Chemical doping of organic field-effect transistors

Chemical doping of organic field-effect transistors

F. van Seijen

Master’s thesis in applied physics

August 10, 2010

Supervision: D.M. de Leeuw and J.J. Brondijk

Abstract

Two doping methods for organic field-effect transistors (OFETs) are investigated to get an understanding of the doping mechanisms. Regioregular poly(3-hexylthiophene-5,6-diyl) (P3HT) is doped in solution by adding 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) molecules in different concentrations. The changes in the electrical properties of the devices are mapped out. Furthermore P3HT FETs are exposed to a silane-gas (TCFOS) and also from these devices the changes in electrical properties are examined. We also investigated the dedoping process by keeping the devices in vacuum and annealing them. It turns out that TCFOS doped devices are stable in time when they are fully doped. This is probably caused by the crosslinking of the silanol groups.

Contents

1 Introduction 3

2 Theory 4

2.1 Organic field-effect transistor . . . 4

2.2 Characteristics of an OFET . . . 7

2.3 Doped organic field-effect transistor . . . 8

2.3.1 Doping in solution . . . 8

2.3.2 P-type doping with an acid . . . 9

2.4 Determination of dopant density from the IV -curve . . . 9

2.5 Determination of the bulk mobility . . . 13

2.6 Determination of the doping efficiency . . . 13

3 Experimental 15 3.1 Devices . . . 15

3.2 Doping in solution . . . 16

3.3 Doping with TCFOS . . . 16

3.4 Measurements . . . 16

4 Results and Discussion 17 4.1 Doping with DDQ . . . 17

4.2 Doping with TCFOS . . . 20

4.3 Thermally dedoping and evacuation of doped devices . . . 27

4.3.1 Dedoping of DDQ doped devices . . . 27

4.3.2 Dedoping of TCFOS doped devices . . . 27

5 Conclusions and outlook 34 5.1 Conclusions . . . 34

5.2 Outlook . . . 34

Chapter 1

Introduction

Since the discovery of highly conducting polymers, chemically doped polyacetylene, in 1977 [1]

huge research effort is done for electrical applications of conjugated polymers. The advantages of plastics that they are strong, lightweight and cheap make them interesting for several electrical applications. The combination of these properties with the opto-electronic functionality of polymers make them suitable for various opto-electronic devices, such as light-emitting diodes, solar cells and integrated circuits. One of the newer research fields for organic semiconductors is the use of field- effect transistors as chemical and biological sensors. The ability to transduce chemical reactions into electrical signals make OFETs very suitable for a broad range of detection applications. However, to make a reliable sensor the characteristic changes in the OFET due to the reactions should be known. In this research project we investigated the response of a polymer to chemically added dopant molecules. We used two different methods for this investigation. Firstly we doped the polymer by adding dopant molecules already in solution to get a better understanding of the doping process. Secondly we doped OFETs by exposing them to an acidic gas. In this way we investigated the response of an OFET exposed to gasses, for the use as a gas sensor.

Chapter 2

Theory

2.1 Organic field-effect transistor

An organic field-effect transistor (OFET), also called an organic thin-film-transistor (OTFT) is a transistor with a similar structure as transistors made of inorganic materials, MOS-FETs (Metal oxide semiconductor-FETs). The structure of a TFT is shown in figure 2.1. A gate electrode is seperated from the semiconductor by an insulating layer. Two other electrodes, the source and the drain electrode, are in direct contact with the semiconductor.

Gate Insulator

Source Drain

Semiconductor

Figure 2.1: A field-effect transistor.

In figure 2.2 the working mechanism of a p-type TFT is shown schematically. If there is 0 V applied to all three contacts of the transistor (figure 2.2(a)), there is a small amount of positive charges uniformly distributed in the semiconductor, the bulk charge. If a negative drain bias, V_d, is applied, an electric field will arise and charge carriers will move through the semiconductor, resulting in a small and ohmic source-drain current, I_sd, corresponding to the conductivity σ of the semiconductor [4]:

σ ∼=

 L

Zdsemi

Isd

Vd

  _V

g=0,V_d→0 (2.1)

with L the channel length, Z the channel width, and dsemithe thickness of the semiconductor.

If a negative gate bias, Vg, is applied (figure 2.2(b)) positive charges, supplied by the source and drain contacts, are accumulated at the semiconductor/insulator interface (accumulation region).

+ + +

+ ++

+ ++ + + + +

+ + +

+ + + + + ++ + ++ + + +

++

+ ++ + + + + + + + + +

+

++ + + + ++ +

+ +

(a) Vg= Vd= Vs= 0 V + +

+ + ++

+ ++ + + + +

+ + +

+ + + + + ++ + ++ + + +

++

+ ++ + + + + + + + + +

+

+ + + + + ++ +

+ +

- - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

(b) Vg< 0 V, Vs= Vd= 0 V + +

+ +

+ ++

++ + +

+ +

+ + + ++ + +

+ + + + + + +

+

++ ++

+ +

---

++++++++++++++++++++++++++++++++++++++++++++++

(c) Vg> 0 V, Vs= Vd= 0 V + +

+ + ++

+ ++ + + + +

+ + +

+ + + + + ++ + ++ + + +

++

+ ++ + + + + + + + + +

+

++ + + + + + +

+ +

-- - - - - - - - - - - - - -

++ + + + + + + + + + + + + + + + + +

(d) Vg< Vd< 0 V, Vs= 0 V

Figure 2.2: The working mechanism of a field-effect transistor with a p-type semiconductor. + indicates positive charges. 2.2(a) The semiconductor only contains bulk charge. 2.2(b) Additional positive charges, supplied by the source and drain contacts are accumulated at the semiconductor/insulator interface by a negative gate voltage, Vg. 2.2(c) A positive Vg results in a depletion regime, where the positive charge carriers are depleted out of the semiconductor. With increasing Vg the depletion width increases until eventually all the positive charge carriers are depleted out of the semiconductor. 2.2(d) A negative Vd

results in non-uniformly located charges through the conducting channel. The density of charge carriers decreases from source to drain.

Figure 2.3: The band bending of the semiconductor in an organic field-effect transistor. Figure from [5].

This phenomenon can be explained by the energy band diagram of the gate-metal, insulator and semiconductor (see figure 2.3). When the Fermi energy levels, Ef, of the metal and the semicon- ductor are equal, the energy bands of the semiconductor are flat (flat-band condition). If a negative voltage is applied to the metal, the energy bands of the semiconductor will bend upwards at the semiconductor/insulator interface in an attempt to again align with the energy bands of the metal.

Hence additional positive charges are accumulated at the semiconductor/insulator interface. On the other hand, if a positive voltage is applied to the metal, opposite band bending will occur and all the positive charges will be depleted from the semiconductor/insulator interface (depletion regime). Most semiconductors do not have energy bands aligned with the metal, so there will be accumulation or depletion already in the initial state. In order to reach the flat-band condition a voltage equal to the difference in Fermi levels (the flat-band voltage, V_{f b}), should be applied to the gate.

In accumulation, the accumulated charge per unit area is VgCi, where Ci is the capacitance per unit area of the insulator. An equal amount of negative charge is stored on the gate at the gate/insulator interface. The semiconductor now contains two kinds of positive charges, the bulk charge and the charge from the ’field-effect’. If the latter kind is mobile and not trapped and a negative drain bias, Vd, is applied, the field-effect charge will contribute to a larger source-drain current than without gate bias. If a small incremental bias, δVg, is applied, this will give rise to an incremental charge increase of ZLCiδVg. The total incremental current δIsd is then [4]:

δIsd= Z

LµCiVdδVg (2.2)

where µ is the mobility, the ease with which the accumulated charge can move under the influence of an electric field. If the drain current is measured as a function of the gate bias at low drain biases (the linear regime), the transconductance, gm, is

g_m= ∂I_sd

∂Vg

= Z

LµC_iV_d. (2.3)

The field-effect mobility of the accumulated charge as a function of the gate bias, may then be

calculated as:

µ(Vg) = L ZC_iV_d

∂Isd

∂V_g. (2.4)

If a positive voltage is applied to the gate as shown in figure 2.2(c), opposite band bending occurs (see figure 2.3) and the positive bulk charges will flow out of the semiconductor/insulator interface via the source-drain contacts. The region without positive charges is called the depletion zone. The width of this zone, Wdepl, will increase with increasing positive gate bias, until eventually the entire thickness of the semiconductor is depleted.

If a large negative gate bias and a small negative drain bias are applied the charge accumulated in the semiconductor will be uniformily spread over the conducting channel. This is called the linear regime of the FET. However, if the negative drain bias is increased, the accumulated charge will be a function of the position along the channel. At the source electrode the charge will remain the same, but at the drain electrode the voltage drop will decrease, which results in a lower accumulated charge density. The accumulated charge density will thus decrease from source to drain electrode, as shown in figure 2.2(d). When |Vd| > |Vg| the FET is in the saturation regime and positive charge carriers will be depleted out of the region near the drain electrode.

2.2 Characteristics of an OFET

For the characterization of an organic field-effect transistor several parameters are of importance.

One of the most important characterization parameters is the switch-on voltage Vso. The switch-on voltage is defined as the gate voltage at which there is no band bending in the semiconductor, also known as the flat-band voltage in figure 2.3. Below V_sothere is no variation of the channel current with V_g, while above V_sothe channel current increases with V_g.

Another important parameter is the field-effect mobility as mentioned in equation 2.4. The mobility of the semiconductor device is defined as the ease of the charge transport in the semicon- ductor. The charge transport in an organic semiconductor occurs by the hopping of charges between localized sites of the polymer chain. The ease of this process depends on the charge carrier density and thus on the distance between the hopping sites, which depends on the structure of the polymer.

In a highly disordered polymer, the average distance between localized sites is relatively big, while in a more ordered polymer the distance between sites will be smaller and hence the hopping will be easier. In the calculation of the mobility the charge-carrier density p should therefore be taken into account.

There are several theories about the behaviour of µ. One of the theories is described in [6]. Here a description of the mobility dependence on the charge carrier density and the temperature T is given:

µ(p, T ) = µ_h(0, T ) + σ0

e

(T0/T )⁴sin(πT /T0) (2α)³Bc

T0/T

p^{T0/T −1}, (2.5)

where σ0is a prefactor for the conductivity, e is the elementary charge, α⁻¹ is the effective overlap parameter between localized states, T0is a measure of the width of the exponential density of states, and Bc is the critical number for the onset of percolation.

Another theory about the behaviour of µ is developed in [7]. Here a unified description of the full T , p and E (electric field) dependence of µ is derived. First the mobility is determined from

the master equation representing hopping of charge carriers:

X

j6=i

[W_ijp_i(1 − p_j) − W_jip_j(1 − p_i)] = 0, (2.6)

where pi is the probability that site i is occupied by a charge and Wij is the transition rate for hopping from site i to site j. Equation 2.6 is solved by an iteration procedure, which gives a mobility µ as µ = Σi,jWijpi(1 − pj)Rijx/pEV , with Rijx the distance between sites i and j in the x-direction, p = hpii/a³ (with a the lattice constant of a regular cubic lattice) and V the system volume. The p dependence of µ for different temperatures is parametrized by the following scheme:

µ(T, p) = µ0(T ) exp[1

2(ˆσ²− ˆσ)(2pa³)^δ], (2.7a) µ0(T ) = µ0c1exp[−c2ˆσ²], (2.7b) ,

δ ≡ 2ln(ˆσ²− ˆσ − ln(ln 4)) ˆ

σ² , µ₀≡ a²ν0e

σ , (2.7c)

with ˆσ ≡ kBT (with σ the width of a Gaussian and kBthe Boltzmann constant), c1= 1, 8×10⁻⁹and c2= 0.42. Now only the E dependence is left to be taken into account. This can be approximately modeled by a density-independent prefactor f (T, E):

µ(T, p, E) ≈ µ(T, p)f (T, E), (2.8)

with the prefactor being:

f (T, E) = expn

0.44(ˆσ^3/2− 2.2)hr

1 + 0.8Eea σ

2

− 1io

. (2.9)

Although both equations 2.5, 2.8 describe the behaviour of the mobility of a semiconductor, for the field-effect mobility of an organic FET we can use equation 2.4 to calculate this mobility.

2.3 Doped organic field-effect transistor

Organic field-effect transistors are very sensitive for their environment. An acid or oxygen molecule can already dope the semiconductor and thus change the characteristics of the semiconductor.

When a dopant molecule is added to the polymer, a chemical reaction takes place, resulting in extra charge carriers in the semiconductor. The additional charge carriers will be distributed in the entire semiconductor and will change the electrical properties of the device. In this research project we have investigated the changes in the electrical properties of OFETs. We have used two doping methods, which will be explained later in this chapter; doping in solution and exposing the device to an acidic gas.

2.3.1 Doping in solution

To accomplish p-type doping in a semiconductor, free positive charge carriers should be added to or electrons should be removed from the semiconductor in the device. This can be done by adding a dopant molecule with high reduction potential (the tendency of acquiring electrons). The addition of

(a) P3HT (b) F4-TCNQ (c) DDQ Figure 2.4: The structureformulas of (a) P3HT, (b) F4-TCNQ, (c) DDQ.

the dopant will cause a redox reaction between the dopant molecule and the polymer. The polymer will repel electrons which will be accepted by the dopant molecules. In this report the polymer regioregular poly(3-hexylthiophene-5,6-diyl) (P3HT) with a reduction potential between −0, 6 eV and −1, 0 eV ([10]) and the dopant molecule 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) are used. At first the dopant molecule 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F4- TCNQ) was used, but although theoretically this molecule should dope very well, experimentally we could not observe any doping. The structure formulas of the molecules used are shown in figure 2.4.

DDQ is a large organic molecule and is known as a strong oxidizing agent. It has two oxygen atoms double-bonded to its benzene ring, which makes it a divalent dopant (every DDQ molecule can accept two electrons). The redox reactions of DDQ and P3HT are shown in figure 2.5.

2.3.2 P-type doping with an acid

Another method of doping a polymer semiconductor is to dope it with an acid. In this case a reaction between the silane molecule and water takes place resulting in silanol and HCl. The H⁺ of the HCl or the OH group of silanol reacts with P3HT, thereby adding an extra positive charge carrier. In our experiments we used vaporized trichloro(1H,1H,2H,2H-perfluorooctyl)silane (TCFOS), of which the structure formula is shown in figure 2.6.

2.4 Determination of dopant density from the IV -curve

When a transistor is doped, extra charge carriers are added to the semiconductor layer. These charge carriers will be located in the conducting channel as well as in the bulk of the semiconductor.

Because of the extra charge carriers two changes in the transfer curve (the plot of the drain current versus the gate voltage) will occur. The first feature is the appearance of an additional current, in the form of a shoulder, in the depletion regime of the transistor. This current flows in the bulk layer

(a) Half reaction of DDQ.

(b) Half reaction of P3HT.

Figure 2.5: The two half reactions of DDQ and P3HT. The two double bonded oxygen atoms make DDQ a divalent dopant. The reduction potential P3HT is between −0, 6 and −1, 0 eV.

Figure 2.6: The structure formula of TCFOS.

-40 -20 0 20 40 1^x10^-10

1^x10^-9 1^x10^-8 1^x10^-7 1^x10^-6 1^x10^-5

no silane t=1 minute t=10 minutes t=20 minutes

t=1 minute, without pinch-off voltage

Vg (V) I sd

(A)

Vpinch ΔVso

Figure 2.7: Transfer characteristics of a P3HT transistor doped with TCFOS in time. The shoulder in the transfer curve indicates extra charge in the bulk. Also observable is the shift of the switch-on voltage.

and originates from the increase in conductivity in the bulk and the charge dependent mobility.

The additional current will thus increase with dopant density. The voltage at which all the charge carriers are depleted from the semiconductor layer and at which no more current will flow is called the pinch-off voltage, Vpinch.

The second feature that appears is a shift of the swith-on voltage, ∆Vso, to more positive gate voltages. This shift is caused by interfacial charges at the semiconductor/insulator interface. Both features are shown in figure 2.7.

From the pinch-off voltage we can directly determine the amount of doping added to the semi- conductor. First, we have to correct for the observed shift in V_soas in figure 2.8. We can do this by shifting the transfer characteristics along the V_g axis until the field-effect current lies on top of the original undoped curve. The initial curve is taken as a reference measurement and Vsocan be determined. We can now also estimate Vpinch with respect to Vso.

For small Vd we assume that the depletion of the semiconductor takes place uniformly over the

-60 -40 -20 0 20 40 60 1x10^-10

1x10^-9 1^x10^-8 1x10^-7 1x10^-6 1^x10^-5

No silane t=1 minute t=10 minutes t=20 minutes

Vg - ΔVso (V) I sd

(A)

Figure 2.8: Transfer characteristics of a P3HT transistor doped with TCFOS in time, corrected for the shift in Vso.

entire channel length (see figure 2.2(c)) and that the dopants are uniformly distributed throughout the semiconductor layer. The depletion width in a doped semiconductor is given by [3]:

Wdepl=0semi

C_i

"s

1 +2C_i²(Vg− Vso) qN_A₀_semi − 1

#

, (2.10)

with 0 the permittivity of vacuum, semi the relative dielectric constant of the semiconductor, q the elementary charge and NA the dopant density. From [11] we know that, by using the insulator capacitance per unit area,

C_i=₀_ins dins

, (2.11)

where insis the relative dielectric constant of the insulator and dinsis the thickness of the insulator layer, and the semiconductor capacitance,

C_semi=₀_semiA dsemi

, (2.12)

where A is the transistor area (length times width), we can rewrite equation 2.10 to get an equation for the dopant density:

NA= 2Vpinch0

q ^d_^2semi

semi +^2dsemi_ ^dins

ins

 , (2.13)

where we made use of V_pinch = V_g− V_so at which the depletion layer width is equal to the semi- conductor layer thickness (Wdepl= dsemi).

2.5 Determination of the bulk mobility

As earlier stated, the additional current flowing through the semiconductor layer is originated in the bulk of the semiconductor. If we want to determine the mobility of the charge carriers in the bulk (the bulk mobility) we thus need to determine this from the additional current. At V_g= V_so there is no band bending in the semiconductor, so the measured current at Vsomust be flowing in the bulk of the semiconductor. The behaviour of the bulk is ohmic as the current values at Vso

vary linearly with the applied source-drain voltage. From the current at Vsowe can calculate the bulk charge carrier mobility as follows [11]:

µbulk= LIsd

NAqZdsemiVsd

  _V

g=V_so (2.14)

2.6 Determination of the doping efficiency

In the case of DDQ doping the amount of doping added to the P3HT solution is known. As it is also possible to calculate the experimental amount of doping with equation 2.13, we are able to calculate the doping efficiency of the added dopant molecules.

We assume that the density of a spincoated polymer (or dopant) is ρpol= 1 g/cm³. With the weight percentage of the added doping we calculate the density of the spincoated dopant molecule

in the semiconductor layer, ρdop. To calculate the molar concentration of the dopant molecule in the layer, cdop, we use:

cdop= ρdop/Mdop, (2.15)

with Mdop the molar mass of the dopant molecule.

The number density of doping per cubic centimeter, ndop, can be calculated by using:

ndop= Navocdop (2.16)

with Navo the Avogadro constant. Hereby we assumed that the dopant molecule is a monovalent dopant. In the case of DDQ the number density should be multiplied by 2.

To calculate the efficiency, η, we make use of the experimental obtained doping concentration N_A:

η = N_A ndop

· 100% (2.17)

We can also calculate the amount of DDQ per mol P3HT with the weight percentage by using:

moldop= 1 100/Msemi

· wp Mdop

(2.18) with mol_dop the amount of mols dopant per mol semiconductor, M_semi the molar mass of the semiconductor and wp the weight percentage of the dopant.

Chapter 3

Experimental

Different materials and methods are used to dope the transistors. In this chapter the experimental details of the devices, methods of processing and measurement methods are given.

3.1 Devices

In the experiments we used heavily doped silicon wafers with a 200 nm thick layer of thermally grown SiO2as the gate-insulating layers. On top of the SiO2 layer 10 nm titanium, as an adhesion layer, followed by 90 nm of gold is evaporated and patterned by standard optical lithographic techniques.

To make the surface of the SiO2layer hydrophobic, it is treated with hexamethyldisilazane (HMDS).

The geometry used for the source- and drain contact is a ring geometry as shown in figure 3.1. The advantage of this concentric geometry is that the drain electrode (where the current is measured) is shielded, so no parasitic currents from outside the transistor area will be measured.

On top of the wafer, the polymer is spincoated. The standard polymer used in our experiments is regioregular P3HT (see figure 2.4). This is dissolved in chloroform with concentrations varying from 10 mg/ml to 15 mg/ml, depending on the thickness needed. The P3HT has a regioregularity

S D

Figure 3.1: Ring geometry of the transistors. The advantage of a ring geometry is that the drain current is shielded from parasitic currents from outside the transistor area.

of 97%, its molecular weight is 33, 200 g/mol and it is supplied by Imperial College London. The solutions are filtered with PTFE filters with a pore size of 5 µm. Doping of the transistors is experimentally different for both methods and is explained in the sections 3.2 and 3.3. The devices are spincoated in a nitrogen environment with varying programs. The standard program consists of two steps:

step 1: 1000 rpm, 5 s, step 2: 200 rpm, 60 s.

3.2 Doping in solution

DDQ and F4-TCNQ (both purchased from Aldrich) are dissolved in the polar solvent N-methyl-2- pyrrolidone (NMP) with concentrations varying between 5 mg/ml and 10 mg/ml. After filtering the solutions with PTFE filters with a pore size of 5 µm a precise amount of doping solution (varying between 0, 2 µl and 6 µl) is added to 0, 5 ml of P3HT solution. Several weight ratios are used to get different doping concentrations. The doped solutions are spincoated on the samples as decribed in section 3.1 Transfer curves are measured immediately. To obtain absorption spectra the solutions are spincoated on cleaned glassplates.

3.3 Doping with TCFOS

TCFOS doping is added in the measurement system. Firstly, the P3HT transistor is annealed in a vacuum oven at 150 °C for 2 hours. After cooling down it is brought in a homebuilt three-probe measurement system. This system is pumped down to a pressure of 5 · 10⁻⁶ mbar. Then the vacuumpump is turned off and amounts of 20 µl or 60 µl TCFOS are injected via a bypass system.

3.4 Measurements

Transfer and output characteristics (plot of Id versus Vd) are measured in vacuum in the homebuilt three-probe measurement system using a Keithley 4200 Semiconductor Characterization System.

Standard transistors with a length of 10 µm and a width of 2500 µm are used. A standard V_d of

−2 V is used and the Vg sweeps from −50 V to +50 V and back again for the transfer curves, while for the output curves the V_dsweeps from 0 V to −40 V and the V_gsteps from 0 V to −40V in steps of 5 V.

Thicknesses are obtained by measuring the semiconductor layers with a Dektak 150 surface profiler with a resolution of 0.067 µm/sample and a hills and valleys profile.

Absorption spectra are measured by a Perkin Elmer instruments lambda 900. The spectra are measured from 1000 nm to 280 nm with steps of 1 nm.

Chapter 4

Results and Discussion

Transfer curves of various P3HT samples doped with DDQ or TCFOS are measured. DDQ doped solutions are also spincoated on glassplates to measure the absorption spectra.

4.1 Doping with DDQ

The reaction between the polymer and the added dopant molecule is already observable in solution, because the charge-transfer complexes between the dopant and the polymer change the absorption spectrum and therefore the color of the solution. To see the differences in the spectra of doped and undoped polymers, we measured the absorption spectra of a series of glass plates spincoated with doped P3HT with increasing DDQ concentration. The result of this study is illustrated in figure 4.1.

Although the spectrum does not change as much as expected from literature ([14], [16]), there is a clear noticeable difference between the samples. The spectrum is normalized to 518 nm which is the absorption maximum of undoped P3HT. Two peaks at wavelengths around 560 nm and 610 nm increase and shift slightly to higher wavelengths with increasing dopant concentration.

We also made a series of transistors spincoated with P3HT with increasing amount of DDQ added to the solution. The result of this series is shown in figure 4.2.

Two features are observed when adding the dopant molecule. The first is the shift of the pinch- off voltage, from which we can calculate the experimental dopant concentration with equation 2.13 derived in chapter 2 and the second is the shift of the switch-on voltage. In figure 4.3 the transfer curves corrected for ∆Vsoare shown. The device with a concentration of 100 : 0, 01 (7, 5 · 10⁻⁵mol DDQ per mol P3HT) does not show a shoulder, which indicates that experimentally the DDQ did not result in any additional mobile charge carriers. However, the current in this transistor is higher than in the undoped one while the thickness has not significantly been changed. This could be an experimental error. The different solutions are spincoated on different substrates, possibly causing small differences in output. For the dopant concentrations 100 : 0, 05 and 100 : 0, 1 (3, 75 · 10⁻⁴ and 7, 5 · 10⁻⁴ mol DDQ per mol P3HT respectively) N_A = 4, 5 · 10¹⁶ cm⁻³ and N_A = 7, 5 · 10¹⁶ cm⁻³respectively. Compared to the theoretically calculated dopant densities from the added DDQ concentrations (n_dop= 2, 65 · 10¹⁸ cm⁻³ and n_dop= 5, 31 · 10¹⁸cm⁻³) this is a doping efficiency of η ' 1, 5%. Another study with a doping concentration of 100 : 0, 05 resulted in a doping efficiency of η ' 6, 5%, which is the highest we have attained. The low efficiency rates are probably the result

300 400 500 600 700 800

Wavelength (nm)

0 0.2 0.4 0.6 0.8 1

Normalized absorption

p3ht

p3ht:ddq=100:0,05 wr p3ht:ddq=100:0,1 wr p3ht:ddq=100:0,3 wr p3ht:ddq=100:0,5 wr p3ht:ddq=100:0,8 wr

Figure 4.1: The normalized absorption spectrum measurements for P3HT doped with increasing concentra- tions of DDQ spincoated on glassplates. The spectra are normalized to 518 nm, the absorption maximum of undoped P3HT.

-40 -20 0 20 40 60 80 1x10^-10

1x10^-9 1x10^-8 1^x10^-7 1^x10^-6

p3ht, 97 nm

p3ht:ddq=100:0,01, 103 nm p3ht:ddq=100:0,05, 107 nm p3ht:ddq=100:0,1, 143 nm p3ht:ddq=100:0,3, 155 nm

Vg

(V)

I

sd (A)

Figure 4.2: Transistors spincoated with P3HT and various amount of DDQ added in solution. As well the pinch-off voltage as the switch-on voltage shifts to the right for higher DDQ concentrations. The doping efficiency is around 1, 5%.

-40 -20 0 20 40 1x10^-10

1x10^-9 1x10^-8 1^x10^-7 1^x10^-6

p3ht, 97 nm

p3ht:ddq=100:0,01, 103 nm p3ht:ddq=100:0,05, 107 nm p3ht:ddq=100:0,1, 143 nm

Vg - ΔVso (V) I sd

(A)

Figure 4.3: Transfer curves of P3HT transistors doped with various amount of DDQ and corrected for

∆Vso.

of the weak reducing properties of DDQ. To get a higher doping efficiency a dopant with stronger reducing properties could be used. In theory F4-TCNQ would be a proper dopant, because of its high reduction potential, but in our experiments we did not observe any doping from F4-TCNQ.

Except for the low efficiency rates it is found that the processing window is very narrow, because of the small amounts of DDQ that have to be added to the P3HT solutions.

4.2 Doping with TCFOS

The dependence of doping a polymer by exposing it to a silane gas on the pressure in the measure- ment room (directly linked to the amount of TCFOS) and the thickness of the semiconductor layer is not yet known. Several experiments were done to contribute to the understanding of the doping process. Investigated semiconductor thicknesses were 117 nm, 197 nm and 205 nm. Amounts of 20 µl or 60 µl were injected in the bypass system, resulting in pressures of 1, 8 · 10⁻¹ mbar and 3, 2 · 10⁻¹ mbar respectively. Transfer curves of these devices were measured in time.

In figure 4.4 the result of doping a 197 nm P3HT layer based transistor with 20 µl TCFOS for

-40 -20 0 20 40 60 80 1^x10^-10

1^x10^-9 1^x10^-8 1^x10^-7 1^x10^-6 1^x10^-5 1^x10^-4

no silane t=1 minute t=2 minutes t=5 minutes t=10 minutes t=15 minutes t=20 minutes t=26 minutes t=30 minutes t=35 minutes t=41 minutes t=46 minutes t=52 minutes t=57 minutes t=70 minutes t=101 minutes t=141 minutes t=282 minutes

Vg (V) I sd

(A)

Figure 4.4: Transfer characteristics of a P3HT transistor doped with TCFOS in time. The shoulder in the transfer curve indicates additional charge in the bulk. Vpinchas well as Vsoshift to higher voltages.

-60 -40 -20 0 20 40 60 1^x10^-10

1x10^-9 1x10^-8 1^x10^-7 1x10^-6 1x10^-5

No silane t=1 minute t=2 minutes t=5 minutes t=10 minutes t=15 minutes t=20 minutes t=26 minutes t=30 minutes

Vg - ΔVso (V) I sd

(A)

Figure 4.5: Transfer curves of a TCFOS doped P3HT transistor in time corrected for the switch-on voltage shift. An increase of Vpinch in time can be seen.

4 hours and 42 minutes is shown. As can be noticed, the switch-on voltage is shifted to higher gate voltages in time. To calculate the dopant density with equation 2.13 we correct for this voltage shift, which is shown in figure 4.5.

To characterize the transfer curves we plot the dopant density versus the exposure time in figure 4.6. The result shows a faster increase of NA in time for the transistor measured in an almost two times higher pressure (3, 2 · 10⁻¹ mbar instead of 1, 8 · 10⁻¹ mbar), but at the same time the two transistors measured with the same pressure in the room have a similar difference in increase as well. From this graph the dopant density does not seem to be only pressure dependent. However, a pressure difference of only a factor two might not be enough to conclude anything, especially not because the dopant density increases exponentially in time. In other studies ([11]) where the variation in pressure is much more, the dopant density increases faster in time for higher pressures.

We also examined the dependence of the switch-on voltage shift on time and dopant density, of which the results are shown in figure 4.7. For comparison, data from a DDQ doped transistor is plotted in this graph as well. From the figure we can see that ∆Vso increases in time for every single TCFOS doped transistor, but a statement about the dependence can not yet be made.

More devices with various thicknesses should be made and measured under different pressures

100 1000 10000 1x10¹⁷

117 nm, p=1,8e-1 mbar 197 nm, p=1,8e-1 mbar 205 nm, p=3,2e-1 mbar

N A (cm-3 )

Time (s)

Figure 4.6: The dopant density calculated with equation 2.13 versus the time in TCFOS for different thicknesses and pressures.

0 1000 2000 3000 4000 5000 6000 7000 5

10 15 20 25

30 117 nm, p=1,8e-1 mbar 197 nm, p=1,8e-1 mbar 205 nm, p=3,2e-1 mbar

Time (s) ΔV so (V)

5^x10¹⁶ 1^x10¹⁷1.5^x10¹⁷2^x10¹⁷2.5^x10¹⁷ 0

5 10 15 20 25

30 117 nm, p=1,8e-1 mbar 197 nm, p=1,8e-1 mbar 205 nm, p=3,2e-1 mbar DDQ doped

NA (cm-3) Δ Vso (V)

Figure 4.7: ∆Vsoplotted versus the time in TCFOS for different thicknesses and pressures. The inset shows the graph of ∆Vsoversus NAfor different thicknesses and pressures and also for a DDQ doped sample.

to investigate the dependence of ∆V_so on time, thickness and pressure. From the DDQ doped transistors in comparison with the TCFOS doped transistors we can conclude that ∆V_so does not directly depend on doping density. The shift in switch-on voltage differs significantly for the two doping methods for the same dopant density and for the DDQ doped samples we do not observe a significant difference in ∆Vso.

In figure 4.8 the bulk mobilties as a function of the dopant density for TCFOS and DDQ doped samples are plotted. For DDQ two samples are plotted; one with a length of 40 µm and width of 1000 µm and one with the same length and width as standard. As comparison the field-effect mobility of undoped annealed, undoped not annealed, TCFOS doped and DDQ doped samples, calculated as in equation 2.4 as a function of the induced charge carriers are plotted in the same graph. To calculate the amount of induced charge carriers we made use of the fact that the induced charge per unit area is VgCi and thus Ninduced= ^V_qW^gCi

c, where Wcis the thickness of the conducting channel, taken as 1 nm. In this plot it becomes clear that not every dopant has the same doping mechanism. The mobility of the TCFOS doped samples is much higher than the mobility of the DDQ doped samples for the same experimentally induced dopant density. Also the bulk mobilities increase with dopant density for TCFOS doped samples, while for the DDQ doped samples we can not observe this. The differences can not be caused by the spincoating process, because the field-effect mobility caused by the induced charges is the same. The original undoped TCFOS

1x10¹⁷ 1x10¹⁸ 1x10¹⁹ 1x10²⁰ 0.0001

0.001 0.01

117 nm, p=1,8e-1 mbar, bulk mobility 197 nm, p=1,8e-1 mbar, bulk mobility 205 nm, p=3,2e-1 mbar, bulk moblity 117 nm, field-effect mobility

197 nm, field-effect mobility 205 nm, field-effect mobility

110 nm, field-effect mobility, low bias DDQ doped, bulk mobility

100 nm, field-effect mobility TCFOS doped, field-effect mobility DDQ doped, field-effect mobility DDQ doped, L40, bulk mobility

NA, Ninduced (cm-3)

μ bulk, μ fe (cm2 /Vs)

Figure 4.8: The bulk mobility plotted as a function of the dopant density and the field-effect mobility plotted versus the induced charge density. The field-effect causes a higher density of charge carriers resulting in a higher mobility.

samples and the undoped samples processed in the same way as the DDQ doped samples, have the same field-effect mobility as illustrated in figure 4.8. It is also remarkable that the field-effect mobilities are the same for all the doped devices. A possible explanation for the reduced mobility for DDQ doped samples lies within the DDQ molecule itself. The large molecule might disturb the ordening in the P3HT molecule, causing a higher difficulty of the hopping process, which thus reduces the mobility. Another remarkable feature is that for the TCFOS devices the increase of the bulk mobility decreases with increasing NA. We suggest that the TCFOS molecules fill up the existing states in the polymer instead of creating new ones.

Except for the absolute difference in mobilities for the different doping methods in our study, the dependence of the mobility on the dopant density is in contradiction with reported literature ([11],[4]) as well. We found a bulk mobility dependence on the dopant density varying between µbulk∼ N_A^0,2−0,5, which is far off of the reported µbulk∼ N_A^2,3. To visualize this result we plot our data in the same plot as several datasets from other experiments [11], [18], [19] as in figure 4.9.

In this plot it becomes clear that the mobility of the TCFOS doped devices is not only higher than our DDQ doped devices, but it is much higher than the mobilities from other studies as well.

1^x10¹⁴ 1^x10¹⁵ 1^x10¹⁶ 1^x10¹⁷ 1^x10¹⁸ 1^x10¹⁹ 1^x10²⁰ 1^x10^-7

1^x10^-6 1x10^-5 1x10^-4 0.001 0.01

FET, F. Maddalena, literature data FET doped, F. Maddalena, literature data Hole only diode, F. Maddalena, literature date Hole only diode, C. Tanase, literature data FET, C. Tanase, literature data FET doped, E. Meijer, literature data FET, E. Meijer, literature data

FET TCFOS doped, 117 nm, p=1,8e-1 mbar FET TCFOS doped, 197 nm, p=1,8e-1 mbar FET TCFOS doped, 205 nm, p=3,2e-1 mbar FET, undoped, 117 nm, field-effect mobility FET, undoped, 197 nm, field-effect mobility FET, undoped, 205 nm, field-effect mobility FET, undoped, low bias

FET, DDQ doped

FET, undoped, 100 nm, field-effect mobility

NA, Ninduced (cm-3)

μ bulk, μ fe (cm2 /Vs)

Figure 4.9: Mobilities from several studies [11], [18], [19], plotted as a function of dopant density or charge denstiy.

The difference in mobility from a FET and a hole-only diode originates from the strong dependence of the mobility on the charge carrier density in an organic device. The other differences could probably be explained by a different doping method, different morphology of the polymer and a better quality of the material used in our study. Not only the mobility is different, but also the dependence of the mobility on the charge density. The shapes of the graphs for the doped devices in our study vary significantly from the other shapes in figure 4.9. We suggest this is caused by the filling of existing states in the DOS of the semiconductor instead of creating new ones.

4.3 Thermally dedoping and evacuation of doped devices

To get a better understanding of the doping process we annealed and placed doped samples in vacuum to see the difference in transfer curves when the doping density decreases instead of in- creases. In this section the results of evacuation and annealing of samples doped by DDQ, TCFOS (completely and partially doped) and HCl are shown.

4.3.1 Dedoping of DDQ doped devices

In figure 4.10 the result of annealing a P3HT:DDQ=100:0,5 weight ratio doped transistor is shown.

The dedoping occurs after 23 hours and 48 minutes at 50°C and 36 minutes at 100 °C in vacuum.

The transistor is completely dedoped. The curves have a significant amount of hysteresis, but the shapes are comparable with the shapes of the transistors doped with different amounts of DDQ.

The shoulder and the shift of the switch-on voltage decrease in time and thus for decreasing dopant densities. Furthermore, the forelast measured curve, when only an inappreciable amount of DDQ is left, does not show a shoulder anymore, but still shows a shift in the switch-on voltage. This is comparable with the transistor doped with a 100:0,01 weight ratio, as shown in figure 4.2.

4.3.2 Dedoping of TCFOS doped devices

The results of annealing TCFOS doped transistors are shown in figure 4.11. The graph shows that the TCFOS is still present after thermal treatment. Allthough the current is less after several hours it is more or less stable in time. Even after 28 hours the transistor is still doped.

In figure 4.12 the result of keeping a TCFOS doped transistor in vacuum (p ' 8 · 10⁻⁶ mbar) at room temperature is shown. Also in this case the TCFOS remains in the semiconductor, resulting in a stable doped transistor. After 141 hours in vacuum the transistor is still completely doped.

In both cases the doping remains in the semiconductor even when the sample is kept in vacuum or annealed and kept in vacuum for a long time. This indicates a change in the structure of the doped semiconductor. Probably the TCFOS molecules crosslink, which results in a very stable structure. To verify this we doped a P3HT transistor with HCl, which could be the same doping mechanism as with TCFOS, and try to dedope it by annealing the sample and keeping it in vacuum.

In figure 4.13 the results are shown. The sample is doped by placing a P3HT transistor upside down above a bottle of HCl. The HCl vapor dopes the semiconductor.

In figure 4.13 we do not obviously observe doping. The dedoping process shows an immediate disappearence of the shoulder and thus of the bulk doping, while the shift of the switch-on voltage takes about 100 minutes in vacuum at 100°C to disappear. This not only indicates that the bulk doping is immediately forced out of the semiconductor by annealing, but it also again confirms that

-40 -20 0 20 40 1x10^-10

1^x10^-9 1^x10^-8 1x10^-7 1x10^-6 1^x10^-5

Room Temperature in vacuum T=50 degr C in vacuum 2 min at 50 degr C in vacuum 31 min at 50 degr C in vacuum 137 min at 50 degr C in vacuum 329 min at 50 degr C in vacuum 512 min at 50 degr C in vacuum 695 min at 50 degr C in vacuum 878 min at 50 degr C in vacuum 1061 min at 50 degr C in vacuum 1244 min at 50 degr C in vacuum 1428 min at 50 degr C in vacuum

1428 min at 50 degr C and 24 min at 100 degr C in vacuum 1428 min at 50 degr C and 36 min at 100 degr C in vacuum

Vg (V) I sd

(A)

Figure 4.10: Dedoping of an originally doped transistor with a weight ratio P3HT:DDQ of 100:0,5. The dedoping takes place in more than 24 hours in vacuum. The transistor is completely dedoped after the thermal treatment. Allthough there is a significant amount of hysteresis in the curves, the shapes are more or less comparable to the shapes of the doping process as shown in figure 4.2; the shoulder and ∆Vso

decrease in time and thus with decreasing doping density.

-40 -20 0 20 40 1x10^-7

1^x10^-6 1^x10^-5

Not annealed in vacuum

8 minutes at 80 degr C in vacuum

4 hours and 45 min at 80 degr C in vacuum

Other transistor, 6 hours and 1 min at 80 degr C in vacuum Other transistor, 24 hours and 1 min at 80 degr C in vacuum

24 hours and 16 min at 80 degr C and 4 hours and 16 min at 120 degr C in vacuum

Vg (V) I sd

(A)

Figure 4.11: Annealing of TCFOS doped P3HT transistors in vacuum. The transistors do not dedope, not even after 28 hours thermal treatment.

-40 -20 0 20 40 1x10^-7

1^x10^-6 1^x10^-5

Vacuum

63 min in vacuum

Other transistor, 17 hours and 31 min in vacuum 21 hours and 57 min in vacuum

39 hours and 15 min in vacuum

Other transistor, 40 hours and 40 min in vacuum 48 hours and 27 min in vacuum

Other transistor, 141 hours and 2 min in vacuum

Vg (V) I sd

(A)

Figure 4.12: Storing completely doped TCFOS transistors in vacuum does not affect the amount of doping in the transistor. Even after 141 hours of evacuation the transistors remain completely doped.

-30 -20 -10 0 10 20 30 1^x10^-10

1x10^-9 1^x10^-8 1x10^-7 1^x10^-6 1x10^-5

HCl doped in air, R.T.

HCl doped in vacuum, R.T.

6 min vacuum, 100 degr C 11 min vacuum, 5 min 100 degr C 16 min vacuum, 10 min 100 degr C 22 min vacuum, 16 min 100 degr C 39 min vacuum, 33 min 100 degr C 49 min vacuum, 43 min 100 degr C 66 min vacuum, 60 min 100 degr C 97 min vacuum, 91 min 100 degr C 18 hours and 56 min vacuum, R.T.

V

g (V)

(A)

Figure 4.13: A P3HT transistor doped with HCl is dedoped by annealing the sample. The transfer curves do not show a shoulder, but the shift of Vso decreases in time, which indicates the decrease of doping density.

-40 -20 0 20 40 1^x10^-10

1x10^-9 1^x10^-8 1x10^-7 1^x10^-6 1x10^-5

vacuum (p=8e-2mbar)

1,5 min in vacuum (4,6e-4mbar) 2,5 min in vacuum (2,4e-4mbar) 4,5 min in vacuum (1,6e-4mbar)

14,5 min in vacuum (less than 1e-5 mbar) 35,5 min in vacuum

1 hour and 6,5 min in vacuum 2 hours and 3,5 min in vacuum 3 hours and 49,5 min in vacuum 8 hours and 10,5 min in vacuum 22 hours and 13,5 min in vacuum Original before TCFOS doping

Vg (V) I sd

(A)

Figure 4.14: A P3HT transistor is partially doped with TCFOS and dedoped by evacuating the sample.

The transfer curves show a shoulder and a shift of Vsodecreasing in time, which indicates the decrease of dopant density in time.

the shift of the switch-on voltage does not depend on the amount of doping in the bulk of the semi- conductor. The origin of ∆V_so is still unknown. Probably charges on the semiconductor/insulator interface cause the shift in the switch-on voltage.

To be certain about the crosslinking of the TCFOS molecules in the doped samples, we partially dope a P3HT transistor with TCFOS, to pump down and anneal it afterwards. The results of these experiments are shown in figure 4.14 and 4.15, respectively. These two plots show that in these samples the doping does not remain in the semiconductor, so we suggest that the TCFOS molecules do not crosslink in this case. This could be because there are only a few TCFOS molecules present in the semiconductor layer, too little to form crosslinks. Further invesigation is needed to confirm that crosslinking causes the stable doped devices from figures 4.11 and 4.12. Stable doped devices could be used in several applications, such as polymer sensors and light-emitting diodes.

-40 -20 0 20 40 1^x10^-10

1x10^-9 1x10^-8 1^x10^-7 1^x10^-6 1x10^-5

TCFOS doped, air, R.T.

1 min in vacuum, 48 degr C 2,5 min in vacuum, 74 degr C 4 min in vacuum, 80 degr C 5,5 min in vacuum, 80 degr C 13 min in vacuum, 80 degr C 24 min in vacuum, 80 degr C Before doping with TCFOS

Vg (V) I sd

(A)

Figure 4.15: A P3HT transistor is partially doped with TCFOS and dedoped by annealing the sample. The shoulder from the initially doped transfer curve decreases very fast, which indicates the dedoping of the sample. The shift in the switch-on voltage disappears after 24 minutes resulting in a transfer curve very close to the original shape.

Chapter 5

Conclusions and outlook

5.1 Conclusions

Various doping methods were investigated. DDQ was used to dope P3HT in solution and a series of OFETs were spincoated with different weight percentages of the dopant. The doping efficiency was very low (6,5 % as maximum), probably caused by the weak reducing properties of DDQ and the very narrow processing window, resulting in a low reproducibility.

Furthermore doping by exposing an OFET to a chlorosilane gas (TCFOS) was investigated.

The dopant density in the semiconductor increased in time, but the dependence on thickness and pressure could not be found. Furthermore we can conclude that the shift in the switch-on voltage is not directly dependend on the dopant density. Also the doping mechanisms are different for different materials and dopants. The mobilities in our devices are very high, compared to other studies. The ”universal” relation of µbulk ' N_A^2,3 was not observed for our devices. In our study the TCFOS doping seems to fill up already existing states in the DOS instead of creating new ones.

From the annealing of doped samples and keeping them in vacuum we can conclude that TCFOS appears to crosslink resulting in stable doped devices.

5.2 Outlook

The reducing properties of DDQ seems to be too weak to efficiently dope P3HT. Another dopant molecule with a stronger reducing properties could be used to dope more efficiently.

The origin of the shift in the switch-on voltage and the dependence of the shift on thickness and pressure is not clear. This could be investigated by making a range of samples with various thicknesses and varying pressure in the measurement room.

The material we used in our study has a much higher mobility than the materials used in other studies, which was shown in figure 4.9. To examine the complete picture of how the mobility depends on the dopant density a hole-only diode from the same material could be made. This would fill up figure 4.9 and might give interesting information about this dependence.

Furthermore the stable doped devices, which were probably stable due to the crosslinking of TCFOS, give interesting opportunities for the use in sensors, OLEDs or other applications. Further investigation is needed to make this useful.

Acknowledgements

First of all, I would like to thank Johan Brondijk for supervising me during my research project.

He gave me instructions, had a lot of patience to answer my questions and gave direction to my research project. I would also like to thank Dago de Leeuw for supervising and sharing all his knowledge with me. Of course I would also like to thank Paul Blom for giving me the opportunity to do my master research project in his group. Furthermore I would like to thank Hennie Valkenier- Van Dijk for supplying me with materials and for accepting me in her lab for a week. Of course many thanks to Jan Harkema and Frans van der Horst for their technical support, without them it would have been impossible to do a proper research project. And last but not least all the group members for their suggestions and help.

Bibliography

[1] C.K. Chiang et al.: Electrical conductivity in doped polyacetylene, Physical Review Letters 39 (1977), 1098-1101

[2] M.E. Roberts et. al.: Material and device considerations for organic thin-film transistor sen- sors, Journal of Materials Chemistry 19 (2009), 3351-3363

[3] S.M. Sze: Physics of semiconductor devices, 2ⁿd edition, John Wiley & Sons, 868 pp. (1981) [4] A.R. Brown et al.: Field-effect transistors made from solution-processed organic semiconduc-

tors, Synthetic Metals 88 (1997), 37-55

[5] P. Fonteijn: Towards a plastic neuronal signal sensor, Master thesis in Physics July 2008 [6] C. Tanase et al.: Origin of the enhanced space-charge-limited current in poly(p-phenylen viny-

lene, Physical Review B 70 (2004), 193202: 1-4

[7] W.F. Pasveer et al: Unified Description of Charge-Carrier Mobilities in Disordered Semicon- ducting Polymers, Physical Review Letters 94(2005), 206601

[8] M.C.J.M. Vissenberg et al.: Theory of the field-effect mobility in amorphous organic transistors, Physical review B 57 (1998), 12964-12967

[9] C. Tanase et al.: Unification of the Hole Transport in Polymeric Field-Effect Transistors and Light-Emitting Diodes, Physical review letters 91 (2003), 216601: 1-4

[10] Private communication with D.M. de Leeuw

[11] E.J. Meijer: Charge transport in disordered organic field-effect transistors, PhD thesis in nanophysics June 2003

[12] Y.-Y. Lin et al.: Pentacene-Based Organic Thin-film Transistors, IEEE Transactions on Elec- tron Devices 44 (1997), 1325-1331

[13] C.P. Jarett et al.: Field effect measurement in doped conjugated polymer films: Assesment of charge carrier mobilities, Journal of Applied Physics 77 (1995), 6289-6294

[14] C.Y. Kao et al.: Doping of conjugated polythiophenes with alkyl silanes, Advanced Functional Materials 19 (2009), 1-6

[15] S.M. Goetz et al.: Organic field-effect transistors for biosensing applications, Organic Elec- tronics 10 (2009), 573-580

[16] J.J. Apperloo: Iinteractions between π-Conjugated Molecules in their Charged and Photoexcited States, PhD thesis in Macromolecular and Organic Chemistry September 2001

[17] A.R. Brown et al.:A universal relation between conductivity and field-effect mobility in doped amorphous organic semiconductors, Synthetic Metals 68 (1994), 65-70

[18] C. Tanase: Unified Charge Transport in Disordered Organic Field-Effect Transistors and Light- Emitting Diodes, PhD thesis in molecular electronics May 2005

[19] F. Maddalena et al.: Doping kinetics of organic semiconductors investigated by field-effect transistors, not published yet

[20] D.M. de Leeuw: Stable solutions of doped thiophene oligomers, Synthetic Metals 57 (1993), 3597-3602