reasonable to conclude that also the
semiconductor very close to the insulator-semiconductor interface is amorphous. The
X-ray diffraction scans therefore confirm a higher degree of structural disorder in the in-
terface region of DPCX films on bare SiO2 accounting for the low field-effect mobility
without OTS.
4.3 Results and discussion 83
Figure 4.5: Crystal structure of DPCX. The compound crystallizes in the monoclinic space groupC2/c with a = 18.961(5) Å, b = 6.058(2) Å, c = 20.229(5) Å and β = 116.351(9), Z = 4. Themolecules are ordered in slipped stacks along the crystallographic b-axis. The backbone of DPCXin the crystal is flat, and the dihedral angle between the phenyl ring and the backbone is 70.22.
5 10 15 20 25 30 35
500
1000
1500
2000
2500 OTS bare SiO2
Subtrate
Single crystalphase (004)
Thin-filmphase
Single crystalphase (002)
Thin-filmphase
Cou
nts
2Theta (°)
Figure 4.6: X-ray diffraction patterns from DPCX thin films grown on OTS-treated SiO2 (redline) and on bare SiO2 (black line). The films were deposited in the same deposition run. Thefilm on OTS is polycrystalline, while the film on bare SiO2 is amorphous. For the OTS-treatedsamples, the single crystal phase coexists with a thin-film phase.
84 Quinoid heteropentacenes as promising organic semiconductors
The interlayer spacings calculated from the two peaks at 2θ1 = 7.616 and
2θ2 = 9.649 are d = 11.6 Å and d = 9.16 Å. Upon close inspection, a higher order of
these two peaks can also be observed in Fig. 4.6. One of the peaks (and its higher order)
can be related to the single crystal structure. From the structure determination we have
ccos(β−90) = 18.13 Å ≈ 2×9.07 Å. The second peak in Fig. 4.6 therefore is the (002)
peak from the single crystal phase. The first peak cannot be related to the single crystal
structure. One would conclude that the single crystal phase coexists with a phase which
is exclusively present in DPCX thin films. The ratio of the two peaks indicates that the
thin-film phase even dominates. As a matter of fact, polymorphism is a common feature
of organic semiconductors due to the weak van der Waals type interaction between the
molecules. [31, 32, 167]
4.4 Conclusions
The quinoid heteropentacene 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX) was syn-
thesized. Single crystal and thin-film transistors were fabricated with this material.
DPCX is a p-type organic semiconductor and field-effect mobilities of 0.16 cm2/Vs were
achieved so far. The transistor characteristics reveal near zero onset/threshold voltages,
low off-currents at zero applied gate bias and a small current hysteresis.
For the TFT’s, the treatment of the SiO2 surface with octadecyltrichlorosilane (OTS)
leads to a drastic improvement of the device characteristics. The self-assembled mono-
layer of OTS significantly reduces the growth-related structural disorder in thin films of
DPCX, as seen by X-ray diffraction. Eventually, the best subthreshold swing and field-
effect mobility were obtained with single crystal devices that have the lowest degree of
structural disorder. This highlights the crucial importance of structural defects on the
device performance. The situation is, however, complicated by a non-ideal connectivity
between grains in the case of the thin films. This is revealed by the unusual dependence
of the field-effect mobility on the substrate temperature during the thin-film deposition.
The conjugation of the π-electrons in quinoid heteroacenes is of the quinoid type,
as in oxidized aromatic molecules. Quinoid heteroacenes are therefore expected to be
less susceptible to oxidation. A somewhat similar strategy for the stabilization of organic
semiconductors has very recently been disclosed. [177] Furthermore, both the synthesis
and the purification of DPCX are inexpensive. The phenyl groups of the DPCX molecule
4.4 Conclusions 85
result in an increased solubility if compared to the virtually insoluble pentacene. Further
functionalization at the phenyl rings may open the way for low-cost solution processing.
5 Organic small molecule field-effecttransistors with fluoropolymer gatedielectric: Eliminating gate bias stresseffects
In this chapter we describe organic field-effect transistors with unprecedented resistance
against gate bias stress. The single crystal and thin-film transistors employ an organic
fluoropolymer gate dielectric (CytopTM). CytopT M is easy to be used and can be de-
posited in air from solution. The fluoropolymer is highly water repellent, has a very low
dielectric constant εi = 2.1−2.2 and shows a remarkable electrical breakdown strength.
The single crystal transistors are consistently of very high electrical quality: near zero
onset, very steep subthreshold swing (up to 0.75 nF V/(dec cm2)) and negligible current
hysteresis. Furthermore, extended gate bias stress only leads to marginal changes in the
transfer characteristics. The experimental work described in this chapter suggests that
there is no conceptual limitation for the stability of organic semiconductors in contrast to
hydrogenated amorphous silicon. We also address the issue of contact resistance, which
becomes crucial in high quality organic single crystal devices.1
1 Some of the studies described in this chapter are the basis for patent application WO/2008/077463.The results are published inW. L. Kalb, T. Mathis, S. Haas, A. F. Stassen, B. Batlogg, Appl. Phys. Lett. 90, 092104 (2007) andW. L. Kalb, T. Mathis, S. Haas, A. F. Stassen, B. Batlogg, Proc. of SPIE, 6658, 665807 (2007).
88 Eliminating gate bias stress effects
5.1 Introduction
For convenience, organic field-effect transistor studies often employ Si/SiO2 substrates
(Fig. 3.3(a)). As described in Sec. 3.3.2, the SiO2 surface can be rendered hydrophobic
with the self-assembling agent octadecyltrichlorosilane (OTS). [113,114] The experimen-
tal results in Chap. 4 have shown that the OTS-treatment can have a large effect on the per-
formance of organic thin-film transistors, due to a better growth of the vacuum evaporated
organic film on the low-energy surface. In addition, we have seen that the sole presence of
the gate dielectric is expected to influence the charge transport in a field-effect transistor.
In Sec. 2.4.3 we referred to the chemical structure of the gate dielectric, adsorbed water
and the polarity of the gate dielectric as possible causes of traps in the vicinity of the
insulator-semiconductor interface. Consequently, to fully exploit the potential of organic
semiconductors in field-effect transistors, it is of great importance to carefully choose and
prepare the gate dielectric. In addition, the use of easily processable (organic) gate insula-
tors is mandatory for the implementation of low-cost electronics. That is why the search
for suitable organic dielectrics has intensified in recent years. [61, 62, 90]
One of the last obstacles to be overcome for a commercialization of organic thin-film
transistors is gate bias stress effects. Switching the devices on for some time generally
leads to a reduction in current at a given gate voltage. Gate bias stress effects can result
in a significant difference between the forward and reverse sweep in the measurement of
a transfer/output characteristic. [100, 103, 133] Gate bias stress effects have often been
studied by applying a fixed gate voltage for an extended time, followed by a measurement
of the shift of the transfer characteristic. [152] Possible causes of gate bias stress effects
were discussed in Sec. 2.5.4. The effects are thought to be caused by the trapping and
release of charge carriers on a time scale comparable to the measurement time. [103,108]
Mounting evidence indicates that water in the dielectric-semiconductor interface region
can cause gate bias stress effects. [106,107,178] It has also been suggested that electrical
instability is an intrinsic phenomena caused by the formation of bipolarons. [108, 109]
In this chapter we describe combinations of small molecule organic semiconduc-
tors with an organic spin-on dielectric that yield field-effect transistors with exceptionally
high quality characteristics and stability. The transistors have a bottom gate structure with
an amorphous fluoropolymer (CytopTM) as gate dielectric. The fluoropolymer has a ring
structure in the repeat units (Fig. 3.4 and Fig. 3.5), in contrast to ordinary polytetrafluo-
roethylene (PTFE). This renders the material amorphous and leads to desirable properties
5.2 Experimental 89
such as solubility in some perfluorinated solvents and a high optical transparency. Its
dielectric constant is εi = 2.1− 2.2. [179] CytopTM has previously been used by Veres
et al. in field-effect transistors with a solution processable semiconductor in a top gate
structure. [61, 62] We demonstrate this favorable material in combination with two small
molecule semiconductors: rubrene and pentacene. The device stability was evaluated by
applying a gate bias for extended periods of time.
5.2 Experimental
Organic field-effect transistors were fabricated as follows. Glass slides with an indium tin
oxide coating served respectively as substrate and gate electrode. The fluoropolymer so-
lution was prepared by mixing 3 parts of Cytop CTL-809M with two parts of the solvent
CT-Solv.180. The glass slides were cleaned in hot acetone and hot isopropanol and the
fluoropolymer solution was spin-coated onto the substrates at 500 rpm for 10 s followed
by 1000 rpm for 20 s. Eventually, the samples were dried for 1 h at 90 C on a hotplate. All
the processing steps were carried out in air. The thickness of the layers was determined
with a surface step profiler (Tencor Alpha-Step 500) and the gate capacitance was calcu-
lated from these measurements, assuming a dielectric constant of εi = 2.15. In addition,
the dielectric films were characterized by atomic force microscopy (surface roughness),
static water contact angle measurements and leakage current measurements.
Single crystal field-effect transistors (SC-FET’s) were made by evaporating 30 nm
thick gold electrodes onto the fluoropolymer. Rubrene and pentacene single crystals were
grown by physical vapour transport with argon as the inert carrier gas. The single crystals
were then placed onto the gold electrodes in air. [135] The typical channel length of the
completed transistors was L = 50 µm. Pentacene thin-film transistors (TFT’s) were made
by evaporating pentacene onto the fluoropolymer through a shadow mask in high vacuum
and were completed by evaporating gold top contacts. The device structure of the TFT’s
and SC-FET’s is illustrated in Fig. 3.3(c) and Fig. 3.7.
The devices were characterized in a glovebox with a dry He atmosphere (H2O,
O2 < 0.5 ppm) using a HP 4155A semiconductor parameter analyzer. The voltage was
increased in steps of 0.5 V and the integration and delay time were respectively 20 ms
and 0 ms. The devices were characterized by measuring transfer characteristics both in
the linear regime (drain voltage Vd = −5 V) and in the saturation regime (Vd = −80 V).
90 Eliminating gate bias stress effects
Figure 5.1: AFM image (scan size 4 µm×4 µm) of a spin-coated CytopTM film. The surface isrelatively smooth: a root mean square roughness of 0.63 nm was calculated from this image.
Additionally, output characteristics were measured. Forward and reverse sweeps were
obtained in all cases.
We also measured the temperature dependence of the transfer characteristics in some
cases. This was done by slowly cooling a sample down from room temperature and by
measuring a transfer characteristic about every 10 K.
5.3 Results
5.3.1 Properties of the gate dielectric
The CytopT M films are from 430 to 700 nm thick, which results in a gate capacitance of
Ci = 4.4−2.7 nF/cm2. The films are highly hydrophobic: static water contact angles are
on average 112 with a very small standard deviation of about 2.2 In Fig. 3.8 we show a
water droplet on a CytopT M surface with a water contact angle of 112. The water contact
angles are comparable or even slightly better than the water contact angles on methyl-
group terminated high quality self-assembled monolayers. [69] Fig. 5.1 shows a typical
AFM topography image of a CytopT M film. The root mean square roughness from this
image and similar images is ∼ 0.6 nm which is rather smooth.
2 A water droplet was placed on six samples and the contact angle was measured on both “sides” of thedroplet. This gave contact angles of (111, 116), (109, 113.5), (110, 114), (111, 112), (112.5,110) and (114, 112). This results in an average contact angle of 112±2.
5.3 Results 91
0 100 200 300 400 50010-13
10-12
10-11
10-10
10-9
10-8
10-7
10-6
0 2 4 6 8 10
10-10
10-9
10-8
10-7
10-6
10-5
10-4
Leak
age
curr
ent d
ensi
ty [A
/cm
2 ]
Applied field [MV/cm]
451V
Leak
age
curr
ent [
A]
Applied voltage [V]
Figure 5.2: The CytopT M fluoropolymer is an outstanding electrical insulator. This 457±10 nmthick film breaks down at 451 V, which corresponds to a dielectric breakdown field of 9.8 MV/cm.The experiment was carried out with an evaporated circular gold electrode, A = 0.985 mm2.
In Fig. 5.2 we show the leakage current through a 457±10 nm thick film as a func-
tion of applied voltage. The measurement was carried out with an evaporated, circular
gold electrode with an area of A = 0.985 mm2. The fluoropolymer is an outstanding elec-
trical insulator: the leakage current is below 6.5×10−11 A for an applied voltage ≤ 70 V
and below 1 µA up to 450 V. The dielectric breaks down at 451 V for this sample, which
corresponds to a dielectric breakdown field of 9.8 MV/cm. This breakdown strength is
remarkable for an organic dielectric and is better than the SiO2 that we generally use.
5.3.2 Comparison of different devices
The excellent performance of organic field-effect transistors with a CytopT M gate dielec-
tric is shown in Fig. 5.3. The transfer characteristics from a rubrene SC-FET, a pen-
tacene SC-FET and a pentacene TFT, measured in saturation with Vd =−80 V, are given
for the forward and the reverse sweep. Most remarkable is the absence of any hystere-
sis for the SC-FET’s. We emphasize that no additional steps, such as electrical aging
or pre-stressing, were taken to obtain these curves. A further mark of the high qual-
ity of the devices is the steep subthreshold swing S (Eq. 2.37). The subthreshold swing
is S = 0.50 V/dec for the rubrene SC-FET and 0.29 V/dec for the pentacene SC-FET.
92 Eliminating gate bias stress effects
20 0 -20 -40 -6010-13
10-11
10-9
10-7
10-5
10-3
Dielectric:amorphousfluoropolymer
Pentacene TFT
Pentacene SC-FET
RubreneSC-FET
|I d| [A
]
Vg [V]
Figure 5.3: Transfer characteristics in saturation (Vd =−80 V) for a rubrene SC-FET, a pentaceneSC-FET and a pentacene TFT. Forward and reverse sweeps are shown but are indistinguishableover the entire operating range for the SC-FET’s. The TFT shows only a small current hysteresisnear the onset.
This gives normalized subthreshold swings CiS of respectively 1.6 nF V/(dec cm2) and
1.3 nF V/(dec cm2). Higher subthreshold swings and a large current hysteresis are mea-
sured when we place nominally identical crystals on OTS surface-treated SiO2. [103,106]
Remarkable as well is the very small (slightly positive) onset voltage of the two single
crystal devices (+3 V for rubrene and +1 V for pentacene). The saturation field-effect mo-
bilities from the crystals as calculated with Eq. 2.35 are 5.7 cm2/Vs and 1.4 cm2/Vs. The
thin-film transistor shows a notable but small hysteresis in the transfer characteristic close
to the onset voltage, and the onset is more negative than in the case of the single crystals,
i.e. −13 V (Fig. 5.3). The field-effect mobility is 0.26 cm2/Vs.3
5.3.3 Reproducibility
We have produced 17 rubrene SC-FET’s. The single crystals were grown in four different
runs and were placed on seven different substrates. Apart from an obviously malfunction-
ing device, 16 transistors had a near zero onset voltage (average: +0.8 V), a very small
current hysteresis and a steep subthreshold swing as low as 0.75 nFV/(dec cm2). The val-
ues are summarized in Table 5.1. A steep subthreshold swing and a near zero onset are
3 The TFT’s were transported through air, but were kept in the dry He atmosphere for several days priorto electrical characterization. The storage led to a slight improvement in hysteresis.
5.3 Results 93
Table 5.1: Onset voltage, subthreshold swing S and normalized subthreshold swing CiS from 16rubrene-based SC-FET’s. The onset voltage is essentially zero in all cases and we achieved arecord subthreshold swing of 0.75 nF V/(dec cm2) with device 16. The values were extracted fromthe saturation regime characteristics.
Device Onset S CiS [nF V/ Device Onset S CiS [nF V/
No. [V] [V/dec] (dec cm2)] No. [V] [V/dec] (dec cm2)]
1 +0.5 0.26 1.1 9 +1.0 0.46 1.3
2 +0.5 0.26 1.1 10 +1.0 0.45 1.3
3 +0.5 0.39 1.7 11 +1.0 0.34 1.0
4 +0.5 0.35 1.5 12 +0.5 0.35 1.0
5 +3.0 0.50 1.6 13 +1.0 0.28 1.1
6 -0.5 0.42 1.3 14 +0.5 0.40 1.6
7 0.0 0.32 1.0 15 +1.5 0.64 2.6
8 +0.5 0.41 1.2 16 +1.0 0.28 0.75
Average value: +0.8 1.3
most favorable since they facilitate a low power operation of the device. In fact, the sub-
threshold swing of the SC-FET’s would be sufficiently steep for a low power operation
(e.g. |Vg| ≤ 10 V) even with the relatively thick fluoropolymer films. We recall that the
theoretical limit for the subthreshold swing is 0.06 V/dec at T = 300 K (Eq. 2.40). We
have achieved a subthreshold swing which is higher by a factor of ≈ 4 only (device 16 in
Table 5.1). The subthreshold swing is a measure of the density of deep traps. According
to Eq. 2.39, a subthreshold swing of S = 0.28 V/dec (device 16) is related to a trap density
of N2 ≈ 5×1010 cm−2eV−1.4 If we assume the thickness of the accumulation layer to be
1.5 nm (approximate thickness of one molecular layer) we have a volume trap density of
N = 3×1017 cm−3eV−1.
5.3.4 Gate bias stress experiments
The advantages of the material combinations become striking in gate bias stress studies.
We have applied a gate voltage to the three devices in Fig. 5.3 for a prolonged time. After
the initial transfer characteristic measurement, a gate bias of Vg =−70 V was applied for
4 The gate capacitance of device 16 is Ci = 2.7 nF/cm2.
94 Eliminating gate bias stress effects
20 0 -20 -40 -601x10-13
1x10-11
1x10-9
1x10-7
1x10-5
1x10-30
3x10-4
6x10-4
2,8 2,6 2,4 2,2
1x10-11
2.8 2.6 2.4 2.2
2x10-11
8x10-12
6x10-12
|I d| [A
]
Vg [V]
Initial Vg = -70 V, 2h Vg = +70 V, 2h
|I d| [A
]
Figure 5.4: The rubrene single crystal device is highly stable against gate bias stress. The figureshows the transfer characteristic measured at Vd = −80 V prior to the stress sequence (full blackline), after two hours of gate bias stress at Vg =−70 V (dashed red line) and after subsequent gatebias stress at Vg = +70 V for two hours (dotted green line) both on a linear scale (upper panel)and on a logarithmic scale (lower panel). The graph includes the forward and reverse sweep in allthree cases. The inset shows the drain currents close to the onset voltage.
two hours. After a two hour relaxation period, a gate bias of Vg = +70 V was applied.
During the stress periods, the source was grounded and the drain potential was held at
0 V to ensure a homogenous gate stress. The drain current quadratically depends on the
effective gate voltage for an ideal transistor (Eq. 2.35) and is very sensitive to changes
induced by a two hour stress period. Bias stress experiments were carried out in the
dark.5
5.3 Results 95
For the rubrene SC-FET, Fig. 5.4 shows the initial characteristic, the characteris-
tic measured after 2 hours of negative bias and after 2 hours of positive bias. The de-
vice is hardly influenced by the long application of a gate bias. There are only marginal
changes in the transfer characteristic. After negative stress, there is a very small shift of
the onset voltage to more positive voltages, accompanied by a small increase in current
hysteresis and a small decrease in on-current. For the pentacene SC-FET, the observa-
tions are similar. When compared to the rubrene device, the shift of the onset voltage
due to bias stress is even smaller, but the decrease in on-current is somewhat more pro-
nounced (3.8 % at Vg =−70 V). In contrast, in similar experiments with single crystals of
rubrene or pentacene on OTS-treated SiO2, large shifts of the transfer characteristics are
observed. [103,106] For the pentacene TFT, a gate voltage of Vg =−70 V applied for two
hours leads to a rigid shift of the curve by −5.2 V to more negative voltages.
5.3.5 Contact effects in SC-FET’s
In Fig. 5.5 we show the transfer characteristics from the rubrene SC-FET of Fig. 5.3 in the
linear regime (Vd = −5 V, upper panel) and in the saturation regime (Vd = −80 V, lower
panel). Also in the linear regime and when plotted on a linear scale, there is no current
hysteresis. The normalized subthreshold swing in the linear regime is even steeper than in
the saturation regime. In the saturation regime, the drain current depends approximately
quadratically on gate voltage (see Fig. 5.5, lower panel, linear scale). This is in accordance
with Eq. 2.35 which is valid for a device with a low trap density and negligible contact
resistances. In the linear regime, however, there is a deviation from the expected linear
current-voltage relationship (Fig. 5.5, upper panel, linear scale). This finding becomes
more apparent in Fig. 5.6, where we plot the same data as in Fig. 5.5 but, for the saturation
regime, the square root of the drain current is shown. In addition, the same plot is given
for the pentacene/CytopTM SC-FET in the lower panel of the graph (saturation field-effect
mobility: 1.4 cm2/Vs). Clearly, the shape of the transfer characteristics is closer to ideal
as the drain current is increased. In the case of the pentacene SC-FET, the characteristic is
very close to ideal in the saturation regime. The output characteristic of the rubrene SC-
FET is shown in Fig. 5.7. The output characteristic further confirms that the deviations
from the ideal transistor behaviour are more pronounced at low Vd . The non-idealities are
5 Both the negative and the positive bias stress period were shortly interrupted (for 70 s) after 10, 20, 40,and 60 minutes in order to measure a transfer characteristic.
96 Eliminating gate bias stress effects
20 0 -20 -40 -60
10-12
10-10
10-8
10-6
10-4
0
2x10-4
4x10-4
6x10-4
8x10-4
10-12
10-10
10-8
10-6
10-4
0
2x10-5
4x10-5
6x10-5
Vg [V]
|I d| [A
]
Vd = -80V
Vd = -5V
|I d| [A
]
Figure 5.5: Transfer characteristics from the rubrene SC-FET. The upper panel shows the linearregime transfer characteristic (Vd =−5 V) and the lower panel the saturation regime characteristic(Vd = −80 V) both on a linear and on logarithmic scale. Also when measured at a low drainvoltage and plotted on a linear scale, there is no current hysteresis.
most likely caused by parasitic contact resistances and are not related to the high quality
gate dielectric.
5.3.6 Temperature-dependent measurements
We also measured the temperature dependence of the transfer characteristics for rubrene-
based SC-FET’s. The drain current is found to decrease with decreasing temperatures in
the whole range of gate voltages, as shown in Fig. 5.8 for a sample with a 594 nm thick
gate dielectric (Ci = 3.2 nF/cm2) and a channel length of L = 50 µm. In principle, mea-
suring the temperature dependence of the transfer characteristic can reveal information
about the trap DOS and other transport parameters including the intrinsic mobility (see
5.3 Results 97
0
1x10-2
2x10-2
3x10-2
0
3x10-5
6x10-5
9x10-5
0 -20 -40 -600
2x10-3
4x10-3
6x10-3
8x10-3
0
-2x10-6
-4x10-6
-6x10-6
RubreneSC-FET
|I d|1/2
Vd = - 80V
Vd = - 5V
|I d| [A
]
Vd = - 5V
Vd = - 80V
Vg [V]
|I d|1/2
|I d| [A
]
PentaceneSC-FET
Figure 5.6: Near-ideal transfer characteristics in the saturation regime (Vd =−80 V) of a rubreneSC-FET (upper panel) and of a pentacene/CytopTM SC-FET (lower panel). The square root of thedrain current is shown for the saturation regime. For the linear regime characteristics (Vd =−5 V)the drain current increases sub-linearly.
0 -20 -40 -60 -800
-200
-400
-600
-800
Vg = -30V
Vg = -50V
Vg = -70V
I d [A
]
Vd [V]
Figure 5.7: Output characteristic of a rubrene SC-FET. Forward and reverse sweeps are shown,but are indistinguishable due to the very high electrical stability of the device. At low Vd , thedeviations from the ideal transistor behaviour are more pronounced.
98 Eliminating gate bias stress effects
10 0 -10 -20 -30 -40 -50 -60 -700
2x10-5
4x10-5
10 0 -10 -20 -30 -40 -50 -60 -7010-12
10-10
10-8
10-6
10-4
Rubrene / CytopTM SC-FET 296 K
251 K
261 K
270 K276 K
288 K
|I d| [A
]
Vg [V]
Vd = -5V
|I d| [A
]
Vg [V]
Figure 5.8: Linear regime transfer characteristics (Vd = −5 V) of a rubrene/CytopT M SC-FETmeasured at various temperatures. The drain current decreases as the temperature is decreased.The gate capacitance of this transistor is Ci = 3.2 nF/cm2.
0.0036 0.0040
-22
-20
-18
-16
-14
Vg = -15 VVg = -20 VVg = -30 VVg = -40 VVg = -50 VVg = -60 VVg = -70 V
Vg = -10 V
Vg = -7 V
Vg = -5 V
ln
T -1 [K -1]
Figure 5.9: Arrhenius plots of the field-effect conductivity for various gate voltages from themeasurements in Fig. 5.8. The fit of the experimental data to straight lines is passable, but couldbe better.
Chap. 7). More specifically, we show in Chap. 7 that, for trap-controlled band transport,
the field-effect conductivity
5.3 Results 99
σ(Vg) =LW
Id
Vd(5.1)
depends on temperature according to
σ(Vg) = Aexp(−Ea(Vg)
kT
). (5.2)
Ea is approximately equal to the energetic difference between the Fermi level at the
insulator-semiconductor interface E ′F and the valence band edge EV , i.e.
Ea ≈ EV −EF − eV0 = EV −E ′F . (5.3)
According to Eq. 5.2, we can determine the activation energy Ea for a given gate voltage
Vg from a linear regression of the ln(σ) vs. 1/T data. Fig. 5.9 shows the corresponding
Arrhenius plots for the measurement in Fig. 5.8. The fit of the straight lines to the exper-
imental data is quite acceptable, but could be better. Fig 5.10 finally shows the function
Ea(Vg). At low gate voltages, Ea steeply rises as a function of Vg. This is to be expected
for a sample with a low trap density (see Chap. 7). The non-monotonic dependence of
Ea on Vg and the large and almost constant value of the activation energy for |Vg|> 20 V
is surprising at first glance. This is because the Fermi level must approach the valence
band for an increased gate voltage/charge density. We suggest that for |Vg| > 20 V, the
temperature dependence of the drain current is controlled by the contact resistance and
not by the rubrene crystal.
A rough estimate of the trap density can be obtained from the activation en-
ergies Ea at low gate voltages as follows. From Fig. 5.10 we see that a change
of the gate voltage by ∆Vg = (8 − 5)V = 3 V leads to a shift of the activation
energy Ea (i.e. the Fermi level E ′F at the insulator-semiconductor interface) by
∆E ′F ≈ ∆Ea = (0.53− 0.23)eV = 0.3 eV. The change in gate voltage ∆Vg corresponds
to a charge carrier density of P = Ci∆Vg/e = 6× 1010 cm−2. By taking the thickness of
the accumulation layer as one molecular layer (d = 1.5 nm), we have a volume density of
p = P/d = 4×1017 cm−3. For the moment we assume a trap density that does not depend
on energy and we can finally estimate the trap density to be
N = p/∆Ea =Ci
ed
(∆Ea
∆Vg
)−1
= 1.3×1018 cm−3eV−1. (5.4)
100 Eliminating gate bias stress effects
This is in good agreement with the value of N = 3×1017 cm−3eV−1, as estimated from the
excellent subthreshold swing of device 16 in Sec. 5.3.3. The latter estimation also rests on
the assumption that the trap densities do not depend on energy. The trap density, however,
is expected to critically depend on energy (see e.g. Chap. 6 and Chap. 7 of this thesis).
The simple estimates here have to be considered as averages. They are an overestimate
of the correct trap density at large EV −E ′F (i.e. further away from the valance band edge
EV ) and an underestimate at reduced EV −E ′F .
A first step towards improving the calculation of the trap density would be to ex-
change the difference quotient in Eq. 5.4 with the respective derivative [160] which gives
N(E) =Ci
ed
(dEa
dVg
)−1
, (5.5)
i.e. the trap densities N as a function of energy E = Ea. However, the accumulation
layer thickness d is not constant but decreases with increasing gate voltage, which would
still be a significant source of error. In Chap. 7 we will introduce a more sophisticated
concept in order to calculate the trap density as a function of energy from the gate voltage
dependence of the activation energy.
5.3.7 Trap-controlled transport in TFT’s
Fig. 5.11 shows the transfer characteristic of the pentacene TFT in the linear regime
(Vd = −5 V) and the square root of the drain current in the saturation regime
(Vd = −80 V). The current hysteresis is much less apparent on a linear scale. The drain
current increases faster than lineraly in the linear regime and more than quadratic in the
saturation regime (Fig. 5.11). In Fig. 5.12 we show the field-effect mobility as a function
of gate voltage as calculated from the linear-regime transconductance (∂Id/∂Vg)Vd with
µe f f =L
WVdCi
(∂Id
∂Vg
)
Vd
. (5.6)
Eq. 5.6 is discussed in the beginning of Chap. 6. From Fig. 5.12 we see that the effective
field-effect mobility increases with gate voltage. This is indicative of charge transport
in the TFT to be controlled by “fast” traps within the pentacene layer (Sec. 2.5.3 and
[100–102]).
5.3 Results 101
0 -10 -20 -30 -40 -50 -60 -700.6
0.5
0.4
0.3
0.2
0.1
Vg [V]
E a [eV
]
Figure 5.10: Activation energy Ea(Vg) as derived with linear regressions according to Eq. 5.2and as illustrated in Fig. 5.9. For negligible contact resistances, Ea is approximately equal to theenergetic difference between the Fermi level at the insulator-semiconductor interface E ′F and thevalence band edge/mobility edge. Ea steeply rises with gate voltage at low gate voltages, whichmay be seen to further confirm the low trap density in SC-FET’s with CytopT M gate dielectric.The non-monotonic dependence of the activation energy on gate voltage and the large and almostconstant value of Ea for sufficiently large gate voltages (> 0.2 eV) indicates that the tempera-ture dependence of the drain current is dominated by the temperature dependence of the contactresistance at sufficiently high gate voltages.
0 -20 -40 -600.0
4.0x10-4
8.0x10-4
1.2x10-3
1.6x10-3
0
2x10-7
4x10-7
6x10-7
8x10-7
|I d|1/2
Vg [V]
Vd = - 80V
Vd = - 5V
|I d| [A
]
Figure 5.11: Thin-film transistor operation of a pentacene/CytopTM TFT with a saturation field-effect mobility of 0.26 cm2/Vs. The square root of the drain current in the saturation regime anddrain current in the linear regime are shown.
In Fig. 5.13 we show the output characteristic of the pentacene TFT. Again, the trap-
controlled transport is reflected in the superlinear dependence of the drain current at low
Vd and in a more than quadratic dependence at high Vd . It is worthwhile to compare this
characteristic with the characteristic from the rubrene SC-FET in Fig. 5.7.
102 Eliminating gate bias stress effects
-30 -40 -50 -60 -700.0
0.1
0.2
0.3
0.4
Vg [V]
eff [
cm2 /V
s]
Figure 5.12: Effective field-effect mobility µe f f from a pentacene TFT with CytopT M gate dielec-tric. The increase in mobility with gate voltage indicates that the charge transport in the TFT iscontrolled by multiple trapping and release.
0 -20 -40 -60 -800
-1
-2
Vg = -50 V
Vg = -70 V
I d [A
]
Vd [V]
Figure 5.13: Output characteristic from a pentacene/CytopT M TFT with a field-effect mobility of0.26 cm2/Vs. There is no current hysteresis.
5.4 Discussion
The surface of the amorphous fluoropolymer proves to have a highly desirable quality:
essentially no electrically active trap states form in combination with the organic semi-
conductors. Bias stress effects in SC-FET’s are marginal and, thus, long-lived states for
holes are (almost) non-existent at the insulator surface. The absence of energetically deep
5.4 Discussion 103
insulator surface states can account for an improved subthreshold swing. It is remark-
able that the insulator works very well with two different semiconductors, i.e. rubrene
and pentacene. This may indicate that the absence of surface traps is due to the absence
(or low density) of a specific chemical species on the insulator surface. The highly hy-
drophobic CytopTM surface (water contact angle of∼ 112) leads to an (almost) complete
elimination of gate bias stress effects. Water is a known cause of gate bias stress effects
which can be eliminated by employing a highly hydrophobic CytopTM gate insulator.
The high reproducibility of the excellent device performance matches the reproducibly
high water contact angle of the CytopTM films. In Sec. 2.4.3 we have described how wa-
ter could lead to charge carrier trapping. Water molecules may act as traps themselves.
Localization of charge carriers may also be caused by the dipolar nature of the water
molecules. [92,93] Moreover, an electrochemical reaction of water with the organic semi-
conductor [86] and/or with the gate dielectric [91] may impede charge transport. In the
latter case, OH-groups on the surface of the gate dielectric are expected to play a key
role and the CytopT M gate dielectric does not contain these groups. It has also been sug-
gested that a polar gate dielectric leads to an unfavorable broadening of the density of
states function due to dipolar disorder caused by randomly oriented dipoles within the
gate dielectric (see Sec. 2.4.3 and in particular Fig. 2.17). [61, 62] CytopT M has a very
low dielectric constant εi = 2.1−2.2 which may suppress a broadening of the density of
states in the interface region.
For the pentacene SC-FET and TFT we have combined the same semiconductor and
insulator. However, we observe that the TFT is less stable against bias stress than the SC-
FET. The shape of the TFT transfer characteristics reflects the presence of a substantial
density of traps. The field-effect mobility monotonically increases with gate voltage and
also the onset voltage of the TFT is more negative. These effects cannot be attributed to
insulator surface states but should be caused by growth-related structural defects within
the pentacene layer close to the dielectric-semiconductor interface.
For high quality organic single crystals the channel is highly conducting and thus
the contact resistance also needs to be small. The non-idealities of the transistor char-
acteristics of the SC-FET’s are likely due to a contact resistance that decreases with an
increase in drain voltage and thus might be treated as a Schottky diode. [155, 180] We
suggest that the non-monotonic dependence of Ea on Vg and the large and almost constant
value of the activation energy for |Vg|> 20 V is due to the temperature dependence of the
drain current controlled by the contact resistance and not by the rubrene crystal. For the
104 Eliminating gate bias stress effects
TFT’s, the quality of the contacts is not as important as for the SC-FET’s because of the
increased channel resistance. While the dependence of the drain current on gate voltage
is dominated by the contacts for the rubrene SC-FET, it is dominated by trapping in “fast”
traps for the pentacene TFT.
5.5 Conclusions
Our experiments highlight the intrinsically high performance and high stability of
small molecule organic semiconductors when combined with a suitable gate dielectric.
CytopT M is easy to handle (i.e. deposition in air from solution), gives an interface with
very few electrically active defects and is a good electrical insulator. SC-FET’s with
rubrene and pentacene in combination with a fluoropolymer show excellent electrical
characteristics with a high field-effect mobility, and they are hardly affected by long-term
gate bias stress. It seems that there is no conceptual limitation to the stability of organic
semiconductors in contrast to a-Si:H, where the diffusion of hydrogen leads to gate bias-
induced metastable defects. [181] However, since gate bias stress effects are somewhat
more pronounced in pentacene thin-film transistors than in the highly stable pentacene
single crystal transistors with the same gate dielectric, we can conclude that structural
disorder within the semiconductor is a cause of electrical instability.
CytopT M works well with rubrene and pentacene. It seems that the nature of the
gate dielectric is much more important than the chemistry of the organic semiconduc-
tor in order to obtain transistors with a high electrical stability. It is very likely that the
fluoropolymer leads to outstanding transistors with many other small molecule semicon-
ductors.
We have seen that parasitic contact resistances can significantly affect the transfer
characteristics and the temperature dependence of the drain current. It is thus very im-
portant to correct for parasitic contact resistances when extracting the trap DOS and other
transport parameters from the characteristics of organic field-effect transistors. The study
highlights the need to produce electrical contacts that are matching the good semiconduc-
tor/dielectric combination, i.e. that are not limiting current injection and extraction.
6 Defect healing at room temperaturein pentacene thin films and improvedtransistor performance
We now describe a trap reduction process in the organic semiconductor pentacene which
we ascribe to the healing of structural defects at room temperature. This peculiar effect is
a direct consequence of the weak intermolecular interaction which is characteristic of or-
ganic semiconductors. Pentacene thin-film transistors were fabricated and characterized
by gated four-terminal measurements without any unintentional air exposure. Under high
vacuum conditions (base pressure of the order of 10−8 mbar), the device performance is
found to improve with time. The effective field-effect mobility increases by as much as
a factor of two and mobilities up to 0.45 cm2/Vs were achieved. In addition, the con-
tact resistance decreases by more than an order of magnitude and there is a significant
reduction in current hysteresis. Oxygen and nitrogen exposure, as well as annealing ex-
periments, show the improvement of the electronic parameters to be driven by a thermally
promoted process and not by chemical doping. In order to extract the spectral density of
trap states from the transistor characteristics, we have implemented a scheme which al-
lows for a calculation of the trap densities with high accuracy in a straightforward fashion.
We show the performance improvement to be due to a reduction in the density of shallow
traps ≤ 0.15 eV from the valence band edge, while the energetically deeper traps are es-
sentially unaffected. This chapter contributes to an understanding of the shallow traps in
organic semiconductors and identifies structural defects as a major cause.1
1 The results in this chapter are published inW. L. Kalb, F. Meier, K. Mattenberger, B. Batlogg, Phys. Rev. B 76, 184112 (2007).
106 Defect healing at room temperature in pentacene thin films
6.1 Introduction
In Chap. 2 we have seen that the charge transport in crystalline organic semiconductors
such as pentacene can be described by assuming a mobility edge which separates extended
and localized states. The charge carriers are transported in the extended states above the
mobility edge, but are trapped by and thermally released from localized trap states below
the mobility edge. The mobility edge may be identified with the valence or conduction
band edge. We have also seen that, in highly disordered organic semiconductors, a de-
scription of the charge transport by variable range hopping may be more appropriate. Im-
portantly, this situation can be described by trap-controlled transport in a transport level
with a distribution of localized states below the transport level. This description is thus
very similar to the mobility edge picture. From these models, the conduction of charge
in organic semiconductors and thus the transistor characteristics are expected to critically
depend on trap states. Trap states in organic semiconductors have been the subject of
extended investigations. [21] This topic is currently attracting a lot of attention because of
the crucial importance of trap states for the emerging applications of organic field-effect
transistors. Microscopic causes of trap states in organic semiconductors are described in
Chap. 2.
Organic field-effect transistors are excellent devices to study the charge trans-
port in organic semiconductors, since the position of the Fermi level at the dielectric-
semiconductor interface can be fine-tuned by applying a gate voltage. The density of
trap states as a function of energy can be derived from the transistor characteristics. On
the one hand, a density of states function can be postulated a priori and the correspond-
ing transistor characteristic can be simulated by means of a suitable computer program.
[152, 153, 155] On the other hand, the density of states function can be calculated from
the linear regime transfer characteristics in a straightforward fashion. [52, 53, 128, 160]
Organic thin-film transistors (TFT’s) are most often characterized after the samples
have been exposed to ambient air and electrical characterization without air exposure is
very rare. [131, 132] The organic semiconductors, however, are generally presumed to be
sensitive to water vapour and oxygen. Moreover, the transistors are often characterized by
two-terminal measurements which do not allow to distinguish between contact effects and
effects of the semiconducting layer. As we will see below, the gated four-terminal method
yields the field-effect conductivity and the effective field-effect mobility free from contact
effects. It also allows for an extraction of the device contact resistance.
6.2 Experimental methods 107
For the experimental work described in this chapter, pentacene served as a prototyp-
ical oligomeric semiconductor; we have investigated the performance of pentacene thin-
film transistors by gated four-terminal measurements without ever exposing the samples to
ambient air. The corresponding trap states functions were derived with a straightforward
extraction scheme which had been successfully used to study trap states in hydrogenated
amorphous silicon.
6.2 Experimental methods
6.2.1 Device fabrication
Pentacene from Aldrich (purum) was sublimation purified twice (see Sec. 3.2) and was
introduced into the evaporation chamber of the device fabrication and characterization
system immediately after the purification. As substrates we used heavily doped Si wafers
with a 260 nm thick SiO2 layer. The substrates were cleaned with hot acetone and hot
isopropanol (MOS grade) in an ultrasonic bath. Immediately after the cleaning, the sub-
strates were mounted on a sample holder and were introduced into the device fabrication
and characterization system via the load lock (Fig. 3.9(a)). The evaporation chamber
and the prober station were both separated from the load lock by a gate valve and were
constantly kept under vacuum (base pressure ≈ 3×10−8 mbar). The vacuum in the evap-
oration chamber and in the prober station was maintained respectively with a cryopump
and a turbo pump. The substrates were introduced into the evaporation chamber with
transfer rod 1 (Fig. 3.9(a)) and were placed on a shadow mask for the pentacene evapo-
ration. A high precision mask positioning mechanism allowed for a proper adjustment of
the mask with respect to the substrates.
Prior to the device fabrication, the substrates were kept in high vacuum for ap-
proximately 24 h. After that time, also the pressure in the turbo-pumped load lock was
≈ 3× 10−8 mbar. A 50 nm thick film of pentacene was evaporated through the shadow
mask onto the Si/SiO2 substrates, while the substrates were kept at room temperature.
After the pentacene evaporation, the sample holder was positioned on a shadow mask
for the gold evaporation with transfer rod 1 and the pentacene TFT’s were completed by
evaporating 40 nm thick gold electrodes.
The resulting transistor test structures are schematically shown in Fig. 3.13. The
transistors consisted of a well-defined stripe of pentacene and had voltage sensing elec-
trodes with little overlap to the pentacene film. It has been demonstrated that the use
108 Defect healing at room temperature in pentacene thin films
of a “masked” pentacene film and a proper alignment of the electrodes is important for
the four-terminal measurement. [134] The channel length and width of the devices were
L = 450 µm and W = 1000 µm. The voltage sensing electrodes were situated at (1/6)L
and (5/6)L, such that the distance between these electrodes was L′ = 300 µm.
After the completion of the device fabrication, the samples were transported to the
prober station by means of transfer rod 1 and transfer rod 2 (Fig. 3.9(a)).
6.2.2 Electrical characterization
For the electrical measurements we used a HP 4155A semiconductor parameter analyzer
connected to five microprobers at the prober station. Transfer characteristics in the lin-
ear regime were measured in steps of 0.2 V (drain voltage Vd = −2 V). In all cases, the
forward and the reverse sweeps were measured. The integration time was 20 ms and the
delay time was 0 s. In addition to the drain current Id , the voltage drops V1 and V2 between
the grounded source and the two voltage sensing electrodes were measured at each gate
voltage Vg (gated four-terminal measurement). All electrical measurements were carried
out in the dark.
A device was initially measured ≈ 4 h after the completion of the pentacene evap-
oration. Subsequently, the same device was regularly measured (normally twice a
day) for approximately one week while being kept in the prober station in the dark at
≈ 3×10−8 mbar.
In some experiments, we investigated the effect of oxygen or nitrogen exposure on
the device characteristics. This was done by introducing a continuous flow of gas (purity
≥ 99.9999 Vol.-%) into the prober station through a leak valve, thus adjusting the partial
pressure of the gas within the prober station. A few days prior to the device fabrication for
the gas exposure experiments, the prober station was filled with the respective gas through
the leak valve in order to flush the gas supply line and the valve. The prober station was
then re-evacuated and the supply lines were held at an overpressure of 0.5 atm until the
leak valve was opened in the experiment.
In other experiments, the influence of thermal annealing was explored by means of
an electrical heating element at the cryostat in the prober station.
6.3 Parameter extraction 109
6.3 Parameter extraction
In this section, we describe the extraction of key parameters from the transistor char-
acteristics of trap-controlled devices. Our approach consists of measuring the transfer
characteristics at a low drain voltage (Vd = −2 V). We can therefore assume the “un-
perturbed” situation where charge is accumulated by a gate voltage in a metal-insulator-
semiconductor (MIS) structure but no drain voltage is applied in order to calculate the
charge distribution as a function of the distance from the insulator-semiconductor inter-
face. [101] We begin by specifying the basic parameters to be extracted from the transfer
characteristics.
6.3.1 Basic parameter extraction
In single crystalline inorganic MOSFET’s above a threshold voltage, essentially all the
trap states are filled and the charge induced by the gate appears in the valence/conduction
band. [102] Both the constant mobility in the above threshold regime and the threshold
voltage are important device parameters. This approach is described in detail in Sec. 2.5.2.
It may also be valid in the case of organic single crystal transistors, i.e. organic field-effect
transistors with a low trap density. However, we have already seen in Chap. 4 that contact
effects often severely affect the characteristics of these devices due to the low channel
resistance. This is in agreement with other studies. [180]
In Chap. 2 we have also mentioned that the equations developed for single crystalline
MOSFET’s are not suitable to describe organic TFT’s with an increased trap density.
[100,101] Depending on the density of trap states, the majority of charge carriers induced
by the gate may be trapped even at relatively high gate voltages. [155] Provided that
contact effects are negligible, the drain current of an organic TFT in the linear regime
may be written as
Id = (W/L)σVd. (6.1)
σ is the field-effect conductivity, which is the effective field-effect mobility µe f f multi-
plied by the total gate induced charge per unit area Ci(Vg−VFB), i.e
σ = µe f fCi(Vg−VFB). (6.2)
110 Defect healing at room temperature in pentacene thin films
VFB is the flatband voltage and Ci the capacitance per unit area. In Sec. 2.5.2 we have dis-
cussed the origins of a non-zero flatband voltage. The flatband voltage is approximately
equal to the onset voltage of the device. The onset voltage is defined as the gate voltage,
where the drain current, as a function of gate voltage, rises sharply if plotted on a loga-
rithmic scale. In the present work we assume that the flatband voltage is equal to the onset
voltage. The effective field-effect mobility is one of the most important device parameters
and, for a p-type semiconductor such as pentacene, µe f f can be written as
µe f f =Pf ree
Pf ree +Ptrappedµ0, (6.3)
where Pf ree and Ptrapped are respectively the density of free and trapped holes per unit
area. µ0 is the extended state mobility. The effective field-effect mobility µe f f is ex-
pected to increase with gate voltage even at relatively high gate voltages, because the
ratio Pf ree/(Pf ree + Ptrapped) increases as the valence band is bent towards the Fermi en-
ergy. [100]
From a technological point of view, it may be useful to define a threshold voltage
which marks the transition between the regime below threshold and the regime above
threshold. In the above-threshold regime, the deep traps are filled and the field-effect mo-
bility is less strongly dependent on gate voltage. The above-threshold regime in an organic
TFT can be understood as being in between the below-threshold regime and the above-
threshold regime of a single crystalline MOSFET. [102] It can be demonstrated that the
field-effect mobility in the above-threshold regime follows a power law µe f f = κ(Vg−Vt)α
and this allows for a refined extraction of the field-effect mobility and the threshold volt-
age Vt . [101, 102, 159]
The field-effect conductivity can be calculated from
σ(Vg) =LW
Id
Vd. (6.4)
With Eq. 6.2 and Eq. 6.4, µe f f can be approximated as
µe f f (Vg) =1Ci
(∂σ∂Vg
)
Vd
=L
WVdCi
(∂Id
∂Vg
)
Vd
. (6.5)
Since this approach is frequently used, it has the advantage that the values of µe f f can
easily be compared. Moreover, the definition and extraction of a threshold voltage is not
6.3 Parameter extraction 111
necessary. However, small errors in the absolute value of the field-effect mobility and its
gate voltage dependence are to be expected since the derivation rests on the assumption
of a weak dependence of the field-effect mobility on gate voltage. [101]
In an organic field-effect transistor, a significant fraction of the drain voltage Vd may
drop at the contacts; this can introduce significant errors when extracting the field-effect
conductivity and the field-effect mobility. From gated-four terminal measurements, the
conductivity can be derived without error with Id = (W/L′)σV ′d . L′ is the distance between
the voltage sensing electrodes and V ′d = V1−V2 the voltage drop between these electrodes
(Fig. 3.13). The contact-corrected field-effect conductivity is then given by
σ(Vg) =L′
WId
(V1−V2). (6.6)
The effective field-effect mobility µe f f is not influenced by contact effects when calcu-
lated from
µe f f (Vg) =L′
W (V1−V2)Ci
(∂Id
∂Vg
)
Vd
. (6.7)
In the following, we use the expressions “two-terminal conductivity” and “two-
terminal mobility” as short hand for Eq. 6.4 and Eq. 6.5. We furthermore use the ab-
breviations “four-terminal conductivity” and “four-terminal mobility” for the quantities
defined in Eq. 6.6 and Eq. 6.7.
The device contact resistance Rcontact was extracted from the four-terminal measure-
ment and was compared to the device channel resistance Rchannel . We now assume a linear
voltage drop all along the channel, i.e. from the source to the drain. With this assumption,
the voltage drop across the transistor channel is (V1−V2)L/L′ and the voltage drop at the
contacts is Vd− (V1−V2)L/L′. The contact resistance is thus given by
Rcontact(Vg) =Vd− (V1−V2)L/L′
Id(6.8)
and the channel resistance by
Rchannel(Vg) =(V1−V2)L/L′
Id. (6.9)
112 Defect healing at room temperature in pentacene thin films
6.3.2 Advanced parameter extraction
The gate voltage dependence of the field-effect mobility reflects the spectral density of
trap states close to the valence band. A number of direct extraction schemes has been
suggested to obtain the underlying density of states function. [52, 53, 128, 160] In these
approaches, the relevant energy scale is derived from the activation energy Ea(Vg) of
the current (i.e. the field-effect conductivity) that is obtained from temperature depen-
dent measurements. If, however, the electrical characteristics of a transistor change on a
timescale that is comparable to the time of a temperature dependent measurement (hours),
this approach is not suitable.
Grünewald et al. have suggested an extraction scheme of high accuracy for amor-
phous silicon thin-film transistors which allows to convert a single linear regime transfer
characteristic into the underlying density of states function. [140, 142, 145] It is based on
surprisingly few simplifying assumptions including:
1. the charge density is homogenous all along the transistor channel,
2. the semiconductor is homogenous perpendicular to the insulator-semiconductor in-
terface, and
3. insulator surface states only introduce an initial band bending without applied field,
i.e. contribute to a non-zero flatband voltage VFB.
Extraction schemes are often based on the abrupt approximation: all the charge is as-
sumed to reside in a region of depth λ(Vg) close to the dielectric-semiconductor interface.
Grünwald’s method, however, is not based on this simplification, but takes proper account
of the gate-induced band bending. In the following, we present the key equations of the
extraction scheme.
Within the Boltzmann approximation, the field-effect conductivity can be written as
σ(Ug) =σ0
d
Z d
0exp
(eV (x)
kT
)dx. (6.10)
Now
Ug = |Vg−VFB| (6.11)
is the gate voltage above the flatband voltage, d is the thickness of the pentacene film,
eV (x) is the band shift as a function of the distance x from the insulator-semiconductor
6.3 Parameter extraction 113
Figure 6.1: Potential drop across the gate insulator (thickness l, dielectric constant εi) and thepentacene thin film (thickness d, dielectric constant εs). Most of the gate voltage drops across thegate dielectric. At the insulator-semiconductor interface the potential is V0.
interface and σ0 = σ(Ug = 0) is the conductivity at flatband. σ0 can be approximated as
σ0 = eµ0d NV exp(− EV −EF
kT
)= eµ0 Pf ree. (6.12)
NV is an effective density of extended states and EV the energy of the valence band edge
far from the insulator-semiconductor interface. The situation is depicted in Fig. 6.1 and
Fig. 6.2. The complicated dependence of the band shift on the space coordinate x in
Eq. 6.10 can be eliminated and an equation can be derived, which implicitly contains the
interface potential V (x = 0) = V0 as a function of gate voltage:
exp(
eV0
kT
)− eV0
kT−1
=e
kTεid
εslσ0
[Ugσ(Ug)−
Z Ug
0σ(U ′
g)dU ′g
]. (6.13)
l is the thickness of the gate insulator and εi and εs are the dielectric constants of the
insulator and the semiconductor (Fig. 6.1). A derivation of Eq. 6.13 can be found in [140].
For each gate voltage, Eq. 6.13 can be numerically evaluated with the measured data
σ(Ug) and a value for the interface potential V0 is obtained. Eventually, we have the
complete function V0 = V0(Vg).
We now outline how a straightforward conversion of Poisson’s equation along with
boundary conditions eventually leads to the key equation which can be used to calculate
the trap DOS from the interface potential V0. We assume that the electrical potential V (x)
at the surface of the pentacene film of thickness d vanishes under all biasing conditions,
114 Defect healing at room temperature in pentacene thin films
Figure 6.2: Sketch of the energetics near the SiO2/pentacene interface: the application of a gatevoltage Vg above the flatband voltage leads to a bending of the valence band (VB) and of theconduction band (CB). At the interface (x = 0), the band shift is eV (x = 0) = eV0. Under theseconditions, the energy E ′ of specific trap states (dashed line) is raised at the interface so that it coin-cides with the Fermi energy EF of the sample. The energy of these trap states relative to the energyof the mobility edge EV is EV −EF − eV0. In Chap. 7, the energy EV −EF − eV0 is approximatedby the experimentally determined activation energy Ea of the field-effect conductivity.
i.e.
V (x = d) = 0. (6.14)
The electric field F at this position is also assumed to drop to zero:
F =−(
dVdx
)
x=d= 0. (6.15)
This is reasonable as long as the pentacene film is thicker than the decay length of the po-
tential. [140] The situation is depicted in Fig. 6.1. The dielectric strength at the insulator-
semiconductor interface must be continuous, i.e. for a zero flatband voltage
εiVg−V0
l=−εs
dVdx x=0
. (6.16)
A conversion of Poisson’s equation with these boundary conditions eventually leads to an
expression for the total hole density p as a function of the interface potential V0. [140]
This expression is
p(V0) =ε0ε2
iεsl2e
Ug
(dV0
dUg
)−1
. (6.17)
6.3 Parameter extraction 115
p (lower case) denotes a volume density of holes. The volume density depends on the
distance x from the insulator-semiconductor interface, i.e. on the electrical potential V (x)
in the semiconductor. The volume density p and the area hole density P are related by
integrating over the depth of the whole film, i.e. P =R d
0 p(x)dx.
Eq. 6.17 yields the functional dependence of the volume density of holes on the
potential V0. Since the total hole density p can be written as
p(V ) =Z +∞
−∞N(E) [ f (E + eV )− f (E)]dE, (6.18)
its derivative is given by
1e
d p(V )dV
=Z +∞
−∞N(E)
∣∣∣∣d f (E + eV )d(E + eV )
∣∣∣∣dE. (6.19)
Eq. 6.19 is a convolution of the density of states function N(E) with the derivative of
the Fermi function. Several deconvolution methods exist to solve this type of equation
for N(E), e.g. with cubic spline functions. [51, 182] However, for a slowly varying den-
sity of states function (absence of monoenergetic states), the difference between N(E)
and d p/edV is expected to be relatively small on a logarithmic scale. [182] We approxi-
mate the Fermi function with a step function according to the common zero-temperature
approximation. [52,183,184] Its derivative then is a delta function and we eventually have
1e
d p(V0)dV0
≈ N(EF + eV0). (6.20)
Within this zero temperature approximation, the band shift at the interface eV0 is equal
to the energy of the respective traps relative to the Fermi energy EF of the sample, i.e.
eV0 = E ′−EF (Fig. 6.2).
We have used Grünewald’s method to interpret the current voltage characteris-
tics from pentacene TFT’s. We used the four-terminal conductivity as a starting point
which permits to extract a density of states function free from contact artifacts. A sim-
ple MATLABr code allowed for the calculation of the density of states function from
Eq. 6.13, 6.17 and 6.20.
116 Defect healing at room temperature in pentacene thin films
0
2x10-7
4x10-7
6x10-7
10 0 -10 -20 -30 -40 -5010-12
10-11
10-10
10-9
10-8
10-7
10-6140 h
Vd = -2 V
140 h
4 h
| Id |
(A)
4 h
| Id |
(A)
Vg (V)
Figure 6.3: Transfer characteristic of a pentacene TFT measured 4 h and 140 h after the comple-tion of the pentacene evaporation. The storage under high vacuum conditions leads to an increasedon-current and to a reduced current hysteresis.
6.4 Results
6.4.1 Improvement of the device performance with time
Fig. 6.3 shows the transfer characteristic of a pentacene TFT measured 4 h and 140 h
after the completion of the pentacene evaporation. The device was always kept at
3× 10−8 mbar. After 140 h, the device shows an increased on-current. In addition, the
current hysteresis is reduced: at a current level of 10−10 A, the difference between the
forward and the reverse sweep is 3.8 V after 4 h and 1.2 V after 140 h. The subthreshold
swing is essentially unaffected by the high vacuum storage. There is a small shift of the
onset voltage to more positive voltages from -6.4 V after 4 h to -4.9 V after 140 h.
An increase in on-current can be caused by changes of the pentacene film and/or by
a reduction of the device contact resistance. The gated four-terminal method can disentan-
gle the field-effect conductivity and the device contact resistance. In Fig. 6.4 we show the
four-terminal conductivity after 4 h and after 140 h, as derived from the forward sweeps
6.4 Results 117
0 -10 -20 -30 -40 -500.00
0.05
0.10
0.15 Four-terminal Two-terminal
140 h
4 h
(S)
Vg (V)
Figure 6.4: The four-terminal conductivity increases with time. The graph shows the four-terminal conductivity (circles) after 4 h and after 140 h. The dashed lines indicate the two-terminalconductivity for comparison.
(Eq. 6.6). The two-terminal conductivity (Eq. 6.4) is given for comparison. The four-
terminal conductivity is increased after 140 h, which reveals changes of the pentacene
film. The difference between the four-terminal conductivity and the two-terminal con-
ductivity is reduced after 140 h, indicative of an additional contact resistance reduction.
Fig. 6.5 shows the four-terminal mobility derived with Eq. 6.7 and the two-terminal
mobility calculated from Eq. 6.5. As expected, the mobility monotonically increases
with gate voltage. When comparing both measurements, a significant improvement in
mobility can be ascertained. At Vg ≈−50 V, the mobility is µ = 0.22 cm2/Vs after 4 h and
µ = 0.45 cm2/Vs after 140 h, i.e. µ has increased by a factor of 2.2
In Fig. 6.6 we show the width-normalized contact resistance RcontactW according to
Eq. 6.8 and, for comparison, the width-normalized channel resistance RchannelW (Eq. 6.9).
There is a drastic reduction in contact resistance. At Vg = −50 V the contact resistance
decreases by a factor of ≈ 11 from 1.95× 105 Ωcm to 1.81× 104 Ωcm. The channel
resistance decreases by a factor of ≈ 2. Importantly, the channel resistance is always
higher than the contact resistance: at Vg = −50 V and after 4 h the channel resistance is
≈ 3 times larger than the contact resistance, and after 140 h it is ≈ 17 times the contact
resistance. Thus, the device is always dominated by the channel resistance.
2 We compare mobilities for comparable total charge densities. Since we only observe small onset voltageshifts, a correction of the gate voltage by the onset voltage is not necessary.
118 Defect healing at room temperature in pentacene thin films
0 -10 -20 -30 -40 -500.0
0.1
0.2
0.3
0.4
0.5 Four-terminal Two-terminal
Vg (V)
4 h
140 h
eff (
cm2 /V
s)
Figure 6.5: Four-terminal mobility (circles) as a function of gate voltage. There is a significantimprovement in mobility with time. At Vg ≈−50 V, the mobility increases from µ = 0.22 cm2/Vsafter 4 h to µ = 0.45 cm2/Vs after 140 h, i.e. by a factor of two. The dashed lines show the two-terminal mobilities for comparison.
-30 -35 -40 -45 -50104
105
106
140 h
4 h
Contact resistance
Channel resistance
4 h140 h
Vg (V)
R*W
(*c
m)
Figure 6.6: Width-normalized contact resistance (triangles) after 4 h and after 140 h. The contactresistance is drastically reduced: at Vg = −50 V it decreases by a factor of ≈ 11. The graph alsocontains the respective width-normalized channel resistances (circles) for comparison. The contactresistance is always lower than the channel resistance.
6.4.2 Influence on the density of states function
In Sec. 6.4.4 and 6.4.5 we show that the performance improvement is not due to doping by
a residual gas, but due to a healing of structural defects at room temperature. In order to
investigate the energetic position of these defects, we have applied the scheme described
in Sec. 6.3.2. The first step was to calculate the interface potential V0 as a function of
gate voltage with Eq. 6.13. We have assumed a dielectric constant of εi = 3.9 for SiO2
6.4 Results 119
and εs = 3.0 for pentacene. [185, 186] Fig. 6.7(a) shows the result of the extraction for
the measurement after 4 h and after 140 h. In a second step, the density of states was cal-
culated by two numerical differentiations of V0(Ug) according to Eq. 6.17 and Eq. 6.20.
Some degree of data smoothing was applied in order to obtain a reasonably smooth den-
sity of states function. Fig. 6.7(b) shows the final result for the measurement after 4 h and
after 140 h. A step width of 0.2 V in the gate voltage sweeps leads to a good resolution of
the deep states.
The Fermi energy of the sample coincides with the zero point of the energy scale
in Fig. 6.7(b). At high gate voltages, the valence band edge at the interface is close to
the Fermi level. Consequently, the Fermi level is approximately 0.45 eV from the valence
band edge at flatband. For comparison, the bandgap of pentacene is about 2.2 eV. [187]
The interface potential in Fig. 6.7(a) reflects the spectral density of trap states. At low gate
voltages, bands bend easily, and this implies a low trap density. At high gate voltages,
however, band bending is more difficult and this is indicative of a high trap density very
close to the valence band edge.
From Fig. 6.7(b) it is clear that it is the shallow traps with energies approximately
0.15 eV from the valence band edge which are reduced by the high vacuum storage. It
is the density of these states that influences the value of the field-effect mobility µe f f . A
relatively small reduction causes a significant improvement in field-effect mobility. The
traps which are deeper in energy are essentially unaffected, resulting in an almost identical
subthreshold swing of the transfer characteristics.
After 4 h as well as after 140 h, the density of states function can reasonably well be
approximated by a single exponential function
N(E) = N0 exp(
EE0
). (6.21)
It is however slightly steeper than exponential. Fitting the curves in Fig. 6.7(b) for
eV0 ≥ 0.25 eV to Eq. 6.21 yields the parameters E0 = 32 meV for the measurement af-
ter 4 h and E0 = 37 meV for the measurement after 140 h.
120 Defect healing at room temperature in pentacene thin films
010
2030
4050
0.0
0.1
0.2
0.3
0.4
0.00.1
0.20.3
0.410
17
1018
1019
1020
1021
(b)(a)
4 h
140 h
V0 (V)
|Vg -V
FB | (V)
0.00.1
0.20.3
0.4
1x1021
2x1021
3x1021
4x1021
4 h140 h
dp/edV N(E) (cm-3eV-1)
4 h
140 h
eV0
E-EF (eV)
Figure6.7:
(a)Interface
potentialV0
asa
functionof
gatevoltage
aboveflatband|V
g −V
FB |for
them
easurementafter
4h
andafter
140h.
(b)D
ensityof
trapstates
asa
functionof
energy.T
hem
ainpanelshow
sthe
dp/edV
dataas
afunction
ofthe
bandshiftatthe
interfaceeV
0on
alogarithm
icscale.
The
bandshiftatthe
interfaceis
approximately
equaltothe
energyofthe
trapsabove
theFerm
ienergyofthe
sample,i.e.eV
0 ≈E−
EF .T
hequantity
dp/edV
isa
goodapproxim
ationofthe
densityoftrap
statesN
(E).T
hehigh
vacuumstorage
leadsto
areduced
densityoftraps
closeto
thevalence
bandedge.T
heinsetshow
sthe
trapdensities
ona
linearscale.
6.4 Results 121
Table 6.1: Four-terminal mobility µ1, at Vg ≈ −50 V, width-normalized contact resistance R1Wat Vg =−50 V and onset voltage Von from the initial measurements of different experiments. Themobility increased by as much as a factor of 2 in the course of an experiment, so that mobilitiesup to 0.45 cm2/Vs were achieved.
Run µ1 (cm2/Vs) R1W (Ωcm) Von (V)
1 0.22 1.95×105 -6.4
2 0.24 1.10×105 -4.1
3 0.14 3.03×105 -5.1
4 0.14 3.35×105 -6.4
Oxygen 0.12 3.98×105 -6.2
Nitrogen 0.10 4.70×105 -7.0
Annealing 0.10 4.55×105 -5.7
6.4.3 Comparison of several experiments
The effects described above, i.e. a significant increase in the four-terminal mobility, a
drastic reduction in the contact resistance and a reduction in the current hysteresis, have
been observed in all experiments. Fig. 6.8 shows the evolution of the four-terminal mo-
bility and the contact resistance with time for four different runs. The values are for
Vg ≈−50V and are normalized by the value obtained after ≈ 4 h.3 A time span of at least
4 h was allowed in between subsequent measurements. The absolute values for the mobil-
ity and the contact resistance from the initial measurement are summarized in Table 6.1.
Table 6.1 also contains the respective onset voltages. Initially, the onset voltage is
between −4.1 V and −6.4 V and shifts in all cases by less than 2.5 V during the course
of an experiment. The reduction of the current hysteresis takes place in the early stages
of the experiments. In all four runs, the current hysteresis is significantly reduced after
the first ≈ 24 h to 1.0−1.5 V at a current level of 10−10 A. Subsequently, there is only a
small further reduction of the current hysteresis.
3 The value for the contact resistance was taken at Vg =−50V . For the mobility, an average in the rangebetween Vg = −45 V and −50 V was taken, since in some cases the derivative of the drain current wasmore noisy.
122 Defect healing at room temperature in pentacene thin films
1.0
1.2
1.4
1.6
1.8
2.0
0 30 60 90 120 150 180
0.1
1
4
3
2
/
1
R /
R1
time (h)Figure 6.8: Upper panel: four-terminal mobility at Vg ≈−50V normalized by the mobility µ1 ofthe initial measurement for four different runs. The lower panel shows, on a logarithmic scale,the respective values for the contact resistance R at Vg = −50V relative to the contact resistanceR1 from the initial measurement. The field-effect mobility increases with time and the contactresistance reduces with time in all experiments. There is some variation in the rate of the effects.
We now proceed by providing experimental evidence that the performance improve-
ment is not due to doping of the pentacene thin films by residual oxygen or nitrogen and
show that the performance improvement is a thermally promoted process.
6.4.4 Influence of oxygen and nitrogen
Even at a pressure of the order of 10−8 mbar, the time for the formation of a monolayer
of residual gas molecules is less than ten minutes. [188] In the case of semiconducting
polymers, experimental evidence indicates that doping leads to an increased field-effect
mobility. [96] This may be understood if we assume the charge transport to take place by
variable-range hopping and the doping to increase the density of states/hopping sites. [96]
In crystalline small molecule organic semiconductors, however, charge transport in ex-
tended states has to be considered. Doping raises the Fermi level of the sample. This
6.4 Results 123
Figure 6.9: Effect of oxygen exposure: the graph shows the four-terminal mobility at Vg ≈−50V(upper panel) and the contact resistance at Vg = −50V (lower panel) as a function of time. Thedashed red line is an estimate of the time-dependence of the contact resistance without oxygenexposure. The drastic increase in oxygen partial pressure neither leads to a more rapid increase inmobility nor to a more rapid decrease in contact resistance. On the contrary, the oxygen exposureslows down the decrease in contact resistance. In the inset we show the device onset voltage asa function of time. The increase in oxygen partial pressure does not cause a sudden shift of theonset voltage, which would be indicative of doping.
results in an increase of the (flatband) conductivity, since the density of free holes is in-
creased (Eq. 6.12). It would, however, not immediately be obvious how chemical doping
would increase the field-effect mobility in pentacene.
Both oxygen and nitrogen have been reported to have the capability of doping pen-
tacene. Gas exposure is found to lead to a shift of the transistor transfer characteristic
to more positive voltages, which corresponds to a shift of the Fermi level. [132, 189] In
other studies, an increase in conductivity of gas exposed pentacene thin-films or single
crystals was observed in two-terminal measurements without gate-field induced charge.
The increased conductivity is ascribed to an increased charge carrier density caused by
doping. [190–192] There seems to be no evidence that doping leads to an increased ef-
fective mobility in pentacene. Recent measurements on rubrene single crystals also show
doping not to increase the field-effect mobility. [193]
124 Defect healing at room temperature in pentacene thin films
Figure 6.10: Effect of nitrogen exposure: a nitrogen partial pressure of 10−4 mbar leads to aslowing of the increase in mobility and decrease in contact resistance. The dashed red line is anestimate for the time-dependence of the contact resistance if the sample had not been exposed tonitrogen. The inset shows the device onset voltage. When the nitrogen partial pressure is increasedto 10−6 mbar, there is a sudden shift of the onset voltage which is indicative of doping.
The issue of reversibility and the importance of light for the doping is not com-
pletely clear. Doping is observed when pentacene thin films are exposed to relatively
high partial pressures (0.01 atm-1 atm) of oxygen or nitrogen in the presence of light.
[189, 190, 192] The doping effect by oxygen is reported to be negligible in the absence
of light. [189, 190] Ultraviolet photoelectron spectroscopy (UPS) before and after the ex-
posure to 5× 10−6 mbar oxygen could not detect a lasting effect on the position of the
energy levels. [194] Another study shows that the exposure of pentacene TFT’s to an
oxygen partial pressure of 10−5 mbar leads to a doping of the films. [132]
Although the mobility increase in crystalline/polycrystalline samples cannot be eas-
ily linked to chemical doping, a reduction of the contact resistance might be reconciled
with doping. From inorganic semiconductor physics it is well known that doping close to
a contact reduces the contact resistance. Moreover, it has been demonstrated with UPS
that oxygen, at a low partial pressure, can lead to a lowering of the injection barrier for
holes at an Au/pentacene interface. [194]
6.4 Results 125
The results of our studies concerning the effect of gas exposure on the field-effect
mobility and on the contact resistance are discussed in the following. If e.g. oxygen were
responsible for the increase in mobility and/or reduction in contact resistance, an increase
in the oxygen partial pressure should accelerate the rate of the respective process. Fig. 6.9
shows the influence of oxygen on the four-terminal mobility and on the contact resis-
tance (see Table. 6.1 for the initial parameters). After 41 h, the oxygen partial pressure
was raised from a total turbo-pumped background pressure of the order of 10−8 mbar to
10−6 mbar. After 69 h, the partial pressure was increased to 10−4 mbar and, eventually,
after 98 h, the chamber was re-evacuated. The increase in oxygen partial pressure does
not accelerate the gradual increase in mobility or the reduction in contact resistance. The
dashed red line in Fig. 6.9 is an estimate for the time-dependence of the contact resistance
if the sample had not been exposed to oxygen. Consequently, the increase in oxygen par-
tial pressure even causes a delay of the reduction in contact resistance. It is noteworthy
that neither the small current hysteresis nor the subthreshold swing of the devices are ef-
fected by the oxygen exposure. Since oxygen at a partial pressure of 10−6 mbar for 28 h
and at a partial pressure of 10−4 mbar for 29 h appears not to accelerate the development
of the device parameters with time, we can conclude that oxygen at the base pressure is
not responsible for the performance improvement.
The inset in Fig. 6.9 shows the onset voltage for each of the measurements. The
overall shift of the onset voltage is small and smooth (≈ 2.6 V over≈ 140 h). In particular,
there is no sudden shift of the transfer characteristic after an increase in oxygen partial
pressure. Therefore, there is no evidence that oxygen at partial pressures ≤ 10−4 mbar
leads to a doping of the films.
Fig. 6.10 presents the analogous experiment with nitrogen. After 50 h the nitrogen
partial pressure was increased to 10−6 mbar, followed by an increase to 10−4 mbar after
69.5 h. After 98 h, the system was pumped down to the base pressure of the order of
10−8 mbar. The increase in nitrogen partial pressure does not accelerate the increase in
mobility or the reduction in contact resistance. Similarly to the oxygen experiment, we
have estimated the time-dependence of the contact resistance if the sample had not been
exposed to nitrogen (dashed red line in Fig. 6.10). Clearly, the nitrogen exposure leads to
a slowing down of the decrease in contact resistance. The nitrogen exposure leaves the
small current hysteresis and the subthreshold swing unaffected. Residual nitrogen is not
responsible for the performance improvement.
126 Defect healing at room temperature in pentacene thin films
Figure 6.11: Effect of annealing at 320 K: the increased temperature accelerates the increase inmobility and the decrease in contact resistance, as indicated by the dashed red lines.
The inset in Fig. 6.10 shows the device onset voltage as a function of time. When
the nitrogen partial pressure is increased to 10−6 mbar, there is a sudden shift of the onset
voltage to more positive voltages by 1.2 V. We take this as evidence for doping by nitrogen
at a partial pressure of 10−6 mbar. When the pressure is increased to 10−4 mbar, there is
no additional marked shift of the onset voltage.
6.4.5 Annealing at slightly elevated temperatures
As pointed out in Chap. 2, annealing pentacene thin films at moderate temperatures (e.g.
50 C) results in an improved crystallinity of the films, as seen by XRD. [76,77,79] When
concentrating on the channel region adjacent to the gate dielectric in a field-effect tran-
sistor, the annealing is found to leave the mobility unchanged or to lead to an increased
mobility. [76, 79] Since the interaction between the pentacene molecules is of the weak
Van der Waals type, it is not too surprising that annealing at moderate temperatures causes
an improved crystallinity of the films. Moreover, it seems plausible that structural defects
6.5 Discussion 127
can be “annealed” even at room temperature, which will account for the effects we ob-
serve. In this scenario we would expect an acceleration of the performance improvement
at higher temperatures. Fig. 6.11 shows the influence of such an annealing at 320 K (see
Table 6.1 for the initial parameters). After 44.5 h, the sample temperature was slowly
raised from RT to 320 K at a rate of 0.2 /min. Transfer characteristics were measured
after the temperature of 320 K had been reached. After 104.5 h, the heating was switched
off, followed by a slow cooling of the sample to room temperature. Two effects can be
discerned in the annealing process. First, the overall level of the mobility and the contact
resistance are affected by the increase in temperature since both quantities depend on tem-
perature. Second, and more significant in the present context, is the rate of change: after
raising the temperature, the increase in mobility and the decrease in contact resistance is
significantly accelerated. This is indicated by the dashed red lines in Fig. 6.11.
In an additional experiment, the sample temperature was raised from room temper-
ature to 310 K after 45.5 h and was increased to 320 K after 74 h. Even at 310 K, which
is not that much above room temperature (≈ 297K), the increase in mobility and the
decrease in contact resistance is noticeably accelerated.
All these results taken together clearly indicate an improvement of the electronic
parameters driven by a thermally promoted process, and not by chemical doping. We
suggest this process to be a healing of structural defects. We now discuss the microscopic
origin of the relevant traps and suggest how a reduced trap density can lead to a reduced
contact resistance.
6.5 Discussion
6.5.1 Defect healing at room temperature
We keep in mind that pentacene thin films on Si/SiO2 substrates are known to have a
layered structure and the layers are parallel to the substrate. Within these layers, the
molecules are arranged in a herringbone pattern and are oriented almost perpendicular
with respect to the substrate. It has been shown by high impedance STM that some of
the pentacene molecules in the layers are displaced along their long molecular axis, while
the two-dimensional packing is not disturbed. [75] With electronic structural calculations
it could be shown that the displaced molecules result in traps ≤ 0.1 eV from the valence
band edge. [75] Fig 2.11 in Sec. 2.4.1 illustrates the key result of this study. On the other
hand, calculations reveal that structural defects that form during the growth of the film
128 Defect healing at room temperature in pentacene thin films
can relax into the ideal crystal structure at a later point in time during the completion of
the film growth. [74] This theoretical study is described in more detail in Sec. 2.4.1 and
one of the findings is depicted in Fig. 2.10.
In Sec. 6.4.2 we have shown that only the shallow traps ≤ 0.15 eV from the valence
band edge are reduced during the high vacuum storage at room temperature. We suggest
that a major cause of the shallow traps in pentacene thin films are pentacene molecules
that are slightly misplaced, i.e. various types of structural defects located within the
grains or at grain boundaries. Some of these defects are in a metastable state before
relaxation. They require only a small amount of energy in order to align, which can
be provided by the thermal energy at room temperature. This is a direct manifestation
of the weak intermolecular interaction which is characteristic of small molecule organic
semiconductors. The annealing of shallow traps at room temperature can easily explain
the increase in effective field-effect mobility.
6.5.2 Defects and contact resistance
In a simplistic view, the contact resistance is given by the energetic difference between
the work function of the metal and the ionization energy of the pentacene. In reality,
however, a clear correlation between the metal work function and the contact resistance
is often not observed. [187] Interface states can significantly affect the energetics at the
metal-pentacene interface. [133] In a top contact device, also the film resistance should
contribute to the contact resistance. The film resistance in pentacene devices with gold top
contacts has even been suggested to dominate the contact resistance. [134] The situation
is illustrated in Fig. 6.12: the hole injection at the source/pentacene interface is good,
but the holes must cross the pentacene film in order to reach the channel at the insulator-
semiconductor interface. The intrinsic mobility perpendicular to the molecular layers is
lower than parallel to the layers. Combined with high trap densities, this can result in
a large resistance between the gold electrodes and the transistor channel. If the contact
resistance is dominated by the film resistance, a reduction in the degree of structural
disorder is therefore expected to lead to a reduced contact resistance.
6.6 Conclusions
Pentacene thin-film transistors were made by thermal evaporation, employing a high pre-
cision mask alignment mechanism. The devices were characterized electrically by gated
6.6 Conclusions 129
Figure 6.12: Sketch of the different contributions to the total contact resistance. The holes mustovercome a barrier associated to the gold/pentacene interface. The resistance of the 50 nm thickpentacene film adds to the total contact resistance both at the source and at the drain.
four-terminal measurements without breaking the high vacuum (base pressure of the or-
der of 10−8 mbar). Under high vacuum conditions, the device performance is found to
improve with time. Within approximately one week, the contact-corrected field-effect
mobility improves by a factor of up to two and the device contact resistance typically
decreases by more than an order of magnitude. In addition, the current hysteresis signif-
icantly reduces. We have shown that an increased partial pressure of oxygen or nitrogen
does not accelerate the performance improvement. On the contrary, the gas exposure de-
lays the decrease in contact resistance. Annealing at a slightly elevated temperature (e.g.
320 K), on the contrary, leads to an acceleration of the performance improvement.
Some defects within the pentacene “anneal” even at room temperature. This is a
peculiarity of the physics of organic semiconductors, which is governed by the weak Van
der Waals type interaction between the molecules. We have derived the spectral density
of trap states from the four-terminal conductivity. The calculations show shallow traps
≤ 0.15 eV from the valence band edge to be significantly affected by the defect heal-
ing. We suggest these traps to originate from structural defects, i.e. slightly misaligned
molecules within the polycrystalline film. The effective field-effect mobility critically de-
pends on the number of these shallow traps and a relatively small reduction results in a
significant improvement of the mobility. The contact resistance is likely to be dominated
by the film resistance and also depends on the active traps within the film.
The method to calculate the spectral density of traps is a powerful tool to further elu-
cidate the origin of trap states in organic semiconductors, provided that contact effects are
130 Defect healing at room temperature in pentacene thin films
properly taken into account. It is particularly suitable to study metastable defects in or-
ganic semiconductors, because the density of states function can be derived from a single
transfer characteristic in an unambiguous fashion and with a minimal set of simplifying
assumptions.
7 Oxygen-related traps in pentacenethin films: Energetic position andimplications for transistor performance
In this chapter we describe a detailed study of the influence of oxygen on the electronic
trap states in a pentacene thin film. The study was done by carrying out gated four-
terminal measurements on thin-film transistors as a function of temperature and without
ever exposing the samples to ambient air. Photo-oxidation of pentacene is shown to lead
to a peak of trap states centered at 0.28 eV from the mobility edge, with trap densities of
the order of 1018 cm−3. As the gate voltage is ramped up, these trap states need to be oc-
cupied at first and cause a reduction in the number of free carriers at a given gate voltage.
Moreover, the exposure to oxygen reduces the mobility of the charge carriers above the
mobility edge. We correlate the change of these transport parameters with the change of
the essential device parameters, i.e. subthreshold performance and effective field-effect
mobility. The experiments described below support the assumption of a mobility edge for
charge transport, and contribute to a detailed understanding of an important degradation
mechanism of organic field-effect transistors: deep traps in an organic field-effect transis-
tor reduce the effective field-effect mobility by reducing the number of free carriers and
their mobility above the mobility edge.1
1 The results in this chapter are published inW. L. Kalb, K. Mattenberger, B. Batlogg, Phys. Rev. B 78, 035334 (2008)
132 Oxygen-related traps in pentacene thin films
7.1 Introduction
Critical issues of organic field-effect transistors are electrical stability and environmen-
tal stability. The experimental results in Chap. 5 clearly show that the electrical stability
of organic field-effect transistors can be very high, if suitable gate dielectrics are used
and if the semiconductor has a high degree of structural order (see also e.g. [15]). The
environmental stability of the organic semiconductor, thus, is an urgent issue to be ad-
dressed. [195, 196] Studies of the degradation of organic field-effect transistors are rare
and indicate that, in the case of p-type organic semiconductors, atmospheric oxygen or
ozone is a major cause. [126–130] It is crucial to understand in detail the way in which an
oxidation of the organic semiconductor impedes the charge transport and thus degrades
the transistor characteristics.
We characterized pentacene thin-film transistors by temperature dependent measure-
ments without ever exposing the samples to ambient air and after controlled exposure to
oxygen and light. The effect of oxygen can only be clarified with pristine samples. More-
over, since the field-effect conductivity depends (approximately) exponentially on tem-
perature, temperature is a very sensitive parameter. Again, another distinct feature of our
approach are the gated four-terminal measurements instead of the commonly employed
gated two-terminal measurements.
We have seen that charge transport in organic semiconductors can be described with
a mobility edge or transport level concept. In the mobility edge picture, the parameters
dominating charge transport are the trap densities as a function of energy relative to the
mobility edge, the number of delocalized charge carriers above the mobility edge and the
mobility of the latter charge.
In order to assess the fundamental transport parameters, we have developed a scheme
for organic field-effect transistors that is easy to use. The approach readily reveals all
the key parameters with high accuracy in a straightforward and unambiguous fashion.
The scheme is based on the method developed by Grünewald et al. which we used in
Chap. 6. [140] Instead of estimating the interface potential from the transistor character-
istic measured at a single temperature as in the original scheme, we extract the interface
potential from the temperature-dependence of the field-effect conductivity.
7.2 Experimental 133
7.2 Experimental
As in Chap. 6 we used the device fabrication and characterization system which is de-
scribed in Sec. 3.6.1 (Fig. 3.9 and Fig. 3.10). The device fabrication was the same as for
the experiments in Chap. 6. Heavily doped Si wafers with a 260 nm thick SiO2 layer were
cut, cleaned with hot solvents, fixed on a sample holder and were then introduced into the
cryo-pumped evaporation chamber of the device fabrication and characterization system.
After approximately 24 h, two times sublimation purified pentacene was evaporated onto
the samples through a shadow mask at a base pressure of the order of 10−8 mbar. The sub-
strates were kept at room temperature during the evaporation and the final film thickness
was 50 nm. The sample holder with the samples was then placed on a shadow mask for
the gold evaporation without breaking the high vacuum and gold electrodes were evapo-
rated onto the pentacene. Again, the completed thin-film transistors had a channel length
of L = 450 µm and a channel width of W = 1000 µm. Moreover, two voltage sensing
electrodes were situated at (1/6)L and (5/6)L and had little overlap with the “masked”
pentacene film, as schematically shown in Fig. 3.13. The sample holder was then trans-
ferred to and clamped on the cryostat in the turbo-pumped prober station of the device
fabrication and characterization system without breaking the high vacuum (10−8 mbar,
Fig. 3.9(b)).
In Chap. 6 we have seen that the device performance improves with time when
pentacene thin-film transistors are kept in high vacuum. Therefore, the devices were kept
in the prober station at a pressure of ≈ 2× 10−8 mbar for approximately three weeks
before starting the study. After that time, the device characteristics were found to be
stable on the timescale of several days.
The prober station is equipped with five micro-probers (Fig. 3.9(a)). The prober
arms are connected to the cryostat with thick copper braids and are thus cooled when the
cryostat and the sample holder with the samples is cooled down. For the temperature-
dependent gated four-terminal measurements, the source, the drain and the voltage sens-
ing electrodes were contacted with thin gold wires attached to four of the micro-probers.
By means of an electrical feedthrough to the cryostat, a gate bias could be applied. In
order to measure the temperature on the surface of the samples, an AuFe/Chromel ther-
mocouple was attached on the 5th micro-prober and was carefully pressed against the
surface of the sample at each temperature.
134 Oxygen-related traps in pentacene thin films
As always, the electrical measurements were carried out with an HP 4155A semicon-
ductor parameter analyzer. Transfer characteristics were measured with a drain voltage of
Vd =−2 V in steps of 0.2 V with an integration time of 20 ms and a delay time of 100 ms.
For each gate voltage Vg, the drain current Id and the potentials V1 and V2 between the
grounded source electrode and the respective voltage sensing electrode were measured
(Fig. 3.13). All electrical measurements were done in the dark.
The devices were exposed to oxygen by introducing 1 bar of oxygen (purity
≥ 99.9999 vol%) into the prober station through a leak valve. The pressure of the oxygen
in the prober station was measured with a mechanical pressure gauge. In addition, the
samples were exposed to a combination of 1 bar of oxygen and white light from a cold
light source (colour temperature 3200 K) trough a quartz window.
7.3 Charge transport parameters
As described for the experiments in Chap. 6, the transfer characteristics were measured
at a low drain voltage (Vd = −2 V). This results in a drastic simplification of the device
physics. The two-dimensional problem is approximated by two one-dimensional equa-
tions, i.e. Poisson’s equation for the charge distribution perpendicular to the insulator-
semiconductor interface and a simple Ohmic current-voltage relationship.
7.3.1 Field-effect conductivity, field-effect mobility and contact resis-tance
Once more, we use the terms “two-terminal conductivity” and “four-terminal conductiv-
ity” for the expressions defined respectively in Eq. 6.4 and Eq. 6.6. Eq. 6.5 is the “two-
terminal mobility” and Eq. 6.7 is the “four-terminal mobility”. We already mentioned that
µe f f , as calculated with Eq. 6.5 or Eq. 6.7, is an effective mobility. For a p-type semicon-
ductor such as pentacene, it is a rough estimate of the ratio of the free surface hole density
Pf ree to the total surface hole density Ptotal = Pf ree + Ptrapped multiplied by the mobility
µ0 of the holes in the valence band, i.e.
µe f f ≈Pf ree
Ptotalµ0. (7.1)
The device contact resistance Rcontact is given by Eq. 6.8 and should be compared to the
channel resistance Rchannel as given by Eq. 6.9.
7.3 Charge transport parameters 135
In order to gain a deeper insight into the physics of the organic semiconductor in
a field-effect transistor, a more sophisticated description is developed in the following
sections.
7.3.2 Spectral density of trap states and free hole density
We treat the polycrystalline pentacene layer as uniform and assume that trap states on the
surface of the gate dielectric only contribute to a non-zero flatband voltage. Consequently,
the trap densities to be determined are an average over in-grain and grain boundary regions
and may also be influenced, to some extend, by trap states on the surface of the gate
dielectric. The same simplifications were used in Chap. 6.
In the previous chapter we have already seen that the total hole density p as a func-
tion of the interface potential V0 may be written as
p(V0) =ε0ε2
iεsl2e
Ug
(dV0
dUg
)−1
. (7.2)
εi and εs are the dielectric constants of the gate insulator and the semiconductor and l and
d are the thickness of the gate dielectric and semiconducting layer (Fig. 6.1). p (lower
case) denotes a volume density of holes. Moreover,
Ug = |Vg−VFB| (7.3)
in Eq. 7.2 is the gate voltage above the flatband voltage VFB. Again, we approximate the
Fermi function with a step function (zero-temperature approximation) and we have an
expression for the density of states N(E)
1e
d p(V0)dV0
≈ N(EF + eV0). (7.4)
From Eq. 7.2 and Eq. 7.4 we can see that the interface potential V0 as a function of
gate voltage is the key to the density of states function (DOS). Since the change of the
interface potential and the change of the drain current with gate voltage are linked (see
Eq. 6.13 in Chap. 6), it is possible to extract the interface potential from the transfer char-
acteristic measured at a single temperature. [140] Once the interface potential is known,
136 Oxygen-related traps in pentacene thin films
the trap DOS in the mobility gap can be estimated with Eq. 7.2 and Eq. 7.4. Thus, the
trap densities can be plotted as a function of the band shift eV0 at the interface, which is
the energy of the traps relative to the Fermi energy of the sample (Fig. 6.2). This method
was applied to four-terminal conductivity data in Chap. 6. Here we have advanced the
extraction scheme. As described in the following, we used gated four-terminal measure-
ments at various temperatures to estimate the interface potential. We then used Eq. 7.2
and Eq. 7.4 to calculate the DOS. The consistency of the assumption of charge transport
above a mobility edge with the temperature-dependent measurements provides a greater
degree of confidence to any conclusion. Moreover, the latter approach has the advantage
of eventually giving the DOS as a function of energy relative to the mobility edge.
We now show that the activation energy Ea(Vg) of the field-effect conductivity as
defined by
σ(Vg) = Aexp(−Ea
kT
)(7.5)
and as determined with Arrhenius plots is related to the band shift eV0 at the insulator-
semiconductor interface. Following Boltzmann’s approximation, the field-effect conduc-
tivity
σ(Vg) = eµ0Pf ree (7.6)
may be written as
σ(Vg) = eµ0
Z d
0p f reedx =
= eµ0NV exp(−EV −EF
kT
)×Z d
0exp
(eV (x)
kT
)dx. (7.7)
NV is the effective (volume) density of extended states, EV is the energetic position of the
mobility edge and EF is the Fermi energy. For the moment we consider an exponential
trap DOS
N(E) = N0 exp(
EkT0
)(7.8)
with a characteristic slope of kT0. If we assume that all the gate-induced charge is trapped,
the integration of the exponential trap DOS leads to a simple exponential dependence of
7.3 Charge transport parameters 137
the total hole density p on the potential V (x), which is
p ∝ exp(
eVkT0
). (7.9)
This approximation is not expected to cause serious errors as long as the majority of
the gate-induced charge is trapped. [139] With Eq. 7.9 it can be shown that, except for
small values of V , also the electric field F perpendicular to the insulator-semiconductor
interface exponentially depends on V , as
F ∝ exp(
eV2kT0
). (7.10)
The electric field as in Eq. 7.10 results in an expression for the free surface hole density
Pf ree in Eq. 7.6. [52] This expression is
Pf ree ≈ 2kT ε0εs
eCiUg
l2l−1
NV exp(−EV −EF
kT
)×
×[
exp(
eV0
kT
)− exp
(eV0
2lkT
)], (7.11)
where
l =T0
T. (7.12)
The second exponential term in Eq. 7.11 can safely be neglected and so we have
Pf ree ≈ Lal
2l−1NV exp
(−EV −EF − eV0
kT
), (7.13)
with
La =2kT ε0εs
eCiUg. (7.14)
The factor l/(2l− 1) is expected to be close to unity and La may be understood as the
effective thickness of the accumulation layer. [159] Deviations from an exponential trap
DOS might be considered with a variable parameter l. However, the variation of l can be
138 Oxygen-related traps in pentacene thin films
ignored when compared to the exponential term in Eq. 7.13. If we compare Eq. 7.13 with
Eq. 7.5 and Eq. 7.6, we eventually have
Ea ≈ EV −EF − eV0 = EV −E ′F . (7.15)
The measured activation energy of the field-effect conductivity Ea(Vg) is approximately
equal to the energetic difference between the Fermi level EF and the mobility edge at the
interface, as indicated in Fig. 6.2. E ′F , as defined in Eq. 7.15, is the Fermi level at the
insulator-semiconductor interface. By substituting dV0 =−dEa/e in Eq. 7.2 and Eq. 7.4,
we finally have the DOS
N(E)≈ ddEa
[ε0ε2
iεsl2 Ug
(dEa
dUg
)−1]
(7.16)
as a function of the energy E = EV −E ′F ≈ Ea(Vg) relative to the mobility edge.
7.3.3 Fraction of free holes and band mobility
The fraction of free holes Pf ree/Ptotal is of crucial importance since it is proportional to
the effective field-effect mobility as described by Eq. 7.1. It can readily be extracted from
temperature dependent measurements. From Eq. 7.13 and Eq. 7.15 and with the total
surface hole density
Ptotal = CiUg/e, (7.17)
we eventually have
Pf ree
Ptotal=
LaeCiUg
l2l−1
NV exp(−Ea
kT
). (7.18)
Finally, from Eq. 7.6, we see that the band mobility µ0 can be estimated with
µ0 = σ/(ePf ree). (7.19)
7.4 Results 139
7.4 Results
7.4.1 Extraction method and the influence of the contact resistance
In this section, we demonstrate the extraction of the DOS and the hole densities from a set
of gated four-terminal measurements. Moreover, we analyze the influence of the contact
resistance on these functions.
In a first step, we derived the activation energy Ea(Vg) of the four-terminal conduc-
tivity as a function of the gate voltage, according to Eq. 7.5. Fig. 7.1 shows Arrhenius
plots and the corresponding linear regression lines. We found that only currents equal to
or above ≈ 1 nA are usable in the sense that the corresponding four-terminal conductivity
follows a straight line in an Arrhenius plot. Therefore, we used all currents above 1 nA
for the extraction of the activation energy. At low gate voltages, only the measurements
at the highest temperatures were considered as a consequence of the 1 nA limit. The final
result is given in Fig. 7.2. The activation energy Ea(Vg) was then represented by a smooth
fit (red line in Fig. 7.2) in order to suppress the noise in the data.
Finally, the DOS was obtained with Eq. 7.16 and is plotted as a function of the ener-
getic distance to the mobility edge Ea(Vg)≈ EV −E ′F . For the calculations, the dielectric
constant of pentacene was assumed to be εs = 3. In order to determine Ug = |Vg−VFB|in Eq. 7.16, the flatband voltage VFB was assumed to be equal to the device onset volt-
age at room temperature. The same assumption was used in Chap. 6. The onset voltage
is the gate voltage where the drain current sharply rises when plotted on a logarithmic
scale, i.e. where the drain current becomes measurable. The flatband voltage marks the
onset of the accumulation regime and a small difference between the flatband voltage and
the onset voltage may thus exist. A scheme to extract the flatband voltage was devel-
oped for amorphous silicon-based transistors and this scheme involves the temperature-
dependence of the device off-current. [143] However, the scheme cannot be applied to our
devices because the off-currents are due to experimental limitations and are not related to
the conductivity of the pentacene film.
Fig. 7.3 (circles) shows the DOS as derived from the activation energy in Fig. 7.2.
The procedure was also applied to the same data without correcting for the contact re-
sistance, i.e. to the two-terminal conductivity. The dashed line in Fig. 7.3 is the re-
sult, highlighting the necessity to correct for the contacts. Even for long channel devices
(L = 450 µm), neglecting the contact resistance leads to significant errors in the shape and
magnitude of the DOS, even more so closer to the mobility edge.
140 Oxygen-related traps in pentacene thin films
0.0036 0.0040-22
-21
-20
-19
-18
-17
Vg = -45 V
Vg = -35 V
Vg = -25 V
Measured Linear regression
Vg = -50 V
Vg = -30 V
Vg = -40 V
Vg = -20 V
ln[
]
T -1 (K -1)Figure 7.1: Arrhenius plots of the four-terminal conductivity at various gate voltages Vg. Theactivation energy Ea(Vg) was derived from the slope of the linear regression lines.
-10 -20 -30 -40 -50
0.4
0.3
0.2
0.1 Measured Smooth fit
E a (eV
)
Vg (V)
Figure 7.2: Activation energy Ea(Vg) as determined with linear regressions according to Eq. 7.5and as verified with Arrhenius plots (Fig. 7.1). The graph also shows a smooth fit of the activationenergy (red line).
7.4 Results 141
0.4 0.3 0.2 0.1
1019
1020
1021 Four-terminal Two-terminal
DO
S (c
m-3eV
-1)
EV-E'F (eV)
Figure 7.3: Density of traps as a function of energy relative to the mobility edge (circles). Thedashed line was extracted from the same set of temperature dependent measurements, but thecontact resistance was neglected. The contact resistance can lead to significant errors in the densityof states function, especially closer to the mobility edge.
The free hole density, the total hole density and the fraction of free holes were re-
spectively obtained with Eq. 7.13, Eq. 7.17 and Eq. 7.18. We have assumed that the
effective density of extended states NV is equal to the density of the pentacene molecules,
i.e. NV = 3× 1021 cm−3 and Fig. 7.4 is the result. For the given sample at high gate
voltages, ≈ 10 % of the holes that are induced by the gate are free, i.e. only this fraction
actually contributes to the drain current.
7.4.2 Oxygen-related device degradation
This study correlates the oxygen-related degradation of the transistor characteristics with
the change of the fundamental transport parameters. We begin by presenting the char-
acteristic effects of oxygen on the pentacene transistor characteristics. The blue line in
Fig. 7.5 is a transfer characteristic measured as grown (after a high vacuum storage time
of approximately 3 weeks). The sample was then exposed to 1 bar of oxygen for 19 h
and, additionally, to 1 bar of oxygen and white light for 4 h. The red curve in Fig. 7.5 is a
measurement of the same device after the oxidation process and after an evacuation time
of 22 h at a base pressure of the order of 10−8 mbar. Fig. 7.5 contains the forward and the
reverse sweeps in both cases. Fig. 7.6 shows the corresponding four-terminal mobilities
and, for comparison, the respective two-terminal mobilities. The degradation effects are
the following: a significant degradation of the subthreshold performance, a decrease in
142 Oxygen-related traps in pentacene thin films
10 20 30 40 50107
108
109
1010
1011
1012
1013
0
3
6
9
12
15
P (c
m-2)
Ug (V)
Pfree / Ptotal
Pfree
Ptotal
P free /
Pto
tal (
%)
Figure 7.4: Total hole density Ptotal = CiUg/e, free hole density Pf ree and fraction of free holesPf ree/Ptotal at room temperature as derived from the temperature dependence of the four-terminalconductivity. A significant fraction of the total gate-induced charge is trapped even at high gatevoltages. The dashed line is the ratio Pf ree/Ptotal if the contact resistance is neglected.
on-current, a decrease in effective mobility and a shift of the transfer characteristic to-
wards more positive voltages. Also the contact resistance is increased after the oxygen
exposure. At room temperature and Vg = −50 V it increases from the as grown value of
RcontactW = 2.8×104 Ωcm to 9.2×104 Ωcm after the oxygen exposure, i.e. it increases
by a factor of 3.3. The increase in contact resistance may be due to the fact that the
contact resistance is dominated by the film resistance. In a top-contact device, the holes
must pass from the electrodes through the pentacene film to the channel at the insulator-
semiconductor interface (see Chap. 6, Fig. 6.12 and [134]). Importantly, before and after
the oxygen exposure we have a device that is not limited by the contact resistance. At
Vg = −50 V for example, the contact resistance Rcontact is ≈ 14 times smaller than the
channel resistance Rchannel prior to the oxygen exposure and ≈ 9 times smaller than the
channel resistance after the oxygen exposure.
The degradation effects can be observed when the transistor is subjected to oxygen
in the dark. It is, however, much accelerated when the oxygen exposure is carried out in
the presence of light, i.e. in the presence of activated oxygen and oxygen radicals.
The degradation of the subthreshold performance and of the field-effect mobility
is a result of oxygen-related defects that cause electrically active trap states within the
mobility gap. The shift of the transfer characteristic is due to a change of the flatband
7.4 Results 143
10 0 -10 -20 -30 -40 -5010-12
10-11
10-10
10-9
10-8
10-7
10-6
Vd = -2V
After oxygen
As grown
-I d (A)
Vg (V)
Figure 7.5: Linear regime transfer characteristic of a pentacene thin-film transistor measured asgrown (blue line) and after oxidation (red line). The graph shows the forward and the reversesweeps in both cases. The characteristic oxygen-related degradation effects are a decrease insubthreshold performance, a decrease in on-current and a shift of the transfer characteristic tomore positive voltages. The current hysteresis is essentially unaffected.
0 -10 -20 -30 -40 -500.0
0.1
0.2
0.3
0.4
2-t.4-t.
2-t.4-t.
After oxygen
As grown
eff (
cm2 /V
s)
Vg (V)
Figure 7.6: Four-terminal effective mobility (4-t.) from an as grown sample (blue pentagons) andafter the oxygen exposure (red circles). At Vg = −50 V for example, the contact-corrected field-effect mobility decreases from µe f f = 0.35 cm2/Vs to 0.17 cm2/Vs, i.e. it is reduced by a factorof 2.1. The dashed lines represent the respective two-terminal mobility (2-t.) where the contactresistance is neglected.
voltage. It is known that oxygen can cause changes of the flatband voltage in an organic
semiconductor device. [197, 198]
144 Oxygen-related traps in pentacene thin films
0 -10 -20 -30 -40 -500
1x10-7
2x10-7
3x10-7
4x10-7
5x10-7
0 -10 -20 -30 -40 -5010-12
10-11
10-10
10-9
10-8
10-7
0 -10 -20 -30 -40 -500
5x10-8
1x10-7
2x10-7
2x10-7
3x10-7
0 -10 -20 -30 -40 -5010-12
10-11
10-10
10-9
10-8
10-7
Vd = -2V
-I d (A
)
(b)
As grown
-I d (A
)
247 K257 K270 K278 K288 K297 K
-I d (A
)
Vg (V)
Vg (V)
(a)
248 K259 K270 K281 K289 K298 K
After oxygen
-I d (A
)
Vd = -2V
Vg (V)
Vg (V)
Figure 7.7: (a) Linear regime transfer characteristics at various temperatures from an as grownsample, i.e. prior to oxygen exposure. (b) Transfer characteristics at similar temperatures afterexposing the sample to 1 bar of oxygen for 23 h (19 h in the dark and 4 h in white light). Thetemperature dependence of the drain current in the subthreshold regime is much more pronouncedafter the oxygen exposure.
7.4.3 Oxygen-related traps
We now turn to the determination of the trap densities prior to and after the oxygen ex-
posure. For the temperature-dependent gated four-terminal measurements, we used a low
cooling rate (0.2− 0.25 /min) in order not to damage the sample. In Fig. 7.7 we show
the temperature dependence of the transfer characteristics prior to and after the oxygen
exposure. The temperature dependent measurements in Fig. 7.7 are from the same sample
as the measurements in Fig. 7.5 and Fig. 7.6 and were carried out shortly after these latter
measurements. The main difference after the oxygen exposure is that the temperature de-
pendence in the subthreshold regime (drain currents on a logarithmic scale) is much more
pronounced.
The DOS was extracted for both sets of measurements as described in Secs. 7.3.2
and 7.4.1. The main panel of Fig. 7.8 shows the final result on a logarithmic scale. The
difference between two adjacent data points in Fig. 7.8 corresponds to a change of 0.2 V in
the gate voltage. Some gate voltages Ug are indicated in the graph. The spacing between
7.4 Results 145
0.5 0.4 0.3 0.2 0.1
1018
1019
1020
1021
+O2
After oxygen
As grown 0.4 0.3 0.20
1x1020
2x1020
40 V
40 V
30 V
15 V
30 V
15 V
8 VUg = 8 V
DO
S (c
m-3eV
-1)
EV-E'F (eV)
Figure 7.8: Main panel: DOS as a function of energy relative to the mobility edge on a logarithmicscale. The blue pentagons are trap densities measured prior to the oxygen exposure and the redcircles are trap densities measured after the oxygen process. The oxidation of the pentacene filmleads to a significant increase in traps that are somewhat deeper in energy. The corresponding gatevoltage Ug above flatband is indicated. The inset shows the deeper traps on a linear scale.
the data points decreases as the gate voltage is increased, since at high gate voltages it is
increasingly difficult to shift the Fermi level E ′F at the insulator-semiconductor interface
towards the mobility edge due to the increased trap density.
We keep in mind that even in an ideal (trap-free) metal-insulator-semiconductor
(MIS) structure the interface potential increases with gate voltage more rapidly at low
gate voltages, than at high gate voltages. This is a screening effect. The screening de-
pends on the total charge in the device and it increases with gate voltage.
The oxygen exposure leads to a significant increase in the density of traps that are
somewhat deeper in energy (Fig. 7.8). The inset in Fig. 7.8 shows the deeper traps on
a linear scale. In Fig. 7.9 we show the difference of the trap densities prior to and after
the oxygen exposure on a linear scale (full and dash-dotted black line). We assume that
our method allows for a determination of the DOS to an accuracy of 5 %, and this is
indicated by the error bars in Fig. 7.9. At energies ≤ 0.25 eV from the mobility edge,
the difference in the DOS is comparable to or smaller than the estimated error and so, for
energies ≤ 0.25 eV, the DOS is essentially unaffected by the oxygen exposure. At larger
energies, however, the oxygen exposure leads to a broad peak of trap states. The dashed
red line in Fig. 7.9 is a Gaussian fit of the experimental data for energies ≥ 0.25 eV. Good
agreement is achieved with the Gaussian fit. The peak is centered at EC = 0.28 eV and
146 Oxygen-related traps in pentacene thin films
0.40 0.35 0.30 0.25 0.200
1x1019
2x1019
3x1019 Pentacene andoxygen
EC=0.28 eV
Diff
eren
ce in
DO
S (c
m-3eV
-1)
EV-E'F (eV)
Measured Gaussian fit
Figure 7.9: Difference of the DOS prior to and after the oxygen exposure (full and dash-dottedblack line). A relative error of 5 % is assumed for the determination of a trap DOS and this isindicated by the error bars. The oxygen exposure leads to a broad peak of trap states. A Gaussianfit for energies ≥ 0.25 eV (dashed red line) gives good agreement with the measured curve. Thepeak is centered at EC = 0.28 eV. At energies≤ 0.25 eV, the difference in the DOS is comparable toor smaller than the estimated error and the DOS is essentially unaffected by the oxygen exposure.
from the area under the peak we estimate a volume trap density of ≈ 4× 1018 cm−3.
With a density of the pentacene molecules of 3×1021 cm−3, this gives an oxygen-related
impurity concentration of ≈ 0.1 %, provided that each impurity results in one trap.
The DOS close to the mobility edge is well described by a single exponential func-
tion. A fit for energies ≤ 0.22 eV gives essentially identical characteristic slopes prior to
and after the oxygen exposure, i.e. respectively kT0 = 47 meV and kT0 = 48 meV. These
values are in good agreement with characteristic slopes from pentacene-based field-effect
transistors in the literature. A characteristic slope of kT0 = 40 meV is reported, as de-
termined by simulating the measured transfer characteristics of pentacene thin-film tran-
sistors. [152] Characteristic slopes of kT0 = 32− 37 meV were derived from pentacene
thin-film transistors with another device simulation program. [155,157] Yet, another pro-
gram gives a slope of kT0 = 100 meV for pentacene thin-film transistors. [156] However,
the band mobility µ0 needs to be fixed for the simulations and, depending on the choice
of the band mobility, slopes of up to 400 meV are also used. [156] In Chap. 6 we have
extracted characteristic slopes of kT0 = 32 meV shortly after the evaporation of the pen-
tacene and of kT0 = 37 meV in the aged thin film with reduced trap density with the initial
scheme by Grünewald et al. For a pentacene single-crystal device, a characteristic slope
of kT0 = 109 meV is reported. [160]
7.4 Results 147
0 10 20 30 40 500
4
8
12
16
0
4
8
12
16
2x1011
4x1011
6x1011
1.9
P fre
e / P
tota
l (%
)
Ug (V)
After oxygen
As grown
After oxygen
As grown
P free (
cm-2)
9 V
Figure 7.10: Upper panel: free hole density Pf ree prior to and after the oxygen exposure as derivedfrom the temperature-dependent gated four-terminal measurements. After the oxygen exposure thecurve is shifted by 9 V. The magnitude of the shift is closely linked to the density of the additionaltraps. Additional trapped holes with a density of 5× 1018 cm−3 can be estimated from the shift,which is highly consistent with the trap density estimated from an integration of the peak in Fig 7.9(4×1018 cm−3). Lower panel: corresponding fraction of free holes functions Pf ree/Ptotal prior toand after the oxygen exposure. At a given gate voltage Ug, the fraction of free holes is significantlyreduced after the oxygen exposure. At Ug = 40 V for example, the fraction of free holes drops from15 % to 8 %, i.e. it is reduced by a factor of 1.9.
7.4.4 Trap induced changes in the free hole density
The upper panel in Fig. 7.10 shows the free surface hole density (Eq. 7.13) prior to and
after the oxygen exposure from the two sets of temperature dependent measurements in
Fig. 7.7. The parameter l = T0/T was calculated with the characteristic slopes mentioned
above: for room temperature l = 1.9 (l/(2l− 1) = 0.68) prior to and after the oxygen
exposure. At sufficiently high gate voltages, the free hole density as a function of gate
voltage is shifted by 9 V towards higher gate voltages as a consequence of the oxygen
exposure.
The corresponding fractions of the free holes Pf ree/Ptotal were extracted according
to Eq. 7.18 and are shown in the lower panel of Fig. 7.10. The fraction of the free holes
changes significantly due to the oxygen exposure. At Ug = 40 V for example, 15 % of
148 Oxygen-related traps in pentacene thin films
all the induced holes are free prior to the oxygen exposure and this fraction drops to 8 %
after the oxygen exposure. It is reduced by a factor of 1.9. A fraction of free holes of
8− 15 % at 40 V is in good agreement with values for pentacene thin-film transistors
found in the literature. A fraction of free holes around 10 % is specified for comparable
total gate-induced charge densities. [155]
7.4.5 Stability of the oxygen-related defects
The DOS after the oxygen exposure was measured after a re-evacuation time of ≈ 22 h.
In order to elucidate the stability of the oxygen-related traps, the sample was kept in the
prober station at 10−8 mbar for an additional 7 days period. After that time, temperature
dependent gated four-terminal measurements were carried out and the DOS was extracted.
After these measurements, the sample was kept at 10−8 mbar for another 10 days. The
pentacene films were then slowly heated to 50 C at a rate of 0.2 /min with an electrical
heating element at the cryostat. The temperature was held for 2 h and the sample was
then left to cool down. The same procedure was repeated with a final temperature of
70 C. Because of the low heating and even lower cooling rates, the whole process took
3 days and the effective heating time was very long. Fig. 7.11 shows the DOS after
a re-evacuation time of ≈ 1 day (same as in Fig. 7.8), 8 days and 22 days, the latter time
including the heating procedure. The DOS functions are very similar and we can conclude
that the oxygen-related trap states are very stable.
7.5 Discussion
7.5.1 Effect of oxygen on the trap DOS
The DOS as extracted from the measurements of the as grown sample is of particular
interest, since the sample was kept at a pressure of the order of 10−8 mbar all along. The
trap densities are relatively high (1018 − 1021 cm−3eV−1) with a rather smooth depen-
dence on energy. In the case of the measurements on the as grown samples, the effect of
ambient gases can be excluded. It should be kept in mind that we used pentacene powder
that was re-crystallized in high vacuum twice. We conclude that the “amorphouslike”
trap DOS measured with an as grown sample is mainly due to structural defects within
the pentacene. Trap states on the surface of the gate dielectric, caused by certain chemical
groups for example, may also contribute to the states that are deeper in energy.
7.5 Discussion 149
0.5 0.4 0.3 0.2 0.1
1018
1019
1020
1021
DO
S (c
m-3eV
-1)
EV-E'F (eV)
1 day 8 days 22 days + heating
Figure 7.11: Trap densities after oxygen exposure. The re-evacuation time after the oxygen ex-posure is 1 day (full red line), 8 days (dashed black line) and 22 days (dash-dotted green line). Theoxygen-related traps are very stable, i.e. the DOS functions coincide. Prior to the last characteri-zation after 22 days, the sample was slowly heated to temperatures up to 70 C.
When pentacene is exposed to oxygen, the gas migrates into the pentacene film and
interacts with the pentacene molecules. This effect is expected to be accelerated if, in the
presence of light, oxygen is activated and its dissociation is aided. We observe signifi-
cant and irreversible changes in the transfer characteristics and in the DOS caused by the
oxygen exposure. It should be kept in mind, however, that several hours of exposure to
1 bar of oxygen are necessary in order to observe these changes. Consequently, pentacene
thin-films are not very sensitive towards oxidation.
The oxygen exposure leads to a broad peak of trap states centered at 0.28 eV, as
shown in Fig. 7.9. This suggests the degradation mechanism to be dominated by the cre-
ation of a specific oxygen-related defect. The large width of the peak (0.16 eV) is thought
to result from local structural disorder that modifies the on-site energy of the oxygen-
affected molecules. As a matter of fact, very similar arguments are used to explain the
smooth distribution of trap states in hydrogenated amorphous silicon. Even small devia-
tions in the local structure of a defect lead to a different electronic state (see Fig. 2.8(a) in
Chap. 2 and [199]).
Theoretical studies predict various types of oxygen-related defects in pentacene.
[86, 200] For example, one study predicts oxygen defects, in which a H atom of a pen-
tacene molecule is replaced by an oxygen atom to form a C22H13O molecule. [86] The
oxygen forms a double bond with the respective C atom and the pz orbital at this atom
150 Oxygen-related traps in pentacene thin films
no longer participates in the π-electron system of the pentacene molecule. The oxidation
at the middle ring is shown to be energetically most favourable. These oxygen defects
are expected to lead to trap states in the mobility gap. [86] Another theoretical study pre-
dicts other oxygen defects. [200] An example is a single oxygen intermolecular bridge
where a single oxygen atom is covalently bound to the carbon atoms on the center rings
of two neighbouring pentacene molecules. This defect, for instance, is calculated to lead
to electrically active traps at 0.33 and 0.4 eV above the valence band maximum. [200]
7.5.2 Influence of oxygen-related traps on the field-effect mobility
It is immediately plausible that the oxygen-related traps which are somewhat deeper in
energy result in a degradation of the subthreshold performance of the thin-film transistors.
We do, however, also observe a significantly decreased field-effect mobility after oxygen
exposure. This can be understood as follows. The deep traps created by the oxidation
need to be filled at first and the position of the Fermi level lags behind the position of
the Fermi level before the oxygen exposure. This is indicated in Fig. 7.8 by labeling the
corresponding gate voltages Ug. At the same gate voltage (which is proportional to the
total gate-induced hole density), the Fermi level is further away from the mobility edge.
The fraction of free holes, however, exponentially depends on the position of the Fermi
level E ′F at the insulator-semiconductor interface (Eq. 7.18). The field-effect mobility as
described by Eq. 7.1 is proportional to Pf ree/Ptotal and so a reduction in the fraction of free
holes readily affects the field-effect mobility. At Ug = 40 V, for example, the fraction of
free holes is reduced by a factor of 1.9 after the oxygen exposure (main panel of Fig. 7.10).
In addition, it is quite possible that the mobility µ0 of the delocalized charge above
the mobility edge is changed after the oxygen exposure. With Eq. 7.19, this mobility is
estimated to be µ0 = 1.2 cm2/Vs prior to the oxygen exposure and µ0 = 0.95 cm2/Vs after
the oxygen exposure. We have a reduction by a factor of 1.3 and a change of the “intrinsic”
charge transport. Conclusively, the major cause for the reduction of the effective field-
effect mobility is occupancy statistics, and a reduction of the mobility above the mobility
edge also plays a role.
The reduction in the mobility µ0 might be explained by a scattering of charge carriers
at the oxygen-related defects. Another indication that scattering plays a role in organic
field-effect transistors is the fact that the mobilities µ0 that we extract are lower than the
best field-effect mobilities (up to 5 cm2/Vs) from pentacene thin-film transistors. [14] As
already mentioned in Chap. 2, repeated purification of pentacene has been shown to lead
7.5 Discussion 151
to very high mobilities in pentacene single crystals. [87] This effect is attributable to the
reduction of the concentration of the oxidized pentacene species 6,13-pentacenequinone
(Fig. 2.15) which degrades the transport properties by scattering the charge carriers. [87]
7.5.3 Consistency check: trapped holes vs. traps
The upper panel in in Fig. 7.10 shows that the oxidation causes a shift of the curve for
the free hole density Pf ree by ∆Ug = 9 V. The same free hole density Pf ree is realized for
different total hole densities Ptotal . Clearly, for an identical number of free holes, the
difference in the number of total holes must be attributed to a difference in the number
of the trapped holes. Consequently, due to the oxygen-related traps, we have additional
holes that are trapped with a density of Ci∆Ug/e = 7.5× 1011 cm−2. Except at very low
gate voltages above the flatband voltage, the charge in an organic field-effect transistor is
concentrated at the insulator-semiconductor interface. As explained above, our extraction
scheme only considers currents above 1 nA. Therefore, it is reasonable to assume that the
holes are trapped in a region at the insulator-semiconductor interface with a thickness of
the order of one molecular layer (≈ 1.5 nm for pentacene). This gives a volume density
of trapped holes of ≈ 5× 1018 cm−3. From the area under the peak in the trap DOS, we
have derived a trap density of ≈ 4×1018 cm−3 which is in very good agreement with the
density of the trapped holes.
7.5.4 Deep traps and device performance
The results in this chapter reveal how an increase in the density of deeper traps can sig-
nificantly affect the field-effect mobility. The influence of deep traps on the device char-
acteristics is of most general concern because deep traps can have various origins. Trap
states caused by the surface of the gate dielectric, for example, are expected to be elec-
tronically deep traps. In Chap. 5 we have seen that the use of a polymeric gate dielectric
not only leads to an improved subthreshold swing but can also result in improved mobili-
ties. Importantly, in the case of the SC-FET’s, the semiconductor is grown separately, and
growth-related effects can be excluded. Similar observations are reported in [103,180]. In
the light of the present study, it seems possible that these effects can be solely understood
with transport in extended states above a mobility edge and a distribution of trap states:
a reduced number of deep traps leads to an increased number of free carriers above the
mobility edge and to a higher mobility of that charge.
152 Oxygen-related traps in pentacene thin films
7.6 Conclusions
Pentacene-based thin film transistors were characterized without exposing the samples to
ambient air (as grown) and after exposure to oxygen in combination with white light. The
exposure of the pentacene to the oxidizing agent causes a degradation of the subthreshold
performance, a decrease in field-effect mobility, a shift of the flatband voltage and an
increased contact resistance.
Contact-corrected trap state functions were extracted from temperature dependent
gated four-terminal measurements. We show that the exposure to oxygen leads to a broad
peak of trap states centered at 0.28 eV. The emergence of a peak indicates the process to
be dominated by the creation of a specific oxygen-related defect. The large width of the
peak is a result of the energetically different surroundings induced by structural disorder.
The oxygen defects are very stable and are likely to be caused by pentacene molecules
with covalently bound oxygen.
The decrease in field-effect mobility is caused by the oxygen-related deep traps.
These states are filled upon increasing the gate voltage and the Fermi level at the inter-
face lags behind the position it has in as-deposited samples. This leads to a significantly
smaller fraction of free holes. The magnitude of the shift in the free hole function is highly
consistent with the density of the oxygen-related traps (≈ 4× 1018 cm−3), as estimated
from the difference in the trap DOS prior to and after the oxygen exposure. In addition,
the oxygen exposure leads to a decrease of the mobility of the charge carriers above the
mobility edge.
The results can be seen from a more general point of view. At first, the tempera-
ture dependent measurements are self-consistent with the assumption of a mobility edge
or transport level, thus contributing to an understanding of charge transport in organic
semiconductors. Moreover, they are an example of the way in which deeper traps can
influence the effective field-effect mobility.
Theoretical studies may help to identify the oxygen defect and organic synthetic
chemistry may soon find a way to tailor organic semiconductors where the creation of
defects by oxidation is completely inhibited.
8 Summary, conclusions and outlook
We carried out experiments in order to contribute to a quantification, identification and
elimination of electronic trap states in organic field-effect transistors based on oligomeric
semiconductors and to a better understanding of the charge transport mechanism in these
materials.
In essence, the experimental findings and conclusions are the following: we
have seen that it is quite unlikely to obtain field-effect mobilities ≥ 1 cm2/Vs with a
new oligomeric semiconductor, even if single crystals are employed in the transistors
(Fig. 3.2). However, most new oligomers studied for this thesis led to operating transis-
tors with mobilities ranging from 10−7 cm2/Vs to 0.1 cm2/Vs. Mobilities in single crystal
field-effect transistors (SC-FET’s) were found to be always higher than in thin-film tran-
sistors (TFT’s) by a factor of 3− 670, depending on the material (as long as operating
devices of both types could be made). This is likely due to increased structural disorder in
the evaporated films, as compared to the single crystals. As a matter of fact, the detailed
study of 7,14-Diphenyl-chromeno[2,3-b]xanthene (DPCX) in Chap. 4 shows very clearly
that the effective field-effect mobility critically depends on structural order (Fig. 4.2). The
mobility is 0.16 cm2/Vs in SC-FET’s, 0.01 cm2/Vs in polycrystalline films grown on OTS
and 2×10−5 cm2/Vs in amorphous films grown on bare SiO2.
The few known oligomeric semiconductors that do lead to mobilities ≥ 1 cm2/Vs
(e.g. pentacene) may have a very strong tendency to crystallize and to form high quality
crystals and films with few structural defects. This interpretation would be consistent with
the X-ray diffraction measurements in Fig 4.6. For the polycrystalline film of DPCX, the
peaks can only be observed up to the second order and the intensity of the second order
peaks is rather weak. For pentacene films 4-5 orders of the diffraction peaks are readily
observed. [69, 201] The structural disorder would cause traps, thus reducing the effective
mobility. The intrinsic mobility as derived from the orbital overlap of adjacent molecules
154 Summary, conclusions and outlook
in the crystal is also important for the field-effect mobility of a given material. It is not
unreasonable to suggest that the orbital overlap (interaction strength) is related not only
to the intrinsic mobility but also to the tendency to crystallize and to form high quality
crystals and films.
In Chap. 5 we have seen that SC-FET’s with rubrene or pentacene can have a very
high electrical stability if an amorphous fluoropolymer gate dielectric is used. From
these measurements we conclude that electrical instability in organic field-effect tran-
sistors is not an intrinsic phenomena and is not e.g. due to the formation of bipolarons.
The experimental work in Chap. 5 identifies two important causes of gate bias stress ef-
fects in organic field-effect transistors. Since gate bias stress effects are somewhat more
pronounced in pentacene thin-film transistors than in the highly stable pentacene single
crystal transistors with the same gate dielectric, we can conclude that certain structural
defects within the semiconductor are a cause of electrical instability. Moreover, since
SC-FET’s with comparable single crystals and other gate dielectrics are significantly less
stable [103, 106], the choice of an unsuitable gate dielectric is the second cause of gate
bias stress effects. The gate dielectric may lead to electrical instability because of the
presence of electrically active traps on the surface of the gate dielectric. These traps may
also be created by an electrochemical reaction which would be driven by applying a gate
voltage to the device. For example, adsorbed water may react with the organic semicon-
ductor or with the gate dielectric and thus form traps. Moreover, dipolar disorder caused
by a polar gate dielectric may also lead to long-lived traps that cause electrical instability.
The highly hydrophobic, low dielectric constant, amorphous fluoropolymer gate di-
electric CytopT M works extremely well in combination with two organic semiconductors
of different chemical nature, i.e. pentacene and rubrene. We arrive at the conclusion that
the nature of the gate dielectric is much more important than the chemistry of the organic
semiconductor in order to obtain transistors with a high electrical stability. The electrical
stability may to some extend depend on the chemical nature of the organic semiconductor.
For example, passivating the most reactive sites in the pentacene molecule by means of
synthetic chemistry may impeade the electrochemical reaction of the organic semicon-
ductor with water. This leads to better semiconductors in that respect (e.g. DPCX in
Chap. 4).
Another aspect of the study in Chap. 5 is the issue of contact resistance. We have
pointed out the substantial influence of parasitic contact resistances in the case of high
quality single crystal transistors. Parasitic contact resistances can significantly affect the
155
transfer characteristics and the temperature dependence of the drain current. It is therefore
very important to correct for parasitic contact resistances when extracting the trap DOS
and other transport parameters from the characteristics of organic field-effect transistors.
This correction significantly contributes to the quality of the studies in Chap. 6 and 7.
The new device fabrication and characterization system allows for measurements
with a high degree of control. For example, it allows for the electrical characterization
of organic field-effect transistors by temperature-dependent gated four-terminal measure-
ments without any unintentional air exposure between the transistor fabrication and char-
acterization. The experiments with the system in Chap. 6 reveal that the performance of
pentacene-based field-effect transistors improves with time. As for the causes of elec-
trical instability, it is certainly important to understand the origin of these changes for a
successful commercialization of organic field-effect transistors. We show that the perfor-
mance improvement is due to structural changes within the pentacene film. The relevant
structural defects that anneal at room temperature are related to shallow traps with an en-
ergy of less than 0.2 eV from the mobility edge/transport level. The effective field-effect
mobility critically depends on the density of these traps. The exact microscopic nature
of the relevant structural defects is not known, but it seems plausible that some of these
defects are sliding defects. Once more, we can confirm the importance of structural order
of the semiconducting layer.
For the study of oxygen-related traps in pentacene films described in Chap. 7 we
have developed an improved scheme to extract the spectral density of trap states and other
important transport parameters from temperature-dependent gated four-terminal measure-
ments in an unambiguous and straightforward fashion. The measurements on pentacene
thin-film transistors confirm the validity of assuming charge transport above a mobility
edge or in a transport level and a distribution of localized states below the mobility edge.
Moreover, we demonstrate that the trap DOS in pentacene thin films is indeed quite well
described by a single exponential function of energy. Most of the trap states within the
pentacene films originate from structural defects. We show that the exposure of pentacene
films to oxygen results in the formation of a peak of trap states centered at 0.28 eV from
the mobility edge. The traps are thought to result from a specific oxygen-induced impu-
rity where oxygen is covalently bound. The large width of the peak (0.16 eV) is likely
caused by local structural disorder that modifies the on-site energy of the oxygen-affected
molecules.
156 Summary, conclusions and outlook
The oxygen-related deep traps reduce the effective mobility by decreasing the num-
ber of free carriers at a given gate voltage and their mobility above the mobility edge.
This may also be true for deep traps of other origin, e.g. certain chemical groups on the
surface of the gate dielectric. Provided that all other parameters remain constant, we ex-
pect a correlation of general validity between the subthreshold swing and the effective
field-effect mobility.
We would like to suggest several directions of possible studies in order to continue
the experimental work described in this thesis. The environmental and electrical stability
of DPCX could be further tested. This would best be done with DPCX/CytopT M single
crystal transistors and by controlled gas exposure in the prober station of the device fab-
rication and characterization system. It would be interesting to compare the stability of
DPCX with the stability of pentacene in a systematic fashion.
We believe that the extraction scheme that we have developed in Chap. 7 is a valu-
able tool to further elucidate the origin of trap states and the charge transport mechanism
in organic field-effect transistors, both with an oligomeric and with a polymeric semi-
conductor. Specifically, it should be rewarding to use the extraction scheme in order to
calculate the trap DOS and the “intrinsic” mobility of several different organic semicon-
ductors in field-effect transistors. It would be interesting to see to what extend the trap
DOS and the intrinsic mobility vary from material to material, also keeping in mind that
for an intrinsic mobility of e.g. µ0 = 0.01 cm2/Vs trap-controlled band transport would
certainly not be a self-consistent description anymore. The extraction scheme can also
be adapted to the bipolar case and temperature-dependent measurements on bipolar field-
effect transistors e.g. with a polymeric semiconductor [202,203] would thus allow for the
determination of the trap DOS in the whole bandgap. Moreover, it would be possible to
compare the “intrinsic” electron and hole mobilities. Furthermore, it may be worth a try to
correlate the spectral density of trap states from pentacene films with the film morphology
(e.g. grain size), as seen by AFM. Different films could be obtained by varying the depo-
sition conditions (e.g. substrate temperature and deposition rate). Since charge transport
takes place in the first few molecular layers of the organic films, we would however have
to work with ultrathin pentacene films (e.g. 5 nm thick).
Finally, it seems promising to improve the contacts of single crystal/CytopT M tran-
sistors. We have seen that the transistor characteristics from devices made with the “flip-
crystal approach” have serious contact problems. The contacts might be improved by
157
changing the device geometry and/or the contact material. Other metals than gold or
conducting polymers such as PEDOT:PSS [204] might be useful.
Bibliography
[1] M. Pope and C. E. Swenberg. Electronic processes in organic crystals and polymers. 2nd ed., Oxford University Press, New
York (1999)
[2] J. S. E. Lilienfeld. “Method and apparatus for controlling electric currents”. US Patent No. 1,745,175 (Jan. 28, 1930)
[3] S. M. Sze. Physics of semiconductor devices. Wiley, New York (1981)
[4] F. Ebisawa, T. Kurokawa, and S. Nara. “Electrical properties of polyacetylene/polysiloxane interface”. J. Appl. Phys. 54, 3255
(1983)
[5] J. H. Burroughes, C. A. Jones, and R. H. Friend. “New semiconductor device physics in polymer diodes and transistors”.
Nature 335, 137 (1988)
[6] J. H. Burroughes, R. H. Friend, and P. C. Allen. “Field-enhanced conductivity in polyacetylene - construction of a field-effect
transistor”. J. Phys. D 22, 956 (1989)
[7] A. Tsumura, H. Koezuka, and T. Ando. “Macromolecular electronic device: Field-effect transistor with a polythiophene thin
film”. Appl. Phys. Lett. 49, 1210 (1986)
[8] H. Koezuka, A. Tsumura, and T. Ando. “Field-effect transistor with polythiophene thin film”. Synth. Metal. 18, 699 (1987)
[9] A. Tsumura, H. Koezuka, and T. Ando. “Polythiophene field-effect transistor: Its characteristics and operation mechanism”.
Synth. Metal. 25, 11 (1988)
[10] A. Assadi, C. Svensson, M. Willander, and O. Inganäs. “Field-effect mobility of poly(3-hexylthiophene)”. Appl. Phys. Lett.
53, 195 (1988)
[11] R. Madru, G. Guillaud, M. Al Sadoun, M. Maitrot, J.-J. André, J. Simon, and R. Even. “A well-behaved field effect transistor
based on an intrinsic molecular semiconductor”. Chem. Phys. Lett. 145, 343 (1988)
[12] G. Horowitz, D. Fichou, X. Peng, Z. Xu, and F. Garnier. “A field-effect transistor based on conjugated alpha-sexithienyl”.
Solid State Commun. 72, 381 (1989)
[13] J. M. Shaw and P. F. Seidler. “Organic electronics: Introduction”. IBM J. Res. & Dev. 45, 3 (2001)
160 Bibliography
[14] T. W. Kelley, D. V. Muyres, P. F. Baude, T. P. Smith, and T. D. Jones. “High performance organic thin film transistors”. Mat.
Res. Soc. Symp. Proc. 771, L6.5.1 (2003)
[15] X.-H. Zhang, B. Domercq, and B. Kippelen. “High-performance and electrically stable C60 organic field-effect transistors”.
Appl. Phys. Lett. 91, 092114 (2007)
[16] J. Takeya, J. Kato, K. Hara, M. Yamagishi, R. Hirahara, K. Yamada, Y. Nakazawa, S. Ikehata, K. Tsukagoshi, Y. Aoyagi,
T. Takenobu, and Y. Iwasa. “In-crystal and surface charge transport of electric-field-induced carriers in organic single-crystal
semiconductors”. Phys. Rev. Lett. 98, 196804 (2007)
[17] S. K. Park, T. N. Jackson, J. E. Anthony, and D. A. Mourey. “High mobility solution processed 6,13-bis(triisopropyl-
silylethynyl) pentacene organic thin film transistors”. Appl. Phys. Lett. 91, 063514 (2007)
[18] C. D. Sheraw, T. N. Jackson, D. L. Eaton, and J. E. Anthony. “Functionalized pentacene active layer organic thin-film
transistor”. Adv. Mater. 15, 2009 (2003)
[19] A. Salleo, T. W. Chen, A. R. Völkel, Y. Wu, P. Liu, B. S. Ong, and R. A. Street. “Intrinsic hole mobility and trapping in a
regioregular poly(thiophene)”. Phys. Rev. B 70, 115311 (2004)
[20] I. McCulloch, M. Heeney, C. Bailey, K. Genevicius, I. MacDonald, M. Shkunov, D. Sparrowe, S. Tierney, R. Wagner, W. Zhang,
M. L. Chabinyc, R. J. Kline, M. D. McGehee, and M. F. Toney. “Liquid-crystalline semiconducting polymers with high charge-
carrier mobility”. Nature Mater. 5, 328 (2006)
[21] K. H. Probst and N. Karl. “Energy levels of electron and hole traps in the bandgap of doped anthracene crystals”. Phys. Status
Solidi A 27, 499 (1975)
[22] E. A. Silinsh and V. Cápek. Organic molecular crystals. AIP Press, New York (1994)
[23] W. Warta and N. Karl. “Hot holes in naphthalene - high, electric-field-dependent mobilities”. Phys. Rev. B 32, 1172 (1985)
[24] A. Troisi and G. Orlandi. “Charge transport regime of crystalline organic semiconductors: diffusion limited by thermal off-
diagonal disorder”. Phys. Rev. Lett. 96, 086601 (2006)
[25] E. Heilbronner and H. Bock. Das HMO-Modell und seine Anwendung. Verl. Chemie, Weinheim (1970)
[26] C. Mortimer. Chemie. Georg Thieme Verlag, Stuttgart (1996)
[27] W. L. Kalb. Structure-performance relationship in pentacene-based thin-film transistors. Master’s thesis, RWTH Aachen
(2004)
[28] A. Gavezzotti and G. R. Desiraju. “A systematic analysis of packing energies and other packing parameters for fused-ring
aromatic hydrocarbons”. Acta Cryst. B 44, 427 (1988)
[29] G. R. Desiraju and A. Gavezzotti. “Crystal-structures of polynuclear aromatic-hydrocarbons - classification, rationalization
and prediction from molecular structure”. Acta Cryst. B 45, 473 (1989)
[30] R. B. Campbell, J. Trotter, and J. M. Robertson. “Crystal and molecular structure of pentacene”. Acta Cryst. 14, 705 (1961)
Bibliography 161
[31] C. C. Mattheus, A. B. Dros, J. Baas, A. Meetsma, J. L. de Boer, and T. T. M. Palstra. “Polymorphism in pentacene”. Acta
Crystallogr., Sect. C: Cryst. Struct. Commun. 57, 939 (2001)
[32] T. Siegrist, C. Besnard, S. Haas, M. Schiltz, P. Pattison, D. Chernyshov, B. Batlogg, and C. Kloc. “A polymorph lost and found:
the high-temperature crystal structure of pentacene”. Adv. Mater. 19, 2079 (2007)
[33] R. Ruiz, B. Nickel, N. Koch, L. C. Feldman, R. F. Haglund, A. Kahn, and G. Scoles. “Pentacene ultrathin film formation on
reduced and oxidized Si surfaces”. Phys. Rev. B 67, 125406 (2003)
[34] J. L. Brédas, J. P. Calbert, D. A. da Silva Filho, and J. Cornil. “Organic semiconductors: A theoretical characterization of the
basic parameters governing charge transport”. Proc. Natl. Acad. Sci. 99, 5804 (2002)
[35] P. Puschnig and C. Ambrosch-Draxl. “Density-functional study for the oligomers of poly(para-phenylene): Band structures
and dielectric tensors”. Phys. Rev. B 60, 7891 (1999)
[36] D. Nabok, P. Puschnig, C. Ambrosch-Draxl, O. Werzer, R. Resel, and D.-M. Smilgies. “Crystal and electronic structures of
pentacene thin films from grazing-incidence x-ray diffraction and first-principles calculations”. Phys. Rev. B 76, 235322 (2007)
[37] J. E. Anthony. “The larger acenes: Versatile organic semiconductors”. Angew. Chem. Int. Ed. 47, 452 (2008)
[38] T. Holstein. “Studies of polaron motion part 1. The molecular crystal model”. Ann. Phys. 8, 325 (1959)
[39] T. Holstein. “Studies of polaron motion part 2. The small polaron”. Ann. Phys. 8, 343 (1959)
[40] K. Hannewald, V. M. Stojanovic, J. M. T. Schellekens, and P. A. Bobbert. “Theory of polaron bandwidth narrowing in organic
molecular crystals”. Phys. Rev. B 69, 075211 (2004)
[41] G. B. Street and W. D. Gill. “Photoconductivity and drift mobilities in single crystal realgar (As4S4)”. Phys. Stat. Sol. 18, 601
(1966)
[42] D. J. Gibbons and W. E. Spear. “Electron hopping transport and trapping phenomena in orthorhombic sulphur crystals”. J.
Phys. Chem. Solids 27, 1917 (1966)
[43] J. Yamashita and T. Kurosawa. “On electronic current in NiO”. J. Phys. Chem. 5, 34 (1958)
[44] H. Sirringhaus. “Device physics of solution-processed organic field-effect transistors”. Adv. Mater. 17, 2411 (2005)
[45] V. Coropceanu, J. Cornil, D. A. da Silva Filho, Y. Olivier, R. Silbey, and J. L. Bredas. “Charge transport in organic
semiconductors”. Chem. Rev. 107, 926 (2007)
[46] G. Nan, L. Wang, X. Yang, Z. Shuai, and Y. Zhao. “Charge transfer rates in organic semiconductors beyond first-order
perturbation: From weak to strong coupling regimes”. J. Chem. Phys. 130, 024704 (2009)
[47] A. Troisi and G. Orlandi. “Dynamics of the intermolecular transfer integral in crystalline organic semiconductors”. J. Phys.
Chem. A 110, 4065 (2006)
[48] A. Troisi. “Prediction of the absolute charge mobility of molecular semiconductors: the case of rubrene”. Adv. Mater. 19, 2000
(2007)
162 Bibliography
[49] A. Troisi. “Charge dynamics through pi-stacked arrays of conjugated molecules: effect of dynamic disorder in different
transport/transfer regimes”. Molecular Simulation 32, 707 (2006)
[50] H. A. v. Laarhoven, C. F. J. Flipse, M. Koeberg, M. Bonn, E. Hendry, G. Orlandi, O. D. Jurchescu, T. T. M. Palstra, and
A. Troisi. “On the mechanism of charge transport in pentacene”. J. Chem. Phys. 129, 044704 (2008)
[51] C. Krellner, S. Haas, C. Goldmann, K. P. Pernstich, D. J. Gundlach, and B. Batlogg. “Density of bulk trap states in organic
semiconductor crystals: discrete levels induced by oxygen in rubrene”. Phys. Rev. B 75, 245115 (2007)
[52] G. Horowitz, R. Hajlaoui, and P. Delannoy. “Temperature dependence of the field-effect mobility of sexithiophene -
Determination of the density of traps”. J. Phys. III France 5, 355 (1995)
[53] F. Schauer. “Temperature dependent field effect in organic-based thin-film transistor and its spectroscopic character”. J. Appl.
Phys. 86, 524 (1999)
[54] J. M. Marshall. “Carrier diffusion in amorphous semiconductors”. Rep. Prog. Phys. 46, 1235 (1983)
[55] R. J. Chesterfield, J. C. McKeen, C. R. Newman, P. C. Ewbank, D. A. da Silva Filho, J.-L. Brédas, L. L. Miller, K. R. Mann,
and C. D. Frisbie. “Organic thin film transistors based on n-alkyl perylene diimides: Charge transport kinetics as a function of
gate voltage and temperature”. J. Phys. Chem. B 108, 19281 (2004)
[56] P. W. Anderson. “The size of localized states near the mobility edge”. Proc. Nat. Acad. Sci. USA 69 (5), 1097 (1972)
[57] N. F. Mott and E. A. Davis. Electronic processes in non-crystalline materials. Clarendon Press, Oxford (1971)
[58] A. Miller and E. Abrahams. “Impurity conduction at low concentrations”. Phys. Rev. 120, 745 (1960)
[59] V. I. Arkhipov, E. V. Emelianova, and G. J. Adriaenssens. “Effective transport energy versus the energy of most probable jumps
in disordered hopping systems”. Phys. Rev. B 64, 125125 (2001)
[60] V. I. Arkhipov, P. Heremans, E. V. Emelianova, G. J. Adriaenssens, and H. Bässler. “Charge carrier mobility in doped
semiconducting polymers”. Appl. Phys. Lett. 82, 3245 (2003)
[61] J. Veres, S. D. Ogier, S. W. Leeming, D. C. Cupertino, and S. M. Khaffaf. “Low-k insulators as the choice of dielectrics in
organic field-effect transistors”. Adv. Funct. Mater. 13, 199 (2003)
[62] J. Veres, S. Ogier, G. Lloyd, and D. de Leeuw. “Gate insulators in organic field-effect transistors”. Chem. Mater. 16, 4543
(2004)
[63] J. Owen, G. P. Sworakowski, J. M. Thomas, D. F. Williams, and J. O. Williams. “Carrier traps in ultrahigh purity single-crystals
of anthracene”. J. Chem. Soc. Far. II 70, 853 (1974)
[64] J. O. Williams and J. M. Thomas. “Lattice imperfections in organic solids .1. anthracene”. Trans. Far. Soc. 63, 1720 (1967)
[65] J. N. Sherwood. “Lattice defects in organic crystals”. Mol. Cryst. Liq. Cryst. 9, 37 (1969)
[66] S. E. Fritz, T. W. Kelley, and C. D. Frisbie. “Effect of dielectric roughness on performance of pentacene TFTs and restoration
of performance with a polymeric smoothing layer”. J. Phys. Chem. B 109, 10574 (2005)
Bibliography 163
[67] D. Knipp, R. A. Street, A. Völkel, and J. Ho. “Pentacene thin film transistors on inorganic dielectrics: morphology, structural
properties, and electronic transport”. J. Appl. Phys. 93, 347 (2003)
[68] F.-J. Meyer zu Heringdorf, M. C. Reuter, and R. M. Tromp. “Growth dynamics of pentacene thin films”. Nature 412, 517
(2001)
[69] W. Kalb, P. Lang, M. Mottaghi, H. Aubin, G. Horowitz, and M. Wuttig. “Structure-performance relationship in
pentacene/Al2O3 thin-film transistors”. Synth. Metal. 146, 279 (2004)
[70] J. A. Venables, G. D. T. Spiller, and M. Hanbucken. “Nucleation and growth of thin-films”. Rep. Prog. Phys. 47, 399 (1984)
[71] C. D. Dimitrakopoulos and R. L. Malenfant. “Organic thin film transistors for large area electronics”. Adv. Mater. 14, 99
(2002)
[72] K. P. Pernstich, S. Haas, D. Oberhoff, C. Goldmann, D. J. Gundlach, B. Batlogg, A. N. Rashid, and G. Schitter. “Threshold
voltage shift in organic field effect transistors by dipole monolayers on the gate insulator”. J. Appl. Phys. 96, 6431 (2004)
[73] H.-L. Cheng, Y.-S. Mai, W.-Y. Chou, L.-R. Chang, and X.-W. Liang. “Thickness-dependent structural evolutions and growth
models in relation to carrier transport properties in polycrystalline pentacene thin films”. Adv. Funct. Mater. 17, 3639 (2007)
[74] S. Verlaak, C. Rolin, and P. Heremans. “Microscopic description of elementary growth processes and classification of structural
defects in pentacene thin films”. J. Phys. Chem. B 111, 139 (2007)
[75] J. H. Kang, D. da Silva, J. L. Bredas, and X. Y. Zhu. “Shallow trap states in pentacene thin films from molecular sliding”.
Appl. Phys. Lett. 86, 152115 (2005)
[76] T. Komoda, Y. Endo, K. Kyuno, and A. Toriumi. “Field-dependent mobility of highly oriented pentacene thin-film transistors”.
Jpn. J. Appl. Phys. 41, 2767 (2002)
[77] R. B. Ye, M. Baba, K. Suzuki, Y. Ohishi, and K. Mori. “Effect of thermal annealing on morphology of pentacene thin films”.
Jpn. J. Appl. Phys. 42, 4473 (2003)
[78] F. Dinelli, M. Murgia, F. Biscarini, and D. M. De Leeuw. “Thermal annealing effects on morphology and electrical response
in ultrathin film organic transistors”. Synth. Metal. 146, 373 (2004)
[79] S. J. Kang, M. Noh, D. S. Park, H. J. Kim, C. N. Whang, and C. H. Chang. “Influence of postannealing on polycrystalline
pentacene thin film transistor”. J. Appl. Phys. 95, 2293 (2004)
[80] T. Sekitani, S. Iba, Y. Kato, Y. Noguchi, T. Someya, and T. Sakurai. “Suppression of DC bias stress-induced degradation of
organic field-effect transistors using postannealing effects”. Appl. Phys. Lett. 87, 073505 (2005)
[81] D. Guo, S. Ikeda, K. Saiki, H. Miyazoe, and K. Terashima. “Effect of annealing on the mobility and morphology of thermally
activated pentacene thin film transistors”. J. Appl. Phys. 99, 094502 (2006)
[82] A. C. Mayer, M. T. Lloyd, D. J. Herman, T. G. Kasen, and G. G. Malliaras. “Postfabrication annealing of pentacene-based
photovoltaic cells”. Appl. Phys. Lett. 85, 6272 (2004)
164 Bibliography
[83] J. Pflaum, J. Niemax, and A. K. Tripathi. “Chemical and structural effects on the electronic transport in organic single crystals”.
Chem. Phys. 325, 152 (2006)
[84] N. Karl. Festkörperprobleme 14, 261 (1974)
[85] C. C. Mattheus, J. Baas, A. Meetsma, J. L. de Boer, C. Kloc, T. Siegrist, and T. T. M. Palstra. “A 2:1 cocrystal of 6,13-
dihydropentacene and pentacene”. Acta Cryst. E58, o1229 (2002)
[86] J. E. Northrup and M. L. Chabinyc. “Gap states in organic semiconductors: Hydrogen- and oxygen-induced states in
pentacene”. Phys. Rev. B 68, 041202(R) (2003)
[87] O. D. Jurchescu, J. Baas, and T. T. M. Palstra. “Effect of impurities on the mobility of single crystal pentacene”. Appl. Phys.
Lett. 84, 3061 (2004)
[88] M. Ahles, R. Schmechel, and H. von Seggern. “N-type organic field-effect transistors based on interface-doped pentacene”.
Appl. Phys. Lett. 85, 4499 (2004)
[89] 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 434, 194 (2005)
[90] M. H. Yoon, C. Kim, A. Facchetti, and T. J. Marks. “Gate dielectric chemical structure-organic field-effect transistor
performance correlations for electron, hole, and ambipolar organic semiconductors”. J. Am. Chem. Soc. 128, 12851 (2006)
[91] D. Kumaki, T. Umeda, and S. Tokito. “Influence of H2O and O2 on threshold voltage shift in organic thin-film transistors:
Deprotonation of SiOH on SiO2 gate-insulator surface”. Appl. Phys. Lett. 92, 093309 (2008)
[92] J. Sworakowski. “Effect of polar molecules on the transport and localization of charge carriers in molecular crystals”. Brazilian
Journal of Physics 29, 318 (1999)
[93] S. Verlaak and P. Heremans. “Molecular microelectostatic view on electronic states near pentacene grain boundaries”. Phys.
Rev. B 75, 115127 (2007)
[94] A. F. Stassen, R. W. I. de Boer, N. N. Iosad, and A. F. Morpurgo. “Influence of the gate dielectric on the mobility of rubrene
single-crystal field-effect transistors”. Appl. Phys. Lett. 85, 3899 (2004)
[95] I. N. Hulea, S. Fratini, H. Xie, C. L. Mulder, N. N. Iossad, G. Rastelli, S. Ciuchi, and A. F. Morpurgo. “Tunable Fröhlich
polarons in organic single-crystal transistors”. Nature Mater. 5, 982 (2006)
[96] A. R. Brown, C. P. Jarrett, D. M. de Leeuw, and M. Matters. “Field-effect transistors made from solution-processed organic
semiconductors”. Synth. Metal. 88, 37 (1997)
[97] J. Zaumseil and H. Sirringhaus. “Electron and ambipolar transport in organic field-effect transistors”. Chem. Rev. 107, 1296
(2007)
[98] A. Rolland, J. Richard, J. P. Kleider, and D. Mencaraglia. “Electrical properties of amorphous silicon transistors and MIS-
devices: comparative study of top nitride and bottom nitride configurations”. J. Electrochem. Soc. 140, 3679 (1993)
Bibliography 165
[99] M. McDowell, I. G. Hill, J. E. McDermott, S. L. Bernasek, and J. Schwartz. “Improved organic thin-film transistor performance
using novel self-assembled monolayers”. Appl. Phys. Lett. 88, 073505 (2006)
[100] G. Horowitz, R. Hajlaoui, D. Fichou, and A. El Kassmi. “Gate voltage dependent mobility of oligothiophene field-effect
transistors”. J. Appl. Phys. 85, 3202 (1999)
[101] G. Horowitz, P. Lang, M. Mottaghi, and H. Aubin. “Extracting parameters from the current-voltage characteristics of organic
field-effect transistors”. Adv. Funct. Mater. 14, 1069 (2004)
[102] M. Shur and M. Hack. “Physics of amorphous silicon based alloy field-effect transistors”. J. Appl. Phys. 55, 3831 (1984)
[103] C. Goldmann, C. Krellner, K. P. Pernstich, S. Haas, D. J. Gundlach, and B. Batlogg. “Determination of the interface trap
density of rubrene single-crystal field-effect transistors and comparison to the bulk trap density”. J. Appl. Phys 99, 034507
(2006)
[104] A. Salleo, F. Endicott, and R. A. Street. “Reversible and irreversible trapping at room temperature in poly(thiophene) thin-film
transistors”. Appl. Phys. Lett. 86, 263505 (2005)
[105] M. L. Chabinyc, F. Endicott, B. D. Vogt, D. M. DeLongchamp, E. K. Lin, Y. L. Wu, P. Liu, and B. S. Ong. “Effects of humidity
on unencapsulated poly(thiophene) thin-film transistors”. Appl. Phys. Lett. 88, 113514 (2006)
[106] C. Goldmann, D. J. Gundlach, and B. Batlogg. “Evidence of water-related discrete trap state formation in pentacene single-
crystal field-effect transistors”. Appl. Phys. Lett. 88, 063501 (2006)
[107] H. L. Gomes, P. Stallinga, M. Cölle, D. M. de Leeuw, and F. Biscarini. “Electrical instabilities in organic semiconductors
caused by trapped supercooled water”. Appl. Phys. Lett. 88, 082101 (2006)
[108] R. A. Street, A. Salleo, and M. L. Chabinyc. “Bipolaron mechanism for bias-stress effects in polymer transistors”. Phys. Rev.
B 68, 085316 (2003)
[109] R. A. Street, M. L. Chabinyc, and F. Endicott. “Chemical impurity effects on transport in polymer transistors”. Phys. Rev. B
76, 045208 (2007)
[110] M. J. Powell. “The physics of amorphous-silicon thin-film transistors”. IEEE Trans. Electron Devices 36, 2753 (1989)
[111] T. B. Singh, F. Meghdadi, S. Günes, N. Marjanovic, G. Horowitz, P. Lang, S. Bauer, and N. S. Sariciftci. “High-performance
ambipolar pentacene organic field-effect transistors on poly(vinyl alcohol) organic gate dielectric”. Adv. Mater. 17, 2315 (2005)
[112] K. P. Pernstich. The influence of trap states on charge transport in organic transistors. Ph.D. thesis, ETH Zurich, Nr. 17245
(2007)
[113] Y. Y. Lin, D. J. Gundlach, S. F. Nelson, and T. N. Jackson. “Stacked pentacene layer organic thin-film transistors with improved
characteristics”. IEEE Electron Device Lett. 18, 606 (1997)
[114] Y.-Y. Lin, D. J. Gundlach, S. F. Nelson, and T. N. Jackson. “Pentacene-based organic thin-film transistors”. IEEE Trans.
Electron Devices 44, 1325 (1997)
166 Bibliography
[115] W. R. Thompson and J. E. Pemberton. “Characterization of octadecylsilane and stearic-acid layers on Al2O3 surfaces by
raman-spectroscopy”. Langmuir 11, 1720 (1995)
[116] A. Ulman. “Formation and structure of self-assembled monolayers”. Chem. Rev. 96, 1533 (1996)
[117] S. H. Chen and C. W. Frank. “Infrared and fluorescence spectroscopic studies of self-assembled n-alkanoic acid monolayers”.
Langmuir 5, 978 (1989)
[118] Y.-T. Tao. “Structural comparison of self-assembled monolayers of n-alkanoic acids on the surfaces of silver, copper and
aluminum”. J. Am. Chem. Soc. 115, 4350 (1993)
[119] Y.-T. Tao, G. D. Hietpas, and D. L. Allara. “HCl vapor-induced structural rearrangements of n-alkanoate self-assembled
monolayers on ambient silver, copper, and aluminum surfaces”. J. Am. Chem. Soc. 118, 6724 (1996)
[120] POF Atlas, Fachhochschule Nürnberg, http://www.pofac.de/pofac/en/information/glossary.php?term=CYTOP&glossarylang=en.
[121] R. A. Laudise, C. Kloc, P. G. Simpkins, and T. Siegrist. “Physical vapor growth of organic semiconductors”. J. Crystal Growth
187, 449 (1998)
[122] J. Wang, D. J. Gundlach, C. C. Kuo, and T. N. Jackson. “Improved contacts for organic electronic devices using self-assembled
charge transfer materials”. 41st Electron. Mater. Conf. Dig. p. 16 (1999)
[123] G. Horowitz, F. Garnier, A. Yassar, R. Hajlaoui, and F. Kouki. “Field-effect transistor made with a sexithiophene single crystal”.
Adv. Mater. 8, 52 (1996)
[124] J. N. Israelachvili. Intermolecular and surface forces. Academic Press, London (1997)
[125] Z.-T. Zhu, J. T. Mason, R. Dieckmann, and G. G. Malliaras. “Humidity sensors based on pentacene thin-film transistors”.
Appl. Phys. Lett. 81, 4643 (2002)
[126] C. Pannemann, T. Diekmann, and U. Hilleringmann. “Degradation of organic field-effect transistors made of pentacene”. J.
Mat. Res. 19, 1999 (2004)
[127] A. Maliakal, K. Raghavachari, H. Katz, E. Chandross, and T. Siegrist. “Photochemical stability of pentacene and a substituted
pentacene in solution and in thin films”. Chem. Mater. 16, 4980 (2004)
[128] F. De Angelis, S. Cipolloni, L. Mariucci, and G. Fortunato. “Aging effects in pentacene thin-film transistors: Analysis of the
density of states modification”. Appl. Phys. Lett. 88, 193508 (2006)
[129] M. L. Chabinyc, R. A. Street, and J. E. Northrup. “Effects of molecular oxygen and ozone on polythiophene-based thin-film
transistors”. Appl. Phys. Lett. 90, 123508 (2007)
[130] H. Klauk, U. Zschieschang, R. T. Weitz, H. Meng, F. Sun, G. Nunes, D. E. Keys, C. R. Fincher, and Z. Xiang. “Organic
transistors based on di(phenylvinyl)anthracene: Performance and stability”. Adv. Mater. 19, 3882 (2007)
[131] M. Kiguchi, M. Nakayama, T. Shimada, and K. Saiki. “Electric-field-induced charge injection or exhaustion in organic thin
film transistor”. Phys. Rev. B 71, 035332 (2005)
Bibliography 167
[132] D. Knipp, A. Benor, V. Wagner, and T. Muck. “Influence of impurities and structural properties on the device stability of
pentacene thin film transistors”. J. Appl. Phys. 101, 044504 (2007)
[133] P. V. Pesavento, R. J. Chesterfield, C. R. Newman, and C. D. Frisbie. “Gated four-probe measurements on pentacene thin-film
transistors: Contact resistance as a function of gate voltage and temperature”. J. Appl. Phys. 96, 7312 (2004)
[134] P. V. Pesavento, K. P. Puntambekar, C. D. Frisbie, J. C. McKeen, and P. P. Ruden. “Film and contact resistance in pentacene
thin-film transistors: Dependence on film thickness, electrode geometry, and correlation with hole mobility”. J. Appl. Phys.
99, 094504 (2006)
[135] J. Takeya, C. Goldmann, S. Haas, K. P. Pernstich, B. Ketterer, and B. Batlogg. “Field-induced charge transport at the surface
of pentacene single crystals: A method to study charge dynamics of two-dimensional electron systems in organic crystals”. J.
Appl. Phys. 94, 5800 (2003)
[136] C. Goldmann, S. Haas, C. Krellner, K. P. Pernstich, D. J. Gundlach, and B. Batlogg. “Hole mobility in organic single crystals
measured by a “flip-crystal” field-effect technique”. J. Appl. Phys. 96, 2080 (2004)
[137] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers, and M. E. Gershenson. “Intrinsic charge transport on the
surface of organic semiconductors”. Phys. Rev. Lett. 93, 086602 (2004)
[138] L. Bürgi, T. J. Richards, R. H. Friend, and H. Sirringhaus. “Close look at charge carrier injection in polymer field-effect
transistors”. J. Appl. Phys. 94, 6129 (2003)
[139] W. E. Spear and P. G. Le Comber. “Investigation of the localized state distribution in amorphous Si films”. J. Non-Cryst. Solids
8-10, 727 (1972)
[140] M. Grünewald, P. Thomas, and D. Würtz. “A simple scheme for evaluating field effect data”. Phys. Status Solidi B 100, K139
(1980)
[141] W. E. Spear, D. Allan, P. Le Comber, and A. Ghaith. “A new approach to the interpretation of transport results in a-Si”. Philos.
Mag. B 41, 419 (1980)
[142] M. Grünewald, K. Weber, W. Fuhs, and P. Thomas. “Field effect studies on a-Si:H films”. J. Phys. 42, 523 (1981)
[143] R. L. Weisfield and D. A. Anderson. “An improved field-effect analysis for the determination of the pseudogap-state density in
amorphous semiconductors”. Philos. Mag. B 44, 83 (1981)
[144] M. J. Powell. “Analysis of field-effect-conductance measurements on amorphous semiconductors”. Philos. Mag. B 43, 93
(1981)
[145] K. Weber, M. Grünewald, W. Fuhs, and P. Thomas. “Field effect in a-Si:H films”. Phys. Status Solidi B 110, 133 (1982)
[146] F. Djamdji and P. G. Le Comber. “An investigation of the conductivity prefactor in a-Si as a function of fermi level position
using the field-effect experiment”. Philos. Mag. B 56, 31 (1987)
[147] R. Schumacher, P. Thomas, K. Weber, W. Fuhs, F. Djamdji, P. G. Le Comber, and R. E. I. Schropp. “Temperature-dependent
effects in field-effect measurements on hydrogenated amorphous silicon thin-film transistors”. Philos. Mag. B 58, 389 (1988)
168 Bibliography
[148] G. Fortunato and P. Migliorato. “Determination of gap state density in polycrystalline silicon by field-effect conductance”.
Appl. Phys. Lett. 49, 1025 (1986)
[149] P. Migliorato and D. B. Meakin. “Material properties and characteristics of polysilicon transistors for large area electronics”.
Appl. Surface Si. 30, 353 (1987)
[150] G. Fortunato, D. B. Meakin, P. Migliorato, and P. G. Le Comber. “Field-effect analysis for the determination of gap-state
density and fermi-level temperature dependence in polycrystalline silicon”. Philos. Mag. B 57, 573 (1988)
[151] V. Foglietti, L. Mariucci, and G. Fortunato. “Temperature dependence of the transfer characteristics of polysilicon thin film
transistors fabricated by excimer laser crystallization”. J. Appl. Phys. 85, 616 (1999)
[152] A. R. Völkel, R. A. Street, and D. Knipp. “Carrier transport and density of state distributions in pentacene transistors”. Phys.
Rev. B 66, 195336 (2002)
[153] S. Scheinert, G. Paasch, M. Schrödner, H. K. Roth, S. Sensfuss, and T. Doll. “Subthreshold characteristics of field effect
transistors based on poly(3-dodecylthiophene) and an organic insulator”. J. Appl. Phys. 92, 330 (2002)
[154] D. Knipp, P. Kumar, A. R. Völkel, and R. A. Street. “Influence of organic gate dielectrics on the performance of pentacene thin
film transistors”. Synth. Metal. 155, 485 (2005)
[155] D. Oberhoff, K. P. Pernstich, D. J. Gundlach, and B. Batlogg. “Arbitary density of states in an organic thin-film field-effect
transistor model and application to pentacene devices”. IEEE Trans. Electron Devices 54, 17 (2007)
[156] S. Scheinert, K. P. Pernstich, B. Batlogg, and G. Paasch. “Determination of trap distributions from current characteristics of
pentacene field-effect transistors with surface modified gate oxide”. J. Appl. Phys. 102, 104503 (2007)
[157] K. P. Pernstich, B. Rössner, and B. Batlogg. “Field-effect-modulated Seebeck coefficient in organic semiconductors”. Nature
Mater. 7, 321 (2008)
[158] I. Zhivkov, S. Nešpurek, and F. Schauer. “Influence of oxygen on the parameters of a thin film copper phthalocyanine field
effect transistor”. Adv. Mater. Opt. Electron. 9, 175 (1999)
[159] G. Horowitz, M. E. Hajlaoui, and R. Hajlaoui. “Temperature and gate voltage dependence of hole mobility in polycrystalline
oligothiophene thin film transistors”. J. Appl. Phys. 87, 4456 (2000)
[160] D. V. Lang, X. Chi, T. Siegrist, A. M. Sergent, and A. P. Ramirez. “Amorphouslike density of gap states in single-crystal
pentacene”. Phys. Rev. Lett. 93, 086802 (2004)
[161] F. De Angelis, L. Mariucci, S. Cipolloni, and G. Fortunato. “Analysis of electrical characteristics of high performance pentacene
thin-film transistors with PMMA buffer layer”. J. Non-Cryst. Solids 352, 1765 (2006)
[162] N. Kawasaki, T. Nagano, Y. Kubozono, Y. Sako, Y. Morimoto, Y. Takaguchi, A. Fujiwara, C. C. Chu, and T. Imae. “Transport
properties of field-effect transistors with Langmuir-Blodgett films of C-60 dendrimer and estimation of impurity levels”. Appl.
Phys. Lett. 91, 243515 (2007)
Bibliography 169
[163] J. A. Merlo, C. R. Newman, C. P. Gerlach, T. W. Kelley, D. V. Muyres, S. E. Fritz, M. F. Toney, and C. D. Frisbie. “p-channel
organic semiconductors based on hybrid acene-thiophene molecules for thin-film transistor applications”. J. Am. Chem. Soc.
127, 3997 (2005)
[164] J. E. Anthony. “Functionalized acenes and heteroacenes for organic electronics”. Chem. Rev. 106, 5028 (2006)
[165] M. Mas-Torrent and C. Rovira. “Novel small molecules for organic field-effect transistors: towards processability and high
performance”. Chem. Soc. Rev. 37, 827 (2008)
[166] D. J. Gundlach, J. A. Nichols, L. Zhou, and T. N. Jackson. “Thin-film transistors based on well-ordered thermally evaporated
naphthacene films”. Appl. Phys. Lett. 80, 2925 (2002)
[167] R. T. Weitz, K. Amsharov, U. Zschieschang, E. Barrena Villas, D. K. Goswami, M. Burghard, H. Dosch, M. Jansen, K. Kern,
and H. Klauk. “Organic n-channel transistors based on core-cyanated perylene carboxylic diimide derivatives”. J. Am. Chem.
Soc. 130, 4637 (2008)
[168] J. E. Anthony, D. L. Eaton, and S. R. Parkin. “A road map to stable, soluble, easily crystallized pentacene derivatives”. Org.
Lett. 4, 15 (2002)
[169] H. Moon, R. Zeis, E.-J. Borkent, C. Besnard, A. J. Lovinger, T. Siegrist, C. Kloc, and Z. Bao. “Synthesis, crystal structure, and
transistor performance of tetracene derivatives”. J. Am. Chem. Soc. 126, 15322 (2004)
[170] W. L. Kalb, T. Mathis, S. Haas, A. F. Stassen, and B. Batlogg. “Organic small molecule field-effect transistors with CytopTM
gate dielectric: Eliminating gate bias stress effects”. Appl. Phys. Lett. 90, 092104 (2007)
[171] Y. Ma, Y. Sun, Y. Liu, J. Gao, S. Chen, X. Sun, W. Qiu, G. Yu, G. Cui, W. Hu, and D. Zhu. “Organic thin film transistors based
on stable amorphous ladder tetraazapentacenes semiconductors”. J. Mater. Chem. 15, 4894 (2005)
[172] P.-T. T. Pham, Y. Xia, C. D. Frisbie, and M. M. Bader. “Single crystal field effect transistor of a y-shaped ladder-type oligomer”.
J. Phys. Chem. C 112, 7968 (2008)
[173] U. Berens, A. Stassen, B. Schmidhalter, W. Kalb, and F. Bienewald. “Quinoid systems as organic semiconductors”. Patent
application WO 2007/118799 A1
[174] H. Liebermann and J. Barrollier. “über umformungen einiger 1,4-dioxy-tetraaryl-2,5-xylylenglycole”. Liebigs Ann. Chem.
509, 38 (1934)
[175] C. P. Gerlach. “Bis(2-acenyl)acetylene semiconductors”. US Patent No. 7,109,519, Sep. 19, 2006
[176] T. B. Singh, N. S. Sariciftci, H. Yang, L. Yang, B. Plochberger, and H. Sitter. “Correlation of crystalline and structural properties
of c60 thin films grown at various temperature with charge carrier mobility”. Appl. Phys. Lett. 90, 213512 (2007)
[177] N. Kobayashi, M. Sasaki, and K. Nomoto. “Stable peri-xanthenoxanthene thin-film transistors with efficient carrier injection”.
Chem. Mater. 21, 552 (2009)
[178] K. P. Pernstich, D. Oberhoff, C. Goldmann, and B. Batlogg. “Modeling the water related trap state created in pentacene
transistors”. Appl. Phys. Lett. 89, 213509 (2006)
170 Bibliography
[179] Asahi Glass Company. “Cytop amorphous fluoropolymer technical data sheet”
[180] V. Podzorov, S. E. Sysoev, E. Loginova, V. M. Pudalov, and M. E. Gershenson. “Single crystal organic field effect transistors
with the hole mobility ∼ 8 cm2/Vs”. Appl. Phys. Lett. 83, 3504 (2003)
[181] W. B. Jackson, J. M. Marshall, and D. M. Moyer. “Role of hydrogen in the formation of metastable defects in hydrogenated
amorphous silicon”. Phys. Rev. B 39, 1164 (1989)
[182] F. Schauer, S. Nešpurek, and O. Zmeškal. “The bulk trap spectroscopy of solids by temperature-modulated space-charge-
limited currents (TMSCLC): Application to real crystalline and amorphous semiconductors”. J. Phys. C 19, 7231 (1986)
[183] W. L. Kalb, F. Meier, K. Mattenberger, and B. Batlogg. “Defect healing at room temperature in pentacene thin films and
improved transistor performance”. Phys. Rev. B 76, 184112 (2007)
[184] D. Braga, N. Battaglini, A. Yassar, G. Horowitz, M. Campione, A. Sassella, and A. Borghesi. “Bulk electrical properties of
rubrene single crystals: Measurement and analysis”. Phys. Rev. B 77, 115205 (2008)
[185] M. Matters, D. M. de Leeuw, M. J. C. M. Vissenberg, C. M. Hart, P. T. Herwig, T. Geuns, C. M. J. Mutsaers, and C. J. Drury.
“Organic field-effect transistors and all-polymer integrated circuits”. Optical Materials 12, 189 (1999)
[186] M. Schubert, C. Bundesmann, G. Jacopic, H. Maresch, and H. Arwin. “Infrared dielectric function and vibrational modes of
pentacene thin films”. Appl. Phys. Lett. 84, 200 (2004)
[187] 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”. J. Appl. Phys. 100, 024509 (2006)
[188] A. Roth. Vacuum Technology. Elsevier Science B. V., Amsterdam (1990)
[189] S. Ogawa, T. Naijo, Y. Kimura, H. Ishii, and M. Niwano. “Photoinduced doping effect of pentacene field effect transistor on
oxygen atmosphere studied by displacement current measurement”. Appl. Phys. Lett. 86, 252104 (2005)
[190] H. Kuroda and E. A. Flood. “Effect of ambient oxygen on electrical properties of an evaporated film of pentacene”. Can. J.
Chem. 39, 1981 (1961)
[191] O. D. Jurchescu, J. Baas, and T. T. M. Palstra. “Electronic transport properties of pentacene single crystals upon exposure to
air”. Appl. Phys. Lett. 87, 052102 (2005)
[192] P. Parisse, M. Passacantando, S. Picozzi, and L. Ottaviano. “Conductivity of the thin film phase of pentacene”. Org. Electron.
7, 403 (2006)
[193] W.-Y. So, J. M. Wikberg, D. V. Lang, O. Mitrofanov, C. L. Kloc, T. Siegrist, A. M. Sergent, and A. P. Ramirez. “Mobility-
independent doping in crystalline rubrene field-effect transistors”. Solid State Commun. 142, 483 (2007)
[194] A. Vollmer, O. D. Jurchescu, I. Arfaoui, I. Salzmann, T. T. M. Palstra, P. Rudolf, J. Niemax, J. Pflaum, J. P. Rabe, and N. Koch.
“The effect of oxygen exposure on pentacene electronic structure”. Eur. Phys. J. E 17, 339 (2005)
[195] D. M. de Leeuw, M. M. J. Simenon, A. R. Brown, and R. E. F. Einerhand. “Stability of n-type doped conducting polymers and
consequences for polymeric microelectronic devices”. Synth. Metal. 87, 53 (1997)
Bibliography 171
[196] T. D. Anthopoulos, G. C. Anyfantis, G. C. Papavassiliou, and D. M. de Leeuw. “Air-stable ambipolar organic transistors”.
Appl. Phys. Lett. 90, 122105 (2007)
[197] E. J. Meijer, A. V. G. Mangnus, B.-H. Huisman, G. W. ‘t Hooft, D. M. de Leeuw, and T. M. Klapwijk. “Photoimpedance
spectroscopy of poly(3-hexyl thiophene) metal-insulator-semiconductor diodes”. Synth. Metal. 142, 53 (2004)
[198] A. Wang, I. Kymissis, V. Bulovic, and A. I. Akinwande. “Tunable threshold voltage and flatband voltage in pentacene field
effect transistors”. Appl. Phys. Lett. 89, 112109 (2006)
[199] R. A. Street. Hydrogenated amorphous silicon. Cambridge University Press, Cambridge (1991)
[200] L. Tsetseris and S. T. Pantelides. “Intercalation of oxygen and water molecules in pentacene crystals: First-principles
calculations”. Phys. Rev. B 75, 153202 (2007)
[201] D. J. Gundlach, T. N. Jackson, D. G. Schlom, and S. F. Nelson. “Solvent-induced phase transition in thermally evaporated
pentacene films”. Appl. Phys. Lett. 74, 3302 (1999)
[202] J. Zaumseil, R. H. Friend, and H. Sirringhaus. “Spatial control of the recombination zone in an ambipolar light-emitting organic
transistor”. Nat. Mater. 5, 69 (2006)
[203] L. Bürgi, M. Turbiez, R. Pfeiffer, F. Bienewald, H. J. Kirner, and C. Winnewisser. “High-mobility ambipolar near-infrared
light-emitting polymer field-effect transistors”. Adv. Mater. 20, 2217 (2008)
[204] K. Hong, S. Y. Yang, C. Yang, S. H. Kim, D. Choi, and C. E. Park. “Reducing the contact resistance in organic thin-film
transistors by introducing a PEDOT : PSS hole-injection layer”. Org. Electron. 9, 864 (2008)
Acknowledgments
The experiments for this thesis were done in the group of Prof. Dr. Bertram Batlogg at the
Laboratory for Solid State Physics at ETH Zurich. I would like to thank Bertram Batlogg
for the freedom to choose research projects and to take measurement-related decisions.
I would like to thank Bertram for sharing his ideas which has improved the quality of
this work substantially. His constant support has carried this work on a higher level of
quality. Moreover, I am thankful for having had the opportunity to participate in numerous
international conferences and in the collaboration with Ciba Speciality Chemicals. This
has enriched the whole PhD experience significantly.
I would like to thank Prof. Dr. Gilles Horowitz for accepting the responsibility of being
a co-examiner of this thesis. It is a great honor to have one of the pioneers of organic
field-effect transistors as co-examiner. I was influenced and inspired a lot by working
on organic thin-film transistors with Dr. Gilles Horowitz and Dr. Philippe Lang back in
the year 2004 during a 4-month internship at ITODYS, Paris in the frame of my diploma
work at RWTH Aachen.
I owe thanks to my colleagues from the early days Dr. Kurt Pernstich, Dr. Simon Haas and
Dr. Claudia Goldmann who also worked on organic semiconductors. I am thankful for
helping me with adjusting to being a PhD student and for introducing me to the equipment
in the lab. Later on they always found time to pass on their experience in the lab while
finishing up their PhD work.
I thank Dr. David Gundlach who was a PostDoc in the group when I started this the-
sis. Although he left the group only a few month after I started my research I benefited
greatly from meeting this experienced scientist. He helped me to find the right track in
the beginning and I am grateful for that.
Dr. Arno Stassen was a PostDoc in the group in the first year of my PhD project. Arno
Stassen is a chemist and specialist in organic single crystal field-effect transistors. We
worked together on the project with Ciba and made field-effect transistors with all kinds
of new organic semiconductors. I enjoyed his good humor and profited from working
with an experienced scientist in the lab.
Thomas Mathis started in the group as a summer intern. We worked together on Cytop
as gate dielectric in the early days. Later on he became one of my fellow PhD students
working on organic semiconductors. He contributed significantly to a good atmosphere
in the group with his humorous, optimistic and communicative spirit.
I owe many thanks to our technicians Kurt Mattenberger and Hans-Peter Staub for the
excellent support. Having profited from their decade-long experience in a lab has been
most fortunate and essential for the experimental work. The quality of their solutions and
in particular their efficiency never stopped surprising me. Kurt Mattenberger was an in-
dispensable part of setting up and maintaining the device fabrication and characterization
system.
I would like to thank our secretary Gaby Strahm who was most helpful with numerous
administrative concerns. She organized many Apéros and group excursions which were
an integral part of the group’s social life.
I thank Dr. Frank Bienewald, Dr. Ulrich Berens and Dr. Andreas Hafner at Ciba for
the fruitful collaboration on organic field-effect transistors with new organic semiconduc-
tors. It was most fortunate to work with these outstanding chemists and to have access
to organic semiconductors no one ever even thought about making a transistor with. In
addition to that it was interesting to understand the perspective of a scientist working
in industry in general. Specifically, it was enriching to understand the challenges to be
overcome for a commercialization of organic electronics.
I would like to thank Prof. Dr. Yoshihiro Kubozono, our energetic friend and collaborator
from Japan. It was a great pleasure to work with him in the lab during his visits and to
profit from his experience with organic semiconductors.
Matthias Walser, Fabian Meier, Thomas Brenner, Sieghard Seyrling and Pierre Joris did
their diploma thesis on organic field-effect transistors in the group. Their experimental
work is well appreciated and has contributed to the progress of this thesis.
I am thankful for the company of my fellow PhD students Dr. Benjamin Rössner, Dr.
Markus Brühwiler, Jakob Kanter, Philip Moll, Florian Pfuner, Andrey Belousov and
Michela Lavagnini. It was a pleasure to spend time together.
My very special thanks goes to my parents Gisela and Wilhelm Kalb who have supported
me all along the way.
Finally, I would like to dedicate this thesis to my partner in life Ludovica Stortini. The
extent of her support and patience is beyond words.
Curriculum Vitae
5.12.1977 Born in Neuss, Germany
1984-1988 Primary School in Dormagen, Germany
1988-1997 Gymnasium in Dormagen
June 1997 Abitur (general qualification for university entrance)
1997-1998 Civil service in the county hospital in Dormagen
1998-2004 Study of Physics at RWTH Aachen, Germany
and at the University of Manchester, Great Britain
2003-2004 Diploma work at the
III. Physikalischen Insitut, RWTH Aachen
and at ITODYS, Universite Paris 7, France
Diploma thesis: Structure-performance relationship
in pentacene-based thin-film transistors
August 2004 Diploma in physics from RWTH Aachen
2005-2009 Research and teaching assistant in the research group of
Prof. Dr. B. Batlogg at the
Laboratory for Solid State Physics, ETH Zurich, Switzerland
Ph.D. thesis: Trap states in organic field-effect transistors:
Quantification, identification and elimination
Publication list
Publications in journals
1. W. L. Kalb, A. F. Stassen, B. Batlogg, U. Berens, B. Schmidhalter, F. Bienewald, A.
Hafner, and T. Wagner. Quinoid heteropentacenes as promising organic semicon-
ductors for field-effect transistor applications. J. Appl. Phys. 105, 043705 (2009)
2. M. P. Walser, W. L. Kalb, T. Mathis, T. J. Brenner, and B. Batlogg. Stable comple-
mentary inverters with organic field-effect transistors on Cytop fluoropolymer gate
dielectric. Appl. Phys. Lett. 94, 053303 (2009)
3. W. L. Kalb, K. Mattenberger, and B. Batlogg. Oxygen-related traps in pentacene
thin films: Energetic position and implications for transistor performance. Phys.
Rev. B 78, 035334 (2008)
4. Y. Kubozono, S. Haas, W. L. Kalb, P. Joris, F. Meng, A. Fujiwara, and B. Batlogg.
High-performance C60 thin-film field-effect transistors with parylene gate insulator.
Appl. Phys. Lett. 93, 033316 (2008)
5. W. L. Kalb, F. Meier, K. Mattenberger, and B. Batlogg. Defect healing at room
temperature in pentacene thin films and improved transistor performance. Phys.
Rev. B 76, 184112 (2007)
highlighted in New Scientist, Vol. 196 No. 2632, p. 27 (1. Dec. 2007) and
Nanomaterials News, Vol. 3, Issue 21, p. 7 (29. Jan 2008)
6. W. L. Kalb, T. Mathis, S. Haas, A. F. Stassen, and B. Batlogg. High performance
organic field-effect transistors with fluoropolymer gate dielectric. Proc. of SPIE
6658, 665807 (2007)
7. W. L. Kalb, T. Mathis, S. Haas, A. F. Stassen, and B. Batlogg. Organic small
molecule field-effect transistors with CytopT M gate dielectric: Eliminating gate bias
stress effects. Appl. Phys. Lett. 90, 092104 (2007)
Patent applications
1. W. Kalb and T. Mathis. Organic field-effect transistors with polymeric gate dielec-
tric and method for making same. Patent application WO/2008/077463
2. U. Berens, A. Stassen, B. Schmidhalter, W. Kalb, and F. Bienewald. Quinoid sys-
tems as organic semiconductors. Patent application WO/2007/118779
Contributions at conferences
Talks as presenting author
1. Oxygen-related traps in pentacene thin films: Energetic position and implications
for transistor performance. APS March Meeting, March 2009, Pittsburgh, USA
2. A close look at oxygen-related traps in pentacene thin films: Energetic position
and implications for transistor performance. MRS Fall Meeting, December 2008,
Boston, USA
3. Defect healing at room temperature in pentacene thin films and improved transistor
performance. MSE 08, September 2008, Nürnberg, Germany
4. Room temperature defect healing in pentacene based thin-film transistors. E-MRS
Spring Meeting, May 2008, Strasbourg, France
5. Defect healing at room temperature in pentacene thin films and improved transistor
performance. APS March Meeting, March 2008, New Orleans, USA
6. High performance organic field-effect transistors with a cyclic fluoropolymer gate
dielectric. ECOER’07, October 2007, Varenna, Italy
7. Eliminating gate bias stress effects in organic field-effect transistors. APS March
Meeting, March 2007, Denver, USA
8. New organic semiconductors for flexible electronics. MRC Graduate Symposium,
June 2006, Zürich, Switzerland
9. New organic semiconductors for flexible electronics. EUROFET Meeting, June
2006, Berlin, Germany
Poster presentations
1. Highly stable organic field-effect transistors with a fluoropolymer gate dielectric.
MRC Graduate Symposium, May 2008, Zürich, Switzerland
2. Defect healing at room temperature in pentacene thin films and improved transistor
performance. Alpine Workshop, December 2007, Braunwald, Switzerland
3. Highly stable organic field-effect transistors with a fluoropolymer gate dielectric.
Alpine Workshop, December 2007, Braunwald, Switzerland
4. Field-effect in new small molecule organic semiconductors. ECOF 10, August
2006, Riga, Latvia