Molecular organic semiconductors for electronic devices Jurchescu, Oana Diana

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2006 Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Jurchescu, O. D. (2006). Molecular organic semiconductors for electronic devices s.n.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 23-01-2017

Molecular Organic Semiconductors for Electronic Devices

Oana Diana Jurchescu

MSC PhD thesis series 2006-15 issn: 1570-1530 isbn: 90-367-2708-1 The work described in this thesis was performed in the research group Solid State Chemistry of the Materials Science Centre at the University of Groningen, the Netherlands. Printed by: PrintPartners Ipskamp B.V., Enschede, the Netherlands

Rijksuniversiteit Groningen

Molecular Organic Semiconductors for Electronic Devices

Proefschrift ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op vrijdag 6 oktober 2006 om 14:45 uur

door

Oana Diana Jurchescu geboren op 1 maart 1979 te Timi¸soara, Roemeni¨e

Promotor:

Prof. Dr. T.T.M. Palstra

Beoordelingscommissie:

Prof. Dr. Y. Iwasa Prof. Dr. G. G. Malliaras Prof. Dr. P. Rudolf

Contents

1 Introduction 1.1 Organic conductors - general aspects . . . . . . . . . . . . . . . . . 1.2 Charge transport in organic semiconductors . . . . . . . . . . . . . 1.3 Organic electronic devices . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 Field effect transistors - operation principles . . . . . . . . . 1.3.2 Fabrication of OFETs . . . . . . . . . . . . . . . . . . . . . 1.3.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Organic molecular crystals . . . . . . . . . . . . . . . . . . . . . . . 1.5 Single crystals - model systems for investigation of intrinsic properties 1.6 Aim of the present research . . . . . . . . . . . . . . . . . . . . . . 1.7 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 3 8 8 11 14 15 18 18 19 26

2 The Effect of Impurities on the Mobility of Single Crystal Pentacene 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Control of the defects and impurity states in organic crystals . . . 2.2.1 Spectroscopic evidence for the presence of impurities . . . . 2.2.2 Purification of the starting material . . . . . . . . . . . . . 2.2.3 Single crystal growth . . . . . . . . . . . . . . . . . . . . . . 2.2.4 Quantitative analysis of 6,13-pentacenequinone content . . 2.3 Electronic response to structural and chemical properties . . . . . 2.3.1 Space charge limited currents . . . . . . . . . . . . . . . . .

27 28 28 29 30 31 32 35 35

v

vi

Contents 2.3.2 Evaluation of the bulk mobility of the high quality crystals 2.3.3 Band transport in pentacene single crystals . . . . . . . . . 2.3.4 Origin of traps . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 Interface Controlled High-mobility Organic Transistors 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Current status of the field . . . . . . . . . . . . . . . . . . 3.3 Novelty . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Single crystal growth . . . . . . . . . . . . . . . . . . . . . 3.5 Surface morphology of pentacene single crystals . . . . . . 3.5.1 Surface mapping . . . . . . . . . . . . . . . . . . . 3.5.2 AFM measurements . . . . . . . . . . . . . . . . . 3.6 Comparison between different FETs . . . . . . . . . . . . 3.6.1 FETs fabricated with conventional dielectrics . . . 3.6.2 FETs fabricated with pentacenequinone dielectric 3.7 Device fabrication . . . . . . . . . . . . . . . . . . . . . . 3.8 Properties of FETs fabricated with pentacenequinone gate dielectric . . . . . . . . . . . . . . 3.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

38 39 40 43 46

. . . . . . . . . . .

47 48 48 49 49 50 50 50 53 53 54 54

. . . . . . . . . . . . . . .

56 59 63

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

4 Cross-over from 1D to 2D - Space Charge Limited Conduction in Pentacene Single Crystals 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Space-charge-limited-current theories for semiconductors . . . . . . 4.2.1 Mott-Gurney model . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Geurst model . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Space charge limited currents in pentacene single crystals with planar contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Experimental procedure . . . . . . . . . . . . . . . . . . . . 4.3.2 Charge transport in the SCLC regime . . . . . . . . . . . . 4.3.3 Evaluation of mobility for sandwich-type and gap-type structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.4 Resistivity anisotropy effects - effective thickness . . . . . . 4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

65 66 66 66 67 68 68 69 70 72 73 76

vii

Contents

5 Electronic Transport Properties of Pentacene Single Crystals upon Exposure to Air 77 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 5.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 5.3 Exposure to dry and ambient air . . . . . . . . . . . . . . . . . . . 79 5.3.1 I-V characteristics in the dark - in vacuum and after exposure to air . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.3.2 Gas diffusion in pentacene single crystals . . . . . . . . . . 80 5.3.3 Effect of Oxygen and water in dark . . . . . . . . . . . . . 84 5.3.4 Effect of Oxygen and water in light . . . . . . . . . . . . . 86 5.3.5 Effect of high O2 pressure . . . . . . . . . . . . . . . . . . . 87 5.4 O2 and H2 O effects - experimental observations . . . . . . . . . . . 88 5.4.1 Electrical measurements . . . . . . . . . . . . . . . . . . . . 88 5.4.2 UPS spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 89 5.5 Mechanism of oxygen doping . . . . . . . . . . . . . . . . . . . . . 91 5.5.1 Formation of the charge transfer complex between pentacene and O2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 5.5.2 Reversible oxidation of pentacene . . . . . . . . . . . . . . . 92 5.5.3 Calculation of the oxygen-induced dipole moment . . . . . 93 5.6 Quantitative analysis of the effect of O2 and H2 O . . . . . . . . . . 96 5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6 Low Temperature Crystal Structure of Rubrene Single Crystals Grown by Vapor Transport 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Polymorphism in rubrene . . . . . . . . . . . . . . . . . . . 6.3 Growth of rubrene single crystals . . . . . . . . . . . . . . . 6.4 Single crystal diffraction . . . . . . . . . . . . . . . . . . . . 6.4.1 Experimental details . . . . . . . . . . . . . . . . . . 6.4.2 Orthorhombic rubrene . . . . . . . . . . . . . . . . . 6.5 Relation between molecular stacking and electronic mobility 6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

103 104 104 105 106 106 106 109 113 116

A Crystal structure of pentacene

117

B Crystal structure of rubrene

121

viii

Contents

Summary

131

Samenvatting

135

List of Publications

139

1

Introduction

1.1

Organic conductors - general aspects

Organic semiconductors are a fascinating class of materials, with a wide range of properties [1–7]. The interest in organic electronics is motivated by increasing demands of supplementing Si-based electronics with materials that offer easier processing methods completed with functionalization by chemical manipulation. Typical deposition techniques for organics are readily available or not very difficult to implement (spin-coating, drop-casting, direct printing via stamps or ink-jets). Their compatibility with light-weight, mechanically flexible plastic substrates, completed by new, innovative fabrication routes, make them possible candidates for future electronic devices. Moreover, due to different molecular tailoring possibilities via chemical synthesis, organic materials present an infinite variety in functionality. The design of the molecular structures can be engineered to enhance particular properties (solubility in different solvents, color of light emission, crystal packing). In particular, addition of polar groups in polymers (e.g. polyvinylidene fluoride and its copolymers with trifluoroethylene and tetrafluoroethylene) leads to rich systems for investigation of ferroelectricity [8]. By incorporating them in transistors (ferroelectric field-effect transistors - FeFETs), combined functionality can be achieved for non-volatile memory devices that are compatible with flexible plastic substrates [9]. Modification of the chemical end-groups also allows fabrication of large transistor arrays for organic sensors. Different methods are developed to increase the sensitivity and selectivity of organic sensing transistors to different chemical or biological species. Applications include environmental issues (e.g. air pollution), pH indicators [10], food freshness, toxic compounds, 1

2

Chapter 1. Introduction

stress and pressure indication in clothes [11]. Rapid progress is being made in the industrial development of the organic electronic devices. Organic devices can be implemented on large scale application area when their operation offer high performance, reliability, stability, long life time, good control and reproducibility. Impressive steps have been accomplished towards incorporation of organic conductors in devices, as constituents of plastic electronic components, by continued innovation in materials, processing methods and device design. For example, Philips and Polymer Vision already presented their prototype rollable display based on pentacene FETs that will be used for fabrication of mobile phones [12]. OLED1 technology has already been incorporated in commercial applications such as small displays for mobile phones, car radios, and digital cameras [13]. It was suggested that if ”the field continues to progress at its current, rapid pace, electronics based on organic thin-films materials will soon become a mainstay of our technological existence” (S. Forrest [14]). In the academic community, the interest in electrical properties of organic materials has grown enormously in the last years [16–26]. Different materials and structures are actively being investigated. On the one hand, the research is motivated by new applications for organic (plastic) electronics that can overcome limitations of inorganic materials. On the other hand, the effort is driven by the intellectual challenges that the investigations of new generation of functional materials offer. This implies addressing new questions, concepts and models that allow the manipulation of chemical, structural, physical properties and device performance. The field is moving very fast; at the moment researchers are able to control the material properties at the molecular level. They can impose different molecular packing during growth, add side groups that enhance particular properties, as a result of the understanding of the various microscopic contributions to the charge transport. Past investigations were hindered by material instabilities (to environmental conditions and/or processing methods) and structural defects that prevented the measurement and understanding of the intrinsic properties of the organic materials. In the last years, by careful design of new methods that enable the exploitation of the fascinating properties of carbon-based semiconductors, good control of the structural properties and charge density distribution in organic materials is achieved (both by electrostatic and chemical doping). The degradation is avoided as a consequence of a better understanding of the material behavior in different conditions (temperature, moisture, light) and interactions during processing. 1 OLED

- organic light emitting diodes

1.2. Charge transport in organic semiconductors

1.2

3

Charge transport in organic semiconductors

Organic semiconductors are wide-band-gap and small bandwidth semiconductors. The HOMO2 -LUMO3 gap is in the range of 1-4 eV [27]. With such a large gap, it might be expected that the organic materials are insulators (an electron would have to acquire a large thermal energy to make the jump from the valence band to the conduction band). There are some effective methods that can generate charge carriers in the organic semiconductors: ⋄ injection of carriers from metallic electrodes; ⋄ optical excitation (creation of electron-hole pairs); ⋄ electrostatic or chemical doping. Different organic molecular structures are of current interest for the continued search for novel and useful properties. Carbon nanotubes (CNTs) are onedimensional, cylindrical systems that are attractive for nanometer-sized electronics because they combine the high electrical and thermal conductivity with good mechanical strength and flexibility [28]. The physical properties of nanotubes are imposed by their diameter, length, and chirality, or twist. For example, they can be both metallic and semiconducting, depending on the chirality [29]. Nanotubes are attractive also for the field emission properties [30] because they are long and very thin, conductive, with high mechanical strength, and have a low work-function. As the CNTs are generally accepted as being the bridge between the nano- and microscopic length scales, a step further was done in coupling them to biological systems aiming the fabrication of biosensors [31, 32]. More recently, superconductivity has been observed in nanotubes [33]. When the carbon sheet is not rolled into a tube, it forms a graphene sheet. Graphene is a candidate for nanometer-scale electronic devices that combine the properties of carbon nanotubes with the compatibility with established microelectronics manufacturing techniques. Low scattering yields mobilities up to 104 cm2 /Vs [34]. This value gives ultra-fast-switching transistors for electronics. Also, the quantum Hall effect was recently reported in graphene [35]. The most common 3D structure in the fullerene structural family is C60 , the spherical buckyball (Fig. 1.1(d)). Because of its particular electronic structure, C60 exhibits semiconducting properties, but it reveals one amazing property after another upon doping [36]. It can be tuned to a metallic or insulating 2 HOMO 3 LUMO

- highest occupied molecular orbital - lowest unoccupied molecular orbital

4

Chapter 1. Introduction

state by transferring charge on the molecule through doping. At low temperature this metallic state often exhibits superconductivity, with transition temperatures only exceeded by the high-temperature superconducting copper oxides [37]. This is not an unique type of organic superconductor. Several organic superconductors have been identified, such as the quasi-one-dimensional Bechgaard salts ((TMTSF)2 X4 and (TMTTF)2 X5 , with X= PF6 − , FSO3 − , ClO4 − , etc.), and quasi-two-dimensional salts derived from the donor group BEDT-TTF6 (Fig. 1.1(a)) [38]. Organic superconductors are charge transfer compounds, unconventional superconductors, in which the interchain transfer integral can be changed with pressure, temperature, and composition (the nature of the anion X controls the chemical pressure), leading to a delicate balance of interactions that gives rise to a rich variety of physical phenomena. These include charge ordering, spin-Peierls ground state, antiferromagnetism, spin density waves (SDW), metallicity and superconductivity [38]. In polymers, the electrical properties are dominated by disorder that localize the charges at low temperature. Still, these materials present fascinating properties that make them attractive for organic electronic devices [1,2,6]. We presented in Section 1.1 the potential of polymers, that were already incorporated in commercial devices, as components of the OLEDs. Their semiconducting properties are also used in FETs and solar cells [16–18]. In OLEDs and solar cells an exciton (excited electron - hole pair) is generated in the organic material. In OLEDs light is generated through radiative recombination of electrons and holes and it is emitted through the transparent electrode. Two distinct processes, equally important, govern the operation of the device: charge transport and recombination. The color of the emitted light is tuned by the band gap of the active material that is used. In solar cells light is absorbed through the transparent electrode and the photons absorbed in the semiconductor create mobile excited electron-hole pairs. These excitons subsequently undergo dissociation yielding free electrons and holes that are collected at the contacts. The efficiency of the solar cell is given by the photon absorption efficiency, exciton dissociation efficiency and charge collection efficiency. Organic photovoltaic devices are limited by exciton lifetime, and low charge carrier mobilities; only a fraction of the light-induced excitons contribute to the current generation. Tang et al. demonstrated that the efficiency can be increased considerably if a heterojunction of a electron donor and electron acceptor material is manufactured [39]. To enhance the quantum separation efficiency, a 4 TMTSF

stands for tetramethyltetraselenafulvalene stands for tetramethyltetrathiafulvalene 6 BEDT-TTF stands for bis(ethylenedithio)tetrathiafulvalene 5 TMTTF

1.2. Charge transport in organic semiconductors

5

Figure 1.1: Chemical formulae of organic molecules with different functionalities.

6

Chapter 1. Introduction

network of internal heterojunctions, with a large contact area between donor and acceptor species (referred to as ”bulk heterojunctions”) was proposed [40]. Despite the fast developments in the field of organic electronics and dramatic improvement in knowledge and manipulation of the charge injection and transport in organic semiconductors, a reliable relation between microscopic properties and their effect on the physical properties is still under development. There is significant work done in understanding the mechanism of conduction in organic semiconductors. Correct correlations between morphology, molecular packing and the resulting electronic properties are essential, in order to elucidate fundamental questions regarding charge transport in these materials. Small molecule conductors allow these type of studies. Because of a higher degree of order than solution-processed polymers, they can be structurally characterized straightforward using diffraction, and microscopy techniques. Complementary to this, the theoretical modelling can be easier implemented because of the lower complexity of these systems. In organic molecular solids, the intermolecular forces are weak (van der Waals and electrostatic type). A detailed description of these issues will be developed in Section 1.4. The bonding energies are considerably lower than in covalent and ionic inorganic semiconductors [27, 41]. For this reason, the mechanism of charge transport is fundamentally different. In organic conductors, the charge carriers interact strongly with the lattice environment leading to polarization effects and tendency of charge carrier localization. The weak van der Waals interactions result in a small electronic bandwidth, strong electron-lattice interaction, and polaron formation. For example, calculations performed on organic molecular crystals that are free of defects, yield bandwidths in the order of 0.1-0.5 eV [42–47]. This is more that one order of magnitude lower than the bandwidth in silicon (∼ 10 eV, [48]). The small bandwidths is reflected in low charge carrier mobilities (10−5 10 cm2 /Vs [16–26] for organic semiconductors, compared to 50-500 cm2 /Vs in silicon [49]), and strong interactions between free charge carriers and the lattice. This interaction facilitates the localization of the charge carriers and narrowing of the bandwidths even further, and thus are expected to crucially affect the transport properties. Considerable effort is involved in describing the polaron dynamics, lifetime, binding energies, and diffusion in the lattice [50, 51]. The mobility of the charge carriers reflects the drift velocity of the charges in the lattice, thus it is influenced by all interactions that it encounters: µ=

vd E

(1.1)

where µ is the drift mobility, vd is the drift velocity and E the electric field.

1.2. Charge transport in organic semiconductors

7

Scattering by defects, impurities and phonons (lattice vibrations) will decrease the drift velocity of charges, thus the electronic mobility. The conductivity of the material is given by charge carrier density n and charge carrier mobility µ. σ = neµ

(1.2)

where e is the elementary charge. The mobility is influenced by the scattering events, so that [52]: eτ (1.3) µ= ∗ m where τ is the time between two consecutive scattering events, and m∗ is the effective mass. This provides a direct relation between the morphology of the materials and their mobility. In polymers the conduction (µ = 10−8 − 10−4 cm2 /Vs) [16, 17] is limited by disorder and hopping of charges between polymer chains, and thus is lower than in oligomers (µ = 10−3 − 10 cm2 /Vs). Even for the class of oligomer devices, the mobility increases with the degree of structural order. Structural defects and chemical impurities present in the crystal lattice promote the localization of the charge carriers. They can either form new states in the semiconductor band-gap, leading to electronic traps, or scatter the charges. Both processes result in a decrease of the mobility. The value of the mobility directly affects the performance of the material in devices, as it is related with the switching speed. Dramatic improvement in control and understanding of the transport mechanism in organic materials, on the molecular level, together with the knowledge on the influence of intrinsic and extrinsic factors on a good performance, has been achieved lately (the value of the mobility increased 5 orders of magnitude in the last 15 years, reaching the value of amorphous silicon [49]). This is very important, as some of the possible applications, like switching devices for active-matrix flat-panel displays (AMFPDs) based on OLED displays, active-matrix backplanes of OTFTs (organic thin film transistors) for ”electronic paper” displays, or radiofrequency identification tags (RFID) require mobilities greater than 1 cm2 /Vs [53], values already exceeded in devices built on organic single crystals [19]. Parallel to organic electronic devices, organic-inorganic hybrids emerge as new applications by coupling high carrier mobilities inorganic semiconductors with the flexibility of organic materials [54]. Most organic semiconductors behave as either p-type or n-type semiconductors (they have either holes, or electrons as majority carriers) (see Fig. 1.1 c, d). The absence of ambipolar behavior is a severe limitation for the organic electronic devices for the fabrication of CMOS7 -like circuits. Different strategies were proposed to overcome this problem. Most of them involve separate steps to fabricate 7 CMOS

- complementary metal oxide semiconductor

8

Chapter 1. Introduction

n-type and p-type transistors [17,55]. However, for many materials it is generally accepted that this is not an intrinsic property, but rather a result of trapping of one type of charge carriers [16] or due to a high energetic barrier for either electron or hole injection from the metal electrodes, which is caused by the relatively large bandgap of organic semiconductors.

1.3 1.3.1

Organic electronic devices Field effect transistors - operation principles

Organic semiconductors are active materials in different devices. Field effect transistors, light emitting diodes and solar cells are intensely studied. Organic materials are soft, fragile and relatively reactive, thus the conventional semiconductors device fabrication technologies are not always compatible with these compounds. For this reason the intrinsic electronic properties could not be reached in devices for a long period of time. They were explored with different techniques (e.g. time of flight [56]). Only lately, using innovative approaches, the fabrication of organic electronic devices was successful and a good reproducibility between research groups was achieved [19, 24, 25].

Figure 1.2: Structure of a field effect transistor (FET) with organic semiconductor as active material. The source electrode is connected to the ground. This convention is valid for all the devices presented in this thesis.

1.3. Organic electronic devices

9

The field effect transistor (FET) is a three terminal device. The three contacts are referred to as gate (G), drain (D) and source (S, connected to ground). The schematic picture of a a FET is drawn in Figure 1.2. The active channel forms at the semiconductor-insulator interface. The gate insulator acts like a capacitor and the electric field applied at the gate electrode determines the density of the charge carriers accumulated at the interface. The current between source and drain is modulated by the gate voltage (VG ). The operation principle of a FET was introduced by Lilienfeld [57], in 1930, and later developed by Shockley and Pearson [58]. In 1947 Bardeen, Brattain and Shockley (Bell Laboratories) discovered the transistor effect and fabricated the first device [59]. They were awarded the Nobel Prize in physics in 1956 for their discovery. The first metal-oxide-semiconductor field-effect transistor (MOSFET) was introduced in 1960 by Khang and Atalla [60]. MOSFETs, built at the surface of inorganic semiconductors, were intensively studied, and they are incorporated in integrated circuits [41]. Owing to similar experimental I-V characteristics between organic field-effect transistors (OFETs) and MOSFETs, the theory developed for MOSFETs is used as starting point in modelling the OFET behavior. However, the electrical transport in organic semiconductors is different than in the covalently bonded inorganic semiconductors. Some attempts to describe the operation principle in OFETs were performed [61, 62]. The transistor channel is active only when the gate voltage (VG ) value exceeds the value of the threshold voltage (VT ). Below this point, the transistor is turned off, and there is no conduction between drain and source (sub-threshold region). In the operation of a FET, two distinct regimes can be distinguished (Fig. 1.3(a)). In the linear regime (small drain-source VD voltages, VD ≪ VG − VT ), the current between drain and source (ID ) depends linearly on the applied voltage (Eq. 1.4). The device acts as a gate voltage - controlled variable resistor. The value of the drain current, ID , is given by: ID =

W µCi (VG − VT )VD L

(1.4)

where L is the channel length, W is the gate width, Ci is the capacitance per unit area of the gate insulator. This model assumes constant velocity, electric field, and inversion layer charge density between the source and the drain. A more realistic approach accounts for the variation of the inversion layer charge between source and drain, and yields the following expression for the current: h V2i W (1.5) µCi (VG − VT )VD − D ID = L 2

10

Chapter 1. Introduction

Figure 1.3: I −V curves for a FET operation. (a) Output characteristics (ID −VD ) for different applied gate voltages (VG ). The curves are obtained from simulating the I − V curves that correspond to expression 1.5. The linear and saturation regimes are indicated. The dotted line points the pinch-off that separates the linear region of operation on the left from the saturation region on the right. (b) Transfer characteristics (ID − VG ) for different drain voltages. The curves correspond to experimental points for a pentacene transistor.

At higher VD voltages, the channel is not continuous, but a depletion area forms at the drain contact. The onset of this region is called pinch-off (Fig. 1.3(a)). Beyond this point, the operation regime is referred to as saturation regime. The drain current is now relatively independent of the drain voltage, being only controlled by the gate voltage, and varies quadratically with the field: ID =

W µCi (VG − VT )2 2L

(1.6)

Equations 1.4, 1.5, and 1.6 represent expressions that yield the value of the mobility of the semiconductor. The mobility can also be estimated from the gate voltage sweep (Fig. 1.3(b)), at low drain voltages VD . Here, the expression for the transconductance (gm ) is: ∂ID (1.7) gm = ∂VG From this equation, the value of the mobility can be extracted: µ=

L 1  ∂ID  W Ci VD ∂VG VD →0

(1.8)

1.3. Organic electronic devices

11

However, all these expressions assume a field independent mobility, and that all the charges induced by the gate voltage are mobile. Key parameters in the operation of a FET are the electronic mobility of the charge carriers and the on/off ratio. The first determines the switching speed and the maximum current, and the later impose the switching of the device from a non-conducting (off ) to a conducting state (on). Good performance organic field-effect transistors (OFETs) were fabricated at the surface of the organic single crystals using deposition techniques that minimize damage at the interface [19, 24]. Fabrication of devices with competitive characteristics remains an ambitious task for large-scale applications.

1.3.2

Fabrication of OFETs

Owing to the fragility of organic materials, processing to incorporate them in electronic devices represents a challenge at this early stage. However, they trigger interest to develop revolutionary methods that are simple and efficient to be used for large scale applications. In this section we will describe recent advances in organic electronic devices, focusing on OFETs fabricated on molecular crystals. Different device structures were proposed to study the electrical transport at the surface of organic crystals, in order to measure a high intrinsic mobility, that is not diminished by disorder introduced during processing. There are many factors involved in the good performance of the device. In spite of the better reproducibility that is archived lately, the results reported by different groups are still not always consistent. This can be attributed in part to the fact that the performance of the electronic devices depends critically on the quality of the crystals and the interfaces, as well as on the extrinsic factors (like, for example the environmental conditions in which the experiments were performed). Organic semiconductor quality, dielectric properties, contacts, and interface properties are equally important. Deposition of the organic semiconductor The most common method used for the deposition of small molecule conductor films is vacuum sublimation. The macroscopic electronic properties of the films are imposed by their crystallinity [53]. Better crystallinity and larger grain size facilitates higher mobilities. Values of 1 cm2 /Vs were reached in vacuum sublimated pentacene TFTs, after optimization of the fabrication process [85]. We mentioned in Section 1.1 that the highest impact that the organic electronics can produce over traditional Si-based technology, is the relatively easy

12

Chapter 1. Introduction

processing techniques that their deposition requires. Polymers are attractive because they are soluble in organic solvents, thus they can be spin coated or printed on flexible substrates, forming amorphous or polycrystalline films. Still, the highest mobilities are achieved in devices build with small molecules. A drawback is that their solubility is limited and they require deposition methods like vacuum sublimation, or physical vapor deposition. These demands are not straightforward to accomplish. Because a high electronic mobility is not sufficient for a material to be competitive for large scale applications, scientists develop different methods to facilitate the compatibility with cheap and easy solution processing techniques. Herwig et al. proposed a synthetic concept for fabrication of a soluble pentacene precursor [63]. The precursor is converted to pentacene via thermal [63] or irradiative [64] treatment, and the obtained thin film transistors (TFTs) exhibit mobilities of 0.2 cm2 /Vs. A different route to increase the solubility of oligomers is the attachment of flexible side groups. At the molecular design, careful attention is payed to the interplay between the degree of solubility and the molecular stacking that the side group induces. Field-effect transistors with maximum mobilities of 0.01 cm2 /Vs were fabricated from quaterthiophene and hexathiophene end-substituted with 3-butoxypropyl groups [65]. The above mentioned directions represent a compromise, a balance between performance and cost, because the mobility of materials deposited from solution remains lower than that of the thermal evaporated material [19, 20]. Moreover, in polycrystalline films [20, 21, 23], the mobility is lower than in single crystals [19, 24–26]. This is partially caused by a large grain boundary resistance. Gate dielectric materials In field-effect transistors the conduction takes place at the surface of the semiconductor, thus the performance is limited by the quality of the interface between organic and dielectric, and only in part by the bulk properties. This is evident from experiments that demonstrate that the gate insulator can modify the charge density at the interface, having a crucial effect on the operation of both polymer [66] and small molecule [67] devices. Particularly important are the roughness of the semiconductor/dielectric interface, and the density of defects and impurities present in this region. Different treatments of the dielectric were proposed in order to decrease the trap density (e.g. OTS8 treatment [68]). General requirements for a high quality dielectric include several parameters. The introduction of a large capacitance, that governs the magnitude of charge 8 OTS

denotes octadecyltrichlorosilane

1.3. Organic electronic devices

13

Figure 1.4: Different geometries of the OFETs. Source (S), drain (D), and gate electrodes are indicated. The transport takes place at the interface between semiconductor layer and gate insulator layer. (a) bottomgate geometry; (b) top-gate geometry.

induced in the channel by gate effect, can be done using high-k dielectrics or by varying their thickness. Besides this, a good insulator in FETs should account for a large breakdown voltage and low leakage currents, excellent thermal and chemical stability. There are three classes of dielectric materials incorporated in OTFTs [69]: inorganic dielectric materials (e.g. SiO2 , Ta2 O5 , Al2 O3 ), organic dielectric materials (e.g. parylene, PS9 , PMMA10 , Fig. 1.1(e)) and self-assembled mono-and multilayers. In FETs with top-gate configuration (Fig. 1.4(a)), the deposition of inorganic dielectrics partially damages the surface of the organic crystals and introduces chemical and structural disorder, due to violent processing methods (sputtering, e-beam, plasma-enhanced chemical vapor deposition). This is the reason why organic dielectrics are used only in bottom-gate devices (Fig. 1.4(b)). The deposition of an organic dielectric requires lower temperatures (typical deposition techniques are spin-coating, casting, and printing), are not destructive, and yield remarkably high mobilities (8 cm2 /Vs in rubrene single crystals transistors with parylene gate dielectric [24]). Contacts Contact issues are amply discussed in the field of molecular electronics. Different models are proposed, as it is believed that the charge transport is either limited by injection problems due to poorly defined contacts, or by bulk conductivity. Baldo 9 PS

denotes polystyrene denotes polymethylmethacrylate

10 PMMA

14

Chapter 1. Introduction

et al. elaborated an interface injection model, that accounts for two distinct injection steps [70]. During the first step, charges are injected from the contacts into an interface region that contain a broad distribution of states induced by interface dipoles. In the second, limiting step, the charges migrate from the interface to the bulk region. Two types of device geometries are used in the fabrication of the field-effect transistors: top-contact and bottom-contact structures. In the first case, it was observed that the deposition of contacts on the organic semiconductors inflicts with their chemical and mechanical fragility, and introduce traps locally, at the metal/organic interface [71]. Different deposition methods are used, depending on the nature of the contact. Metal contacts are usually evaporated or sputtered using shadow masks, or painted (the latter method is also applied for colloidal graphite paste). The choice of the metal contact gives the value of the Schottky barrier, thus the deviation from the ohmic regime, and also the type of the majority carriers. Alternative contact materials, with specific deposition requirements were proposed and successfully applied. Room temperature lamination of metal coated elastomeric stamps (e.g. PDMS11 [19]) represents a non-destructive, high resolution method to fabricate contacts. The polymer PANI12 and the copolymer PEDOT:PSS13 (Fig. 1.1(b)) are currently widely used in organic devices, especially OLEDs and solar cells because of their excellent transparency in the visible region, but also in FETs due to good electrical conductivity, and environmental stability [72].

1.3.3

Outlook

Reviewing the recent advances in materials, methods and emerging applications, two very important issues can be mentioned. Firstly, organic semiconductor research is an exciting and interdisciplinary area of current research activity that raised the interest of theorists, chemists, physicists, and device scientists and is developing extremely fast. Secondly, organic materials impose a new way of thinking because they exhibit a wide range of electrical properties, being tunable from insulator (in gate dielectrics) to semiconductor (as active layers in devices), to metal, and even superconductor. A scheme with the wide range of functionalities that the organic materials offer can be found in Figure 1.1(a-e). These diverse properties are spectacular and give the opportunity of building ”all-organic devices”. 11 PDMS

denotes polydimethylsiloxane denotes polyaniline 13 PEDOT:PSS denotes polyethylene(3,4-dioxythiophene)/polystyrene sulfonate 12 PANI

1.4. Organic molecular crystals

1.4

15

Organic molecular crystals

Characteristic of organic semiconductors are the intramolecular bonds. The alternation of single (σ) and double (π) bonds, referred to as conjugation, is typical for organic conductors. The Carbon atoms involved in this type of bonding are sp2 hybridized. The three hybrid sp2 orbitals form the σ bonds, with the σ-electrons being highly localized. The remaining nonhybridized pz orbitals of adjacent Carbons form the π- orbitals perpendicular to the σ bonds and delocalized over the molecule. The σ bonds are very strong (they are positioned lower in energy). The molecular orbitals that are filled (π-bonding orbitals) form the valence states. The filled orbital with the highest energy is called the Highest Occupied Molecular Orbital. The vacant orbitals (π ∗ -antibonding orbitals) form the conduction band, with LUMO being the Lowest Unoccupied Molecular Orbital. The π-electrons are responsible for an important part of the intermolecular conduction in these materials. Besides this, the molecular packing in the solid, imposed by the physical interactions, determine the physical properties, e.g. the electronic transport. The intermolecular forces are weak, and result from the cooperation and competition between π − σ and π − π interactions. The interaction energy between molecules originates from several contributions. The electrostatic interaction between permanent multipoles (usually quadrupoles), and the dispersion (van der Waals) forces between induced dipoles represent the dominant force. The interaction between the permanent multipoles and the induced multipoles, called induction, is generally treated as a second order term. In a simple pictorial model, the van der Waals forces between molecules originate from the instantaneous polarization of the neutral molecules into dipoles [27]. The temporary generated dipoles promote the formation of new induced dipoles in neighboring molecules and spread in the lattice. For interplanar separation that characterize organic molecular crystals (d > 3.4 ˚ A [32]), the van der Waals forces are always attractive. The quadrupolar interactions emerge from the coupling between the permanent quadrupolar moments of the molecules [73, 74], and can be attractive or repulsive, depending on the relative orientation of the quadrupoles. The crystal structure is determined by an interplay between van der Waals forces and quadrupolar interactions. The two competing interactions promote different molecular stacking that minimize the energy. The van der Waals interactions are optimized for ”face-to-face” orientation of the molecules, that result in the maximum π− overlap. On the other hand, the presence of quadrupolar moments favor the ”edge-to-face” stacking, in which the hydrogen on one ring

16

Chapter 1. Introduction

Figure 1.5: Crystal structure of pentacene single crystal. Left panel: view along the [100] axis (the layered structure can be seen). Right panel: the herringbone arrangement within the layer, view along the long axis of the molecule. The unit cell orientation is also drawn.

encounter the π− network of the adjacent molecule. The most common crystal packing that results from the afore mentioned mechanisms is the herringbonearrangement (Fig. 1.5). Oligomers often crystallize in a layered fashion, and the molecular arrangement within the layers is imposed by the van der Waals and quadrupolar interactions. Gavezzotti et al. distinguished four possible herringbone modes in which polynuclear aromatic hydrocarbons pack: herringbone structure (naphthalene, anthracene, pentacene ), sandwich herringbone structure (pyrene, perylene), γ structure (benzopyrene, coronene), β structure (trybenzopyrene, tetrabenzoperylene) [75]. Cofacial π-stacking in molecular crystals is hardly ever found. Anthony et al. synthesized these types of structure by substitution of different groups to pentacene backbone [76]. They showed that different type and amount of π-overlap can be controlled by the nature, size and position of the substituent, leading to various stacking motifs. Although seldom encountered, cofacial configurations are particularly attractive because they accommodate the largest electronic splitting in the HOMO and LUMO levels, promoting the highest theoretical predicted charge carrier mobilities [77]. In the local picture (as opposed to the band picture) of the charge transport in organic conjugated materials, the efficiency of this process reflects how easy the charges are transferred

1.4. Organic molecular crystals

17

between neighboring sites, and, as expected, is very sensitive to orientation of the molecules with respect to each other. The electronic coupling between adjacent molecules, quantified by the transfer integral t, is modulated by the molecular arrangement and directly associated with the electronic mobility [45, 46, 77]. In the framework of these calculations, the amplitude of the electronic coupling is influenced by the intermolecular separation distance, the molecular overlap, the length of the molecule, and, in the case of herringbone structures, the rotation of molecular planes [77]. Additionally, the relevance of thermal motions in the modulation of the electronic coupling between molecules in an an organic solid was demonstrated [45]. Owing to the weak interaction forces between molecules in the solid, small variations in the crystal packing can be present, leading to polymorphism. The term polymorphism refers to the existence of more than one crystal structure for a particular compound. This phenomenon is frequently displayed by organic crystals and is driven by the growth conditions and/or subsequent treatment. In rubrene, different polymorphs can be obtained, depending on the pressure in the system in which the starting material is sublimed (see Chapter. 6 in this thesis and the references therein). In quaterthiophene (α−4T) [79] and sexithiophene (α−6T) [80, 81] different molecular arrangements are induced by the source temperature. The general picture is even more complex in pentacene films, where four polymorphs were detected [78, 82]. The obtained polymorph is dictated by the substrate type, the substrate temperature and the thickness of the film. For pentacene single crystals, only one polymorph exists [78]. The existence of different polymorphs for some molecular crystals provides a unique opportunity to understand the influence of the crystal packing on the electronic properties, since these systems only differ with the orientation of the constituent building blocks with respect to each other, the molecule being the same [42, 46, 81]. For example, Troisi and Orlandi performed band structure calculations on the four pentacene polymorphs and found that the mobility tensor is highly anisotropic for three of the four considered polymorphs [83]. This result is relevant for understanding the fundamental mechanisms of charge transport in organic crystals.

18

1.5

Chapter 1. Introduction

Single crystals - model systems for investigation of intrinsic properties

The progress in the field of organic electronics requires a good fundamental understanding of factors that influence the electronic behavior. This leads to a better control of the microscopic properties that determine the conduction of the materials used in the devices. A correct insight on the interplay between the effects of chemical structure and molecular orientation on the transport process can only be accomplished when single crystals are used. Single crystals are not meant to be incorporated in applications, but to provide a well defined structure, in which the intrinsic electronic properties of the material can be measured [19,24–26,84]. They serve as model systems, for which structureproperties correlations can be explored. Although thin films (polycrystalline or amorphous) are more attractive for organic electronic devices, here the intrinsic properties are generally masked by grain boundaries where structural defects localize and trap the charge carriers. The major limitation of thin-film transistors (TFTs) comes from the fact that their performance is severely dependent on the fabrication conditions. Moreover, even films that are grown in an identical manner can exhibit very different electrical properties [85]. The electrical properties of organic single crystals have been successfully probed by time-of-flight (TOF) [56] and space-charge-limited current (SCLC) methods [26], as well as field-effect transistor (FET) experiments [19, 24, 25].

1.6

Aim of the present research

Motivated by the various properties that organic semiconductors reveal, we evaluate in this thesis a critical analysis of the different factors that determine the electrical conduction in these materials. Although the progress is fast, both in performance and reliability, there are still numerous questions that lack an answer, or for which inconsistent explanations were given. The questions we address in this thesis concern intrinsic and extrinsic factors that determine the charge transport in organic conductors. In order to answer these questions we focus our work on single crystals (the advantages of working with single crystals are outlined in Section 1.5). We systematically study the effect of structural defects and impurities (Chapter 2, Chapter 3), geometrical factors (Chapter 4), molecular stacking (Chapter 6) and exposure to ambient conditions

1.7. Outline of the thesis

19

(Chapter 5). While the organic field-effect transistor characteristics were improved in the past years by implementation of novel processing techniques, the charge carrier mobility remains limited. In this context, our efforts focus on the understanding of the microscopic processes that determine the value of the mobility in molecular crystals, in particular the generation and migration of defects. The work aims not only at improving the performance of organic materials for possible incorporation in devices, but also at satisfying the pressing need for insight into the relevant physical processes that govern the electrical conduction in these materials. We show that even small defect densities are critical and act as charge traps, consequently decreasing the mobility significantly. By systematic measurements, we demonstrate that the origin of defects is of dual nature. The trapping sites such as crystal defects and chemical impurities are created during crystal growth. Additional traps are introduced during processing or exposure to ambient conditions. Moreover, we propose successful methods that allow the study of high quality materials with desired properties. Furthermore, we incorporate the investigated materials into field-effect devices. Here, the conduction channel is limited to the vicinity of the interface. We show that the conductivity is strongly influenced by disorder and charge traps near the interface, and propose a reliable method to overcome this limitation. Hence, our efforts can broadly be defined as contributions in the area of fundamental and applied studies on the organic molecular crystals properties. By means of correlations between electrical measurements and structural properties, determined via X-ray diffraction, or thermodynamic response via thermogravimetric experiments, we wish to cover important features of the chemical, physical and technological problems that arise in these materials.

1.7

Outline of the thesis

This thesis is organized as follows: Chapter 2 emphasizes the importance of the control of defects and impurity states in organic molecular crystals in order to obtain a high electronic mobility. We measure record electronic mobilities as a result of the reduction of the number of traps by careful crystal growth and subsequent handling. The temperature dependence of the mobility is consistent with the band model for electronic transport. In recent studies, little attention is paid to the concentration, distribution of impurities and their consequences on electronic properties. Thus far, it has

20

Chapter 1. Introduction

been generally assumed that the impurities are evenly distributed throughout the lattice. In Chapter 3 we will show that this is not the case for pentacene single crystals, where the impurities are are located preferentially at the surface. Moreover, we describe a new, reliable method, and demonstrate that in our fieldeffect transistor devices the impurity states at the interface can be made inactive by incorporating them into the dielectric gate barrier. In Chapter 4 we give a geometrical description of the electric field distribution in organic crystals. We report the cross-over from 1D to 2D - space charge limited conduction in pentacene single crystals with planar contacts. Furthermore, we establish the parameters for which the conduction is dominated by bulk charges and show that the transition to a surface-governed transport takes place gradually. These results incorporate corrections for the anisotropic resistivity. Chapter 5 is dedicated to the effect of air exposure on the electronic properties of pentacene single crystals. We show experimentally that air can diffuse reversibly in and out of the crystals. This process is reflected in the electrical properties. We are able to distinguish two competing mechanisms that modulate the electronic transport: the doping effect of oxygen and the trapping caused by the presence of water vapor. By combining the gravimetric and electric measurements, we can describe quantitatively these effects. The investigations that yield the results presented in Chapters 2, 3, 4, and 5 focused on the study of pentacene single crystals, the first organic crystalline material already incorporated in prototypes of commercial devices [20]. The measurements discussed in Chapter 6 were motivated by intriguing changes observed in the values of electronic mobility of rubrene single crystals at low temperatures. We relate this changes with the structural transformations in the material.

References

[1] R. H. Friend, R. W. Gymer, A. B. Holmes, J. H. Burroughes, R. N. Marks, C. Taliani, D. D. C. Bradley, D. A. Dos Santos, J. -L. Br´edas, M. Logdlund, and W. R. Salaneck, Nature 397, 121 (1999) [2] H. Sirringhaus, P. J. Brown, R. H. Friend, M. M. Nielsen, K. Bechgaard, B. M. W. Langeveld-Voss, A. J. H. Spiering, R. A. J. Janssen, E. W. Meijer, P. Herwig, and D. M. de Leeuw, Nature 401, 685 (1999) [3] A. Dodabalapur, H. E. Katz, L. Torsi, and R. C. Haddon, Science 269, 1560 (1995) [4] B. Crone, A. Dodabalapur, Y. Y. Lin, R. W. Filas, Z. Bao, A. LaDuca, R. Sarpeshkar, H. E. Katz, and W. Li, Nature 403, 521 (2000) [5] C. R. Kagan, D. B. Mitzi, and C. D. Dimitrakopolous, Science 286, 945 (1999) [6] G. Malliaras, and R. Friend, Physics Today May 2005, 53 (2005) [7] S. Coe, W. K. Woo, M. Bawendi, and V. Bulovic, Nature 420, 800 (2002) [8] K. Tashiro, Ferroelectric Polymers, H. S. Nalwa ed. (Marcel Dekker, New York, 1995) [9] R. C. G. Naber, C. Tanase, P. W. M. Blom, G. H. Gelinck, A. W. Marsman, F. J. Touwslager, S. Setayesh, and D. M. de Leeuw, Nat. Mat. 4, 243 (2005)

21

22

References

[10] X. Guo, J. P. Small, J. E. Klare, Y. Wang, M. S. Purewal, I. W. Tam, B. H. Hong, R. Caldwell, L. Huang, S. O’Brien, J. Yan, R. Breslow, S. J. Wind, J. Hone, P. Kim, and C. Nuckolls, Science 311, 356 (2006) [11] A. Bonfiglio, D. De Rossi, T. Kirstein, I. R. Locher, F. Mameli, R. Paradiso, and G. Vozzi, IEEE Trans. on Info. Tech. in Biomed. 9, 319 (2005) [12]

http://www.research.philips.com/newscenter/archive/2004/rollabledisplay.html

[13] Components fabricated on organic materials and introduced on the market include displays for mobile phones (Philips 639 - a clamshell with a PolyLED external display), portable digital music players (Philips SA1784-SA179 ), car radios, men’s shaver (Philishave 8894) and digital cameras (Kodak LS633) [14] S. F. Forrest, Nature 428, 911 (2004) [15] C. K. Chiang, C. R. Fincher, Jr., Y. W. Park, A. J. Heeger, H. Shirakawa, E. J. Louis, S. C. Gau, and A. G. MacDiarmid, Phys. Rev. Lett. 39, 1098 (1977) [16] L. L. Chua, J. Zaumseil, J. F. Chang, E. C. W. Ou, P. K. H. Ho, H. Sirringhaus, and R. H. Friend, Nature 434, 194 (2005) [17] E. J. Meijer, D. M. de Leeuw, S. Setayesh, E. van Veenendaal, B. H. Huisman, P. W. M. Blom, J. C. Hummelen, U. Scherf, and T. M. Klapwijk, Nat. Mat. 2, 678 (2003) [18] R. J. Kline, M. D. McGehee, E. N. Kadnikova, J. S. Liu, and J. M. J. Frechet, Adv. Mat. 15, 1519 (2003) [19] V. C. Sundar, J. Zaumseil, V. Podzorov, E. Menard, R. L. Willett, T. Someya, M. E. Gershenson, and J. A. Rogers, Science 303, 1644 (2004) [20] G. H. Gelinck, H. E. A. Huitema, E. van Veenendaal, E. Cantatore, L. Schrijnemakers, J. B. P. H. van der Putten, T. C. T. Geuns, M. Beenhakkers, J. B. Giesbers, B. H. Huisman, E. J. Meijer, E. M. Benito, F. J. Touwslager, A. W. Marsman, B. J. E. van Rens, and D. M. de Leeuw, Nat. Mat. 3, 106 (2004) [21] Y. J. Zhang, J. R. Petta, S. Ambily, Y. L. Shen, D. C. Ralph, and G. G. Malliaras, Adv. Mat. 15, 1632 (2003) [22] D. V. Lang, X. Chi, T. Siegrist, A. M. Sergent, and A. P. Ramirez, Phys. Rev. Lett. 93, 086802 (2004)

References

23

[23] D. J. Gundlach, T. N. Jackson, D. G. Schlom, and S. F. Nelson, Appl. Phys. Lett. 74, 3302 (1999) [24] V. Podzorov, S. E. Sysoev, E. Loginova, V. M. Pudalov, and M. E. Gershenson, Appl. Phys. Lett. 83, 3504 (2003) [25] R. W. I. de Boer, T. M. Klapwijk, and A. F. Morpurgo, Appl. Phys. Lett. 83, 4345 (2003) [26] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 84, 3061 (2004) [27] M. Pope, and C. E. Swenberg, Electronic Processes in Organic Crystals and Polymers, 2nd ed. (Oxford University Press, New York, 1999) [28] S. J. Tans, M. H. Devoret, H. Dai, A. Thess, R. E. Smalley, L. J. Geerligs, and C. Dekker, Nature 386, 474 (1997) [29] T. W. Ebbesen, H. J. Lezec, H. Hiura, J. W. Bennett, H. F. Ghaemi, and T. Thio, Nature 382, 54 (1996) [30] N. de Jonge, Y. Lamy, K. Schoots, and T. H. Oosterkamp, Nature 420, 393 (2002) [31] S. S. Wong, E. Joselevich, A. T. Woolley, C. Li Cheung, and C. M. Lieber, Nature 394, 52 (1998) [32] K. A. Williams, P. T. M. Veenhuizen, B. G. de la Torre, R. Eritja, and C. Dekker, Nature 420, 761 (2002) [33] Z. K. Tang, L. Zhang, N. Wang, X. X. Zhang, G. H. Wen, G. D. Li, J. N. Wang, C. T. Chan, and P. Sheng, Science 292, 2462 (2001) [34] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A. Firsov, Science 306, 666 (2004) [35] Y. B. Zhang, Y. W. Tan, H. L. Stormer, and P. Kim, Nature 438, 201 (2005) [36] M. Mehring, K. F. Thier, F. Rachdi and T. de Swiet, Carbon 38, 1625 (2000) [37] A. F. Hebard, M. J. Rosseinsky, R. C. Haddon, D. W. Murphy, S. H. Glarum, T. T. M. Palstra, A. P. Ramirez, and A. R. Kortan, Nature 350, 600 (1991) [38] D. Jerome, Science 252, 1509 (1991)

24

References

[39] C. W. Tang, App. Phys. Lett. 48, 183 (1986) [40] G. Yu, J. Gao, J. C. Hummelen, F. Wudl, and A. J. Heeger, Science 270, 1789 (1995) [41] S. M. Sze, Physics of Semiconductor Devices (Wiley, New York, 1981) [42] R. C. Haddon, X. Chi, M. E. Itkis, J. E. Anthony, D. L. Eaton, T. Siegrist, C. C. Mattheus, and T. T. M. Palstra, J. Phys. Chem. B 106, 8288 (2002) [43] R. Silbey, J. Jortner, S. A. Rice, and M. T. Vala, Jr., J. Chem. Phys. 42, 733 (1965) [44] G. Brocks, J. van den Brink, and A. F. Morpurgo, Phys. Rev. Lett. 93, 146405 (2004) [45] A. Troisi, and G. Orlandi, Phys. Rev. Lett. 96, 086601 (2006) [46] Y. C. Cheng, R. J. Silbey, D. A. da Silva Filho, J. P. Calbert, J. Cornil, and J.-L. Br´edas, J. Chem. Phys. 118, 3764 (2003) [47] K. Hannewald, V. M. Stojanovic, J. M. T. Schellekens, P. A. Bobbert, G. Kresse, and J. Hafner, Phys. Rev. B 69, 075211 (2004) [48] http://www.eecs.umich.edu/∼singh [49] J. M. Shaw, and P. F. Seidler, IBM J. Res.& Dev. 45, 3 (2001) [50] K. Hannewald, and P. A. Bobbert, Phys. Rev. B 69, 075212 (2004) [51] R. W. Munn, and R. Silbey, J. Chem. Phys. 83, 1843 (1985) [52] N. W. Ashcroft, and N. D. Mermin, Solid state physics (Saunders College Publishing, 1979) [53] C. D. Dimitrakopoulos, and D. J. Mascaro, IBM J. Res. & Dev. 45, 11 (2001) [54] C. R. Kagan, D. B. Mitzi, and C. D. Dimitrakopoulos, Science 286, 945 (1999) [55] Y. Y. Lin, A. Dodabalapur, R. Sarpeshkar, Z. Bao, W. Li, K. Baldwin, V. R. Raju, and H. E. Katz, Appl. Phys. Lett. 74, 2714 (1999) [56] W. Warta, and N. Karl, Phys. Rev. B 32, 1172 (1985) [57] J. E. Lilienfeld, U.S. patent 1745175 (1930)

References

25

[58] W. Shockley, and G. L. Pearson, Phys. Rev. 74, 232 (1948) [59] W. Shockley, U.S. patent 02569347 (1951) [60] D. Khang, and M. M. Atalla, IRE Solid-State Devices Res. Conf., Carnegie Institute of Technology, Pittsburg, Pa. (1960) [61] G. Horowitz, R. Hajlaoui, R. Bourguiga, and M. Hajlaoui, Synth. Met. 101, 401 (1999) [62] G. Horowitz, and P. Delannoy, J. App. Phys. 70, 469 (1991) [63] P. T. Herwig, and K. M¨ ullen, Adv. Mat. 11, 480 (1999) [64] A. Afzali, C. D. Dimitrakopoulos, and T. O. Graham, Adv. Mat. 15, 2066 (2003) [65] H. E. Katz, A. J. Lovinger, J. Johnson, C. Kloc, T. Siergist, W. Li, Y.-Y. Lin, and A. Dodabalapur, Nature 404, 478 (2000) [66] J. Veres, S. D. Ogier, S. W. Leeming, D. C. Cupertino, and S. M. Khaffaf, Adv. Func. Mat. 13, 199 (2003) [67] A. F. Stassen, R. W. de Boer, N. N. Iosad, and A. F. Morpurgo, Appl. Phys. Lett. 85, 3899 (2004) [68] C. Goldmann, S. Haas, C. Krellner, K. P. Pernstich, D. J. Gundlach, and B. Batlogg, J. App. Phys. 96, 2080 (2004) [69] A. Facchetti, M. -H. Yoon, and T. J. Marks, Adv. Mat. 17, 1705 (2005) and the refferences therein [70] M. A. Baldo, and S. R. Forrest, Phys. Rev. B, 64, 085201 (2001) [71] R. W. I. de Boer, and A. F. Morpurgo, Phys. Rev. B, 72, 073207 (2005) [72] B. J. de Gans, P. C. Duineveld, and U. S. Schubert, Adv. Mat. 16, 203 (2004) [73] C. A. Hunter, and J. K. M. Sanders, J. Am. Chem. Soc. 112, 5525 (1990) [74] C. R. Newman, C. D. Frisbie, D. A. da Silva Filho, J.-L. Br´edas, P. C. Ewbank, and K. R. Mann, Chem. Mat. 16, 4436 (2004) [75] G. R. Desiraju, and A. Gavezzotti, Acta. Cryst.B 45, 473 (1989) [76] J. E. Anthony, D. L. Eaton, and S. R. Parkin, Org. Lett. 4, 15 (2002)

26

References

[77] J. Cornil, D. Beljonne, J.-P. Calbert, and J.-L. Br´edas, Adv. Mat. 13, 1053 (2001) [78] C. C. Mattheus, A. B. Dros, J. Baas, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Acta. Cryst. C 57, 939 (2001) [79] T. Siegrist, C. Kloc, R. A. Laudise, H. E. Katz, and R. C. Haddon, Adv. Mat. 10, 379 (1998). [80] G. Horowitz, B. Bachet, A. Yassar, P. Lang, F. Demanze, J.-L. Fave, and F. Garnier, Chem. Mat. 7, 1337 (1995) [81] T. Siegrist, R. M. Fleming, R. C. Haddon, R. A. Laudise, A. J. Lovinger, H. E. Katz, P. Bridenbaugh, and D. D. Davis, J. Mater. Res. 10, 2170 (1995) [82] C. C. Mattheus, G. A. de Wijs, R. A. de Groot, and T. T. M. Palstra, J. Am. Chem. Soc. 125, 6323 (2003) [83] A. Troisi, and Giorgio Orlandi, J. Phys. Chem. B 109, 1849 (2005) [84] C. Reese, and Z. Bao, J. Mater. Chem. 16, 329 (2006) [85] S. F. Nelson, Y.-Y. Lin, D. J. Gundlach, and T. N. Jackson, Appl. Phys. Lett. 72, 1854 (1998)

2

The Effect of Impurities on the Mobility of Single Crystal Pentacene∗

We have obtained a hole mobility for the organic conductor pentacene (in the single crystal form) of µ = 35 cm2 /Vs at room temperature increasing to µ = 58 cm2 /Vs at 225 K. These high mobilities result from a purification process in which 6,13-pentacenequinone was removed by vacuum sublimation. The number of traps is reduced by two orders of magnitude compared with conventional methods. The temperature dependence of the mobility is consistent with the band model for electronic transport.

∗ This Chapter is adapted from O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 84, 3061 (2004).

27

28

2.1

Chapter 2. The Effect of Impurities on the Mobility

Introduction

Organic materials are presently being investigated and incorporated in semiconductor devices for a new era of the electronics industry. The understanding of the electrical conduction mechanism in these materials and at their interfaces represents a challenge, for which various, often conflicting models have been proposed. Molecular crystals are formed by relatively weak van der Waals interaction between molecules, and the molecular packing determines the electronic behavior. Thus the charge carrier transport must be described using completely different models than for covalently bonded semiconductors. Of the many molecular conductors, pentacene is a promising candidate for future electronic devices and an interesting model system. Recent improvements in electronic applications showed that this material exhibits mobilities higher than 1 cm2 /Vs for TFTs made from highly ordered films [1,2]. A mobility up to 8 cm2 /Vs was measured in single crystals of rubrene using a complete organic field-effect transistor [3]. The anisotropy of mobility, and a band-like temperature dependence was demonstrated in rubrene FETs with PDMS flexible elastomeric stamps as substrate, and air-gap as gate dielectric. This fabrication method yields mobilities as high as 15 cm2 /Vs at room temperature [4] and 20 cm2 /Vs using the 4-probe measurements [5]. The above mentioned values for mobilities reflected hole transport. By using the last mentioned approach, electron mobilities of 1.6 cm2 /Vs were measured for TCNQ (7,7,8,8-tetracyanoquinodimethane) single crystals [6]. The importance of impurities for the limitations in device performance has been emphasized during the last few years. However, little quantitative analysis concerning the consequences of impurities is incorporated in recent studies [7].

2.2

Control of the defects and impurity states in organic crystals

We report a mobility of µ = 35 cm2 /Vs at room temperature increasing to µ = 58 cm2 /Vs at 225 K for ultra-pure pentacene single crystals. The crystals were obtained by vapor transport growth in argon flow after purification of the material by a vacuum sublimation technique designed to remove pentacenequinone. The content of the quinone impurity in pentacene was determined using high pressure liquid chromatography technique (HPLC), indicating a reduction by almost one order of magnitude. The structures of pentacene and 6,13-pentacenequinone, respectively, are shown in Figure 2.1.

2.2. Control of the defects and impurity states in organic crystals

2.2.1

29

Spectroscopic evidence for the presence of impurities

Figure 2.1: Molecular structure of pentacene (a) and 6,13-pentacenequinone (b).

Figure 2.2: Left panel: IR spectrum of pentacene powder without any additional purification. The presence of the peak around 1697 cm−1 is due to quinone impurities. Right panel: mass spectrum of the starting material. The peak at m/z = 278 corresponds to pentacene and the one at m/z = 308 is a result of the presence of 6,13-pentacenequinone. The starting material for the experiment was pentacene obtained from Aldrich. Infrared absorption measurements (Nicolet Nexus spectrometer) show that 6,13pentacenequinone is present as an impurity. The evidence for this is the absorption peak at 1697 cm−1 which is assigned to a C=O bond vibration (Fig. 2.2 - left panel). The pure material does not have significant absorption in this region. The infrared experiments were complemented by mass spectrometry analysis, which

30

Chapter 2. The Effect of Impurities on the Mobility

confirmed that the C=O vibration originates from a pentacenequinone molecule (Fig. 2.2- right panel).

2.2.2

Purification of the starting material

We have used vacuum sublimation under a temperature gradient as purification method. This technique is effective for the separation of impurities from a solid if these impurities have a vapor pressure that is sufficiently different from the desired product [8]. The cleaning tube is shown in Figure 2.3. The pentacene powder is placed in an alumina boat inside a glass tube that is thoroughly cleaned chemically, and then heated in a furnace under vacuum to remove the solvents used for cleaning. A controlled temperature gradient is applied along this tube. The purification takes place at T = 430 K for 70 h under a dynamic vacuum of a membrane pump. Special attention is paid to avoid contamination due to vacuum connections. The sublimated molecules will condense in the cold zone of the tube. The entire set-up is placed in the dark to prevent UV degradation of the acene molecules. The carbonyl groups at each side of the middle ring reduce the sublimation enthalpy compared to the host molecule, thus at T = 430 K pentacene will not sublime and only quinone will be removed. This can be detected as a brown powder on the walls of the tube. The violet powder that did not sublime is purified pentacene that is used as the starting material for the single crystal growth. We repeat this purification process several times to minimize the impurity content. We will present in Section 2.2.4 the effect of the different sublimation steps on the degree of purity.

Figure 2.3: Schematic overview of the vacuum sublimation set-up used to remove the 6,13-pentacenequinone from pentacene.

2.2. Control of the defects and impurity states in organic crystals

2.2.3

31

Single crystal growth

Pentacene single crystals were obtained using physical vapor transport in a horizontal glass tube [9] under a stream of argon. The use of ultra-pure argon without hydrogen as the transporting gas is motivated by the need to prevent the introduction of other impurities, like 6,13-dihydropentacene in the crystal, which can form by the hydrogenation of the acene at the middle ring (most reactive positions).

Figure 2.4: Crystal growth setup. The source material is placed in the boat in the hot part of the tube and it is transported by the inert gas to the crystallization region. The temperature profile of the growth tube is shown in the lower panel. The gas was obtained from AGA, with purity of 99.999%. A drying column was inserted in the system for additional purification of the gas. We paid attention to the quality of the transporting gas because the quinone can be re-introduced by residual water or oxygen as ppm impurities in the carrier gas during growth. The growth tube set-up is presented in Figure 2.4 (upper panel). Prior to growth, the inner tube was cleaned with soap, type II water (ρ ∼ = MΩcm at 25◦ C), acetone and alcohol. The empty tube was heated in argon flow at T = 620 K for 15 hours

32

Chapter 2. The Effect of Impurities on the Mobility

to remove possible water and other solvents. 30 − 40 mg of the source pre-cleaned material was placed at the end of the tube in an alumina boat. A temperature gradient was applied by resistive heating of two heater coils around the tube. The temperature profile is shown in Figure 2.4 (lower panel). The temperature was controlled using two thermocouples, placed at positions marked with labels T1 and T2 . The sublimation temperature, Ts = 545 K, was kept as low as possible, in order to obtain a low crystallization rate that ensures a minimum formation of defects and avoids side reactions [10]. The setup was placed in the dark to protect against oxidation. The molecules, in the vapor phase, are transported by the argon gas and will crystallize in the cold part of the tube (approx. 30 cm from the sublimation point. The crystals are platelet-shaped and their dimensions can be controlled by the growth time. Typical growth time is around 3 days. This yields crystals of maximum 4 × 4 mm in plane and 50 µm thickness. After obtaining ultra-pure crystals, they were annealed for 50 h at a lower temperature (Th = 450 K) than the crystallization temperature (Tc = 490 K). This treatment diminishes the number of dislocations and the stress in the crystal. The crystal structure of pentacene single crystals at room temperature is given in the Appendix A of this thesis. Without pre-purification of the pentacene, pentacenequinone will sublime together with the host molecule. Part of it will be introduced in the pentacene matrix in the crystallization process at the low temperature part of the tube.

2.2.4

Quantitative analysis of 6,13-pentacenequinone content

We used HPLC (Agilent 1100 LC/MSD) [11] to determine the impurity concentration of the pentacenequinone in pentacene single crystals. Pentacene crystals were dissolved in 1,2,4-trichlorobenzene by stirring for 24 h at 45◦ C under inert atmosphere (in the glove-box). Pentacene (P) and pentacenequinone (PQ) were separated on a silica column. The selectivity for retention was controlled with a 3 : 1 v/v mixture of 1,2,4trichlorobenzene and cyclohexane used as mobile phase, at 80◦ C. The volume of the sample injected was 50 µl. We used the Beer’s law [12] to determine the concentration of pentacenequinone impurity in pentacene single crystals. The absorbance A at a particular wavelength λ is proportional to the concentration c of the constituent species: A(λ) = bε(λ)c (2.1) where b is optical path length and ε(λ) is the extinction coefficient at λ. Com-

2.2. Control of the defects and impurity states in organic crystals

33

Figure 2.5: Retention times for pentacene (P) and pentacenequinone (PQ) at λ = 390 nm. Upper panel: P with 0.4% PQ. Lower panel: P enriched with 30% PQ

bining the expressions for the absorbance of P and PQ, corresponding to the two peak areas in Figure 2.5, we obtain the relative concentration (c%) of PQ in P. The expression that allows the calculation of concentration involve the extinction coefficients (εP , εP Q ), peak areas (AP , AP Q ) and molar masses (MQ , MP Q ) of the two components: H (2.2) cP Q (%) = 100 · 1+H where we define H: AP Q εP MP Q · · (2.3) H= AP εP Q MP The amount of quinone was determined from the integrated intensity of the chromatogram, using a diode array UV-Vis detector that can be tuned at different absorption lines (Fig. 2.5). We have chosen λ = 390 nm because it corresponds to a low extinction coefficient for pentacene, and much higher value for pentacenequinone [13]: ( log(εP ) = 2.75 (2.4) log(εP Q ) = 4.389 Figure 2.5 presents typical HPLC measurements from which pentacenequinone concentration in pentacene was calculated. The retention times for pentacene and pentacenequinone were ≃ 2 min, and ≃ 11 min respectively. The upper panel corresponds to 0.4% PQ in P. We deliberately introduced pentacenequinone in

34

Chapter 2. The Effect of Impurities on the Mobility

Figure 2.6: 6,13-pentacenequinone concentration in pentacene in different stages of purification: 1- as received, 2- single sublimation clean, 3- double sublimation cycle, 4- crystal grown from untreated powder, 5- crystal grown from doubly cleaned powder.

the solution obtained from the solvation of a pentacene single crystals. This can be seen as an enhanced signal at time = 11 min in the lower panel of Figure 2.5, that corresponds to 30% PQ content. Additional peaks, of lower intensity, arising from other impurities can be detected in both panels of Figure 2.5. However, in this study we did not concentrate on the quantitative evaluation of their content, but we focus only on the majority impurity. The quinone concentration was reduced from the as received material containing 0.68% in two sublimation steps to 0.17%. Subsequent crystal growth reduces the quinone concentration to 0.028%(±0.004) compared with 0.11%(±0.006) in crystals grown from untreated powder (Fig. 2.6). The characteristic absorption of C=O is observable even for the purest crystals (stage 5 in Fig. 2.6). We will show in Chapter 3 that this small quantity of pentacenequinone impurity still present in pentacene single crystals is critical for the fabrication and operation of the field-effect transistors.

2.3. Electronic response to structural and chemical properties

2.3

35

Electronic response to structural and chemical properties

2.3.1

Space charge limited currents

We determined the electrical properties of the pure pentacene single crystals using space-charge-limited current (SCLC) measurements (Fig. 2.7). Gold was used as hole-injecting electrode. The samples were measured in dark and a vacuum of 2 · 10−7 mbar. In the analysis of the evolution of the current density J with respect to the applied electric field E several assumptions are made. Following the standard analysis [7, 14, 15], we consider: • a 1-dimensional, unipolar current flow (holes are injected from the contact placed at position x = 0 and collected at the contact at x = L); • the contacts are ohmic; • the mobility of free charge carriers is independent of the magnitude of the applied electric field; • the diffusion of charge carriers inside the crystal is neglected; • the injecting contact is an infinite source of charge carriers; • the traps are homogenously distributed in space and all correspond to one discrete energy level; • the density of free charge carriers (nf ) is described by Boltzmann statistics, and the density of the trapped (localized) carriers (nt ) follows the FermiDirac statistics, as follows:  E (x)  F nf = NV exp − kT

(2.5)

nt =

(2.6)

h(E)   F (x) 1 + exp Ei −E kT

Here NV is the effective density of states in the valence band, EF the Fermi level, Ei stands for the energy, h(E) describes the energetic distribution of localized states, and k is Boltzmann’s constant.

36

Chapter 2. The Effect of Impurities on the Mobility

Figure 2.7: Current density (J) vs. electric field (E) for pentacene single crystal at room temperature. The solid lines represent the fits for the three different regimes. The value of the electric field that corresponds to the transition to the trap-free regime (ET F L ) is shown. The inset shows experimental configuration of the a, b, and c*-axis, and the contacts.

Four distinct regimes can be distinguished in the evolution of the current with the applied electric field (see Fig. 2.7). At small electrical fields the transport is ohmic, and the current density depends linearly on the electric field. The current is space charge limited (SCLC) at high electric fields. In the first part of the SCLC regime, the injected carriers are trapped and the current is reduced by a factor Θ, which represents the ratio between free and total number of charge carriers introduced in the solid (Eq. 2.7): Θ=

nf nf = ntot nf + nt

(2.7)

where nf and nt are the free and trapped carriers density, respectively, and ntot is the total carrier density. At VT F L /ET F L (trap-filling voltage) the current rises abruptly. After this point, all the traps are filled and the trap-free regime is reached (Θ = 1). The expression that describes the current-voltage (density of current - electric field) are obtained upon combining several expressions:

2.3. Electronic response to structural and chemical properties

37

• the continuity equation (Eq. 2.8) [14, 16]: J(x) = eµnf (x)E(x)

(2.8)

where J(x) is the current density at distance x from injecting electrode, e is the elementary charge, µ is the charge carrier mobility and E is the electric field,

• Poisson’s equation (Eq. 2.9): dE(x) e e ntot (x) = [nf (x) + nt (x)] = dx ǫ0 ǫr ǫ0 ǫr

(2.9)

Combining equations 2.7, 2.8, and 2.9, the expression of the current becomes: J = µΘǫ0 ǫr E

dE dx

(2.10)

By integrating this expression, one obtains: Jx E 2 (x) − E 2 (0) = µΘǫ0 ǫr 2

(2.11)

Taking into account the assumption made previously, the electric field at the injecting electrode (x=0) is E(0)=0, Eq 2.11 becomes: s 2Jx (2.12) E(x) = µΘǫ0 ǫr Integration of Eq 2.12, together with the use of the boundary conditions imposed in the beginning, yields the expression of the density of current versus electric field (Mott-Gurney law) [7, 14]: JSCLC =

9 V2 ǫ0 ǫr Θµ 3 8 L

(2.13)

where JSCLC is the current density in the space-charge-limited regime, V is the applied voltage across a length L, ǫr is the dielectric constant of the conductor (for which we use the literature value ǫr =3), and Θ the ratio between the concentration of free carriers and the total numbers of carriers (see Eq. 2.7). The mobility in our experiments was calculated from the trap-free region of the space-charge-limited-current regime, using the Mott-Gurney model (Eq. 2.13). We note that this formula was derived for the sandwich-type electrode geometry,

38

Chapter 2. The Effect of Impurities on the Mobility

Figure 2.8: Temperature dependence of the electrical hole mobility for a pentacene single crystal using the actual crystal thickness (◦) and the effective thickness (•).

whereas we use a gap-type geometry. A description of the two types of structure, and a detailed study on the evaluation of the current-voltage characteristics will be presented in Chapter 4. The value of the number of traps Nt can be calculated from the value of the trap-filling voltage (VT F L ): VT F L =

2 eL2 Nt 3 ε0 ǫr

(2.14)

The SCLC measurements show that the pentacene single crystals grown after precleaning of the starting material are very pure, with Nt = 1.74 · 1011 traps/cm−3 . This is almost two orders of magnitude lower than the number of traps obtained for crystals grown with the conventional procedures [17].

2.3.2

Evaluation of the bulk mobility of the high quality crystals

If we assume a homogeneous current flow through the sample, the mobility is µ = 11.2 cm2 /Vs. However, as the mobility in the basal plane ab is much larger than the mobility along the c* -axis (perpendicular to ab plane), the current will be confined to the contact’s side of the crystal. The measurements of the ohmic

2.3. Electronic response to structural and chemical properties

39

regime of the current-voltage characteristic showed different values for the resistivity for different directions (ρa = 1.3 · 106 Ωm, ρb = 4.7 · 105 Ωm, ρc∗ = 2.1 · 108 Ωm). We have used Montgomery’s method [18, 19] for analyzing anisotropic materials, transforming an anisotropic sample with resistivities ρa , ρb , and ρc∗ and dimensions x, y, and z, to an isotropic solid with dimensions x′ , y ′ and z ′ . For the isotropic solid, the normalized effective thickness is determined to be zef f ′ /(x′ y ′ )1/2 = 0.7 for normalized sample thickness z ′ /(x′ y ′ )1/2 = 2.19 and the ratio for the in-plane directions y ′ /x′ = 0.535. Using this procedure, the relation between the effective thickness (zef f ) and the real thickness of the crystal (z) is calculated. Eq. 2.15 expresses the conversion for the two dimensions: zef f = 0.32 · z.

(2.15)

With the effective thickness introduced in the current density JSCLC,tf in Eq. 2.13, for the trap-free regime, the calculated mobility increases by more than a factor of three (see Fig. 2.8). Therefore, a more accurate value for the mobility is µ = 35 cm2 /Vs at 290 K and µ = 58 cm2 /Vs at 225 K. We note that the Montgomery method is only valid in the linear part of the I − V regime. For I ∝ V 2 the effective thickness (zef f ) should be considered as the upper limit. Thus, this analysis provides a lower limit of the intrinsic mobility.

2.3.3

Band transport in pentacene single crystals

Fig. 2.8 shows that below room temperature the mobility increases with decreasing temperature following the relation: µ = C · T −2.38

(2.16)

This behavior is noticed in several samples and it is consistent with a band model for charge transport in pentacene [14], with the interaction of the delocalized carriers with the phonons, the main scattering process. Above room temperature a different transport mechanism dominates the mobility. This low temperature dependence was earlier suggested by time-of-flight experiments on single crystals of naphthalene and anthracene [20]. In this Chapter we present the low-temperature electrical measurements performed on organic single crystals pentacene, that also point to a ”band-type” conduction. We were able do reproduce the early TOF experiments in electrical measurements using crystals with a high degree of purity (Nt = 1.74 · 1011 traps/cm−3 ). The ”metallic-like” behavior was also demonstrated in field-effect transistors built at the surface of high quality rubrene single crystals [4, 5]. Previous electrical measurements on organic crystals were dominated by defects and impurities that severely localized the charges. This was

40

Chapter 2. The Effect of Impurities on the Mobility

reflected in a lower mobility than the intrinsic one, and a thermally activated charge transport, which is also not of intrinsic origin. In the band-like picture of the transport in molecular crystals (Fig. 2.9(a)), the mobility decreases with increasing temperature due to electron-phonon coupling. The scattering events involve interactions with defects and lattice vibrations. The vibration of the lattice increases with temperature, causing more scattering, and thus reducing the mobility. Moreover, following Holstein’s theory of small-polarons, at low temperatures the electronic mobility decreases with increasing temperature due to bandwidth narrowing. Hannewald et al. expanded the local approach developed by Holstein with nonlocal (Peierls-type) couplings using first-principles densityfunctional calculations [22, 23]. Their theory is in good agreement with the low temperature behavior of charge-carrier mobility shown in Figure 2.8. At high temperature, due to phonon-assisted hopping (Fig. 2.9(b)), the mobility increases with temperature (as seen in Fig. 2.8 above room temperature) [21].

2.3.4

Origin of traps

In the following, we will focus on the origin of the trapping factor (Θ) in Eq. 2.7. The traps in the crystal are mainly caused by structural imperfections and chemical impurities (see the scheme in Fig. 2.10). In our crystals Θ varies from Θ = 0.3 at 225 K to Θ = 0.82 at 340 K. With increasing electric field the density of injected carriers will increase, and above the trap-filled limit voltage (VT F L ) the mobility is not affected by impurity states and defects [24], as shown in Fig. 2.7. This part of the curve was used to calculate the mobility of pentacene. Extended defects, such as edge dislocations or screw dislocations modify the available energy levels in their vicinity, often leading to the presence of accessible vacant orbitals in the band gap. We minimized the number of traps by a careful crystal growth and subsequent handling. Heating the as grown crystals in an inert argon atmosphere will reduce the dislocation density. These defects are introduced during the crystallization process and are thermodynamically unstable. Thus their number decreases by annealing. Dislocations will also enhance the chemical reactivity in their vicinity. Under the influence of light and temperature, reactions that oxidize pentacene to pentacenequinone will occur preferentially at dislocations [7]. So, even if the quinone is not present after crystal growth, it can be formed at defects after exposure to air and light. We found that 6,13-pentacenequinone is the dominant chemical impurity. We did not observe C22 H15 and C22 H13 O impurities that were calculated to form gap states in our pentacene, C22 H14 [25]. Moreover, we argue that these molecules

2.3. Electronic response to structural and chemical properties

41

Figure 2.9: Charge transport mechanisms in organic conductors. (a) Band-type conduction. In a perfect crystal, a free charge carrier is delocalized. As the temperature is increased, the lattice vibrations scatter the charges. This limits the charge carrier mobility. The mobility µ for band transport increases with decreasing temperature. (b) Hopping conduction. If the carrier is localized (e.g. due to defects), the lattice vibrations promote a carrier to ”hop” over barriers of height Eb between the localized sites. For hopping transport the mobility µ increases with increasing temperature. The figure is adapted from ref. [27].

42 Chapter 2. The Effect of Impurities on the Mobility

Figure 2.10: Origin of traps for the injected charges in pentacene single crystals

2.4. Conclusions

43

are irrelevant as these radicals are highly reactive. We were able to prevent the formation of the dihydropentacene C22 H16 by using argon as transport gas during the crystal growth. We have shown that the reduction of 6,13-pentacenequinone (C22 H12 O2 ) impurities in pentacene by a factor five reduces the number of traps by almost two orders of magnitude. These impurities have different energy levels from pentacene, but they are energetically inert as a hole trap because their HOMO level is positioned below that of the host molecule ( [25] for C22 H16 and [26] for C22 H12 O2 ). This is distinctly different from experiments performed on smaller acenes, where the impurities yield states in the gap [7]. For this reason, the number of traps in our measurements is different from the number of chemical impurities. Although the impurity molecules do not act as trapping centers, they will induce a local deformation by distorting the pentacene lattice locally and create a scattering center. The quinone molecule is non planar and larger than pentacene. The middle ring has a flattened-chair shape with the bond length of 1.216 ˚ A and planar inclined at an angle of 3.1◦ to the molecular plane [27]. Thus, it will induce a local deformation leading to an increase in potential energy because of changes in molecular density. The quinone will strongly influence the number of such scattering sites, and thus the charge transport through organic single crystal.

2.4

Conclusions

In conclusion, we have reduced the impurity concentration of pentacenequinone in pentacene by a pretreatment consisting of vacuum sublimation of the impurity under a temperature gradient. The crystals exhibit a trap-free space charge limited current behavior. The mobility increases with decreasing temperature with a power law µ ∝ T −n from µ = 35 cm2 /Vs at room temperature to µ = 58 cm2 /Vs at 225 K, indicating band transport. These results incorporate corrections for the effective thickness of the crystal for the anisotropic resistivity, where the effective thickness is at least three times smaller than the crystal thickness. Our results emphasize the importance of the control of defects and impurity states in molecular organic crystals in order to obtain a high electronic mobility, and allow studies of the band transport regime [28].

References

[1] D. J. Gundlach, T. N. Jackson, D. G. Schlom, and S. F. Nelson, Appl. Phys. Lett. 74, 3302 (1999) [2] D. Knipp, R. A. Street, B. Krusor, R. Apte, and J. Ho, J. Non-Cryst. Solids 299-302, 1042 (2002) [3] V. Podzorov, V. M. Pudalov, and M. E. Gershenson, Appl. Phys. Lett. 82, 1739 (2003) [4] V. C. Sundar, J. Zaumseil, V. Podzorov, E. Menard, R. L. Willett, T. Someya, M. E. Gershenson, and J. A. Rogers, Science 303, 1644 (2004) [5] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers, and M. E. Gershenson, Phys. Rev. Lett. 93, 086602 (2004) [6] E. Menard, V. Podzorov, S.-H. Hur, A. Gaur, M. E. Gershenson, and J. A. Rogers, Adv. Mat. 16, 2097 (2004) [7] M. Pope, and C. E. Swenberg, Electronic Processes in Organic Crystals and Polymers 2nd ed. (Oxford Univ. Press., New York, 1999) [8] A. R. McGhie, A. F. Garito, and A. J. Heeger, J. Cryst. Growth 22, 295 (1974) [9] R. A. Laudise, C. Kloc, P. Simpkins, and T. Siegrist, J. Cryst. Growth 187, 449 (1998)

45

46

References

[10] L. B. Roberson, J. Kowalik, L. M. Tolbert, C. Kloc, R. Zeis, X. L. Chi, R. Fleming, and C. Wilkins, J. Am. Chem. Soc. 127, 3069 (2005) [11] The measurements are done in collaboration with P. van ’t Hof, and J. C. Hummelen (RuG) [12] J. R. Dyer, Applications of Absorption Spectroscopy of Organic Compounds Prentice-Hall, Inc. (1965) [13]

http://webbook.nist.gov/chemistry/

[14] K. C. Kao, and W. Hwang, Electrical Transport in Solids, vol. 14, Pergamon Press, New York, (1981) [15] S. Neˇspurek, and J. Sworakowski, Phys. Stat. Sol. (a) 41, 619 (1977). [16] S. M. Sze, Physics of Semiconductor Devices Wiley, New York, (1981) [17] C. C. Mattheus, A. B. Dros, J. Baas, G. T. Oostergetel, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Synth. Met. 138, 475 (2003) [18] H. C. Montgomery, J. Appl. Phys. 42, 2971 (1971) [19] J. D. Wasscher, Philips Res. Repts. 16, 187 (1961) [20] W. Warta, and N. Karl, Phys. Rev. B 32, 1172 (1985) [21] T. Holstein, Ann. Phys. 8, 343 (1959) [22] K. Hannewald, and P. A. Bobbert, Phys. Rev. B 69, 075211 (2004) [23] K. Hannewald, and P. A. Bobbert, Phys. Rev. B 69, 075212 (2004) [24] G. Horowitz, M. E. Hajlaoui, and R. Hajlaoui, J. Appl. Phys. 87, 4456 (2000) [25] J. E. Northrup, and M. L. Chabinyc, Phys. Rev. B 68, 041202(R) (2003) [26] G. de Wijs, private communication [27] A. V. Dzyabchenko, V. E. Zavodnik, and V. K. Belsky, Acta Cryst. B35, 2250 (1979) [28] V. Y. Butko, X. Chi, D. V. Lang, and A. P. Ramirez, Appl. Phys. Lett. 83, 4773 (2003)

3

Interface Controlled High-mobility Organic Transistors∗

We report the fabrication of Organic Field-Effect Transistors (OFETs) in which it is possible to reach the intrinsic bulk mobility. This was accomplished by minimizing the trap density at the semiconductor-insulator interface. We obtain low defect density interfaces by converting impurity scattering centers into the gate dielectric. High mobilities (µ= 14−40 cm2 /Vs), large on/off rations (Ion /Iof f = 106 ) and good reproducibility of the measurements characterize the transistors fabricated with present method.

∗ This Chapter is adapted from O. D. Jurchescu, M. Popinciuc, B. J. van Wees, and T. T. M. Palstra, submitted

47

48

Chapter 3. Interface Controlled High-mobility Organic Transistors

3.1

Introduction

Electronic devices based on organic semiconductors have gained prominent industrial and academic interest in recent years [1–18]. Organic materials represent new semiconductors for flexible, low-cost, light-weight electronic devices. The device functionality remains limited by the relatively low electronic mobility, which for thin film devices is partially caused by a large grain boundary resistance [19]. Only single crystal based devices allow the determination of intrinsic properties. Rapid technological progress has been made in the fabrication of field-effect transistors (FET) at the organic - semiconductor interface [20–25]. However, the electronic mobility of such FET devices is typically much smaller than can be obtained in bulk materials 1 [26–28]. We demonstrate that for pentacene-derived devices this can be related to surface properties of the crystal. We use a new method to inactivate the impurity states at the interface in the single crystal devices by incorporating them into the dielectric gate barrier. Thus, high mobility devices in which the bulk mobility is reached can be reproducibly constructed. Recent progress in organic single crystal FET devices has shown that the semiconductor interface layer, which forms the conduction channel between the source and drain electrode, contains a broad distribution of interface states [2]. In order to minimize the trap density, organic dielectrics such as parylene have been used as gate dielectric [1, 20, 21, 26]. Previous attempts using more robust dielectrics such as SiO2 and Al2 O3 involve to many chemical reactions and lead to structural disorder. Nevertheless, even organic dielectrics often yield limited mobilities, which is partially attributed to damage below the metal electrodes [6]. However, little detailed knowledge is available on the nature of the defect states and traps.

3.2

Current status of the field

The mobility of high purity organic field-effect transistors, is usually much lower than the value determined from SCLC measurements. This can be attributed to the fact that in FET devices the surface conduction is probed, whether in SCLC the bulk mobility is measured. The mobility in the bulk can be increased by reducing the number of traps through controlled processing. We have chosen to investigate pentacene. This material has become a prominent choice, both for industrial applications and for fundamental studies. Moreover, it is used in rollable displays in prototypes of commercial devices [3], and exhibits the highest 1 By

bulk mobility we denote the mobility obtained from SCLC measurements

3.3. Novelty

49

reported electronic mobility: µ= 35 cm2 /Vs at room temperature in bulk single crystals [28]. However, for FET devices built on pentacene single crystals, a mobility of 0.3 cm2 /Vs was measured when using a parylene gate dielectric [26]. This value was improved to 2.2 cm2 /Vs for purified pentacene [8]. A mobility of 1.4 cm2 /Vs was measured for pentacene single crystal FETs using a SiO2 gate dielectric treated with self-assembled monolayers [27]. The different mobilities of the bulk and FET devices built on pentacene single crystals indicate the relevance of a structurally and chemically clean interface. The morphology, spatial ordering and roughness at the interface are as important as the purity of the active area of the devices [14–16].

3.3

Novelty

Our OFETs, fabricated using the method described in this work, have a typical charge carrier mobility of µ = 15 - 40 cm2 /Vs at room temperature, similar to the value measured in bulk, ultra-pure single crystals [28]. This mobility represents a record value for the FETs based on organic semiconductors. The on/off ratios are as high as Ion /Iof f = 106 . The large values of the mobility and on/off ratio result from a reduction in the number of scattering sites at the interface between the semiconductor and insulator. The method developed in this work leads to remarkably good reproducibility of the devices.

3.4

Single crystal growth

We have grown pentacene single crystals with a high degree of purity using physical vapor transport [7]. The starting material was obtained from Sigma Aldrich and was purified using vacuum sublimation under a temperature gradient. We have described in Section 2.2.3 of Chapter 2 the details concerning this step. This growth method yields platelets of pentacene single crystals with in-plane dimensions of 1 − 4 mm and thicknesses of 10 − 50 µm.

50

3.5 3.5.1

Chapter 3. Interface Controlled High-mobility Organic Transistors

Surface morphology of pentacene single crystals Surface mapping

We have previously established that 6,13-pentacenequinone (PQ) constitutes the largest impurity fraction in pentacene single crystals [28]. The quinone is easily formed upon oxidation of the relatively reactive conjugated pentacene molecule. Moreover, pentacene is routinely synthesized from PQ, thus small fractions of the quinone can remain in the final reaction product. The purity of the material can be improved by vacuum sublimation treatment of the starting material [28] or repeated vapor transport crystallization [21]. In typical crystals, PQ may be present at a concentration of 0.11% and even after repeated distillation we observed a fraction of 0.028% [28]. Thus far, it has been commonly assumed that such an impurity is evenly distributed throughout the lattice. We will show that this is not the case, rather that PQ is located preferentially at the surface. This is reasonable and can be expected from the dynamics of the crystal growth process. At the interface, PQ molecules form scattering centers, reducing the mobility or even preventing transistor action altogether. However, PQ can be incorporated into the gate dielectric by growing additional quinone layers on the surface. Thus, the PQ can form a pinhole-free gate barrier and considerably minimize the trap states. Pentacene is a layered material characterized by a herringbone pattern within the layers, while adjacent layers are bonded by van der Waals forces. Pentacene has several polymorphs, each with a particular d(001)-spacing along the c*-axis (Fig. 3.1). This length varies from 14.1 to 15.5 ˚ A, but for pentacene single crystals it is unique: dP (001)=14.1 ˚ A [29]. This polymorph is already adopted in the first monolayers. The d(001) spacing is significantly different in 6,13 pentacenequinone: dP Q (001)=17.79 ˚ A [30] (Fig. 3.1). This allows us to distinguish between the two types of molecules present at the surface by Atomic Force Microscopy (AFM).

3.5.2

AFM measurements

In order to characterize the surface morphology of pentacene single crystals, we investigated several crystals using the AFM technique and measuring on different areas of the crystals. The AFM measurements were performed in the tapping mode using a MultiMode Scanning Probe Microscope, the NanoScope IV, from Digital Instruments. On the surface of pentacene single crystals we observed

3.5. Surface morphology of pentacene single crystals

51

Figure 3.1: Pentacene crystal packing. Pentacenequinone impurity is located randomly at the surface. The molecular structure, as well as the interlayer separation for pentacene (14.1 ˚ A) and pentacenequinone 17.79 ˚ A are indicated.

Figure 3.2: AFM picture at the surface of pentacene single crystal showing characteristic steps. The step height reveals the inter-layer separation distance d(00l).

52

Chapter 3. Interface Controlled High-mobility Organic Transistors

˚ and 18 ˚ Figure 3.3: Upper panel: histogram showing the occurrence of the 14 A A steps at the surface of the pentacene single crystals. The Gaussians peaking at 14 ˚ A and 18 ˚ A are a guide to the eye. Lower panel: step distributions characteristics for the bulk (obtained after cleaving the crystals), where only the 14.1 ˚ A step is observed.

3.6. Comparison between different FETs

53

terraces (see Fig. 3.2), which exhibit steps heights with predominantly two values: 14 ˚ A and 18 ˚ A. The former value corresponds to the d-spacing of single crystal pentacene, and the latter to that of PQ. Although we have no detailed quantitative analysis over the entire crystal, we observe that the 14 ˚ A step occurs approximately five times more often than the 18 ˚ A step. The results differ slightly between crystals, depending on the surface area that is measured. The occurrence of the 14 ˚ A and 18 ˚ A steps at the surface of the pentacene single crystals, corresponding to pentacene and pentacenequinone, respectively, is shown in Figure 3.3-upper panel. The crystals were then cleaved and the new surface was mapped (bulk material). There, we observed only terraces with 14.1(5) ˚ A steps, characteristic of pentacene (Fig. 3.3-lower panel). We conclude that the impurity PQ molecules agglomerate in patches that are distributed over the surface of the pentacene crystals. Similar effects were reported for acenes with a lower number of benzene rings. Quantitative analysis via Gas Chromatography (GC) on tetracene single crystals allowed the detection of the tetracenequinone surface concentration larger by one order of magnitude with respect to the bulk [31]. Both in pentacene and tetracene single crystals, because the pentacenequinone and tetracenequinone, respectively, are preferentially located at the surface, many more scattering sites are formed in the FET geometry than we determined for the bulk [28]. This is critical for the performance of the FETs because the traps are located at the interface with the gate insulator, where the active channel forms.

3.6 3.6.1

Comparison between different FETs FETs fabricated with conventional dielectrics

In the case of OFETs built on the surface of pentacene single crystals using conventional methods (Fig. 3.4(a)) [8, 26, 27], the charge transport is dominated by disorder, thus the mobility is lower than the bulk value. This reduction can originate from the presence of pentacenequinone molecules at the surface. The interactions between the π−systems of the ’host’ and ’impurity’ molecules introduce additional scattering sites to the inevitable traps formed during the deposition of the gate insulator. The electric field applied at the gate electrode has to fill first these trap states. Only a higher gate bias can populate the density of states (DOS) of pentacene and modulate the drain current for a proper FET operation. Moreover, random dipoles and quadrupoles [32] are locally introduced at the interface due to the presence of the impurity molecules. This leads to energetic disorder and an increase in carrier localization by electronic polarization. It

54

Chapter 3. Interface Controlled High-mobility Organic Transistors

has been shown by Lang et al. [2] that even high quality pentacene single crystals have a relatively broad distribution of states (∼ 1eV ) above the edge of the valence band. The different intermolecular interactions between pentacene, pentacenequinone and the gate insulator material modulate the energy levels of the states. This results in a broadening of the DOS and the formation of an increased number of tail states.

3.6.2

FETs fabricated with pentacenequinone dielectric

We replace the conventional gate dielectrics (SiO2 , parylene, polymers) with ordered 6,13 - pentacenequinone films (P-PQ OFET: Fig. 3.4(b)). This method presents the unique advantage of being able to transform the scattering sites present at the surface of pentacene single crystals into an extremely good layer at the semiconductor-gate interface. A slow deposition rate ensures a small surface roughness. Once the pentacenequinone layer at the surface is completed, an uniform, extremely good-quality heterointerface is built. This ensures the uniformity of the conduction path. In this device, the electric field applied at the gate is able to inject directly into the DOS of pentacene. This is reflected in a smaller value of the threshold voltage (VT = ± 2 V) of devices built with the geometry in Figure 3.4(b), compared with devices fabricated with conventional methods (VT = 5 V in ref. [26] and VT = -10 V in ref. [27]).

3.7

Device fabrication

We paid special attention to the fabrication of the field-effect transistors on the organic crystal surface, in order to avoid damaging the surface of the semiconductor. Silver epoxy source and drain contacts were painted on the crystal at different separation lengths, L. A film of 6,13-pentacenequinone was deposited on top. The deposition of the insulator was carried out in a clean environment, in high-vacuum (10−7 mbar), and the growth was carefully controlled. The source material was obtained from Sigma Aldrich. No further purification was done. The sublimation rate was low (5 ˚ A/min) to ensure the homogeneity of the film. A shutter was inserted between the quinone source material and the pentacene crystal. First the evaporation rate was stabilized, and only after that the device structure was exposed to the molecules. The thickness of the film was determined very precisely during the growth by a quartz crystal balance. The silver epoxy gate electrode was then painted on top of the device.

3.7. Device fabrication

Figure 3.4: Schematic cross-section of a FET fabricated on a pentacene single crystal. The bondline chemical formulae for 6,13-pentacenequinone and pentacene are drawn. (a) Conventional gate dielectric is deposited on the surface of the pentacene crystal. Here, pentacenequinone molecules present at the interface form scattering sites and decrease the mobility. (b) P-PQ OFET: highly ordered pentacenequinone films act as gate insulators and reduce the number of scattering sites at the interface. 55

56

3.8

Chapter 3. Interface Controlled High-mobility Organic Transistors

Properties of FETs fabricated with pentacenequinone gate dielectric

Field-Effect Transistor measurements were performed in the dark and a vacuum of 10−6 mbar [9], in a home-built probe station. The characterization was done using a Hewlett Packard 4155B Semiconductor Parameters Analyzer. We investigated several FETs built on pentacene single crystals obtained from different batches. The electrical properties of the devices were similar. We performed I − V measurements on pentacenequinone single crystals in order to estimate the electrical breakdown field of the gate dielectric. We obtained a value of ∼ 3 MV/cm for the PQ single crystal insulator. Pinhole-free insulation was obtained for film thicknesses greater than 200 nm. We performed capacitance measurements on single crystals of the PQ gate dielectric material and evaluate the dielectric constant εr (PQ)= 3.5. An Andeen-Hagerling 2500A Ultra Precision Capacitance Bridge was used for the dielectric measurements performed on pentacenequinone and a Keithley 237 Source Measure Unit was used to determine the insulator breakdown voltage. The procedure we utilize for pentacene may be applied for the fabrication of FETs of similar materials. Surface oxidation is widely encountered in conjugated organic materials. Pentacene is highly susceptible to oxidation at the most reactive positions, leading to 6,13-pentacenequinone. The larger acenes are even less stable. The attachment of substituents on the acene backbone (e.g. phenyl in the case of rubrene) makes these positions less reactive and protects the molecule from unwanted oxidation. In this case the interaction of the oxygen with the crystal is significantly different; the oxidation is reversible yielding endoperoxide instead of quinones [38]. The transistors built using this method, on the surface of ultra-pure pentacene single crystals, exhibit p-type conduction. We did not observe ambipolar transport for this configuration. The sharp increase in the drain current ID in the transfer characteristics presented in Figure 3.5 is a typical feature of a highperformance field-effect device. We calculate the field-effect mobility µ from the I −V characteristics presented in Figure 3.5 and Figure 3.6. We observe typical mobilities in the range µ = 15 - 40 cm2 /Vs for the transistors fabricated using the PQ gate dielectric. The variations in mobility are caused by different crystal quality, contact injection issues, and contact resistance. The on/off ratio is of the order of 106 . We measured 20 devices built on 8 different crystals and observed consistent behavior. The expression of

3.8. Properties of FETs fabricated with pentacenequinone gate dielectric

57

Figure 3.5: OFET fabricated on a high purity pentacene single crystal with 250 nm thick pentacenequinone film gate dielectric. The geometry of the device is L × W = 250 µm × 4 mm. The graph shows the variation of the drain current (ID ) with respect to the applied gate voltage (VG ) (relative to the source contact) for VD = 1V and VD = -1V. The properties of the transistor are included in the inset.

the mobility from transconductance (Fig. 3.5) is: ∂ID ∂VG

(3.1)

L 1 1  ∂ID  W Ci VD ∂VG

(3.2)

∂(logID ) ∂VG

(3.3)

gm = At low drain voltages VD : µ=

where L is the channel length, W is the gate width, and Ci is the capacitance per unit area of the gate insulator. The expression for the subthreshold slope is: S= The normalized subthreshold slope: Si = S · Ci

(3.4)

has a value Si = 33.7 V nF/dec · cm2 . This value is higher by a factor of 6 - 8 than the value obtained for the best rubrene transistors [20].

58

Chapter 3. Interface Controlled High-mobility Organic Transistors

Figure 3.6: Drain current (ID ) versus drain voltage (VD ) for different negative gate voltages VG for a device with L × W = 250 µm × 4 mm, and 250 nm thick pentacenequinone film gate dielectric.

Typical output characteristics of the P-PQ transistors are plotted in Figure 3.6. The curves are stable with time, even if the device is stored in ambient atmosphere and light. The current in the channel increases as the negative gate bias (VG ) is increased, and saturates at VD = VG − VT . For each curve of Figure 3.6, corresponding to a gate voltage, when the drain voltage reaches the value VD = VG − VT , a depletion region will start to form around the drain contact. As VD ≫ VG − VT , this region increases and the drain current becomes saturated, independent of the applied voltage. We attribute the nonlinearities in the low VD regime to the contact injection issues. This is consistent with previous studies that assigned the superquadratic-like behavior at low drain voltages to the presence of the Schottky barrier between Fermi level of the metal contact and HOMO of the organic semiconductor (LUMO in the case of electron conduction) [33], the (diffusion-limited) thermionic emission and the recombination at the metalorganic interface [34], or the presence of a depletion area in the vicinity of the ”top contacts”, with a high density of localized states, that induces a parasitic resistance in the FET circuit [6, 35, 36]. The quality of the manually deposited contacts is not very good, and this is reflected even in the I − V characteristics of the best devices of both VG and VD sweeps. Moreover, the contacts are not equivalent and the device shows an asymmetry in the VG sweep between VD = +1 V and VD = -1 V.

3.9. Conclusions

59

Small shifts in the threshold voltage were recorded for different devices, between positive and negative values. As for devices fabricated with identical methods, VT is both positive and negative, we suggest that this is not of intrinsic origin (the negative VT is not caused by bulk defects, and the positive VT is not a result of bulk crystal doping). More probably, during the deposition of the contacts and the gate insulator, traps/additional charges are accumulated at the surface of the crystal, randomly. The field-effect-mobility estimated from Figure 3.6, from the linear regime, as well as from the saturation regime (using Eq. 1.5, and Eq. 1.6, respectively), increases with drain and gate voltage. This feature has also been reported by Goldman et al. [27] and Podzorov et al. [37]. This can result from a field dependent mobility, while in all the assumptions we considered that this parameter does not vary with the electric field. Also, in the part of the graph that corresponds to the linear regime of the FET operation, the channel resistance (Rc ) varies linearly with the applied voltage for a standard FET. The parasitic resistance introduced at the contacts is field-independent. The total resistance (Rt ) of the channel is thus: Rt = RS + Rc + RD , where RS , RD are the resistance of the Source, respective Drain contacts, and will not have a linear dependence of the field. This will introduce additional variations from the already available models.

3.9

Conclusions

In conclusion, we have been able to incorporate the interface scattering centers of pentacenequinone into the gate dielectric of Organic Field Effect Transistors built on pentacene single crystals. Thus, we obtain a semiconductor/dielectric interface of extremely high quality. The FET mobility µ = 15 − 40 cm2 /Vs can reproduce the value obtained in ultra-pure single crystal bulk devices. This demonstrates that we are able to measure the intrinsic properties of the single crystal by careful control of the interface properties, which imply minimizing the trap density in the active channel.

References

[1] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers, and M. E. Gershenson, Phys. Rev. Lett. 93, 086602 (2004) [2] D. V. Lang, X. Chi, T. Siegrist, A. M. Sergent, and A. P.Ramirez, Phys. Rev. Lett. 93, 086802 (2004) [3] G. H. Gelinck, H. E. A. Huitema, E. van Veenendaal, E. Cantatore, L. Schrijnemakers, J. B. P. H. van der Putten, T. C. T. Geuns, M. Beenhakkers, J. B. Giesbers, B. H. Huisman, E. J. Meijer, E. M. Benito, F. J. Touwslager, A. W. Marsman, B. J. E. van Rens, and D. M. de Leeuw, Nat. Mat. 3, 106 (2004) [4] N. Stingelin-Stutzmann, E. Smits, H. Wondergem, C. Tanase, P. Blom, P. Smith, and D de Leeuw, Nat. Mat. 4, 601 (2005) [5] A. Troisi, and G. Orlandi, Phys. Rev. Lett. 96,086601 (2006) [6] R. W. I. de Boer, and A. F. Morpurgo, Phys. Rev. B 72, 073207 (2005) [7] A. Laudise, C. Kloc, P. Simpkins, and T. Siegrist, J. Cryst. Growth 187, 449 (1998) [8] L. B. Roberson, J. Kowalik, L. M. Tolbert, C. Kloc, R. Zeis, X. L. Chi, R. Fleming, and C. Wilkins, J. Am. Chem. Soc. 127, 3069 (2005) [9] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 87, 052102 (2005) 61

62

References

[10] F. -J. Meyer zu Heringdorf, M. C. Reuter, R. M. and Tromp, Nature 412, 517 (2001) [11] R. Ruiz, A. Papadimitratos, A. C. Mayer, and G. G. Malliaras, Adv. Mat. 17, 1795 (2005) [12] D. A. da Silva Filho, E. G. Kim, and J. -L. Br´edas, Adv. Mat. 17, 1072 (2005) [13] J. E. Northrup, and M. L. Chabinyc, Phys. Rev. B 68, 041202(R) (2003) [14] R. W. de Boer, N. N. Iosad, A. F. Stassen, T. M. Klapwijk, and A. F. Morpurgo, Appl. Phys. Lett. 86, 032103 (2005) [15] A. F. Stassen, R. W. de Boer, N. N. Iosad, and A. F. Morpurgo, Appl. Phys. Lett. 85, 3899 (2004) [16] J. Veres, S. D. Ogier, S. W. Leeming, D. C. Cupertino, and S. M. Khaffaf, Adv. Func. Mat. 13, 199 (2003) [17] B. Crone, A. Dodabalapur, Y.-Y. Lin, R. W. Filas, Z. Bao, A. LaDuca, R. Sarpeshkar, H. E. Katz, and W. Li , Nature 403, 521 (2000) [18] L.-L. Chua, J. Zaumseil, J.-F. Chang, E. C.-W. Ou, P. K.-H. Ho, H. Sirringhaus, and R. H. Friend, Nature 434, 194 (2005) [19] S. F. Nelson, Y. Y. Lin, D. J. Gundlach, and T. N. Jackson, Appl. Phys. Lett. 72, 1854 (1998) [20] V. C. Sundar, J. Zaumseil, V. Podzorov, E. Menard, R. L. Willett, T. Someya, M. E. Gershenson, and J. A. Rogers Science 303, 1644 (2004) [21] R. W. de Boer, M. E. Gershenson, A. F. Morpurgo, and V. Podzorov, Phys. Stat. Sol. A 201, 1302 (2004) [22] M. Halik, H. Klauk, U. Zschieschang, G. Schmid, C. Dehm, M. Schutz, S. Maisch, F. Effenberger, M. Brunnbauer, and F. Stellacci, Nature 431, 963 (2004) [23] S. Kobayashi, T. Nishikawa, T. Takenobu, S. Mori, T. Shimoda, T. Mitani, H. Shimotani, N. Yoshimoto, S. Ogawa, and Y. Iwasa, Nat. Mat. 3, 317 (2004) [24] G. Horowitz, Adv. Mat. 10, 365 (1998)

References

63

[25] C. D. Dimitrakopoulos, and P. R. L. Malenfant, Adv. Mat. 14, 99 (2002) [26] V. Y. Butko, X. Chi, D. V. Lang, and A. P. Ramirez, Appl. Phys. Lett. 83, 4773 (2003). [27] C. Goldmann, S. Haas, C. Krellner, K. P. Pernstich, D. J. Gundlach, and B. Batlogg, J. App. Phys. 96, 2080 (2004) [28] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 84, 3061 (2004) [29] C. C. Mattheus, A. B. Dros, J. Baas, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Acta Cryst. C 57, 939 (2001) [30] A. V. Dzyabchenko, V. E. Zavodnik, and V. K. Belsky, Acta Cryst. B 35, 2250 (1979) [31] J. Pflaum, J. Niemax, and A. K. Tripathi, Chem. Phys. 325, 152, (2006) [32] C. A. Hunter, and J. M. Sanders, J. Am. Chem. Soc. 112, 5525 (1990) [33] V. I. Arkhipov, E. V. Emelianova, Y. H. Tak, and H. B¨assler, J. Appl. Phys. 84, 848 (1998) [34] J. Campbell Scott, and G. G. Malliaras, Chem. Phys. Lett. 299, 115 (1998) [35] L. B¨ urgi, T. J. Richards, R. H. Friend, and H. Sirringhaus, J. Appl. Phys. 94, 6129 (2003) [36] B. H. Hamadani, and D. Natelson, J. Appl. Phys. 87, 052102 (2005) [37] V. Podzorov, S. E. Sysoev, E. Loginova, V. M. Pudalov, and M. E. Gershenson, Appl. Phys. Lett. 83, 3504 (2003) [38] V. Ramamurthy, and K. Venkatesan, Chem. Rev. 87, 433 (1987)

4

Cross-over from 1D to 2D - Space Charge Limited Conduction in Pentacene Single Crystals∗

We report the cross-over from 1D to 2D - space charge limited conduction in pentacene single crystals with planar contacts. The space charge is confined to the in plane longitudinal direction for L/h ≤ 15, with L the contact separation and h the sample thickness. For L/h > 250 the space charge is dominated by the transverse component of the electric field.

∗ This Chapter is adapted from O. D. Jurchescu, and T. T. M. Palstra, Appl. Phys. Lett. 88, 122101 (2006)

65

66

Chapter 4. Cross-over from 1D to 2D - SCLC in Pentacene Single Crystals

4.1

Introduction

Small molecule organic crystals can exhibit large electronic mobilities at room temperature, if local impurities and extended defects, such as grain boundaries, can be minimized. Thus, single crystals can provide information about the intrinsic electronic transport properties [1–6]. Nevertheless, the evaluation of the transport parameters can be difficult because of the charge injection at the interface, or the non-uniform nature of charge carrier density, current density, and electric field under non-ohmic conditions. The evaluation of the electronic mobility depends strongly on the correctness of the relationship between electronic parameters, including mobility, charge density etc., but also on the associated geometry that is used. In this chapter we analyze the current flow across pentacene single crystals with planar contacts. Our study focuses on the electric field E pattern for parallel planar contacts with different gap separations L, in the space-charge-limitedcurrent (SCLC) regime. We discuss the effect of the geometric parameter L/h, with h the crystal thickness, and the anisotropy of the mobility on the electric field distribution inside the semiconductor. Effects like surface scattering, charge diffusion, surface traps [7] and surface polarization are neglected. We observe a gradual transition from a 1- to 2-dimensional (D) space charge limited conduction as L/h increases.

4.2 4.2.1

Space-charge-limited-current theories for semiconductors Mott-Gurney model

The charge transport in organic conductors is often limited by the emergence of space charges. There are two limiting geometries to evaluate the conduction. The Mott-Gurney theory [8] describes the current-voltage characteristics for sandwichtype contact geometries (Fig. 4.1(a)) for materials with one type of trap (the majority trap) distributed uniformly in space (density Nt ), having one discrete energy level (Et ). This is a 1D theory in which the electric field, as well as the space charge, are confined to the channel. Mott-Gurney model gives an expression of the current density versus applied voltage: J=

V2 9 Θε0 εr µ 3 . 8 L

(4.1)

4.2. SCLC theories for semiconductors

67

-

electrodes crystal L

(a) L (b)

h L

(c)

Figure 4.1: Electrode configuration: (a) sandwich-type geometry-side view, (b) gap-type geometry-side view, (c) gap-type geometry-top view (used in this study).

Here, J is the current density for the applied voltage V , Θ the trapping factor, L the electrode separation, εr the relative dielectric constant of the material, and µ the carrier mobility. A detailed description of this model was presented in Chapter 2, Section 2.3.1.

4.2.2

Geurst model

Geurst analyzed theoretically space-charge-limited-currents in thin semiconductors layers for a gap-structure geometry [9] (Fig. 4.1(b)). Here, the thickness of the film is negligible with respect to the separation between the contacts. For this 2D model the longitudinal component of the electric field is responsible for the charge transport and the transversal component, perpendicular to the conduction channel, is determined by the magnitude of the space charge. This analysis leads to an expression in which the value of the current I depends quadratically on the applied voltage V , is proportional to the relative dielectric constant of the material, εr , and inversely proportional to the square of L, the distance between the contacts: V2 2 (4.2) I = ε0 εr µ 2 π L Zuleeg et al. [10] have introduced the width of the electrodes, W , into the expression of Eq. 4.2. This model improves the approximation of Geurst of infinite long contact length. Thus, the space-charge-limited current at high voltages varies with L−3 in the

68

Chapter 4. Cross-over from 1D to 2D - SCLC in Pentacene Single Crystals

Mott-Gurney 1D theory (Eq. 4.1) and with L−2 in Geurst 2D model (Eq. 4.2). The former was used to estimate the hole mobility in tetracene [11] and rubrene [12] single crystals with parallel plate electrodes. It was also applied for analyzing pentacene single crystals [5,13] and TCNQ-coronene cocrystals [14] in a gap−type geometry, with a conduction channel length comparable to the thickness of the crystals. The model of Geurst was used to describe the charge transport of thin layers of crystalline Se [15], in which the proportionality of the current with L−2 was observed.

4.3

Space charge limited currents in pentacene single crystals with planar contacts

The question we address is for which parameters the 1D-model can be used for planar contacts, and when the transition to the 2D-model takes place. We show experimentally that for small L/h ratios, the Mott-Gurney type behavior is observed. Due to the resistivity anisotropy in different crystallographic directions and short conduction path, the current penetrates only a fraction of the crystal thickness (hef f ), and the electric field is homogenous. We find that for pentacene single crystals the 1D SCLC model is accurate for L/h < 15, whereas the 2D model applies for L/h > 250. The transition takes place gradually and follows the changes in the value of L/h.

4.3.1

Experimental procedure

Platelets of high purity pentacene single crystals were grown from double sublimation cleaned [5] powder from Aldrich. Physical vapor transport in a horizontal glass tube under a stream of argon was used [16], as described in Chapter 2. The geometry of the crystals is maximally 4 × 4 mm in plane and thickness typically of 15 − 20 µm. Silver epoxy contacts were painted parallel on the crystal with a distance varying between 60 µm and 3 mm, limited by the dimensions of the crystals (Fig. 4.1(c)). We used a Keithley 237 Source Measure Unit to perform the electrical characterization. All I − V curves were taken at room temperature, in darkness and a vacuum of 10−6 mbar, to avoid environmental effects [17]. For each single crystal of thickness h, a voltage Vij was applied across different pair of electrodes i and j attached to the sample at a distance Lij and the value of the current Iij was measured. The source contact was kept the same to minimize contact effects [7], and the drain was changed.

4.3. SCLC in pentacene single crystals

69

Figure 4.2: Double logarithmic plot of the current at constant bias for different ratios of the planar electrode distance L with respect to the crystal thickness h. The lines are guides for the eyes and correspond to 1D model (slope -3) and 2D model (slope -2) respectively. The inset represents the derivative d(logI)/d(logL) versus L/h. The positions with values of 2 and 3, indicating the two considered models are marked. The data were obtained for one crystal with multiple contacts at room temperature.

4.3.2

Charge transport in the SCLC regime

The I − V curves are ohmic at small biases (IΩ ∝ V ), and quadratic at high volt
ages ISCLC ∝ V 2 . Figure 4.2 shows the variation of the electrical current I at constant bias voltage for a crystal of h = 20 µm thickness and different contact spacings, Lij (60 µm < Lij < 3 mm), in the SCLC regime. The width of the contacts is approx W = 200 µm. The results are presented as a function of the ratio L/h. The inset shows the logarithmic derivative, d(logI)/d(logL) versus L/h, with a value −3 corresponding to the 1D conduction, and −2 to the 2D model. As it can be observed from this graph, at small L/h ratios the bulk charges are dominant and at high ratios the surface charges determine the conduction. For L/h ≤ 15, there is a good correspondence between the present experiment that we performed on a crystal with planar contacts and the analytical solution described

70

Chapter 4. Cross-over from 1D to 2D - SCLC in Pentacene Single Crystals


by Mott-Gurney for devices with parallel plate electrodes I ∝ L−3 [8]. This is due to the fact that at small electrode spacing L, the electric field inside the semiconductor is relatively uniform (Elong ) and parallel to the conduction path. Moreover, when this distance is small, and considering the anisotropy of mobility in pentacene single crystals, the current is confined to the surface of the conductor [18]. We demonstrate that the 1D theory (Eq. 4.1) has a relatively limited range of validity in case of planar parallel electrodes. It is clear from Figure 4.2 that in this geometry the surface contributions become gradually dominant as L/h increases. The surface charge density gives rise to the normal component of the electric field (Etrans ) and the electric field obtains more pronounced 2D characteristics. Thus the 1D space charge approximation is not valid in this regime and the field distribution within the single crystal must be taken into account [19]. The electric field E becomes not uniform inside the crystal. With increasing L, the transverse electric field results in I ∝ L−α , with α ranging from 3 to 2 (see the inset in Fig. 4.2). We couldn’t reach the I ∝ L−2 regime, corresponding to Geurst’s analysis (Eq. 4.2) [9], as we were limited by the size of the crystals.

4.3.3

Evaluation of mobility for sandwich-type and gaptype structures

In the following section we will use the SCLC technique to calculate the value of mobility in pentacene single crystals. We compare the results obtained for the mobility in the a−b plane with the two types of contact geometry described in this study. Pentacene presents one morphology in the single crystal phase: triclinic, with spacegroup P ¯ 1 [20]. The crystals are formed by herringbone-type packing in layers. The molecular packing determines the electronic behavior. Because of different bonding and antibonding patterns, the effective mass exhibits different values for the three crystallographic directions [21]. This results in a strong anisotropy of the carrier mobility, as is often observed for molecular crystals [1,6]. The mobility is higher along the crystallographic direction in which the bonding or antibonding is dominant. This anisotropy influences the value of the L/h parameter at which the transition between 1D and 2D space charges occurs. We expect that for other molecular crystals the transition from 1D to 2D transport takes place for different geometrical parameters, due to their different mobility anisotropy. Figure 4.3 shows the device configurations and current density J versus applied bias V for one pentacene crystal within the a − b plane, using sandwich-type and gap-type architectures. The ratio L/h = 25 in this case sets the system in the first

4.3. SCLC in pentacene single crystals

71

Figure 4.3: Current density J vs. applied voltage V for pentacene single crystal at room temperature, in the dark and a vacuum of 10−6 mbar for gap structure contact geometry (), and sandwich structure configuration (◦) .

part of the intermediate regime, where the conduction is dominated by the 1D space-charges. This allows us, in first approximation, to use Eq. 4.1 to estimate the value of the mobility (µ) for this crystal. At low voltages, the current is ohmic. At higher voltages, the square law dependence of the current with the applied voltage is observed, corresponding to the SCLC regime (Fig. 4.3). Small deviations from the linear and quadratic regimes can be attributed to the non-linear contribution of Schottky barriers formed at the metal/pentacene interfaces. As the trap-free limit was not reached, µ can not be determined, but only the value of the product Θµ can be calculated. For the gap − type geometry two contacts are deposited on top of the crystal and the I − V curve is measured. When the measurements are completed, the parts of the crystal that were covered with metal are removed and new metal contacts are painted parallel to the thickness of the remaining crystal (h), for the sandwich-type geometry. The calculation of the mobility with sandwich configuration is straight forward. The device geometry is similar to the one of a capacitor and the electric field is constant and oriented along the direction in which the current is measured. From Eq. 4.1, in which the geometrical factors of the crystal are introduced (L × W × h = 500 × 200 × 20 µm), a value of (Θµ)sandwich = 12 cm2 /Vs is obtained. In the case of gap−type structure, we can

72

Chapter 4. Cross-over from 1D to 2D - SCLC in Pentacene Single Crystals

see from Figure 4.2 that, as L/h > 15, the electric field will exhibit deviations from the idealized 1D form, and the use of simple Mott-Gurney theory is not valid. Moreover, this model underestimates the value of the mobility: (Θµ)gap = 7.2 cm2 /Vs. This comes from the fact that the current flow is not homogenously distributed along the entire thickness of the crystal (h), but confined to the surface due to the anisotropy of pentacene.

4.3.4

Resistivity anisotropy effects - effective thickness

We use the Montgomery method [18] to determine the effective thickness (hef f ) penetrated by the electric field lines in the gap configuration. The input for this algorithm are the values of the components of the resistivity tensor along the three crystallographic directions (a, b in-plane and c∗ perpendicular to the plane: ρa = 1.3 · 106 Ωm, ρb = 4.7 · 105 Ωm, ρc∗ = 2.1 · 108 Ωm), and their correspondent electrodes separation on the crystal (x, y, z). We transform this anisotropic solid into the equivalent isotropic system of dimensions x′ , y ′ , z ′ , using van der Pauw’s expression [22]: 1 ′ zef (ρa · ρb ) 4 zef f f · = . (4.3) 1 1 ′ · y ′ ) 21 2 (x · y) 2 (x ρc ∗ ′ Here zef f and zef f are the effective thickness for the anisotropic and isotropic system, respectively. The second factor of the expression is the normalized effective thickness in the isotropic material, which is the abscissa of the graph proposed by Montgomery (Fig. 2 in Ref. [18]). The normalized sample thickness in the isotropic space can be calculated from: 1

z′ 1

(x′ · y ′ ) 2

=

ρc2∗ 1

(ρa · ρb ) 4

·

z 1

(x · y) 2

= 1.036.

(4.4)

The effective thickness depends on the ration x/z, in the anisotropic pentacene single crystal (Eq. 4.4, Eq. 4.5), which corresponds to the order parameter L/h that we have used. Different curves are obtained for different ratios of the in − plane directions. For this particular geometry: y′ ρb 1 y = ( ) 2 · = 0.24. ′ x ρa x

(4.5)

Thus, the normalized effective thickness of this system is 0.715. This leads to the value of zef f = 0.33 · z, corresponding to hef f = 0.33 · h. From Eq. 4.1, the value of (Θµ)gap,hef f = 20 cm2 /Vs can be calculated [18]. This value should

4.4. Conclusion

73

be considered an overestimate of the mobility. This arises mainly from the fact that in this regime small deviations appear from the 1D model that was assumed in the calculations. While the Mott-Gourney theory gives I ∝ L−3 , for our L/h ratio of 25, we should have used I ∝ L−2.7 as better approximation. In the regime L/h ≤ 15,where the current varies with the cube of the contact separation (Fig. 4.2), the values of mobility obtained with the two geometries agree well [5]. The effective thickness that we obtain represents a upper limit. Montgomery’s analysis assumes ohmic conduction in which the equipotential lines are independent of the magnitude of the electric field, while in the SLCL regime the current density capacity increases quadratically with the field, leading to a smaller effective penetration depth.

4.4

Conclusion

In conclusion, we report the cross-over from 1D to 2D type space-charge-limitedcurrent conduction in pentacene single crystals with increasing L/h for gap-type contact geometry. For L/h < 15 the gap type contact is well approximated by 1D-space charges, whereas for L/h > 15 2D-space charges should be taken into account. The 1D-space charges are predominant in pentacene because of the anisotropy in resistivity.

References

[1] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers, and M. E. Gershenson, Phys. Rev. Lett. 93, 086602 (2004) [2] D. V. Lang, X. Chi, T. Siegrist, A. M. Sergent, and A. P. Ramirez, Phys. Rev. Lett. 93, 086802 (2004) [3] C. Goldmann, S. Haas, C. Krellner, K. P. Pernstich, D. J. Gundlach, and B. Batlogg, J. Appl. Phys. 96, 2080 (2004) [4] D. A. da Silva Filho, E. G. Kim, and J. -L. Br´edas, Adv. Mater. 17, 1072 (2005) [5] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 84, 3061 (2004) [6] V. C. Sundar, J. Zaumseil, V. Podzorov, E. Menard, R. L. Willett, T. Someya, M. E. Gershenson, and J. A. Rogers, Science 303, 1644 (2004) [7] R. W. I. de Boer, and A. F. Morpurgo, Phys. Rev. B 72, 073207 (2005) [8] N. F. Mott and R. W. Gurney, Electronic Processes in Ionic Crystals, 2nd Edition, Clarendon Press, Oxford 1948 [9] J. A. Geurst, Phys. Stat. Sol. 15, 107 (1966) [10] R. Zuleeg, and P. Knoll, Appl. Phys. Lett. 11, 183 (1967)

75

76

References

[11] R. W. I. de Boer, M. Jochemsen, T. M. Klapwijk, A. F. Morpurgo, J. Niemax, A. K. Tripathi, and J. Pflaum, J. Appl. Phys. 95, 1196 (2004) [12] V. Podzorov, S. E. Sysoev, E. Loginova, V. M. Pudalov, and M. E. Gershenson, Appl. Phys. Lett. 83, 3504 (2003) [13] C. C. Mattheus, A. B. Dros, J. Baas, G. T. Oostergetel, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Synth. Met. 138, 475 (2003) [14] X. Chi, C. Besnard, V. K. Thorsmolle, V. Y. Butko, A. J. Taylor, T. Siegrist, and A. P. Ramirez, Chem. Mater. 16, 5751 (2005) [15] M. Polke, J. Stuke, and E. Vinaricky, Phys. Stat. Sol. 3, 1885 (1963) [16] A. Laudise, C. Kloc, P. Simpkins, and T. Siegrist, J. Cryst. Growth 187, 449 (1998) [17] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett. 87, 052102 (2005) [18] H. C. Montgomery, J. Appl. Phys. 42, 2971 (1971) [19] S. Hirota, J. Appl. Phys. 50, 3003 (1979) [20] C. C. Mattheus, A. B. Dros, J. Baas, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Acta Cryst. C 57, 939 (2003) [21] G. A. de Wijs, C. C. Mattheus, R. A. de Groot, and T. T. M. Palstra, Synth. Met. 139, 109 (2003) [22] L. J. van der Pauw, Philips Res. Repts. 16, 187 (1961)

5

Electronic Transport Properties of Pentacene Single Crystals upon Exposure to Air∗

We report the effect of air exposure on the electronic properties of pentacene single crystals. Air can diffuse reversibly in and out of the crystals and influences the physical properties. We discern two competing mechanisms that modulate the electronic transport. The presence of oxygen increases the hole conduction, as in dark each four O2 molecules diffused into the crystal introduce one charge carrier. This effect is enhanced by the presence of visible light. In contrast, water, present in ambient air, is incorporated in the crystal lattice and forms trapping sites for injected charges.

∗ This Chapter is adapted from O. D. Jurchescu, and T. T. M. Palstra, Appl. Phys. Lett. 87, 052102 (2005)

77

78

5.1

Chapter 5. Electronic Transport Properties upon Exposure to Air

Introduction

Pentacene is a good material for electronic devices like field effect transistors [1– 8]. However, the results obtained by different research groups are not always consistent. This can be attributed in part to the fact that the performance of the devices depends critically on the environmental conditions [9–11]. In many cases these conditions are not explicitly stated, so a comparison between the experiments is often not appropriate. We report the electronic properties of pentacene single crystals under controlled conditions of pressure, gas composition, illumination, and in time. We provide experimental evidence that the influence of air on the charge carrier conduction consists of two separate, distinguishable contributions. Air diffuses into the pentacene crystals and can be completely removed by re-evacuation. Thus, no irreversible chemical reactions, like oxidation of pentacene to pentacenequinone, can be detected during the diffusion process. We demonstrate that oxygen absorption is responsible for an increase in the conductivity, and its effect is enhanced by the presence of visible light. In contrast, water present in ambient air will form trapping sites for the injected charges. This will reduce the number of free charge carriers, and thus the value of the conductivity. These opposite effects may account for the contradicting reports on the influence of ambient conditions.

5.2

Experimental

Pentacene single crystals of high purity were grown using the vapor transport technique [12]. The growth was preceded by the purification of the starting material using a two-step vacuum sublimation process under a temperature gradient. Details about purification and crystal growth are included in Chapter 2 of this thesis. We performed thermo-gravimetric analysis (TGA) on pentacene single crystals at room temperature to determine the changes in mass. The increase in weight at normal pressure in the presence of different gas flows (air, O2 , N2 , and Ar) were measured using a SDT 2960 from Thermal Analysis Instruments. The crystals are platelets with dimensions of approximately 3 x 3 mm in plane and thickness of 15 − 20 µm. Silver epoxy stripe contacts were painted on the crystal, with a distance of 3 mm. The electrical measurements were performed in a home built micromanipulator probe station, connected to a turbo pump and to gas supplies. The system is equipped with a glass window for optical inspection. This window was covered for the experiments performed in dark. Otherwise, the samples are exposed to ambient fluorescent light, with no additional illumination.

5.3. Exposure to dry and ambient air

79

A Hewlett Packard 4155B Semiconductor Parameters Analyzer was used to perform electrical measurements. The typical hold and delay time for these scans were 10 ms in order to avoid charging of the material. All measurements were performed at room temperature.

5.3 5.3.1

Exposure to dry and ambient air I-V characteristics in the dark - in vacuum and after exposure to air

Figure 5.1: Current density J vs. applied voltage V for pentacene single crystal at room temperature in vacuum and after exposure to dry air (increase in J) and ambient air (decrease in J). The different curves represent different exposure times. In dry air : 250 min, 200 min, 150 min, 100 min, 50 min, 10 min, 0 min, in vacuum, in ambient air : 0 min, 10 min, 20 min, 30 min, 50 min, 100 min, 150 min, 200 min, 250 min.

80

Chapter 5. Electronic Transport Properties upon Exposure to Air

To understand the effect of environmental exposure on the transport in organic crystals, we separate the two competing effects (doping by O2 and trapping by H2 O), by measuring systematically in dry and ambient air. Figure 5.1 shows the evolution in time of the current density J versus applied voltage V for single crystal pentacene in dark, after venting the system with dry and ambient air, respectively. We associate the behavior at low bias with Ohmic behavior: J ∝ V x , with x = 1. The observed values of x range from 0.7 to 1.3 for different samples, which we relate to the quality of the contacts. In this regime, no marked changes in the conductivity are induced by the presence of oxygen or water vapor. Apparently, no free charge carriers are introduced by O2 absorption. However, the space-charge-limited current regime SCLC (J ∝ V 2 ) is influenced by exposure to air. Here, J increases upon exposure to dry air, and decreases upon exposure to ambient air. We interpret this behavior that the induced carriers are located in shallow trapped states and are only released beyond a critical field. The quantitative analysis of the transport changes, introduced by absorbing air, are performed in the SCLC regime (V = 100 V ). In dry air, the number of holes introduced by the O2 molecules increases in time, leading to an increase in the charge carrier density. The system will reach the SCLC regime at smaller bias voltage.

5.3.2

Gas diffusion in pentacene single crystals

We performed in parallel gravimetric and electrical measurements to be able to give a quantitative description of the effect of gas diffusion on the properties of organic crystals. In Figure 5.2 we plot the weight increase of pentacene single crystals upon exposure to 5N dry air (< 17ppm water vapor), at room temperature and atmospheric pressure, expressed as a molar fraction of air in pentacene. For each situation presented in this chapter, we performed 3 experiments and they are reproducible within an accuracy of 15%. The mass of the accumulated air in the crystal structure of pentacene was calculated assuming that the change in mass is caused only by absorption of gases. This assumption is supported by the observation that the process is completely reversible. The same trend is observed for the measurements performed in pure O2 , N2 , and Ar. The time constants are determined by the size and shape of both crystals and diffusing molecules. As the diffusion coefficients have similar values, we conclude that there is no selectivity with respect to any of the gases. The saturation values are within experimental accuracy the same for the different gasses: 2.0% ± 0.3%.

5.3. Exposure to dry and ambient air

81

Figure 5.2: Evolution in time of the molar fraction of air in pentacene upon exposure of single crystals to dry air at room temperature and the fit with a 1D diffusion model. This is a lower limit of the molar fraction, as it is assumed that the gas molecules diffuse homogenously over the crystal.

We used Fick’s second law (Eq. 5.1) to model the diffusion of gases into pentacene single crystals: ∂N ∂2N (5.1) = D( 2 ) ∂t ∂x Here N is the number of gas molecules at a particular time t and position x. We use the solution of this equation for the one-dimensional case, constant diffusion coefficient D, and an inexhaustible source. We use two boundary conditions to obtain a unique solution: • no gas molecules are accumulated in the crystal in the initial stage (N = 0, for x > 0, at t = 0), we will add an extra term later; • at x = 0, Nsource , induced by the gas flow, is maintained constant for t > 0. These boundary conditions are represented by the following expressions: ( N (x, 0) = 0 (5.2) N (0, t) = Nsource

82

Chapter 5. Electronic Transport Properties upon Exposure to Air

The solution of Eq. 5.1, with the boundary conditions of 5.2, is:  x  N (x, t) = Nsource erfc √ 2 Dt

(5.3)

where the erfc (z) function, complementary to the error function, erf(z), is defined as: Z z 2 2 erf c(z) = 1 − erf (z) = 1 − √ e−y dy (5.4) π 0 The evolution of the concentration of the gas molecules versus distance, in time, is shown in Figure 5.3. The distribution of gas molecules, N (x, t), depends on both distance, x, and time, t. The total quantity of gas molecules in the

Figure 5.3: Evolution of the number of gas molecules absorbed in the crystal lattice versus distance and in time. pentacene crystal is obtained by integrating this solution over the length of the diffusion profile. The result is normalized and expressed in molar fraction of gas molecules in pentacene. An extra term needs to be added in the solution (Nstart ). This term reflects the fact that the measurement starts at t = 0 with a quantity of gas that was accumulated in crystal during venting to atmospheric pressure and subsequent handling. Because of experimental limitations, we could not evacuate the gases in the TGA-DSC set-up. The equation that models the diffusion of gasses in pentacene is thus: h   d  N (t) = Nstart + Nsource d 1 − erf √ 2 Dt  d2 i 2 √  Dt 1 − exp − +√ π 4Dt

(5.5)

5.3. Exposure to dry and ambient air

83

where N (t) is the molar fraction of gas at time t, d is the length of the crystal in the direction in which the 1D diffusion occurs, and Nstart is the value of the molar fraction accumulated before the TGA measurement. For the case of dry air (Fig. 5.2), fitting the diffusion curve yields d = 17.5 µm, D = 1.8·10−10 cm2 /s and Nstart = −1.37 · 10−3 molecules air/molecules pentacene. Pentacene has a layered structure with a herringbone arrangement within the layers (Fig. 5.4). Our results show that the diffusion length is similar to the thickness of the crystal. Therefore, the diffusion proceeds via the largest surface area, perpendicular to the layers. We will discuss in Section 5.5 the possible microscopic form adopted by oxygen molecule in pentacene lattice .

Figure 5.4: The herringbone arrangement of the pentacene molecules in the crystal. The alternation of molecular layers is visible. d is the spacing between the layers (d=14.1 ˚ A for pentacene single crystal). The orientation of the c and a + b direction of the unit cell are drawn.

84

5.3.3

Chapter 5. Electronic Transport Properties upon Exposure to Air

Effect of Oxygen and water in dark

Figure 5.5: Pressure dependence of the current at fixed bias voltage for single crystal pentacene in the dark. The effect of dry air is shown in 5.5(a), and that of ambient air in 5.5(b). The experiments are reproducible for different samples with an accuracy of 15%. We investigate the influence of the absorbed gas molecules on the changes in the electrical current. Prior to the experiments, the samples were outgassed in vacuum (p = 5 · 10−7 mbar) in the measurement chamber for at least 100 min. The measurements can be reproduced after re-evacuation, as the gases diffuse reversibly in and out of the crystal. In order to study the influence of environmental conditions on the electrical properties, the pentacene single crystals were exposed to both 5N dry air and ambient air, using a leak-valve connected to the measurement chamber. For the experiments performed in the dark, where the effect is weaker, we measured both while outgassing and venting the chamber to ensure that the changes originate from diffusion and not from charging. The evolution of the electrical properties in time was recorded immediately after venting the system to atmospheric pressure. The initial time, t = 0, is taken when pressure equals atmospheric pressure. Figure 5.5 and 5.6 present the evolution of the electrical current at fixed bias versus pressure (5.5) and versus time (5.6) at atmospheric pressure in the dark,

5.3. Exposure to dry and ambient air

85

Figure 5.6: Time dependence of the current at fixed bias voltage for single crystal pentacene in the dark. The effect of dry air is shown in 5.6(a), and that of ambient air in 5.6(b). The experiments are reproducible for different samples with an accuracy of 15%.

respectively. The effect of oxygen exposure is studied when venting with dry air (Fig. 5.5(a), Fig 5.6(a)). As the pressure increases (Fig. 5.5(a)), oxygen will diffuse into the crystal, and the conductivity increases. The I − V curves are measured immediately after the desired pressure is reached. Once the system is at atmospheric pressure, the current continues to increase (Fig. 5.6(a)) because the diffusion still continues. When venting with ambient air, both water vapor and oxygen are introduced into the chamber (Fig. 5.5(b), 5.6(b)). We noticed that in ambient air, the effect of oxygen exposure is counteracted by the exposure to water. Analyzing the magnitude of the induced changes in the current by combining oxygen and water effects, one can distinguish two pressure regimes (Fig. 5.5(b)). At low pressure (p < 10−2 mbar), the conduction is independent of air pressure. This leads us to conclude that the two effects are opposite and of the same magnitude. At p = 10−2 mbar, a sudden drop in the current is observed. Apparently, for p > 10−2 mbar the introduction of scattering sites exceeds the creation of holes, and this has a major effect on the conduction. This is reflected also in the results at atmospheric pressure, where diffusion of dry air causes a current increase (Fig. 5.6(a)), whereas ambient air diffusion cause a drop (Fig. 5.6(b)).

86

5.3.4

Chapter 5. Electronic Transport Properties upon Exposure to Air

Effect of Oxygen and water in light

Figure 5.7: Pressure dependence of the current at fixed bias voltage for single crystal pentacene in the light. The effect of dry air is shown in 5.7(a), and that of ambient air in 5.7(b). The experiments are reproducible for different samples with an accuracy of 15%. Figures 5.7 and 5.8 show the evolution of the current under the same conditions, but now in the presence of light. The effect of oxygen diffusion in time is stronger than in dark (Fig. 5.7(a), 5.8(a)). As the effect of water is independent of the presence of light, the total influence of ambient air on the crystal properties will be different in the presence of light. Consistent with the measurements in dark, we find two pressure regimes. At low pressure (p < 10−2 mbar), when only an insignificant amount of water vapor is present, the oxygen doping is the dominant factor of the conduction mechanism. At p = 10−2 mbar, when water vapor diffuses considerably into the pentacene crystal, the effect of oxygen is decreased, and therefore a change in slope is observed (Fig. 5.7(b)). However, the presence of oxygen remains the most important factor, as the value of the current increases with pressure. After venting with ambient air, the doping caused by oxygen, activated by illumination, gives a stronger effect than the trapping of charges by water. The hysteresis in figures 5.5b and 5.7b is very small. This indicates that the variation in the values of the current is caused by the diffusion and not by charging.

5.3. Exposure to dry and ambient air

87

Figure 5.8: Time dependence of the current at fixed bias voltage for single crystal pentacene in the light. The effect of dry air is shown in 5.8(a), and that of ambient air in 5.8(b). The experiments are reproducible for different samples with an accuracy of 15%.

5.3.5

Effect of high O2 pressure†

In this section we discuss O2 high-pressure measurements of the electrical properties of pentacene single crystals. The experiments were performed in situ, in a cubic anvil press [13]. The system was first evacuated, then O2 with controlled partial pressure was introduced. 30 min was taken to reach the loading for each pressure, and then I-V curves were taken. In contrast with experiments performed in air, at 10−6 mbar < p < 103 mbar (Section 5.3.3, 5.3.4), at high O2 pressure (103 mbar < p < 2·106 mbar) an increase in current is observed also in the linear regime as the O2 pressure is increased. Here, the resistance decreases with ∼ 20% between the two extreme pressures mentioned above. The relative changes in the resistance of the pentacene crystal, with respect to the resistance in vacuum, versus O2 pressure, is plotted in Figure 5.9. Similar to the case of hole doping by iodine [14, 15], the resistivity decreases, but no metallic state is reached. Contrarily, when doping pentacene films with electron donors (e.g. alkali atoms), a high electrical conductivity and a metallic temperature behavior was reported [16, 17]. † The

measurements are done in collaboration with J. Jun, and J. Karpinski (ETH Zurich)

88

Chapter 5. Electronic Transport Properties upon Exposure to Air

Figure 5.9: Evolution of the resistance with O2 pressure. The results are normalized with respect to the resistance R0 in vacuum. The line is guide for the eye.

For pentacene loading with O2 at high-pressure, after 2·106 mbar, an precipitous decrease in current is observed, both in the linear and quadratic regime. It is possible that pentacene is transformed irreversibly to pentacenequinone, as the electrical characteristics cannot be reproduced after evacuation. Moreover, the sample becomes amorphous.

5.4 5.4.1

O2 and H2 O effects - experimental observations Electrical measurements

The changes in the electrical properties of the organic semiconductors upon exposure to ambient conditions have not been given much attention in the past. Only lately, it was stressed that there is a fundamental need to understand these effects, especially if they are not reversible, because they can affect the performance and the lifetime of the electronic devices in products built for large scale applications [18].

5.4. O2 and H2 O effects - experimental observations

89

We relate the changes that we observe in the electrical properties and describe in Section 5.3.3 and Section 5.3.4 to the diffusion of gases into the crystal. Our model includes two competing mechanisms associated with the two different chemical species that influence the conduction, affected by external factors such as light. On the one hand, water absorbed by exposure to ambient air [19], will create new trapping sites for the charges that are injected from the electrodes [20]. This will decrease the value of the electrical conductivity. The afore mentioned effect is, as expected, independent of the exposure to light. The polar nature of the H2 O molecules causes interactions with the injected charges and increases the energetic disorder [21]. This will lead to a decrease of the conduction. Oxygen present in the air will be incorporated in the crystal structure, as shown in Figure 5.2 and this process will generate holes (a description of the possible doping mechanism will be provided in Section 5.5 and a quantitative analysis in Section 5.6 of this chapter). The density of charge carriers responsible for the conduction will thus increase, and accordingly the value of the current will increase. This process is enhanced by the presence of light. The measurements performed in pure N2 and Ar under identical conditions show insignificant changes in the electrical properties, despite a similar amount of absorption.

5.4.2

UPS spectroscopy‡

We use ultraviolet photoelectron spectroscopy (UPS) to investigate the effect of oxygen and air exposure on the electronic structure of pentacene single crystals [22]. The oxygen doping should induce a shift of the Fermi-level (EF ) in the organic towards the highest occupied molecular orbital (HOMO). Photoemission experiments were performed at the end-station SurICat (beamline PM4) at the synchrotron light source BESSY GmbH. The ultrahigh vacuum (UHV) system consists of interconnected sample preparation (base pressure: 1 · 10−8 mbar) and analysis (base pressure: 5·10−10 mbar) chambers. Sample transfer between chambers proceeded without breaking UHV conditions. Pentacene single crystals were grown ex situ, then mounted on metal sample holders with conducting adhesive tape, and cleaved in situ (in the preparation chamber) by peeling off the top layers with a piece of adhesive tape stuck to the crystal surface. Pentacene molecules crystallize in a layered structure. The interactions between the layers are less strong than the interactions within the layers, as they are van der Waals interac‡ The experiments were performed in collaboration with the groups of N. Koch and P. Rudolf and published as The effect of oxygen exposure on pentacene electronic structure, Eur. Phys. J. E 17, 339 (2005).

90

Chapter 5. Electronic Transport Properties upon Exposure to Air

Figure 5.10: UPS spectra of a pentacene single crystal measured under the following conditions (and in this time-sequence): (a) 3 · 10−8 mbar partial O2 pressure, (b) 2 · 10−9 mbar total residual pressure, (c) 3 · 10−8 mbar partial O2 pressure after exposure to 5 · 10−6 mbar O2 for 30 min., and (d) again at 2 · 10−9 mbar total residual pressure. tions. Therefore, the crystal will cleave easily along the a− b plane (perpendicular to the c∗ direction). The valence region UPS spectra (Fig. 5.10) recorded on a pentacene single crystal under different residual oxygen pressures during the measurement and total oxygen exposure clearly show no influence of molecular oxygen on the position of the energy levels at the given conditions. All four spectra in Figure 5.10 exhibit the same characteristic photoemission features of pentacene, comparable to previous reports [23]. If ”doping” (i.e., p-type) of pentacene by O2 were the case (to a significant degree), a clear shift of all levels towards lower binding energy (BE) would be observed. However, the total in situ exposure of the single crystal to O2 was rather limited (up to a pressure of 3 · 10−8 mbar; for experimental lim-

5.5. Mechanism of oxygen doping

91

itations). Oxygen diffused into pentacene at atmospheric pressure and diffused out again rapidly after re-evacuation, and does not lead to irreversible changes (detectable by UPS) in the electronic structure, like, e.g., p-type doping. In extensive tests we have observed no irreversible reaction of pentacene with molecular oxygen and water, even if pentacene is optically excited. More importantly, it is found that oxygen diffusion through pentacene single crystals is reversible, and does not leave behind - after re-evacuation - electrically active electronic states that would lead to doping of the organic bulk material. This is evidenced by a lack of energy shifts in UPS spectra of pentacene before and after exposure to O2 and to air.

5.5

Mechanism of oxygen doping

In the following section we propose a microscopic description of the effect of oxygen absorption in pentacene. We discuss the nature, on a molecular scale, of oxygen-induced changes in charge transport properties. Under the experimental conditions explored here, these changes are reversible. This is fundamentally different from the experiments in which oxygen creates irreversibly new chemical species in pentacene, leading to trapping states in the energy gap of pentacene [8]. It should be stressed that oxygen doping is widely observed in organic semiconductors. Therefore, it is likely that this interpretation holds for other similar materials. The exact nature of the origin the phenomena encountered here is not known. We propose two possible mechanisms for oxygen doping in pentacene singe crystals. Further experimental and theoretical studies should elucidate which of them is correct, or if they can coexist.

5.5.1

Formation of the charge transfer complex between pentacene and O2

When oxygen is inserted in the pentacene lattice, it can generate a charge transfer complex (CT complex) with the host material. The direction of the charge transfer can be established by comparing the values of the electron affinities of the two neutral molecules. As the value of the pentacene electron affinity (Ea (C22 H14 ) = 2.7 eV [24]) is higher than that of O2 (Ea (O2 ) = 0.45 eV [25]), pentacene is electron-donating in this complex, and O2 is electron-accepting. Because of its electronegativity, oxygen will attract electrons from the electron rich π- system of pentacene molecules, and this process will generate holes. No net charges are generated, but a partial transfer of charge occurs, from the donor to the acceptor.

92

Chapter 5. Electronic Transport Properties upon Exposure to Air

Figure 5.11: The reversible charge transfer formed between pentacene and oxygen.

Figure 5.12: The reversible reaction of pentacene oxidation to endoperoxide. This reaction can yield irreversibly hydroquinone and quinone.

We expect that the degree of charge transfer in our system is much weaker that in typical organic charge transfer complexes (e.g. TTF-TCNQ [26, 27]). Upon formation of the [pentacene-oxygen] charge transfer complex (Fig. 5.11), the properties of pentacene single crystals are perturbed. There are several questions that can be raised. How large is the degree of charge transfer between the two molecules? Does it induce also magnetism, besides changes in electrical properties? Is it detectable with impedance spectroscopy experiments and/or Raman spectroscopy?

5.5.2

Reversible oxidation of pentacene

A second possibility is that molecular oxygen in the crystal forms a 6,13-adduct (pentacene-endoperoxide), via the reversible reaction presented in Figure 5.12. This possibility is analogous to the behavior of carbon nanotubes in oxygen [28]. In pentacene positions 6 and 13 are likely affected because they are the most reactive. This adduct could yield irreversibly, under conditions not explored here,

5.5. Mechanism of oxygen doping

93

via the formation of the hydroquinone, the 6, 13-pentacenequinone, which is conventionally the dominant impurity species [8]. The reaction that pentacene undergoes to endoperoxide in the presence of oxygen is photoinduced [29]. Moreover, it is reversible, in agreement with our experimental observations that the electrical properties in vacuum can be recovered after evacuation. Still, Raman studies that we performed on pentacene single crystals exposed to oxygen do not show the signature of any peak present in the region ∼ 800 cm−1 , that corresponds to the O − O vibration [30]. We find no evidence for the formation of the endoperoxide, in contrast with the case of rubrene [29].

5.5.3

Calculation of the oxygen-induced dipole moment

We measure the frequency dependence of the dielectric constant ε(ω) of pentacene single crystals loaded with oxygen and air (Fig. 5.13). We demonstrate experimentally that a dipole moment is induced in the pentacene crystal, and we determine its magnitude. δ+ The dielectric function ε(ω) is increased by the polarizability of [C22 H14 ·O2δ− ] charge transfer. The charge transfer gives rise to dipoles that couple to the ac electric field. This increases the value of permittivity [31]. We used pentacene single crystals with a high degree of purity, grown from double sublimated powder. Two contacts were deposited, on top and bottom, in a sandwich-type geometry, as drawn in the inset in Figure 5.13. The description of the sandwich-type geometry was explained in Chapter 4. The ac dielectric measurements were performed in a probe station, in air, using a Agilent 4284A. The as grown crystals were measured first in ambient air. The dielectric response is shown in Figure 5.13. A decrease of the dielectric constant by 3% is observed when the frequency is modified from 1 kHz to 200 kHz. The crystals were then exposed to pure O2 (AGA, 5N) for a few hours. This is sufficient for maximum loading, as the diffusion constants for oxygen and air (Fig. 5.2) are similar. The dielectric constant increases markedly at low frequencies for crystals loaded with O2 and decreases by 12% from 1 kHz to 200 kHz (Fig. 5.13). Assuming a Debye response, the dielectric constant will decrease with frequency according to: ε(0) − ε(∞) (5.6) ε(ω) = ε(∞) + 1 + (ωτd )2 where ε(∞) corresponds to the high-frequency permittivity of the material, and ε(0) is the static dielectric constant, ω is the angular frequency (ω = 2πf ), and

94

Chapter 5. Electronic Transport Properties upon Exposure to Air

Figure 5.13: Frequency dependence of the dielectric constant of pentacene single crystals loaded in air and in oxygen. The inset is a schematic representation of the sample geometry, showing also the orientation of the crystal (c* is the axis perpendicular to the a-b plane). The line is a fit of a Debye response with a single relaxation time.

τd represents the relaxation time. As it can be derived from Eq. 5.6, this model considers a single relaxation time. For the measurements performed on pentacene single crystals exposed to pure oxygen, fitting the curve in Figure 5.13 with Eq. 5.6 yields: ( ε(0) = 3.34 (5.7) ε(∞) = 3.07 The fit does not represent very accurately the frequency response of the dielectric constant. At low frequencies, additional contributions increase the permittivity, and the Debye model does not describe the response accurately. Deviations from the model can also be ascribed to the presence of different relaxation times, whereas we assume a single value. We use the Clausius-Mossotti equation [32], that relates the dielectric constant ε with the dipole moment µ inside the solid:

5.5. Mechanism of oxygen doping

Nd µ2 4πNm αm ε(ω) − 1 = + ε(ω) + 2 9ε0 kB T 3

95

(5.8)

Here Nd is the density of dipoles, T is the temperature, Nm is the molecular density in the solid (Nm = 2.92 · 1021 cm−3 for pentacene), αm is the molecular polarizability, and ε0 and kB are the vacuum permittivity and the Boltzmann constant, respectively. We assume one dipole created per each absorbed O2 molecule, thus Nd ≃ 0.02 · Nm . Only 2% of the pentacene molecules form a dipole, corresponding to the 2% molar loading. The distribution of the pentacene species corresponds to the distribution of the oxygen in the lattice (experiments are in progress to determine the spatial distribution of oxygen in pentacene). If we assume that the behavior of ε(ω) is only determined by dipole formation through O2 -loading, we find that: ε(∞) − 1 4πNm αm = 3 ε(∞) + 2

(5.9)

Thus, we can determine value of the dipole moment of this charge transfer complex. At room temperature, we obtain µ = 4 D. This allows the calculation of the degree of charge transfer (q), as µ = q · d, where d represents the distance between the charges. We do not have structural information concerning the position of the O2 molecule in the pentacene crystal lattice. Thus we cannot give an exact value of the distance between the charges. For this reason, we can only extract an order of magnitude for the value of the charge transfer. We obtain q ≃ 0.25e, where e is the elementary charge. Repeating the same algorithm for the measurements performed in air, and considering Nd ≃ 4 · 10−3 · Nm (only 20% from the 2% loading represents O2 molecules), we obtain: ε(0) = 3.2, ε(∞) = 3.08, and µ = 5 D. The correspondent values for the charge transfer is q = 0.3e. We observed that the values of the dipole moment induce by oxygen in pentacene, calculated for the experiments performed in air and in oxygen are similar, as expected. We associate the small differences to additional polarization, induced by the interaction of pentacene molecules with other impurity states in the crystals, and with the metal contacts, and also to experimental errors. The oxygen-induced charge-transfer mechanism presented here is able to describe the main features of the frequency dependence of the dielectric constant: the decrease of the value of ε with frequency, and dependence on the oxygen content.

96

5.6

Chapter 5. Electronic Transport Properties upon Exposure to Air

Quantitative analysis of the effect of O2 and H2 O

A comparison of Figure 5.2 with Figure 5.6(a) reveals a remarkable consistency. The time evolution of the electrical current mirrors that of the oxygen absorption. We can quantitatively assess the influence of O2 and H2 O. The value of the hole current is affected by the number of gas molecules present in the crystal and thus varies in time: Jp (t) = eµp p(t)E (5.10) where e is the elementary charge and µp is the mobility of holes in pentacene. The term p(t) is the density of charge carriers and consists of several terms: p(t) = pinj + pdop − ptraps

(5.11)

Here pinj represents the hole density injected from the contacts, and pdop is the hole density induced by exposure to oxygen, which is a fraction of the number of air molecules absorbed in the solid: pdop = 0.2ηoxygen N (t)

(5.12)

The oxygen efficiency ηoxygen is thus defined as the inverse of the number of oxygen molecules required to generate one hole. The term: ptraps = pdef + pimp + pwater

(5.13)

is given by the density of traps in the band gap. In vacuum a fraction of the injected charges are trapped by crystal defects, pdef , and impurities, pimp . In ambient air, an extra term is added in the trapping sites due to the presence of water molecules, pwater (t) = βN (t) (5.14) which is a fraction of total number of the air molecules N (t) accumulated in the crystal. Combining the measurements performed in dry air (Fig. 5.6(a), 5.8(a)) with N (t) taken from Figure 5.2, we can calculate the oxygen efficiency ηoxygen , which determines the number of holes introduced by one O2 molecule (see upper panel Fig. 5.14). As it can be observed in Figure 5.14, this number is constant and independent of time with ηoxygen ≈ 0.25 in the dark and ηoxygen ≈ 0.5 with light exposure. This means that the efficiency of the oxygen exposure induced doping is independent of oxygen loading and it is increased by light exposure. Roughly two O2

5.6. Quantitative analysis of the effect of O2 and H2 O

97

Figure 5.14: The upper panel shows the evolution in time of the oxygen efficiency in the dark (◮) and with light exposure (•). For ambient light intensity, this efficiency is constant and time independent. The lower panel shows the variation of the trapping factor by introducing water molecules (Θwater ) with time. The values are similar in the dark (◮) and with light exposure (•).

molecules are needed in light to create one hole, compared with four molecules in dark. In ambient air, the number of trapping sites is proportional to the number of water molecules that diffuse into the crystal. We introduce the ratio (Eq. 5.15): Θwater =

pinj − pwater pinj

(5.15)

that quantifies the fraction of untrapped charge carriers: Θwater = 1 in vacuum. After exposure to ambient air, the value of Θwater is reduced. Θwater is calculated by subtracting Figure 5.6(a)/ 5.8(a) from Figure 5.6(b)/ 5.8(b), using appropriate geometrical factors. The lower part of Figure 5.14 shows Θwater versus time. The effects of water in the dark and with light exposure have similar magnitudes, and they follow the diffusion of water, introduced from ambient air. The changes of the conductivity with time and pressure are the electronic response to the diffusion of gases in pentacene crystals. They are fully reversible. The p-doping by molecular oxygen increases the concentration of charge carriers.

98

Chapter 5. Electronic Transport Properties upon Exposure to Air

Water molecules form new electronic states in the gap that trap the injected charges. It should be noticed that experiments under UHV conditions will not be influenced by exposure to an ambient atmosphere, as the absorbed gas can be fully removed by evacuation [22]. We expect that these effects are even larger in thin films, in which the presence of grain boundaries facilitates the diffusion.

5.7

Conclusion

In conclusion, we have investigated the intercalation of gases in pentacene single crystals and its effect on the electronic properties. We differentiate between the exposure to oxygen and water. Absorbed oxygen enhances the conduction by introducing holes near the valance band. Absorbed water molecules create new defect states, which trap the injected charges. This insight in the relationship between the macroscopic diffusion of gas in organic conductors and the effect on the physical properties may lead to an improved control of the device performance.

References

[1] V. Y. Butko, X. Chi, D. V. Lang, and A. P. Ramirez, Appl. Phys. Lett. 83, 4773 (2003) [2] J. Takeya, C. Goldmann, S. Haas, K. P. Pernstich, B. Ketterer, and B. Batlogg, J. Appl. Phys. 94, 5800 (2004) [3] C. Goldmann, S. Haas , C. Krellner, K. P. Pernstich, D. J. Gundlach, and B. Batlogg , J. Appl. Phys. 96, 2080 (2004) [4] G. H. Gelinck, H. E. A. Huitema, E. van Veenendaal, E. Cantatore, L. Schrijnemakers, J. B. P. H. van der Putten, T. C. T. Geuns, M. Beenhakkers, J. B. Giesbers, B. H. Huisman, E. J. Meijer, E. M. Benito, F. J. Touwslager, A. W. Marsman, B. J .E. van Rens, and D. M. de Leeuw, Nat. Mat. 3, 106 (2004) [5] T. Yasuda, T. Goto, K. Fujita, and T. Tsutsui, Appl. Phys. Lett. 85, 2098 (2004) [6] T. Minari, T. Nemoto, and S. Isoda, J. Appl. Phys. 96, 769 (2004) [7] C. C. Mattheus, A. B. Dros, J. Baas, G. T. Oostergetel, A. Meetsma, J. L. de Boer, and T. T. M. Palstra, Synth. Met. 138, 475 (2003) [8] O. D. Jurchescu, J. Bass, and T. T. M. Palstra, Appl. Phys. Lett. 84, 3061 (2004)

99

100

References

[9] B. Crone, A. Dodabalapur, A. Gelperin, L. Torsi, H. E. Katz, A. J. Lovinger, and Z. Bao, Appl. Phys. Lett. 84, 3061 (2004) [10] A. Assadi, G. Gustafsson, M. Willander, C. Svensson, and O. Ingan¨ as, Synth. Met. 37, 123 (1990) [11] L. Tosi, A. Dodabalapur, N. Cioffi, L. Sabbatini, and P. G. Zambonin, Sens. Actuators B 77, 123 (2001) [12] A. Laudise, C. Kloc, P. Simpkins, and T. Siegrist, J. Cryst. Growth 187, 449 (1998) [13] J. Karpinski, S. M. Kazakov, J. Jun, M. Angst, R. Puzniak, A. Wisniewski, and P. Bordet, Physica C 385, 42 (2003) [14] T. Minakata, H. Imai, and M. Ozaki, J. Appl. Phys. 72, 4178 (1992) [15] T. Ito, T. Mitani, T. Takenobu, and Y. Iwasa, J. Phys. Chem. Sol. 65, 609 (2004) [16] Y. Kanedo, T. Suzuki, Y. Matsuo, and S. Ikehata, Synth. Met. 154, 177 (2005) [17] M. F. Craciun, S. Rogge, and A. F. Morpurgo, cond-mat/0603261 [18] C. R. Kagan, A. Afzali, and T. O. Graham, Appl. Phys. Lett. 86, 193505 (2005) [19] Y. Qiu, Y. Hu, G. Dong, L. Wang, J. Xie, and Y. Ma, Appl. Phys. Lett. 83, 1644 (2003) [20] Z. T. Zhu, J. T. Mason, R. Dieckmann, and G. G. Malliaras, Appl. Phys. Lett. 81, 4643 (2002) [21] D. H. Dunlap, P. E. Parris, and V. M. Kenkre, Phys. Rev. Lett. 77, 542 (1996) [22] A. Vollmer, O. D. Jurchescu, I. Arfaoui, T. T. M. Palstra, P. Rudolf, J. Niemax, J. Pflaum, I. Salzmann, J. P. Rabe, and N. Koch, Eur. Phys. J. E 17, 339 (2005) [23] N. Koch, J. Ghijsen, J.-J. Pireaux, J. Schwartz, R. L. Johnson, A. Elschner, and A. Kahn, Appl. Phys. Lett. 82, 70 (2003)

References

101

[24] N. Koch, J. Ghijsen, R. L. Johnson, J. Schwartz, J.-J. Pireaux, and A. Kahn, J. Phys. Chem. B 106, 4192 (2002) [25] R. C. Weast, CRC Handbook of Chemistry and Physics (CRC Press, West Palsm Beach, FL, 1992) [26] J. Ferraris, D. O. Cowan, V. Walatka, and J. H. Perlstein, J. Am. Chem. Soc. 95, 948 (1973) [27] D. J´erome, Chem. Rev. 104, 5565 (2004) [28] G. Dukovic, B. E. White, Z. Zhou, F. Wang, S. Jockusch, M. L. Steigerwald, T. F. Heinz, R. A. Friesner, N. J. Turro, and L. E. Brus, J. Am. Chem. Soc. 126, 15269 (2004). [29] V. Ramamurthy, and K. Venkatesan, Chem. Rev. 87, 433 (1987) [30] Raman spectroscopy in collaboration with D. Fausti, and P. van Loosdrecht (RuG) [31] B. Pevzner, A. F. Hebard, and M. S. Dresselhaus, Phys. Rev. B 55, 16439 (1997) [32] A. K. Jonscher, Dielectric Relaxation in Solids (Chelsea Dielectrics, London, 1983)

6

Low Temperature Crystal Structure of Rubrene Single Crystals Grown by Vapor Transport∗

The study presented in the following chapter is promoted by reports of a precipitous drop in the electronic mobility of rubrene single crystals below 175 K. We assign this change to a phase transition. We perform the crystal structure determination for rubrene single crystals, C42 H28 (5,6,11,12 - Tetraphenyltetracene), in the temperature interval 100 K - 300 K. The crystals are grown by physical vapor transport in an open system. The crystal structure is orthorhombic over the entire temperature range. We find evidence for a phase transition from Differential Scanning Calorimetry measurements, but see no evidence for a structural phase transition in the diffraction experiments.

∗ This Chapter is adapted from O. D. Jurchescu, A. Meetsma, and T. T. M. Palstra, Acta Cryst. B 62, 330 (2006)

103

104

6.1

Chapter 6. Orthorhombic Rubrene Single Crystals

Introduction

The electronic properties of polymers and molecular crystals are of much current interest due to fundamental questions regarding electron transport and associated applications. The present focus is in developing novel chemical structures, as well as improved fabrication techniques that are able to open new fields for electronic applications based on organic materials. A number of molecular conductors are of current interest. Rubrene (Fig. 6.1) exhibits interesting physical properties such as having one of the highest reported electronic mobilities at room temperature [1] (20 cm2 /Vs). Emerging new applications based on rubrene are organic field effect transistors (OFETs) [1–4] and organic light emitting diodes (OLEDs) [5, 6].

Figure 6.1: Bond-line formula of rubrene . Although the technological development is relatively fast, fundamental questions concerning the intrinsic mechanism of conduction have still not been answered. Moreover, for the title compound, little attention has been paid to the interplay between its crystal structure and electronic properties. The interest in the present work is guided by intriguing changes which have been observed at low temperature in the mobility of field-effect transistors built on rubrene single crystals [2, 3].

6.2

Polymorphism in rubrene

In this study, we investigate the structure of rubrene single crystals and correlate it with the electrical properties. The crystals are formed through competition

6.3. Growth of rubrene single crystals

105

between π- stacking and quadrupolar interactions. Owing to these weak interactions, different growth conditions can lead to different polymorphs. This feature is widely encountered for molecular crystals. Several polymorphs of rubrene have been reported. A monoclinic phase was described by Taylor (1936) [7] and a triclinic polymorph was explored by Akopyan et al.(1962) [8]. Unfortunately, these reports do not mention the growth method. An orthorhombic form of rubrene (space group Aba2) was reported by Henn et al. (1971) [9] for crystals grown from vapor in vacuum using sealed ampoules. For crystals obtained in a closed system, in a two-zone furnace, at ambient pressure, Bulgarovskaya et al. (1983) reported a second orthorhombic polymorph, with space group Bbam [10]. For all the polymorphs investigated previously, only the room-temperature structure is reported, thus no relation with the physical properties at low temperature can be made.

6.3

Growth of rubrene single crystals

Rubrene single crystals investigated in this work were grown using physical vapor transport under a temperature gradient, in a horizontal tube [11]. The inner tube, in which the crystallization takes place, was cleaned and heated afterwards at T = 600 K under Ar flow for 10 hours to evaporate the solvents. The source material, with nominal 98% purity, was obtained from Sigma Aldrich. Prior to crystal growth, the powder was purified using vacuum sublimation under a temperature gradient process [12], designed to remove the light impurities1 . The purified powder was placed in the hot part of the growth tube in a alumina boat. The sublimation temperature (T = 553 K) was kept as low as possible (close to the sublimation threshold). This was done to prevent the sublimation of heavy impurities and to ensure the slow sublimation rate that yields crystals with a minimum number of structural defects. Crystals grown at slightly higher temperatures were of significantly lower quality and gave weak diffraction pattern. The growth set-up was placed in dark to avoid oxidation. The transport gas was a mixture of H2 (AGA, 5N ) and Ar (AGA, 5N ), with a volume percentage of 5.25% H2 . The crystals are orange-colored and needle- or platelet-shaped. The typical in-plane dimensions are 200 µ m ×1 cm for the needles and 3 − 5 mm for platelets. The thickness of the crystals is 3 − 20 µ m. 1 We

refer to the species present in the system that have lower/higher molecular masses than the parent compound as ’light’/’heavy’ impurities.

106

6.4 6.4.1

Chapter 6. Orthorhombic Rubrene Single Crystals

Single crystal diffraction Experimental details

The single crystal diffraction experiments were performed on a Bruker SMART APEX CCD diffractometer. A crystal fragment, cut to size to fit in the homogeneous part of the X-ray beam, with dimensions of 0.51 x 0.45 x 0.03 mm was mounted on top of a glass fiber and aligned on the diffractometer. The diffractometer was equipped with a 4K CCD detector set 60 mm from the crystal. Intensity measurements were performed using graphite monochromated Mo-Kα radiation. The crystal was cooled fast using a Bruker KRYOFLEX (1h was taken to reach 100 K) and the measurements were performed during heating, after the temperature was stabilized. For each temperature investigated in this chapter, a total of 1800 frames were collected with an exposure time of 30.0 seconds per frame. The overall data collection time was 18.0 h. During this time, the variations in the temperature were ±0.5 K at 100 K, and at most ±2 K at higher temperatures. The structure was solved by direct methods with SHELXS-97. The positional and anisotropic displacement parameters for the non-hydrogen atoms were refined. Experimental details about the structure determination, data collection and refinement are summarized in the Appendix B of the thesis. The data collection (radiation type, monochromator, data collection method, θ range) and refinement parameters (number of reflections, number of refined parameters, weighting scheme, goodness of fit) are similar for the structures corresponding to the eight temperatures investigated in this study.

6.4.2

Orthorhombic rubrene

Crystals grown from physical vapor transport in an inert gas, in an open system and continuous flow [11] are preferred for electronic applications, because they present the highest purity and thus the largest mobility. Still, no crystal structure determination is available for crystals obtained by this method. In this study we report the crystal structure between 100 K and 300 K of rubrene single crystals grown from physical vapor transport (PVT). This growth method yields orange crystals with orthorhombic symmetry, space group Cmca. We notice that the polymorph of the crystals grown with this method coincides2 with that obtained 2 We used the standard settings of the International Tables for Crystallography. The transformation to Bbam (Cmca → Bbam) is: b → a, c → b, a → c. For the crystal orientation in the electronic measurements [2, 3], the Bbam setting was used.

6.4. Single crystal diffraction

107

Figure 6.2: Perspective ORTEP drawing of the rubrene molecule illustrating the configuration and the adopted labelling scheme for the non-hydrogen atoms at 100 K (a), and room temperature (b). Displacement ellipsoids for non-H are represented at the 50% probability level. .

108

Chapter 6. Orthorhombic Rubrene Single Crystals

Figure 6.3: Lattice parameters a, b and c, and unit cell volume V versus temperature for rubrene single crystals. The lines are guides for the eyes.

for the crystals grown in sealed ampoules by Bulgarovskaya et al. [10]. The molecule presents 2/m symmetry, with the asymmetric unit consisting of one quarter of a rubrene molecule. A twofold axis is located along the C1 − C1b bond. There is a mirror plane perpendicular to the planar tetracene fragment of the molecule, through the inversion center positioned in the middle of the C1−C1b bond. The adopted labelling scheme and the molecular geometry are illustrated in the ORTEP drawings of Figure 6.2(a, b). The plots show graphical representations of the atomic positions and their anisotropic displacement ellipsoids at the minimum and maximum temperatures investigated in this study (T=100 K, and room temperature). These images provide a visual description of the vibrational motions of each atom around its equilibrium position. The crystal structure determination reveals a 2.157% volume thermal expansion of the crystal between 100 K and 300 K, in agreement with typical values measured in organic crystals. The thermal expansion is anisotropic for the three crystallographic directions (∆a = 0.410%, ∆b = 0.320%, ∆c = 1.562%) and oc-

6.5. Relation between molecular stacking and electronic mobility

109

curs predominantly along the weakest bonding c-axis (see Fig. 6.3). The lattice parameters together with other crystallographic parameters for all eight temperature structures investigated in this study can be found in the Appendix B of the thesis.

6.5

Relation between molecular stacking and electronic mobility

We relate the behavior of electronic mobility versus temperature with structural changes. Electronic measurements performed on rubrene single crystals show that the field effect mobility increases with decreasing temperature down to 175 K, consistent with band-like behavior [2, 3] (the results differ slightly between groups). Below 175 K the mobility drops precipitously. The dramatic decrease in the field-effect mobility can be a consequence of a structural phase transition in the material. Our diffraction experiments show that the crystallographic symmetry does not change when passing in this temperature interval. It can also be observed from Figure 6.3 that the unit cell volume and unit cell parameters are continuous with temperature within our resolution. Also, the evolution of the fractional coordinates of the Carbon atoms do not signal the presence of a phase transition. However, differential scanning calorimetry (DSC) performed on the crystals exhibit the signature of a phase transition around T = 175 K (see Fig. 6.4). It is likely that there is a phase transition at this temperature, but it is not observable from our X-ray diffraction experiments. We use the model proposed by da Silva Filho et al. [13] to explain the changes in electronic properties as a response to the structural changes in the crystal. They associate the intrinsic transport properties of rubrene single crystals with the electronic coupling between adjacent molecules. The measure for this interaction is the interchain transfer integral, t, which varies with molecular packing. The electronic coupling is maximal if the molecules are perfect cofacial in the crystal, oscillates when the molecular displacement increases (with positive and negative values) and becomes zero at large shifts. This translates to the direct influence of the molecular stacking on the electronic properties. It can bee seen in Figure 6.5 that the molecules are not perfectly cofacial, but they are shifted with respect to each other. We define the molecular displacement d that quantifies the shift between two molecules. From the calculations of Silva Filho et al. it is obvious that the molecular displacement, d, is the relevant parameter, relating the crystal structure with the electronic mobility, similar to

110

Chapter 6. Orthorhombic Rubrene Single Crystals

Figure 6.4: Differential Scanning Calorimetry of crushed rubrene single-crystals. The measurement has been carried out in N2 atmosphere. The enthalpy associated with the transition is 3.13 kJ/mole. The arrow indicates the position or the peak.

the model proposed by Mattheus et al. [14]. This parameter is complementary to molecular overlap and given by the orientation of the tetracene backbone with respect to the b - crystallographic direction. In order to calculate the value of molecular displacement at each temperature, we show the projection of two rubrene molecules on the b-c plane (see Fig. 6.6). From geometrical considerations, the value of the molecular displacement (d) is given by: (Eq. 6.1): d=r+q (6.1) where r represents half of the length of the rubrene molecule, and q the distance from the extremity of the one molecule to the perpendicular taken from the middle of the molecule to the long axis of the other one. The simplified expression for d at each temperature is: d = b · cos Θ (6.2) where b is the value of the unit cell parameter (Fig. 6.3) and Θ is the angle that the linear part of the molecule makes with this axis. The maximum deviation from planarity in the tetracene backbone of the molecule corresponds to C2 and is 0.0714(11) ˚ A at 100 K and 0.0731(14) ˚ A

6.5. Relation between molecular stacking and electronic mobility

111

Figure 6.5: Schematic representation of two rubrene molecules in the crystal. The π- stacking distance is similar to van der Waals distance. The molecules are shifted with respect to each other along the tetracene backbone. .

Figure 6.6: Arrangement of the rubrene molecules in the unit cell: view along the [100] axis. The drawing includes schematic representation of the parameters (r, q) used in the calculation of the molecular displacement (d). Unit cell b and c axis are also indicated. Θ is the angle between the long axis of the molecule and b-axis.

112

Chapter 6. Orthorhombic Rubrene Single Crystals

Figure 6.7: Evolution of the molecular displacement along the long axis of the rubrene molecule with temperature. The inset represents rubrene molecular packing viewed along the [100] direction, and defines the length d of the molecular displacement.

at room temperature. The interplanar separation between two adjacent parallel molecules, along the π-stack, increases from 3.654(3) ˚ A to 3.715(3) ˚ A between 100 K and 300 K. This length is slightly larger than the typical van der Waals interaction distance in a C − C π-stack [15]. The molecules slide with respect to each other along their long axis when the temperature is modified. Figure 6.7 shows the temperature dependence of the molecular displacement, d along the long axis of the molecule in rubrene single crystal. The value of the shift increases between 100 K and 300 K (Fig. 6.7). The changes of d with temperature are dominated by changes in the lattice parameter b (see Fig. 6.3). However, it can be seen from Figure 6.3 and Figure 6.8, that the temperature dependence of b and cos(Θ) individually do not present any discontinuity or change of slope at T = 175 K. In spite of the fact that a different picture is suggested by their combination in the molecular displacement (see Eq. 6.2), we cannot assign this to a crystallographic phase transition. The variations in molecular displacement induce changes in the transfer integral (t), and determine the electronic mobility (see Figure 4, da Silva Filho

6.6. Conclusions

113

Figure 6.8: Variation of the angle Θ between the tetracene backbone of rubrene molecule and the b crystallographic axis.

et al., [13]). The molecular displacement has a large value due to the effect of the interactions between phenyl side groups of neighboring molecules. They are arranged two by two, on different sides of the planar tetracene part of rubrene and the position is restricted by the inversion center. Their torsion with respect to the acene linear part ranges from 80.30(5)◦ at 100 K to 80.88(7)◦ at room temperature. The phenyl rings positioned on the same part of the molecule make a dihedral angle of 25.32(6)◦ at 100 K and 25.14(10)◦ at 300 K. The connection between the phenyl rings and the tetracene fragment is the C2 − C6 bond. At 100 K this bond makes an angle of 12.57(8)◦ with the tetracene fragment of the molecule and 6.13(9)◦ with the phenyl-plane. The corresponding angles at 300 K are 12.74(9)◦ and 6.00(10)◦, respectively.

6.6

Conclusions

We conclude that rubrene single crystals grown from physical vapor transport in ambient pressure exhibit an orthorhombic symmetry. We find evidence for a phase transition from DSC measurements, but see no evidence for a structural phase transition in the diffraction experiments. Further investigations concerning the mechanism that drives the dramatic changes in electronic mobility at T = 175 K in rubrene are required.

References

[1] V. C. Sundar, J. Zaumseil, V. Podzorov, E. Menard, R. L. Willett, T. Someya, M. E. Gershenson, and J. A. Rogers, Science 303, 1644 (2004) [2] R. W. I. de Boer, M. E. Gershenson, A. F. Morpurgo, and V. Podzorov, Phys. Stat. Sol. A 201, 1302 (2004) [3] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers, and M. E. Gershenson, Phys. Rev. Lett. 93, 086602 (2004) [4] V. Y. Butko, J. C. Lashley, and A. P. Ramirez, Phys. Rev. B 72, 081312 (2005) [5] T. Oyamada, H. Uchiuzou, S. Akiyama, Y. Oku, N. Shimoji, K. Matsushige, H. Sasabe, and C. Adachi, J. Appl. Phys. 98, 074506 (2005) [6] C. N. Li, C. Y. Kwong, A. B. Djurisic, P. T. Lai, P. C. Chui, W. K. Chan, and S. Y. Liu, Thin Solid Films 477, 57 (2005) [7] W. H. Taylor, Z. Kristallogr. 93, 151 (1936). [8] S. A. Akopyan, R. L. Avoyan, and Yu. T. Struchkov, Z. Strukt. Khim. 3, 602 (1962) [9] D. E. Henn, and W. G. Williams, J. Appl. Cryst. 4, 256 (1971) [10] I. Bulgarovskaya, V. Vozzhennikov, S. Aleksandrov, and V. Belsky, Latv. PSR Zinat. Akad. Vestis, Fiz. Teh. Zinat. Ser. 4, 53 (1983) 115

116

References

[11] A. Laudise, C. Kloc, P. Simpkins, and T. Siegrist, J. Cryst. Growth 187, 449 (1998) [12] O. D. Jurchescu, J. Baas, and T. T. M. Palstra, Appl. Phys. Lett 84, 3061 (2004) [13] D. A. da Silva Filho, E. G. Kim, and J.-L. Br´edas, Adv. Mat. 17, 1072 (2005) [14] C. C. Mattheus, G. A. de Wijs, R. A. de Groot, and T. T. M. Palstra, J. Am. Chem. Soc. 125, 6323 (2003) [15] A. Bondi, J. Phys. Chem. 68, 441 (1964)

A

Crystal structure of pentacene

Pentacene — T = 100 K Chemical formula Formula weight Cell setting Space group a (˚ A) b (˚ A) ˚ c (A ) A3 ) V (˚ α β γ Z ρcalc (g cm−3 ) Crystal form, colour Crystal size (mm)

C22 H14 278.35 Triclinic P 1, 2 6.277(1) 7.663(2) 14.363(3) 669.5(2) 76.893(3)◦ 88.182(3)◦ 84.326(3)◦ 2 platelet, violet 0.430 × 0.210 × 0.050

Table A.1: Crystal data and details of the structure determination of pentacene at 100 K

Diffractometer Radiation type Monochromator Data collection method θ range (◦ ) Unique data Data with criterion F0 > 4σ(F0 ) Rint Range of h, k, l

Bruker Smart Apex; CCD area detector Mo Kα Graphite ϕ and ω scans 2.74 − 25.03 2310 1683 0.0245 −7 → h → 7 −9 → k → 9 −16 → l → 17

Table A.2: Data collection for pentacene at 100 K

117

118

Appendix A. Crystal structure of pentacene

No. of reflections No. of refined parameters Final agreement factors: Weighting scheme R(F ) wR(F 2 ), Goodness of Fit (S) Residual electron density ∆ρmin ∆ρmax

2310 255 0.0504 0.1722 1.050 −0.2 (e˚ A −3 ) 0.33 (e˚ A−3 )

Table A.3: Refinement for pentacene at 100 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 H1 H3 H5 H6 H7 H8 H10 H12 H14 H16 H17 H18 H19 H21

x 0.7010(2) 0.5432(2) 0.5778(2) 0.4214(2) 0.4559(3) 0.3002(2) 0.0949(3) 0.0542(2) 0.2132(2) 0.1761(2) 0.6654(2) -0.1816(2) -0.1558(2) -0.3059(2) -0.2788(2) -0.4297(2) -0.3994(3) -0.2153(3) -0.0673(3) -0.0911(2) 0.0573(2) -0.0321(2) 0.843(3) 0.720(3) 0.599(3) 0.325(2) -0.013(3) -0.085(3) 0.030(3) -0.308(3) -0.434(3) -0.555(3) -0.496(3) -0.199(2) 0.062(3) 0.187(3)

y 0.40059(1) 0.37090(9) 0.24511(9) 0.21774(9) 0.0927(2) 0.0696(2) 0.1708(2) 0.2906(2) 0.32033(9) 0.44335(9) 0.52590(9) -0.04326(9) -0.09971(8) -0.19924(9) -0.25490(9) -0.35691(9) -0.4106(2) -0.3661(2) -0.2710(2) -0.21086(9) -0.11257(9) 0.05508(18) 0.332(2) 0.179(2) 0.026(2) -0.0141(9) 0.154(2) 0.363(2) 0.512(2) -0.077(2) -0.231(2) -0.389(2) -0.480(2) -0.405(2) -0.240(2) -0.081(2)

z 0.01612(11) 0.08724(11) 0.17491(11) 0.24535(11) 0.33527(11) 0.40311(11) 0.38655(12) 0.30274(11) 0.22850(11) 0.14271(11) -0.07033(11) -0.04185(11) 0.05702(11) 0.11683(11) 0.21458(11) 0.27553(11) 0.37083(12) 0.41408(11) 0.35924(11) 0.25798(11) 0.20025(11) -0.10063(11) 0.0269(11) 0.1853(11) 0.3450(12) 0.4675(10) 0.4396(11) 0.2899(11) 0.1321(11) -0.0728(11) 0.0867(11) 0.2455(11) 0.4135(12) 0.4857(12) 0.3904(11) 0.2310(11)

Ueq (˚ A2 ) 0.0151(5) 0.0146(5) 0.0155(5) 0.0159(5) 0.0171(5) 0.0189(5) 0.0189(5) 0.0174(5) 0.0150(5) 0.0158(5) 0.0146(5) 0.0148(5) 0.0144(5) 0.0149(5) 0.0150(5) 0.0161(5) 0.0186(5) 0.0189(5) 0.0172(5) 0.0154(5) 0.0155(5) 0.0143(5) 0.022(4) 0.021(4) 0.030(5) 0.007(4) 0.019(4 0.023(4) 0.024(4) 0.020(4) 0.020(4) 0.022(4) 0.029(5) 0.018(4) 0.028(5) 0.019(4)

Table A.4: Fractional atomic coordinates of pentacene at 100 K

119

Pentacene — T = 293 K Chemical formula Formula weight Cell setting Space group a (˚ A) b (˚ A) ˚ c (A ) V (˚ A3 ) α β γ Z ρcalc (g cm−3 ) Crystal form, colour Crystal size (mm)

C22 H14 278.33 Triclinic P 1, 2 6.275(2) 7.791(2) 14.556(4) 688.7(3) 76.434(5)◦ 87.585(5)◦ 84.707(5)◦ 2 1.342 platelet, violet 0.430 × 0.210 × 0.050

Table A.5: Crystal data and details of the structure determination of pentacene at 293 K

Diffractometer Radiation type Monochromator Data collection method θ range (◦ ) Unique data Data with criterion F0 > 4σ(F0 ) Rint Range of h, k, l

Bruker Smart Apex; CCD area detector Mo Kα Graphite ϕ and ω scans 2.70 − 25.02 2376 1417 0.0269 −7 → h → 7 −9 → k → 9 −15 → l → 17

Table A.6: Data collection for pentacene at 293 K

120

Appendix A. Crystal structure of pentacene

No. of reflections No. of refined parameters Final agreement factors: Weighting scheme R(F ) wR(F 2 ), Goodness of Fit (S) Residual electron density ∆ρmin ∆ρmax

2376 255 0.0537 0.1722 1.022 −0.14 (e˚ A−3 ) 0.25 (e˚ A−3 )

Table A.7: Refinement for pentacene at 293 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 H1 H3 H5 H6 H7 H8 H10 H12 H14 H16 H17 H18 H19 H21

x 0.7001(3) 0.5428(3) 0.5768(3) 0.4212(3) 0.4551(4) 0.3000(3) 0.0956(4) 0.0554(3) 0.2134(3) 0.1775(3) 0.6654(3) -0.1807(3) -0.1557(3) -0.3056(3) -0.2795(3) -0.4302(3) -0.4006(4) -0.2173(4) -0.0691(3) -0.0920(3) 0.0554(3) -0.0314(3) 0.842(3) 0.720(3) 0.596(3) 0.326(3) -0.008(4) -0.082(3) 0.037(3) -0.311(3) -0.436(3) -0.557(3) -0.498(4) -0.199(3) 0.064(3) 0.186(3)

y 0.4027(2) 0.3722(2) 0.2485(2) 0.2209(2) 0.0983(3) 0.0755(3) 0.1745(3) 0.2916(3) 0.3212(2) 0.4426(2) 0.5264(2) -0.0442(2) -0.0996(2) -0.1988(2) -0.2532(2) -0.3546(2) -0.4067(3) -0.3616(3) -0.2673(3) -0.2083(2) -0.1107(2) 0.0539(2) 0.338(2) 0.184(2) 0.033(3) -0.004(3) 0.157(3) 0.359(3) 0.507(2) -0.075(2) -0.228(2) -0.387(2) -0.481(3) -0.401(3) -0.235(2) -0.080(2)

z 0.01486(13) 0.08631(13) 0.17321(13) 0.24388(13) 0.33302(15) 0.40070(16) 0.38514(16) 0.30244(15) 0.22806(13) 0.14323(13) -0.07069(12) -0.04027(13) 0.05751(13) 0.11743(13) 0.21392(13) 0.27544(15) 0.36951(16) 0.41105(15) 0.35623(14) 0.25603(13) 0.19804(13) -0.09933(13) 0.0273(12) 0.1848(12) 0.3435(13) 0.4650(15) 0.4381(15) 0.2904(12) 0.1312(11) -0.0693(12) 0.0882(12) 0.2455(13) 0.4120(14) 0.4813(14) 0.3850(12) 0.2262(11)

Ueq (˚ A2 ) 0.0321(6) 0.0313(6) 0.0341(7) 0.0338(7) 0.0420(7) 0.0480(8) 0.0476(8) 0.0410(7) 0.0337(7) 0.0343(7) 0.0309(6) 0.0317(6) 0.0300(6) 0.0327(6) 0.0331(6) 0.0389(7) 0.0472(8) 0.0464(8) 0.0407(7) 0.0344(7) 0.0340(7) 0.0302(6) 0.041(5) 0.041(5) 0.055(6) 0.057(6) 0.068(7) 0.052(6) 0.041(5) 0.041(5) 0.044(5) 0.053(6) 0.056(6) 0.053(6) 0.050(6) 0.036(5)

Table A.8: Fractional atomic coordinates of pentacene at 293 K

B

Crystal structure of rubrene

Rubrene — T = 100 K Chemical formula Formula weight Cell setting Space group a (˚ A) b (˚ A) c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.789(4) 7.170(1) 14.211(2) 2729.6(7) 4 1.296 4621 Platelet, orange 0.51 × 0.45 × 0.03

Table B.1: Crystal data and details of the structure determination of rubrene at 100 K

Diffractometer Radiation type Monochromator Data collection method θ range (◦ ) Tmin Tmax Total data Unique data Data with criterion F0 > 4σ(F0 ) Rint θmax (◦ ) Range of h, k, l

Bruker Smart Apex; CCD area detector Mo Kα Graphite ϕ and ω scans 2.9 − 29.3 0.952 0.998 10125 1424 1201 0.035 26.4 −31 → h → 33 −8 → k → 8 −17 → l → 17

Table B.2: Data collection for rubrene at 100 K

121

122

Appendix B. Crystal structure of rubrene

No. of reflections No. of refined parameters Final agreement factors: R[F 2 > 4(F 2 )], wR(F 2 ), Goodness of Fit (S) Residual electron density ∆ρmin ∆ρmax

1424 124 0.037 0.099 1.06 −0.21 (e˚ A−3 ) 0.26 (e˚ A−3 )

Table B.3: Refinement for rubrene at 100 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02737(6) 0.05314(5) 0.02678(5) 0.05191(5) 0.02646(5) 0.10760(5) 0.14292(5) 0.19199(5) 0.20685(5) 0.17219(5) 0.12302(5) 0.0883(5) 0.0438(5) 0.1328(5) 0.2167(5) 0.2413(6) 0.1825(6) 0.0978(5)

y 0.00000(-) -0.14464(16) -0.29578(16) -0.44754(17) -0.58934(17) -0.13448(16) -0.24817(18) -0.25089(19) -0.1389(2) -0.0251(2) -0.02362(18) -0.447(2) -0.689(2) -0.327(2) -0.332(2) -0.140(2) 0.054(2) 0.054(2)

z 0.00000(-) 0.04914(8) 0.08755(8) 0.13412(9) 0.17534(9) 0.07274(8) 0.02881(9) 0.05979(10) 0.13448(10) 0.17857(10) 0.14816(9) 0.1357(10) 0.2065(10) -0.0256(10) 0.0279(11) 0.1546(10) 0.2311(12) 0.1804(11)

Ueq (˚ A2 ) 0.0135(5) 0.0141(3) 0.0142(4) 0.0170(4) 0.0186(4) 0.0148(3) 0.0185(4) 0.0225(4) 0.0244(4) 0.0233(4) 0.0189(4) 0.016(4) 0.020(4) 0.021(4) 0.028(4) 0.025(4) 0.032(4) 0.021(4)

Table B.4: Fractional atomic coordinates of rubrene at 100 K

123

Rubrene — T = 125 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.789(4) 7.173(1) 14.246(2) 2737.5(7) 4 1.292 4800 Platelet, orange 0.51 × 0.45 × 0.03

Table B.5: Crystal data and details of the structure determination of rubrene at 125 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02723(7) 0.05315(5) 0.02670(5) 0.05192(6) 0.02634(6) 0.10755(5) 0.14287(5) 0.19178(6) 0.20663(6) 0.17202(6) 0.12299(6) 0.0882(6) 0.0442(5) 0.1324(6) 0.2170(6) 0.2416(7) 0.1817(6) 0.0976(6)

y 0.00000(-) -0.14435(18) -0.29541(18) -0.44699(19) -0.58862(18) -0.13436(18) -0.2473(2) -0.2505(2) -0.1387(2) -0.0252(2) -0.0236(2) -0.445(2) -0.688(2) -0.325(2) -0.330(2) -0.141(2) 0.056(2) 0.053(2)

z 0.00000(-) 0.04909(9) 0.08757(8) 0.13405(9) 0.17511(9) 0.07266(9) 0.02915(10) 0.05983(11) 0.13433(11) 0.17824(11) 0.14796(9) 0.1358(10) 0.2072(11) -0.0251(11) 0.0266(11) 0.1545(11) 0.2302(13) 0.1789(10)

Ueq (˚ A2 ) 0.0156(5) 0.0160(4) 0.0162(4) 0.0193(4) 0.0213(4) 0.0168(4) 0.0214(4) 0.0265(5) 0.0285(5) 0.0281(5) 0.0218(4) 0.022(4) 0.027(4) 0.023(4) 0.034(5) 0.033(4) 0.043(5) 0.019(4)

Table B.6: Fractional atomic coordinates of rubrene at 125 K

124

Appendix B. Crystal structure of rubrene

Rubrene — T = 150 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.775(4) 7.1680(10) 14.258(2) 2736.4(7) 4 1.293 4648 Platelet, orange 0.51 × 0.45 × 0.03

Table B.7: Crystal data and details of the structure determination of rubrene at 150 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02727(7) 0.05307(5) 0.02670(5) 0.05186(6) 0.02630(6) 0.10744(5) 0.14283(5) 0.19169(6) 0.20647(6) 0.17191(6) 0.12289(6) 0.0877(6) 0.0447(6) 0.1328(6) 0.2163(6) 0.2412(7) 0.1813(7) 0.0976(6)

y 0.00000(-) -0.14417(18) -0.29515(18) -0.44662(19) -0.58805(19) -0.13413(18) -0.2470(2) -0.2498(2) -0.1379(2) -0.0249(2) -0.0237(2) -0.448(2) -0.688(2) -0.324(2) -0.330(2) -0.142(2) 0.055(3) 0.053(2)

z 0.00000(-) 0.04907(9) 0.08757(8) 0.13409(9) 0.17505(10) 0.07267(9) 0.02912(10) 0.06004(12) 0.13425(11) 0.17798(11) 0.14788(10) 0.1341(11) 0.2071(11) -0.0250(11) 0.0272(12) 0.1537(12) 0.2297(14) 0.1778(11)

Ueq (˚ A2 ) 0.0172(5) 0.0177(4) 0.0181(4) 0.0220(4) 0.0242(4) 0.0191(4) 0.0242(4) 0.0306(5) 0.0335(5) 0.0324(5) 0.0252(4) 0.023(4) 0.031(4) 0.031(4) 0.038(5) 0.043(5) 0.048(5) 0.026(4)

Table B.8: Fractional atomic coordinates of rubrene at 150 K

125

Rubrene — T = 175 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.828(4) 7.181(1) 14.306(2) 2756.1(7) 4 1.284 4431 Platelet, orange 0.51 × 0.45 × 0.03

Table B.9: Crystal data and details of the structure determination of rubrene at 175 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02718(7) 0.05303(5) 0.02673(5) 0.05186(6) 0.02622(6) 0.10736(5) 0.14277(5) 0.19159(6) 0.20620(6) 0.17173(6) 0.12281(6) 0.0886(6) 0.0443(5) 0.1330(6) 0.2168(6) 0.2406(7) 0.1809(6) 0.0975(5)

y 0.00000(-) -0.14399(17) -0.29472(17) -0.44615(18) -0.58729(19) -0.13391(18) -0.24639(19) -0.2491(2) -0.1376(2) -0.0245(2) -0.0235(2) -0.447(2) -0.686(2) -0.324(2) -0.328(2) -0.139(2) 0.056(2) 0.053(2)

z 0.00000(-) 0.04909(8) 0.08762(8) 0.13392(9) 0.17496(9) 0.07267(9) 0.02938(10) 0.06005(11) 0.13407(11) 0.17770(11) 0.14762(9) 0.1352(10) 0.2063(10) -0.0237(11) 0.0272(11) 0.1527(11) 0.2295(13) 0.178(1)

Ueq (˚ A2 ) 0.0197(5) 0.0206(4) 0.0206(4) 0.0251(4) 0.0282(4) 0.0216(4) 0.0278(4) 0.0353(5) 0.0388(5) 0.0379(5) 0.0290(5) 0.031(4) 0.034(4) 0.035(4) 0.048(5) 0.043(5) 0.054(5) 0.027(4)

Table B.10: Fractional atomic coordinates of rubrene at 175 K

126

Appendix B. Crystal structure of rubrene

Rubrene — T = 200 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.838(4) 7.181(1) 14.332(2) 2762.1(7) 4 1.281 4251 Platelet, orange 0.51 × 0.45 × 0.03

Table B.11: Crystal data and details of the structure determination of rubrene at 200 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02718(7) 0.05301(5) 0.02666(5) 0.05171(6) 0.02627(6) 0.10730(5) 0.14271(6) 0.19149(6) 0.20606(7) 0.17164(7) 0.12264(6) 0.0880(7) 0.0447(6) 0.1325(6) 0.2169(7) 0.2407(8) 0.1809(8) 0.0971(6)

y 0.00000(-) -0.14396(19) -0.29437(19) -0.4456(2) -0.5864(2) -0.13351(19) -0.2461(2) -0.2484(3) -0.1373(3) -0.0244(3) -0.0233(2) -0.443(2) -0.686(2) -0.324(2) -0.328(3) -0.139(2) 0.057(3) 0.052(2)

z 0.00000(-) 0.04907(9) 0.08762(9) 0.13397(10) 0.17497(10) 0.07261(9) 0.02964(11) 0.06023(13) 0.13390(13) 0.17747(12) 0.1473(1) 0.1341(12) 0.2055(12) -0.0239(11) 0.0280(13) 0.1536(12) 0.2298(16) 0.1789(12)

Ueq (˚ A2 ) 0.0215(6) 0.0221(4) 0.0224(4) 0.0275(5) 0.0315(5) 0.0236(4) 0.0310(5) 0.0401(6) 0.0442(6) 0.0430(6) 0.0320(5) 0.034(5) 0.038(5) 0.032(4) 0.055(6) 0.053(5) 0.068(6) 0.035(4)

Table B.12: Fractional atomic coordinates of rubrene at 200 K

127

Rubrene — T = 235 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.818(5) 7.174(1) 14.348(3) 2760.5(9) 4 1.282 3849 Platelet, orange 0.51 × 0.45 × 0.03

Table B.13: Crystal data and details of the structure determination of rubrene at 235 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02725(7) 0.05282(5) 0.02673(5) 0.05163(6) 0.02624(7) 0.10725(6) 0.14260(6) 0.19131(6) 0.20583(7) 0.17143(7) 0.12234(7) 0.0880(6) 0.0443(6) 0.1326(6) 0.2163(7) 0.2410(8) 0.1796(7) 0.0971(7)

y 0.00000(-) -0.1437(2) -0.2940(2) -0.4448(2) -0.5852(2) -0.1332(2) -0.2455(3) -0.2471(3) -0.1364(3) -0.0242(3) -0.0230(3) -0.443(2) -0.682(3) -0.324(2) -0.325(3) -0.137(3) 0.052(3) 0.054(3)

z 0.00000(-) 0.04905(11) 0.08752(10) 0.13381(12) 0.17462(12) 0.07259(11) 0.02974(13) 0.06057(15) 0.13366(15) 0.17699(14) 0.14690(12) 0.1359(12) 0.2055(12) -0.0238(12) 0.0277(13) 0.1527(13) 0.2290(15) 0.1768(13)

Ueq (˚ A2 ) 0.0237(7) 0.0246(5) 0.0247(5) 0.0311(5) 0.0355(6) 0.0265(5) 0.0342(6) 0.0459(7) 0.0505(7) 0.0497(7) 0.0364(6) 0.033(5) 0.042(5) 0.032(5) 0.055(6) 0.060(6) 0.060(6) 0.041(5)

Table B.14: Fractional atomic coordinates of rubrene at 235 K

128

Appendix B. Crystal structure of rubrene

Rubrene — T = 275 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.938(5) 7.211(1) 14.461(3) 2809.1(9) 4 1.259 3697 Platelet, orange 0.51 × 0.45 × 0.03

Table B.15: Crystal data and details of the structure determination of rubrene at 275 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02714(7) 0.05290(5) 0.02661(5) 0.05169(7) 0.02612(7) 0.10704(6) 0.14248(6) 0.19109(7) 0.20544(8) 0.17115(8) 0.12231(7) 0.0890(7) 0.0450(6) 0.1322(6) 0.2160(7) 0.2404(9) 0.1794(8) 0.0964(7)

y 0.00000(-) -0.14318(18) -0.29299(18) -0.4439(2) -0.5841(2) -0.13275(19) -0.2446(2) -0.2464(3) -0.1361(3) -0.0238(3) -0.0226(2) -0.441(2) -0.684(2) -0.324(2) -0.324(3) -0.138(3) 0.057(3) 0.050(2)

z 0.00000(-) 0.04901(9) 0.08757(9) 0.13369(10) 0.17468(11) 0.07255(10) 0.03009(12) 0.06060(15) 0.13363(15) 0.17662(14) 0.14648(11) 0.1366(12) 0.2061(12) -0.0219(12) 0.0268(14) 0.1536(15) 0.2296(16) 0.1780(13)

Ueq (˚ A2 ) 0.0294(6) 0.0301(4) 0.0310(4) 0.0387(5) 0.0433(5) 0.0329(4) 0.0425(6) 0.0560(7) 0.0626(8) 0.0599(7) 0.0449(6) 0.053(5) 0.053(5) 0.048(5) 0.078(6) 0.083(7) 0.085(7) 0.054(5)

Table B.16: Fractional atomic coordinates of rubrene at 275 K

129

Rubrene — T = 293 K Chemical formula Formula weight Cell setting Space group a (˚ A) A) b (˚ c (˚ A) V (˚ A3 ) Z ρcalc (g cm−3 ) No. of reflections for cell param. Crystal form, colour Crystal size (mm)

C42 H28 532.68 Orthorhombic Cmca 26.86(1) 7.193(3) 14.433(5) 2788.5(18) 4 1.269 1924 Platelet, orange 0.51 × 0.45 × 0.03

Table B.17: Crystal data and details of the structure determination of rubrene at 293 K

Atom C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 H4 H5 H7 H8 H9 H10 H11

x 0.02705(7) 0.05294(5) 0.02658(5) 0.05147(7) 0.02589(7) 0.10704(6) 0.14252(6) 0.19095(7) 0.20534(8) 0.17085(8) 0.12213(7) 0.0875(6) 0.0444(5) 0.1318(6) 0.2165(7) 0.2409(8) 0.1794(8) 0.0955(6)

y 0.00(-) -0.1429(2) -0.2928(2) -0.4433(2) -0.5831(2) -0.1325(2) -0.2440(2) -0.2455(3) -0.1351(3) -0.0241(3) -0.0226(3) -0.441(2) -0.680(2) -0.324(2) -0.321(3) -0.133(2) 0.060(3) 0.055(2)

z 0.00(-) 0.04897(10) 0.08756(10) 0.13347(11) 0.17451(12) 0.07249(11) 0.03033(13) 0.06085(16) 0.13354(16) 0.17627(15) 0.14638(12) 0.1339(10) 0.2050(12) -0.0213(12) 0.0286(14) 0.1504(12) 0.2267(16) 0.1763(12)

Ueq (˚ A2 ) 0.0327(7) 0.0335(5) 0.0341(5) 0.0424(6) 0.0475(6) 0.0369(5) 0.0467(6) 0.0623(8) 0.0694(9) 0.0659(8) 0.0499(6) 0.036(4) 0.055(5) 0.053(5) 0.085(7) 0.072(6) 0.091(7) 0.054(5)

Table B.18: Fractional atomic coordinates of rubrene at 293 K

Summary

Organic materials are targeted to be implemented as new semiconductors in plastic electronic components for commercial devices. They offer advantages with respect to traditional inorganic semiconductors regarding processability (they require simple techniques such as spin coating, printing) and functionality (modification of the molecules offers enormous variety in functionalization). Rapid progress is being made in the development of organic electronic devices. The use of organic field-effect devices for large scale applications requires high performance, reliability and stability, long lifetimes, good control and reproducibility. One measure of the performance of a semiconductor is the electronic mobility. This is a benchmark for the success of the different organic semiconductors. The achievement of good transport properties is associated with high mobilities. The work presented in this thesis aims to take a step forward towards the understanding of the microscopic processes that determine the electronic mobility in molecular crystals: the orientation of the molecules, intermolecular interactions, defect densities and charge transport. This thesis incorporates the study of the properties of organic molecular crystals at the materials level (Chapters 2, 4, 5, 6), as well as in devices (Chapter 3). The primary goals of our work are to be able to relate the molecular stacking, defect density and device geometry with the electronic properties and further to apply that knowledge in the design of electronic devices with enhanced performance. Past investigations have been hampered by low reproducibility of the exper131

132

Summary

imental results, and by structural defects that prevented measurements of the intrinsic properties of organic materials. In recent years, an unprecedented control of the structural properties at the molecular level and of the charge density distribution has been achieved by better understanding the properties of the organic semiconductor material, and by careful design of device structures. In this general context, we study single crystals of pentacene and rubrene, molecular conductors with the highest reported mobilities. Although in single crystal form they may never match the general requirements for incorporation in large scale applications, they provide unique tool for the investigation of intrinsic properties and model systems. For the materials that we focus on, we perform the crystal growth and crystal structure determination by X-ray diffraction. For pentacene we also investigate the chemical properties using IR spectroscopy, UV-Vis and mass spectroscopy combined with high performance liquid chromatography (HPLC). We use space charge limited current (SCLC) and field-effect transistor (FET) measurements to study the charge transport properties. In Chapter 2 we demonstrate that the purity of the material is critical for highperformance electronic devices. Impurities can form electrically active states in the band gap of semiconductors, or distort the charge distribution in the surrounding crystal introducing states in the gap. Thus, they strongly influence the electronic properties of organic crystals. We perform a quantitative analysis of the pentacenequinone content, the major impurity in pentacene single crystals. Furthermore, we show that the reduction of its concentration by a factor of five (from 0.11% to 0.028%) leads to a decrease of the number of electronically active traps by two orders of magnitude (from Nt = 1013 cm−3 to Nt = 1011 cm−3 ) and a high mobility (µ = 35 cm2 /Vs). In the analysis, we incorporate corrections for the effective thickness of the crystal that result from the anisotropic resistivity and from the distribution of the electric field inside the material. A detailed description of the algorithm is incorporated in Chapter 4. In Chapter 3 we demonstrate that the pentacenequinone impurity is located preferentially at the surface of the crystal and strongly affects the conduction in FET devices by scattering the charges. We overcame this limitation by converting the impurity scattering centers into the gate dielectric. The transistors built in this way are characterized by on/off ratios of 106 and mobilities that reproduce the values determined for the bulk. Organic electronic devices need to operate exposed to air. Thus, the influence of moisture and oxygen are critical in their behavior. Our work concentrates on improving the performance and reliability of organic based devices. We investigate the physical processes that govern the electrical conduction in these materials

Summary

133

in the presence of air. We demonstrate in Chapter 5 that oxygen/water vapor present in air diffuses into the crystal. This process generates holes/traps, leading to enhancement/decrease of the conductivity. The two effects are opposite, they appear together, and they can have comparable magnitudes. By combining electrical and gravimetric measurements, we obtain quantitative information about these effects. We find that on average two O2 molecules are needed in the presence of light to create one hole, compared with four molecules in the dark. For rubrene single crystals, in Chapter 6 we relate the dramatic changes in electronic mobility at 175 K to modifications in the crystal structure via the transfer integrals between neighboring molecules. DSC measurements show evidence for a phase transition at 175 K, but this is not detected by our X-ray diffraction experiments. We have obtained high electronic mobilities in organic crystals by careful control of the material degree of purity. Furthermore, we have successfully fabricated and measured organic FETs in which we are able to reproduce the high mobilities evaluated from the experiments that describe the bulk properties. Our interests involve a broad area of research, and combine crystal growth, crystal structure determination and charge transport measurements. The work presented in this thesis gives insight into physical processes such as the generation and migration of defects in organic crystals, environmental effects and device geometry effects that influence the electrical conduction in molecular crystals.

Samenvatting

Organische materialen komen in aanmerking voor nieuwe halfgeleidertoepassingen in plastic elektronica componenten in commerci¨ele apparatuur. Deze materialen bieden voordelen ten opzichte van de gangbare anorganische halfgeleiders als het gaat om verwerking (er zijn slechts relatief simpele technieken zoals spin coaten en printen nodig) en functionaliteit (modificatie van de moleculen biedt een rijke vari¨eteit in functie). Veel vooruitgang is reeds geboekt in de ontwikkeling van organische elektronische devices. Grootschalig gebruik van organische Veldeffecttransistoren houdt in dat hoge eisen moeten worden gesteld aan hun prestaties, betrouwbaarheid, stabiliteit, levensduur, beheersbaarheid en reproduceerbaarheid. Een maat voor de prestatie van een halfgeleider is de beweeglijkheid van zijn ladingsdragers. Dit is een graadmeter voor het succes van verschillende organische halfgeleiders. Het tot stand brengen van goede transporteigenschappen is inherent aan het bereiken van hoge mobiliteiten. Het onderzoek dat in dit proefschrift aan de orde komt, stelt zich ten doel om te komen tot een beter begrip van de microscopische processen die ten grondslag liggen aan de elektrische geleiding in moleculaire kristallen: het verband tussen moleculen, intermoleculaire interacties, defect dichtheden en ladingstransport. Dit proefschrift omvat naast onderzoek naar eigenschappen van organische moleculaire kristallen op het gebied van materialen (Hoofdstukken 2, 4, 5, 6), tevens onderzoek naar hun eigenschappen wanneer zij zijn ingebouwd in devices (Hoofdstuk 3). De kerndoelen van 135

136

Samenvatting

ons onderzoek zijn enerzijds het in staat zijn om moleculaire stapeling, defect dichtheid en device opbouw in verband te brengen met elektronische eigenschappen en anderzijds om die kennis toe te passen in het ontwerp van elektronische devices met verbeterde prestaties. Onderzoek in het verleden werd bemoeilijkt door de slechte reproduceerbaarheid van meetresultaten en de aanwezigheid van structuurdefecten, zodat de metingen aan intrinsieke eigenschappen ontoegankelijk waren. In de laatste jaren is een ongekende beheersing mogelijk van zowel de structuureigenschappen op moleculaire schaal als van de ladingsdichtheidsverdeling door een beter begrip van materiaaleigenschappen van organische halfgeleiders en een zorgvuldig ontwerp van de device opbouw. Binnen deze algemene context bestuderen we in het bijzonder ´e´enkristallen van pentaceen en rubreen, moleculaire geleiders met de tot dusver hoogst gerapporteerde mobiliteiten. Hoewel dergelijke materialen in de hoedanigheid van ´e´enkristal nooit aan de algemeen geldende eisen voor toepassing op grote schaal kunnen voldoen, verschaffen deze ´e´enkristallen ons toch een uniek instrumentarium voor onderzoek naar intrinsieke eigenschappen en bieden zij ons tevens model systemen. Van de materialen waar wij ons op richten, groeien we ´e´enkristallen en bepalen hun kristalstructuur door middel van R¨ontgendiffractie. Verder doen wij onderzoek naar de chemische en fysische eigenschappen van pentaceen. De bepaling van dergelijke eigenschappen berust op het gebruik van infrarood-, UVVis- en massaspectroscopie in combinatie met vloeistofchromatografie met hoog oplossend vermogen. We maken gebruik van ruimteladingsbegrensde stroom- en veldeffecttransistor metingen om de ladingstransporteigenschappen te bestuderen. In hoofdstuk 2 tonen we aan dat de zuiverheid van het materiaal een sterke invloed heeft op de prestaties van de elektronische devices. Onzuiverheden kunnen elektrisch actieve toestanden in de band gap van halfgeleiders introduceren, of de ladingsverdeling in het aangrenzende kristalvolume verstoren. Op die manier benvloeden zij in sterke mate de elektronische eigenschappen van organische kristallen. We verrichten een kwantitatieve analyse op pentaceenchinon, de belangrijkste onzuiverheid in pentaceen ´e´enkristallen. Bovendien tonen we aan dat het terugbrengen van de concentratie van deze onzuiverheid met een factor vijf (van 0.11% naar 0.028%) resulteert in een afname met twee ordes van grootte van het aantal afgevangen ladingsdragers (van Nt = 1013 cm−3 naar Nt = 1011 cm−3 ) en ook de mobiliteit hoger wordt (µ = 35 cm2 /Vs). In de analyse corrigeren we voor de effectieve kristaldikte die wordt veroorzaakt door de anisotrope weerstand en de verdeling van het elektrische veld binnen het materiaal. Een gedetailleerde beschrijving van het algoritme wordt behandeld in hoofdstuk 4. In hoofdstuk 3 komt aan bod dat pentaceenchinon onzuiverheden

Samenvatting

137

zich bij voorkeur aan het kristaloppervlak bevinden en bovendien daar de geleiding in sterke mate be¨ınvloeden door ladingsdragers te verstrooien. Door onzuiverheden die fungeren als strooicentra om te vormen tot gate di¨electricum, kan deze belemmering worden verholpen. De transistoren die op deze manier vervaardigd zijn, hebben een aan/uit verhouding van 106 en mobiliteiten die overeenstemmen met de waarden die zijn gevonden voor het bulk materiaal. Organische elektronische devices moeten hun taak kunnen blijven vervullen wanneer ze worden blootgesteld aan de lucht. Daarom is de invloed van vocht en zuurstof van belang voor hun gedrag. Ons werk richt zich op het verbeteren van de prestaties en betrouwbaarheid van organische devices. Hiertoe onderzoeken we de fysische processen die de elektrische eigenschappen van deze materialen bepalen in de aanwezigheid van lucht. We tonen aan in hoofdstuk 5 dat de zuurstof/waterdamp die aanwezig is in de lucht diffundeert in het kristal. Dit diffusieproces genereert gaten/vallen, wat leidt tot verbetering/verslechtering van de geleiding. De twee effecten zijn in competitie, treden gezamenlijk op en kunnen van vergelijkbare orde van grootte zijn. Door elektrische en thermogravische metingen te combineren verschaffen we ons een kwantitatief beeld van deze effecten. We hebben vastgesteld dat er in de aanwezigheid van licht gemiddeld twee O2 moleculen nodig om een gat te cre¨eren, vergeleken met vier moleculen in het donker. Hoofdstuk 6 behandelt rubreen ´e´enkristallen. We brengen de enorme verandering in mobiliteit bij 175 K in verband met veranderingen in kristalstructuur met behulp van overdrachtsmatrixelement tussen naburige moleculen. DSC metingen duiden op een fase overgang bij 175 K, maar hiervoor zijn geen aanwijzingen gevonden in een R¨ ontgendiffractie studie. We hebben hoge mobiliteiten bereikt door een uitgebalanceerde beheersing van de zuiverheidsgraad van de gebruikte materialen. Verder hebben we op succesvolle wijze devices vervaardigd en organische FETs gemeten die in staat waren de hoge mobiliteiten te reproduceren zoals die gevonden waren in experimenten die de bulk eigenschappen in kaart plegen te brengen. Onze belangstelling gaat uit naar een breed onderzoeksgebied en combineert kristalgroei, kristal structuurbepaling en ladingstransport metingen. Het werk dat in dit proefschrift aan de aan de orde komt, geeft inzicht in de fysische processen zoals het ontstaan en de migratie van defecten in organische kristallen, omgevingseffecten en effecten op de electrische geleiding van moleculaire kristallen die gerelateerd zijn aan de device opbouw.

List of Publications

• B. B. Van Aken, O. D. Jurchescu, A. Meetsma, Y. Tomioka, Y. Tokura, and T. T. M. Palstra, Orbital-order-induced metal-insulator transition in La1−x Cax MnO3 , Phys. Rev. Lett. 90, 066403 (2003). • 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).

• 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).

• 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).

• O. D. Jurchescu, and T. T. M. Palstra, Crossover from one- to twodimensional space-charge-limited conduction in pentacene single crystals, Appl. Phys. Lett. 88, 122101 (2006).

139

140

List of publications

• O. D. Jurchescu, A. Meetsma, and T. T. M. Palstra, Low-temperature structure of rubrene single crystals grown by vapor transport, Acta Cryst. B 62, 330 (2006).

• O. Ostroverkhova, D. G. Cooke, F. A. Hegmann, J. E. Anthony, V. Podzorov, M. E. Gershenson, O. D. Jurchescu, and T. T. M. Palstra, Ultrafast carrier dynamics in pentacene, functionalized pentacene, tetracene, and rubrene single crystals, Appl. Phys. Lett. 88, 162101 (2006).

• O. D. Jurchescu, M. Popinciuc, B. J. van Wees, and T. T. M. Palstra, Interface controlled high-mobility organic transistors, submitted.

• A. J. C. Buurma, I. Shokaryev, O. D. Jurchescu, A. Meetsma, R. A. de Groot, and T. T. M. Palstra, Crystal growth, structure and electronic band structure of Tetracene-TCNQ, submitted.

• I. Shokaryev, A. J. C. Buurma, O. D. Jurchescu , G. A. de Wijs, T. T. M. Palstra, R. A. de Groot, The electronic band structure of Tetracene - TCNQ and similar acene - TCNQ compounds, in preparation.

Propositions belonging to the thesis: Molecular Organic Semiconductors for Electronic Devices 1. The purity of the molecular organic semiconductor is crucial to achieve high-performance and high-reliability electronic device operation. Chapter 2 of this thesis 2. The greatest challenges at the present stage of the organic electronics research, is device processability rather than new materials. Chapter 3 of this thesis 3. Oxygen diffuses reversibly into pentacene crystals and on average four O2 molecules induce the formation of one charge carrier in the dark. Chapter 5 of this thesis 4. Intentionally use of unnecessary complicated terminology does not imply valuable scientific content. 5. Science results as well as open source software are both communal and competitive: in both cases the property right issue is almost entirely a matter of respecting the authorship of the original work. 6. Contrary to the expectations of some politicians, global warming is the greatest threat to humanity during the next millennium. 7. The impact of an article in the field is more important that the impact factor of the journal in which it is published. 8. In science lately there are more opinions than facts. 9. ”Your focus determines your reality.” Star Wars 10. The mistakes that are most easy to forgive are your own.

Oana Diana Jurchescu October 2006 These propositions are considered defendable and as such have been approved by the supervisor, prof. dr. T.T.M. Palstra.