Eindhoven University of Technology MASTER Degradation of polymer LEDs : the role of oxygen and heating Sturm, J.M.

Eindhoven University of Technology

MASTER

Degradation of polymer LEDs : the role of oxygen and heating

Sturm, J.M.

Award date:

2001

Link to publication

Disclaimer

This document contains a student thesis (bachelor's or master's), as authored by a student at Eindhoven University of Technology. Student theses are made available in the TU/e repository upon obtaining the required degree. The grade received is not published on the document as presented in the repository. The required complexity or quality of research of student theses may vary by program, and the required minimum study period may vary in duration.

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Degradation of polymer LEDs: the role of oxygen and heating

June 2001 J.M. Sturm FfV- TIB I 2001-04

Master thesis

Eindhoven University of Technology Department of Applied Physics

This graduation project was a collaboration of research groups:

Physics and Application of Ion Beams (TIB) Surface and Interface Physics (FOG)

Supervisors:

F .J .J. J anssen L.J. van U zendoom M.J .A. de Voigt

A.W. Denier van der Gon H.H. Brongersma

Summary

Polymer light-emitting diodes (polyLEDs), basedon poly-p-phenylenevinylene (PPV) show great potential for applications. This research project focused on two factors influencing the performance of ITO/PPV /Cal Al polyLEDs: heating of the device during fabrication and oxy- gen exposure of the PPV layer.

lt has been observed that upon transport of a polyLED from UHV to the glove box, current and brightness drop to approximately 50 % of the values in UHV. In this study, it is found that this effect can be fully attributed to a temperature change from 46

oe

oe

in the glove box.

Devices without Al capping layer, show (in UHV) a better performance than devices with capping layer. This effect is found to be caused by heating of the device during deposition of Al. This degradation can be prevented by cleaning the PPV with an annealing step at 65

oe

before evaporation of calcium, which suggest that impurities in the not pre-deposition an- nealed PPV are responsible for the degradation.

PolyLEDs basedon two PPV derivatives, NRS PPV and m-OC₁₀-biphenyl-PPV suffer from large anomalous currents below the normal current on-set in a polyLED. Current-voltage characteristics of NRS PPV polyLEDs with ITO or Au anode and Ca or Al catbode have been compared, anomalous currents are found to be comparable. With the aid of an optical micro- scope, black spots with bright circles can be observed in the light emission of NRS PPV and m-OC₁₀-biphenyl-PPV polyLEDs, but not for OC₁C₁₀-PPV polyLEDs. Locally high electric fields around impurities might be the origin of these bright circles and generate anomalous currents.

The current in polyLEDs consists of holes and electrons. In this report separate investigation of hole and electron transport was established by preparation of single-carrier devices. Both the hole current (in ITO/PPV/Au devices) and electron current (in TiN/PPV/Ca/Al devices) in NRS PPV were found to be space-charge limited with a hole mobility and field dependenee of (5 ± 2)·10-¹² m²/(Vs) and (4.2 ± 0.5)·10-⁴ (m/V)¹¹², respectively. For electrons these values were found to be (3 ± 2)·10-¹³ m²/(Vs) and (7.0 ± 0.5)·10-⁴ (m!V)¹¹².

Exposure of PPV films to oxygen, during light exposure (photo-oxidation), was found to re- sult in a loss of current, brightness and efficiency of the polyLED. With the aid of single- carrier devices, the electron mobility in NRS PPV was found to drop a factor 10 on photo- oxidation for 3.5 h, from (3 ± 2)·10-¹³ m²/(Vs) for untreated PPV to (3 ± 2)·10-¹⁴ m²/(Vs) for photo-oxidised PPV. In the hole mobility, (5 ± 2)·10-¹² m²/(Vs) and the field-dependencies of the mobility, no changes could be observed. Cryogenic elastic recoil detection analysis (ERDA) can determine the oxygen uptake upon photo-oxidation quantitatively.

Unlike NRS PPV, m-OC₁₀-biphenyl-PPV takes up significant amounts of oxygen in the dark.

The oxygen incorporated by oxidation migrates upon deposition of Ca and creates a PPV/CaOx interface, as determined with ERDA. This interface acts as harrier for electron injection and was found to cause an on-set shift for light emission of 1.5 V.

Table of contents

1. Introduetion ... 1

2. Polymer light-emitting diodes ... 3

2.1. Introduetion ... 3

2.2. Conduction mechanism of conducting polymers ... .4

2.3. Device physics of polyLEDs ... 7

2.3.1. A band model for the operation of polyLEDs 8 2.3.2. Injection of charge carriers 10 2.3.3. Carrier transport in the polymer 10 2.3.4. Recombination 12 2.4. Degradation ... 13

3. Analysis techniques ... 15

3.1. Cryogenic Elastic Recoil Detection Analysis ... 15

3.1.1. Collision kinematics 15 3.1.2. Depth-profiling 16 3.1.3. Quantitative analysis 17 3.1.4. Separation of scatters and recoils 17 3.1.5. Elimination of beam damage problems: cryogenic ERDA 17 3.2. Low Energy Ion Scattering ... 18

3 .2.1. Collision kinematics 18 3.2.2. Neutralisation: surface sensitivity 18 3.2.3. LEIS on polymers 19

4. Experimental set-up ... 21

4.1. PolyLED production facility ... 21

4.2. Polymer structure and solution preparation ... 22

4.3. PolyLED fabrication ... 24

5. Influence of evaporation indoeed temperature effects on polyLED performance ... 25

5.1. Characterisation of polyLEDs in UHV and the glove box: effects of temperature .... 25

5.1.1. Introduetion 25 5.1.2. Experimental 26 5.1.3. Results 27 5.1.4. Discussion 28 5.2. PolyLED degradation by heating ... 29

5.2.1. Introduetion 29

5.2.2. Experimental 29

5.2.3. Results 31

5.2.4. Discussion 33

Master thesis Eindhoven University ofTechnology page iii

Degradation of polymer LEDs: the role of oxygen and heating

6. Anomalous currents in polyLEDs ... 37

6.1. Introduetion ... 37

6.2. Experimental. ... 38

6.2.1. Gold anodes 38 6.2.2. Aluminium cathorles 38 6.2.3. Optica} microscopy 38 6.3. Results ... 39

6.3.1. Gold anodes 39 6.3.2. Aluminium cathorles 40 6.3.3. Optica} microscopy 40 6.4. Discussion ... 42

6.4.1. Anode surface roughness 42 6.4.2. Imperfections of electrode-polymer interfaces 42 6.4.3. Optical microscopy 42

7. Controlled study of hole and electron transport ... 45

7 .1. Introduetion ... 45

7 .2. Ex perimen tal. ... 45

7.3. Results ... 46

7 .4. Discussion ... 50

8. (Photo-)oxidation of NRS PPV and m-OC

-biphenyl-PPV ... 55

8.1. Introduetion ... 55

8.2. Experimental. ... 55

8.2.1. Influence of (photo-)oxidation on polyLED performance 55 8.2.2. Influence of (photo-)oxidation on charge transport 8.2.3. ERDA measurements 55 55 8.2.4. LEIS measurements 56 8.3. Results ... 56

8.3.1. PolyLED performance 56 8.3.2. Single-carrier devices 58 8.3.3. ERDA measurements 59 8.3.4. LEIS measurements 63 8.4. Discussion ... 65

9. Conclusions and recommendations ... 69

References ... 73

Dankwoord (Acknowledgement) ... 75

Technology assessment ... 77

1. Introduetion

Since many years, polymers (commonly called "plastics") are widely used in our daily Iife.

The first plastics were of bad quality compared with traditional materials like metals. For this reason the word "plastics" is still associated with low-end products like plastic bags or bad quality toys.

However, advances in chemistry resulted in better quality polymers, or engineering polymers, which can compete or even outclass traditional materials, because of low cost, low weight or high strength compared with traditional materials. Nowadays the car industry uses a wide variety of plastics, for example polycarbonate headlights. Another example of high-end appli- cations of polymers are super strong fibres like Dyneerna or Kevlar.

All polymers listed until now, are known to be insulators. Because of their good insulating properties, polymers are normally used to insulate electrical cables in all kind of applications.

However, notall polymers are insulators; for example the polymer polyacetylene (PAc) has semiconducting properties. Just like inorganic semiconductors (for instanee silicon (Si)), this material can be doped to increase the conductivity to values as high as conductivity in metals.

For inorganic semiconductors, physicist know that the class of ID-V semiconductors like gal- lium arsenide is suitable for fabrication of light-emitting diodes (LEDs), devices showing light-emission when a voltage is applied. A class of semiconducting polymers, the poly-para- phenylenevinylenes (PPVs) can also be used to fabricate LEDs. These polymer LEDs are ex- pected to cause a revolution in applications like back-lights for mobile phone displays, mono- chrome or even full-colour displays, because of their cheapness, flexibility, low weight, ease of processing and the possibility for fabrication of large area devices.

Although the initia} brightness and efficiency of devices from polymers is comparable or bet- ter than performance of devices from inorganic materials, the stability is much less. Oxygen and moisture from ambient air cause rapid degradation of the polymer and the electrodes of the polyLED. Although proper encapsulation can prevent degradation, studying degradation is still important.

The work done in this graduation project can be split in two parts. The research in the first part was done to investigate the influence of temperature effects during fabrication of the de- vices on the performance.

Research in the second part focussed on the properties of two new types of emissive poly- mers, NRS PPV and m-OC₁₀-biphenyl-PPV. The charge transport was studied by comparing polyLEDs and devices where the current is dominated by holes or electrons (single-carrier devices). The influence of oxygen exposure of the polymer was studied by exposing the PPV layer of polyLEDs and single-carrier devices to oxygen and measuring light and current. Be- sides characterisation of devices, analysis techniques Elastic Recoil Detection Analysis and Low Energy Ion Scattering were used to determine the oxygen content of the polymer bulk and surface, respectively.

The outline of this report is as follows. After this introduction, some theory about conducting polymers, polymer LEDs, degradation of PPV and device physics of polymer LEDs is given in chapter 2. Chapter 3 deals with the analysis techniques used in this project; next, in chapter

Master thesis Eindhoven University ofTechnology page I

Degradation of polymer LEDs: the role of oxygen and heating

4, the polyLED fabrication set-up and experimental procedures for fabrication are explained.

Results of the study of temperature effects on polyLED performance are presented in chapter 5. Devices with NRS PPV and m-OC₁₀-biphenyl-PPV, suffer from a problem called anoma- lous currents, chapter 6 contains results of experiments performed in order to elucidate more about these anomalous currents. In chapter 7 the hole and electron transport in NRS PPV is investigated, using single-carrier devices. The influence of oxygen on the performance of polyLEDs based on NRS PPV and m-OC10-biphenyl-PPV is described in chapter 8. Finally conclusions are presented in chapter 9.

2. Polymer light-emitting diodes

The aim of this chapter is to provide background information on polymer light-emitting di- odes (polyLEDs). After an introduetion in section 2.1, the conduction mechanism of semi- conducting polymers is explained in section 2.2. Theory on relevant subjects for this project, device physics and degradation, is reviewed in the last two sections.

2.1. Introduetion

Common polymers, or "plastics", are electrical insulators. A special class of polymers, the so- called conjugated polymers display semiconducting behaviour. The conductivity of semicon- ducting polymers ranges from 10-¹² to 1.43·10⁵ 1/(!l-cm) [1] and depends on the type of mate- rial, synthesis and doping. This range includes the conductivities of polyethylene (1

o-

1/(Q·cm)); silicon (10-⁵ 1/(Q·cm)) and other inorganic semiconductors and even metallic con- ductors as iron or platinum (10⁵ 1/(Q·cm)).

An important application of conjugated polymers is the polyLED, an electroluminescent de- vice basedon a conjugated polymer. The discovery of this electroluminescence in conjugated polymers by Burroughes et al. [2] led to many research activities on the subject of polyLEDs, both at universities and at commercial research labs [2,3,4,5].

~---~

Figure 2.1 provides a schematic picture of a polyLED. The active polymer layer (a PPV (see section 2.2)) is sandwiched between an electron- injecting cathode and a hole-injecting anode. For reasons explained further on in this chapter, the materials calcium (Ca) and indium tin oxide (ITO) are very suitable as cathode and anode, respectively. In the PPV layer, electrons and holes can recombine and form excitons (see sec- tion 2.2). These excitons can decay radiatively, resulting in visible light-emission.

There are numerous possibilities for commercial application of polyLEDs. Possible applications are for instanee back-lights for LCD displays, or pixilated monochrome and even full-colour dis- plays.

The polyLED has a number of advantages over inorganic LEDs of direct semiconductors. The materials are cheaper and easier to process. The flexibility of polymers makes it possible to fabri- cate flexible displays. By adding appropriate side-chains to PPV the colour of the emitted light can be tunedover the entire visible spectrum [3].

Last but not least, the use of polyLEDs in dis-

Master thesis Eindhoven University of Technology

ITO

glass

hv

Figure 2.1: Schematic picture of a poly- mer light-emitting diode (polyLED).

Electroos (e) and holes (h) are injected via electrodes and recombine in the ac- tive light-emitting PPV layer. The pho- tons escape via the transparent ITO electrode and the glass substrate. The aluminium is used to prevent Ca from oxidation.

page 3

Degradation of polymer LEDs: the role of oxygen and heating

plays can reduce the power consumption, which is important for application in portable de- vices like mobile phones or hand-held computers.

Before explaining the operation of polyLEDs, it is necessary to provide background informa- tion on the conduction mechanism of conjugated polymers, this is done in section 2.2. A sim- ple band model of the operation of polyLEDs and some other aspects of the device physics of polyLEDs, relevant for this graduation work, are treated in section 2.3.

2.2. Conduction mechanism of conducting polymers

Polyethylene (PE), the prototype of a linear polymer, is a good insulator. In PE every car- bon atom has 4 single bonds, two with a car- bon and two with a hydrogen atom. Not all polymers are electdeal insulators, so-called conjugated polymers are semiconductors. The prototype of conjugated polymers is trans- polyacetylene (PAc), see figure 2.2.

In conjugated polymers single and double

a ..•.

··

b ·.

Figure 2.2: The two equivalent ground states of trans-polyacetylene (PAc), hy- drogen atoms are omitted for clarity.

bonds between carbon atoms altemate along the backbone, since the electrons of the double bonds form spin-pairs, insteadof distributing along the backbone. The two ways (a and b in the figure) of drawing PAc are equivalent. The semiconducting behaviour is aresult of the Peierls instability [6], which states that a monoatomie one dimensional metal with a half filled band is unstable. The electron pair of the double bond attracts the nuclei of the carbon atoms, resulting in a shorter bond length of a double bond compared to a single bond. When the polymer changes from state a to state b, the lattice of carbon atoms relaxes, since the carbon atoms adapt their positions to the new positions of the electrons of the double bonds. Because of this relaxation, an energy harrier between state a and b exists. This harrier is the origin of the energy gap of P Ac.

In PAc electrical conduction is possible via solitons [ 6], a soliton is a kind of conjugational defect, as shown in figure 2.3. In this case a carbon atom has single bonds with both neigh- bouring carbon atoms instead of a single and a double one, which results in a unbound elec- tron.

The soliton can move to next-nearest neighbour sites by forming a double bond with one of the electrons of the nearest double bond, resulting in an unbound electron at the new site of the soliton.

In inorganic semiconductor physics band pictures are drawn to indicate electronic states. This is often also done when dealing with polymer semiconductors. The valenee band is usually called HOMO, short for highest occupied molecular orbital. The LUMO, lowest unoccupied molecular orbital, is the conduction band of the semiconductor. The energy difference be- tween the HOMO and the LUMO is the energy gap.

The two single honds at the place of a soliton imply a local suppression of the Peierls insta- bility. In a band picture, a soliton is therefore a state at midgap.

2. Polymer light-emitting diodes

LUMO

1 +

HOMO

(a) (b)

Figure 2.3: (a) A soliton (or ''free" electron) in trans-polyacetylene separatesapart of the chain in state a and a part in state b of tigure 2.2. The soliton can move along the chain to next-nearest neighbour sites. (b) In a band picture, a soliton represents a state at half gap, between the Highest Occupied Molecular Orbital (HOMO) and the Lowest Unoccupied Molecular Orbital (LUMO).

PAc is not suitable for production of polyLEDs, since it does not show electroluminescence. The conjugated polymers used for polyLED fabrication are poly-para-phenylenevinylene (PPV) and its derivatives. Figure 2.4 shows two states of PPV, which are clearly not totally equivalent like state a and b of P Ac. State b has a higher energy than state a, so PPV has no twofold degenarate ground state, like P Ac. This implies that solitons cannot exist in PPV [6], because a soliton would separate a part of the chain with high energy and a part with low en- ergy. In this situation the soliton will immediately move to the chain end of the high-energy side.

Figure 2.5: A positive polaron p+ in PPV

a

b

Figure 2.4: In contrast to PAc, poly- para-phenylenevinylene (PPV) does

not have a twofold degenerate ground state. State b bas a higher energy than state a.

The charge carriers in PPV are polarons (see figure 2.5), a positive polaron p+ is formed by removing an electron from the chain (creating a hole). This results in a part of the chain in the quinoid structure (state b) ending with an unbound electron. The hole and the unbound elec- tron, together called polaron, can move along the chain.

Figure 2.6 shows the band structure of PPV, with a positive polaron. The part of the chain in the quinoid structure has a higher energy, so the state where the hole is created shifts from the

Master thesis Eindhoven University ofTechnology page 5

Degradation of polymer LEDs: the role of oxygen and heating

HOMO into the gap. By actding an electron in the valenee band a negative polaron (p-) is cre- ated. In this case a state shifts from the LUMO into the gap.

Like in inorganic semiconductors, the formation of excitons, or bound electron-hole pairs is possible. An exciton can be created starting from a p+, by actding an electron in the conduction band. The same situation is achieved with a p- by actding a hole to the valenee band. The ex- citon can recombine radiatively, creating a pboton with an energy in the order of the magni- tude of the energy gap. Since this gap is in the order of 1 e V, the radiation is visible light.

LUMO

HOMO

Figure 2.6: Band picture of a positive polaron (p+), a negative polaron (p-) and a triplet exciton (X) in PPV. States shift into the gap because of the higher lattice energy of a chain with polaron.

LUMO

HOMO

polaronic effects

Coulomb binding

energy triplet/singlet splitting

Figure 2. 7: Influence of lattice relaxation, Coulomb binding between electron and hole and triplet/singlet splitting on the energy of the emitted photon.

The pboton energy is smaller than the energy difference between the HOMO and the LUMO.

This is caused by three effects (see figure 2.7). The first effect is the Peierls instability, which shifts the polaronic states approximately 0.1 eV into the gap [6]. Second the Coulomb binding (approximately 0.5 eV [6]) between the electron and the hole leadstoa decrease of the pboton energy. The last effect is the triplet splitting of the electron state. Only the singlet state re- combines radiatively.

Variable-range hopping

In order to understand the operation of polyLEDs, band models can be used. Band models should however be approached with extreme care when dealing with organic semiconductors.

A conducting polymer film is a collection of macromolecules, weakly interacting by the Van der Waals force and not a three dimensional lattice of atoms, like an inorganic semiconductor.

2. Polymer light-ernitting diodes

The concept of delocalised electronic states is therefore not valid in conducting polymer films, electronic states are in fact localised.

Ideal conduction of polarons as explained in the previous part, only takes place along small chain segmentsof a few nanometer, because of conjugational defects (single or triple honds), kinks and torsion. For macroscopie conduction, the charge carriers have to hop from segment to segment. Electronic states are thus even localised within one polymer molecule.

On length scales larger than the length of ideal segments, the electronic wave function decays exponentially with the distance tö the segment.

Hopping of charge carriers means tunnelling through the tail of the wave function to another

Figure 2.8: Charge carrier hopping be- state, separated by a distance R and an energy tween localised states

E. This energetica} disorder is caused by dif-

ferent local chemica} states and hence different local energies. The hopping process is plotted in figure 2.8. The energy needed to evereome the energy harrier is supplied by phonons (lat- tice vibrations).

The hopping probability p, taking into account the energy difference E and the distance R can be written as:

R E

poe e ^{L keT'} _(2.1)

with L a typical length scale of ideal segments, ks Boltzmann's constant and T the tempera- ture. The hopping probability decreases with increasing distance, but the probability to find a hop with a low energy E increases with increasing distance, because of the larger amount of hopping sites available. This trade-off between far hops with low energy and close hops with high harriers, makes that this type of hopping is usually called variabie-range hopping.

This type of hopping was first examined by Mott [7], in a study of conduction in disordered inorganic semiconductors. Later it was recognised that this theory can also be applied to or- ganic semi-conductors [4].

The temperature dependenee of the charge carrier mobility J1 fellows Mott's law:

(2.2) with To a characteristic temperature and d the dimension of the hopping process. Since charge carriers make inter- as well as intra-chain hops, the dimension is believed to be 3, which would lead to exp( -(T ofT) ¹¹⁴) temperature dependence. However, in the case of PPV s the dif- ference between behaviour according to this equation and an ordinary activated process, exp(-T ofT), is smaller than the ex perimental uncertainty [8]. In literature fits according to a normal activated process are therefore commonly used.

2.3. Device physics of polyLEDs

Since the construction of the first polyLED in 1990, various researchers are trying to obtain a better understanding of the device physics of polymer LEDs. In order to understand the op-

Master thesis Eindhoven University ofTechnology page 7

Degradation of polymer LEDs: the role of oxygen and heating

eration of a polyLED, a simple band model can be used. This band model is treated in section 2.3.1. In the area of polyLED device physics, three important processes can be distinguished:

charge carrier injection at the electrode-polymer interfaces, carrier transport through the polymer and recombination of carriers. These issues are discussed in the three sections after section 2.3.1.

2.3.1. A band modelfor the operation ofpolyLEDs

After the general part on the conduction mechanism of conjugated polymers (section 2.2), the operation of polyLEDs can be understood in more detail. In order tooperatea polyLED, two types of charge carriers (holes and electroos) must be injected in the device in order to reach exciton formation and light-emission.

An efficient injection of one type of charge carriers can be realised by choosing a conductive electrode, whose Fermi level matches the HOMO or LUMO of the polymer. With the aid of a band picture, as drawn in figure 2.9a, quantitative criteria for matching electrodes to the HOMO and LUMO can be defined. The energy levels of the electrodes and the PPV are commonly expressed relative to the vacuum level. The difference between the Fermi level Ep and the vacuum level is called the work function, indicated with <I>a and <l>c for the anode and cathode, respectively. The energy levels of HOMO and LUMO of the PPV are determined by the ionisation potential I and the electron affinity

x.

In order to achieve a good matching with the polymer LUMO, the negative, electron-injecting electrode (cathode) should have a low work function, a candidate is calcium (Ca). The posi- tive, hole-injecting electrode (anode) should have a high work function. A candidate for the anode is gold (Au), but, like Ca and Al, this material is not transparent for photons. The commonly used anode material is therefore indium tin oxide (ITO), which has a high work function and is transparent for photons in the visible range of the electromagnetic spectrum.

The situation after brioging the electrodes into contact with the PPV is drawn in figure 2.9b.

In equilibrium, the Fermi levels of the electrode will reach the same level. The only way to achieve this situation, is introducing a distance dependent vacuum level, HOMO and LUMO, resulting in a built-in voltage Vbi· Out of the picture, it is clear that e ^Vbi = <I> a- <l>c (with e the elementary charge). When the Fermi levels of the electrodes do not match the HOMO and LUMO exactly, injections harriers <l>s,h

=

=

Due to the built-in voltage, current flow is not possible for applied bias voltages below this built-in voltage. When applying a bias voltage in forward direction (anode positively biased and cathode negatively biased), the Fermi level of the cathode will raise relative to the anode Fermi level over an energy range eV. When the bias voltage equals the built-in voltage Vbi. the situation in figure 2.9c is reached. The HOMO and LUMO are flat, when an electron or hole is injected, it is not anymore hindered to flow through the PPV by the built-in potential. This is the theoretica} on-set voltage for current and light-emission. Upon further increase of bias voltage (figure 2.9d), the hole and electron current will raise due to the higher field in the polymer film.

2. Polymer light-emitting diodes

... ,, ......

I LUMO

HOMO

anode PPV cathode

anode PPV cathode

(a) (b)

vacuum vacuum

I

eV >eVbi

anode PPV cathode

anode PPV cathode

(c) (d)

Figure 2.9: Band model for the operation of polyLEDs: (a) no contact; (b) contact be- tween electrodes and PPV, no bias voltage applied; (c) bias voltage equals the built-in voltage of the device, on-set for current and light-emission; (d) operating polyLED, the bias voltage exceeds the built-in voltage.

As already mentioned in the previous section, band models should be approached with ex- treme care when dealing with molecular semiconductors. Although the simple band model treated in this section can qualitatively explain the operation of a polyLED, the real situation might be much more complex. At the interfaces, dipole layers might for instanee be formed, by charge transfer across the layers [9]. These dipole layers cause shifts in the vacuum level, which can reduce the effective built-in voltage.

Master thesis Eindhoven University of Technology page 9

Degradation of polymer LEDs: the role of oxygen and heating

2.3.2. Injection of charge carriers

In 1994 Parker [5] proposed that the charge transport in polyLEDs is limited by the injection of charges. In order to determine the transport properties of electrans and holes, he used so- called electron-only and hole-only devices, in which the PPV layer is sandwiched between two low work function and two high work function electrodes, respectively.

In a hole-only device a large energy harrier (difference between the catbode Fermi level and the PPV LUMO) for electron injection is created, by replacing the low work function catbode with a high work function materiaL In this way a negligible electron current is injected in the device. In an electron-only, hole-injection is prevented by creating a large harrier for hole injection (difference between the anode Fermi leveland the PPV HOMO). In this way single- carrierdevicescan be used to study transport properties of holes and electrons separately.

Parker fitted current density versus voltage characteristics of single-carrier devices with a model for Fowler-Nordheim tunnelling [5] through a triangular harrier. An injection harrier arises from non-perfect alignment of anode and catbode Fermi levels with the polymer HOMO and LUMO, respectively. Out of the results of the fits, the author concluded that the current in the devices was injection limited.

In more recent work it is believed for several reasons that Fowler-Nordheim tunnelling is not the right mechanism to describe charge injection into organic semiconductors. As mentioned by [4, 10 and 11] the currents predicted by Fowler-Nordheim tunnelling exceed the measured currents by several orders of magnitude (7 orders, according to reference [11]). Another ar- gument for rejecting the tunnelling theory, is the fact that charges have to tunnel into polaron levels, which lie in the bandgap of the polymer (see section 2.2). According to reference [12]

this implies that there is in fact no tunnelling harrier, the contacts have to be ohmic. When comparing measured current-voltage characteristics with simulations [13], thermionic injec- tion is found to describe measurements better than Fowler-Nordheim tunnelling.

2.3.3. Carrier transport in the polymer

Space-charge limited currents

In contrast to the research of Parker, Blom et al. [4], concluded that the current intheir single- carrier devices was bulk-limited instead of injection-limited. For hole-only devices they could fit a space-charge limited current (SCLC), according to equation [ 14]:

v2

J = fêoêrf.l p - 3 '

L ^(2.3)

with J the current density,

t:o

Electron-only devices made by Blom et al. showed other current-voltage characteristics than hole-only devices, which was explained by electron trapping. In contrast to holes, electrans can be severely trapped, for instanee by contaminations with a high electron affinity, like oxygen or carbonyl groups in photo-oxidised chain segments (see section 2.4). Blom et al.

2. Polymer light-emitting diodes

found that the electron current is well described by a modified SCLC model, where an expo- nential distri bution of traps n₁(E) below the conduction level Ec is incorporated:

N E-Ec

n

(E)=

I kT '

I

(2.4)

with Nt the total density of traps; E - Ec the trap depth (E < Ec), kB Boltzmann' s constant and Tt an effective trap temperature. In this way the energy kBTt is an effective trap size.

The J-V-relation according to the SCLC model with trapping incorporated becomes:

( )

r r+l

1 = NcqJ.ln

:~~ ~ 2 ^,+ 1 ^C(r}

^C(r

^{r'(2r +}

^{(r +}

where Ne is the density of states available for conduction per unit volume, q the elementary charge, J.ln the electron mobility and r = I; fT . Electron-only devices fabricated by these authors have a r value of 5, the current is thus proportional to

0

It is not believed anymore that the current-voltage behaviour of the electron current can be explained by trapping, because the electron current does not show the expected temperature dependenee [ 15]. The reason for the different transport properties of electrons and holes is still unknown.

Electron-only devices fabricated by Bozano et al. [8], by spin-coating PPV on titanium nitride (TiN) show a normal space-charge limited current (J oe

V\

Field and temperature dependenee of mobility

The mobility of charge carriers in conductive polymers depends strongly on temperature.

Furthermore, the magnitude of the electrical field in the film also influences the mobility.

Botheffects originate from the hopping conduction in PPV, discussed in section 2.2, where temperature effects are already discussed. An electric field enhances the probability of hops in the field direction [16], which results in an enhanced mobility at higher fields. This behaviour can be described by an exponential dependenee on the square root of the field. Starting from the equation of a normal activated process, the field and temperature dependenee of the mo- bility J.l becomes ([17] or review [18]):

ó.

fl = J.lo e ^ksT ey.[Ë' (2.6)

with J..lo constant,~ an activation energy, kB Boltzmann's constant, T the temperature, ycon- stant and E the electrical field. Typical values are~= 0.48 e V and

r=

temperature, for OC1Cw-PPV. The value of yis temperature dependent, according to the fol- lowing equation:

Master thesis Eindhoven University of Technology page 11

Degradation of polymer LEDs: the role of oxygen and heating

r~a( k~T- k

1

T,}

with G equal to 2.9·10-⁵ eV(Vtmr112

and T₀ approximately 600 K according to studies of Blom et al. using OC1Cw-PPV [10].

Injection limited versus bulk limited currents

Kawabe et al. [ 11, 19] presented numerical calculations of the current in a polyLED, starting from first principles. They combined charge conservation (including recombination) and Poisson's equation with Fowler-Nordheim tunnelling injection contacts. The shape of the nu- merically obtained current-vo1tage characteristics indicates that the current is injection-limited for low fields. Devices with a 100 nm PPV layer and injection harriers of 0.3 e V for holes and electroos showed a bulk-limited current (according to the model of Blom et al.) above 4 V bias.

2.3.4. Recombination

In polyLEDs the hole current normally exceeds the electron current, because of the mobility difference of electroos and holes. In reference [10] a difference of about three orders of mag- nitude is found, by comparing hole-only and electron-only devices. Due to this difference the current characteristics in polyLEDs are govemed by injection and transport properties of the majority charge carriers (holes). The light emission is determined by injection and transport of minority carriers (electrons), since light emission requires both types of carriers.

In a polyLED the description of the current-voltage behaviour is much more complex than in single-carrier devices. Electron and hole current cannot simply be added in order to obtain the total current, because neutralisation and recombination play an important role. The total in- jected current therefore exceeds the sum of the electron and hole current.

The recombination of holes and electroos in polyLEDs is reported to be of the bimolecular type [16]. The recombination rate is thus proportional to both the hole and electron concen- tration. Furthermore the recombination is assumed to be of the Langevin type [ 16], which means that the distance at which the Coulomb interaction between hole and electron equals the thermal energy k₈T is larger than the mean free path of the charge carriers. In the case of PPVs the attraction distance il(4rceof.rk₈T) is= 20 nm, while the mean free path is the typical length of ideally conductive chain segments (2-3 nm). When a hole and an electron feel each other, multiple hops have to be performed before recombination is possible. The recombina- tion is thus limited by the transport properties of the PPV.

In the case of bimolecular recombination, the current density in a double-carrier device can be calculated analytically [14,10]. Such a calculation holds for the case without traps and without a field-dependent mobility. The resulting current-voltage equation, the so-called plasma limit, is given by [14]:

(2.8)

2. Polymer light-emitting diodes

where /lp and Jln are the hole and electron mobility and B the bimolecular recombination con- stant. In the case of Langevin recombination this constant is given by:

B=q{pp+J.LJ.

When including the effects of injection of charge carriers at the electrode-polymer interfaces, field-dependent mobility and trapping, the differential equations descrihing the current in a double-carrier device caooot be solved analytically. A simple approach, by neglecting elec- tron trapping and field-dependent mobility cao be found in [ 11, 19]. Blom et al. presented a more realistic model, including trapping, according to the electron-only theory and field- dependent mobility [4,10]. In this way, they cao fit experimental data very well.

2.4. Degradation

The main problem for commercial application of polyLEDs is the fast degradation during operation. A possible form of degradation is migration of electrode materials into the poly- mer, for instanee indium out of the ITO anode [20], although it is oot known from this refer- ence whether this bas a negative influence on the device performance.

Another form of degradation, oxidation in air, is more related to the work done in this project.

This degradation will cause the lifetime of unencapsulated devices in air to be as short as a few seconds, some minutes, or at maximum hours [21], whereas for commercial applications a lifetime of 10000 - 100000 hours of operation is required.

A type of oxidation during operation is so-called black spot formation [22], caused by oxida- tion of the interface of the reactive, low-work function catbode and the polymer. The active polymer layer itself cao also degrade. Part of this research is focussed on the degradation of the polymer, so the mechanism of this degradation is discussed in more detail.

An important degradation mecha- nism of PPV is photo-oxidation.

This process takes place when the polymer is exposed to oxygen in combination with light. Incident photons with at least the energy of the bandgap cao create an exciton, by moving an electron from the HOMO to the LUMO. Non-

H~

0

radiative decay of this exciton cao Figure 2.10: Photo-oxidation of PPV breaks the excite oxygen to a reactive state. polymer chain in two parts, with carbonyl groups at The excited oxygen breaks the the new chain ends.

polymer chain, making two carbonyl

(C=O) ends [23,24]. According to [24] the photo-oxidation process does not only stop by the creation of carbonyl groups, the reaction cao also proceed, which results in incorporation of oxygen in the side-groups.

Master thesis Eindhoven University of Technology page 13

Degradation of polymer LEDs: the role of oxygen and heating

The broken chain segment is no longer optically active, which results in lower brightness.

Moreover, a carbonyl group can also quench excitons, resulting in a decrease of the photolu- minescence efficiency. When an exciton approaches a carbonyl group, the electron is attracted because of the high electron affinity of carbonyl. As a result, the exciton dissociates and can- not decay radiatively.

When a polyLED is operated, excitons are created by the injection of charges. Based on spin statistics, 25 % of the excitons will be singlet, the remaining 75 % will be triplet. As stated in section 2.2 (figure 2.7), only singlet excitons can recombine radiatively, the majority of the excitons will thus recombine non-radiatively. Operation of a polyLED with residual oxygen from the production process will therefore lead to photo-oxidation of the PPV. Therefore, it is important to avoid oxygen impurities, for instanee by spin-coating in an inert environment as done in this project. Another possibility is using more stabie polymers [21], where the chemi- ca} structure binders diffusion of excitons to quenching sites.

Not only residual oxygen from the spinning environment can cause photo-oxidation when the PPV is exposed to light, but it is also likely that the ITO anode is a souree of oxygen. Re- ported research in literature indicates that the occurrence of this process depends on the treat- ment of the ITO. Reference [24] states that no oxygen migrates from the ITO into the PPV by comparing Fourier Transform InfraRed (FTIR) spectra of polyLEDs as-made and after opera- ti on. The polyLEDs in this research were fabricated with ITO, which was thermally annealed after deposition. In contrast, reference [25] did observe a decrease from the ITO oxygen sig- na} and an increase in the PPV oxygen signal of an operated polyLED, using FTIR as well.

This points out that polyLED stability depends on exact processing conditions.

It can be concluded that the degradation process of polyLEDs depends on processing condi- tions. Although proper encapsulation prevents devices from oxidative degradation [22], it is still important to study degradation processes, in order to gain more insight in the degradation mechanisms and optimal processing conditions. Moreover, the ultimate goal for applications is an all-flexible polyLED, which can be bend like an overhead transparency. For such a flexible device structure, encapsulation is not possible. Whereas actvances in processing tech- nology almost allow commercial fabrication of encapsulated, inflexible devices at this mo- ment, research on degradation by environmental influences is still very important for fabrica- tion of flexible polyLEDs.

3. Analysis techniques

3.1. Cryogenic Elastic Recoil Detection Analysis

Elastic Recoil Detection Analysis (ERDA) is a powerlul nuclear analysis technique to deter- mine element distributions in thin films of light elements, like carbon and oxygen, on relative heavy substrates like glass. This section briefly reviews the theory of ERDA and the problems when performing ERDA on polymer samples.

3.1.1. Collision kinematics

In an ERDA experiment, a beam of high energy i ons (2-20 Me V) is directed onto a sample.

Collisions between incoming ions and sample atoms result in scattered incoming ions and recoiled sample atoms. Roughly speaking, the sample composition is determined out of the energy spectrum of recoiled atoms.

The callision of the ions and atoms can be described by elastic callision theory.

Figure 3.1 shows the situa- tion before and after the col-

before

lision in an ERDA experi- -~•~"'t~-~~0

-

ment. The incoming ion has mass m1 and the atom mass m2 . The atom is assumed to be at rest before the colli-

after

sion, the ion has momenturn

Figure 3.1: Situation before and after collision in an ERDA

Po.

experiment ion is scattered over an angle

iJ, with momenturn

p

p

Conservation of momenturn and energy implies:

Po

Pt

P2

(3.1) (3.2)

where E₀ is the ion energy before callision and E₁ and E₂ the ion and recoil energies after the collision, respectively.

By using the relation E

=

(3.3)

where the mass ratioris defined as m₂/m_1• For given ion mass, energy and scattering geome- try (angle cp), the recoil energy is a function of the atom mass. Therefore out of an energy spectrum of the recoils, the sample composition can be determined.

Master thesis Eindhoven University of Technology page 15

Degradation of polymer LEDs: the role of oxygen and heating

3.1.2. Depth-profiling

When the number of recoils detected in an ERDA experiment is plotted as a function of recoil energy, peak:s appear for surface compounds, while a layer with finite thickness results in a continuum. This continuurn arises from the energy loss of ions and recoils on their path through the sample. The energy loss per unit path length is known as the stopping power.

When an ion penetrates a material to a depth x, under an angle lJ with respect to the surface normal (see figure 3.2), the energy loss M can be written as:

M = ( ) Nx , (3 .4) cos lJ

with N the atomie density of the mate- rial (atoms per volume) and e the so- called stopping cross-section:

1 dE ê = - - - .

N dx

(3.5) Figure 3.2: Energy lossofanion beam, E1 < KE0

For ERDA experiments, the total energy loss of the incoming ion on its path into the material and the recoiled atom on its path out of the material is relevant. Recoiled atoms from colli- sions at the surface of the sample (see figure 3.2) will have energy KE0, with K the kinematic factor for ERDA (E2/E0 in equation (3.3)) and E0 the energy of the incoming ions. Recoils from a collision at depth x in the sample will have an energy E~. given by:

E 1 = K(E ^0- cos(l1êin N

J

1) x - cos(l1

2) x'

with Bin and Bout the stopping cross-section for the incoming ion and the recoiled atom, re- spectively, tJ.1 the angle of the incoming ion with respect to the surface normal and t9:z the an- gle of the recoil, also with respect to the surface normaL Note that the stopping cross-section for the recoil should be calculated with the energy of the recoil directly after the collision (in- dicated with KE in figure 3.2). The total energy difference between recoils from the surface and from a depth x can now be expressed as:

KE E

( Kêin ^{Bout }} N (₃ ?)

o - ¹ = ( )

+ ( )

COS lJ₁ COS iJ.2

with terr the effective stopping cross-section.

Recoils from collisions with atoms at a certain depth in the sample will thus appear at lower energies than recoils from collisions with atoms of the same element at the surface. As shown in this section, the energy scale and depth scale are related, compositional depth-profiling is thus possible.

At this point, it is clearly understood why, when determining light element concentrations on heavy substrates, ERDA is preferred over Rutherford Backscattering Speetrometry (RBS). In RBS, the scattered ions are detected and a substrate will appear as a continuurn in the spec- trum. Light atoms from layers on top of the substrate produce scatters with energies in the range of the continuurn of the substrate, so small features on a large background are meas-

3. Analysis techniques

ured. In an ERDA experiments lighter atoms generate recoils with higher energy (equation 3.3), so light element films are visible as distinct peak.s at higher energy than the continuurn of the heavy elements of the substrate.

3.1.3. Quantitative analysis

Nuclear analysis techniques are popular because of the possibility of reliable quantitative measurements. The reliable quantification arises from the facts that (i) both ions and neutral particles can be detected and (ii) in general the scattering cross-section can be calculated ac- cording to Rutherford, since the interaction between the incoming ion and the target atom can be described by scattering of the ion in the Coulomb field of the target nucleus.

In this work, Heions were used for ERDA experiments. In order to obtain an optimum in the depth resolution of the experiment and a high recoil cross-section, the energy of the incoming He++ ions was chosen to be 13.4 Me V. In our experimental geometry, collisionsof 13.4 Me V Heions and oxygen atoms, produce oxygen recoils of about 6 Me V. At this energy, the stop- ping cross-section for the oxygen recoils has a maximum, which enhances the depth resolu- tion.

The recoil cross section of the callision of 13.4 Me V He++ ions with oxygen atoms does not obey Rutherford cross-section calculations, but has a broad maximum around this energy, which greatly enhances the sensitivity [20]. In this energy range nuclear forces should be taken into account when calculating the cross-section. Therefore, in order to perform quanti- tative measurements, a calibration sample was used [20].

3.1.4. Separation of scatters and recoils

The collisions of ions from the beam and sample atoms result in recoils and scattered ions, which are both detected. Scatters and recoils can have the same energy, so they cannot be separated by standard energy-resolving detection. Therefore in ERDA experiments, other techniques must be applied to separate recoils from scatters.

Different methods to differentiate between scatters and recoils can be used. In this research Pulse Shape Discrimination (PSD) was used. PSD uses a PIPS semiconductor detector for energy-resolving partiele detection. The detector is however operated at low voltages, to keep the depletion zone small.

Heavy recoiled particles are stopped in the depletion zone and deposit their full energy. Light scatters only deposit a small fraction of their energy in the depletion zone and are stopped behind the depletion zone. Since charge coneetion in the depletion zone is faster than charge collection deeper in the detector, the shape of the detector signal is different for recoils and scatters. This difference in peak shape is used to separate recoils and scatters.

3.1.5. Elimination of beam damage problems: cryogenic ERDA

Application of nuclear analysis techniques like RBS or ERDA on polymers is not straightfor- ward, because of ion beam damage. Bonds in the polymer are broken by the incident ions, which leads to the creation of volatile molecules, which evaparate from the sample. When measuring polyLEDs with a metal cathode, gas bubbles will be formed, leading to "explo- sion" of the device.

Master thesis Eindhoven University ofTechnology page 17

Degradation of polyrner LEDs: the role of oxygen and heating

These problems can be circumvented by cooling the sample down to cryogenic temperatures.

Although honds are still broken, volatile species are frozen in place. The depth profile of the atomie composition in the sample therefore remains constant during the measurement.

3.2. Low Energy Ion Scattering

Just like RBS and ERDA, Low Energy Ion Scattering (LEIS) uses an ion beam to investigate sample composition. Like in RBS experiments, backscattered ions are detected, in LEIS how- ever only the outermost atomie layer is studied. The low energy (typically 3 ke V) noble gas i ons used for LEIS measurements have a high neutralisation probability during collisions with sample atoms. By detecting scatters by means of an electrastatic analyser, which is only sen- sitive for ions, it is ensured that LEIS is extremely sensitive to the outermost atomie layer of the sample, since the neutralisation probability of ions colliding in the second monolayer or deeper is approximately 100 %.

3.2.1. Collision kinematics

Collisions of ions and sample atoms in LEIS experiments are govemed by the same collision kinematics as used for ERDA. Solving the equations of conservation of momenturn and en- ergy (equation (3.1) and (3.2)) for the energy of the scattered ion yields:

~ =(cos(b)+~r

E₀ 1+r ^(3.8)

with mass ratio r = m_{2 /} m₁ again the mass of the sample atom over the incoming ion mass.

Only mass ratios larger than 1 are relevant, since backscattering only occurs when the mass of the incoming ion is smaller than the atom mass.

When the incident ion energy and scattering geometry are known, an energy spectrum of backscattered ions can be interpreted as a mass spectrum of the sample atoms of the outer- most atomie layer.

3.2.2. Neutralisation: surface sensitivity

LEIS is carried out with noble gas ions, so the probability that the incoming ion will neutral- ise during the collision is very high. Depending on the sample, typically only 1 in 10000 i ons will not be neutralised during the collision. Since for low energies an electrastatic analyser is used to measure the scatter energy, only ions are detected.

The probability that i ons colliding in the second monolayer of the sample ( or deeper) are not neutralised is negligible. LEIS is thus only sensitive to the outermost atomie layer of the sam- ple.

3. Analysis techniques

3.2.3. LEIS on polymers

lust like in an ERDA experiment, the ion beam damages the surface by recoiling atoms or mole- cules, this process is called sputtering. Therefore, only with very sensitive apparatus, LEIS on polymers is possible.

The LEIS experiments in this project were per- formed with the LEIS apparatus ERISS (figure 3.3). ERISS is short for Energy Resolvent Ion Scattering Spectroscopy. In this machine the incoming ion beam is directed perpendicular to the sample surface, backscattered ions at a fixed angle of 145 ^o are measured by a toroirlal elec- trastatic analyser [26].

The trajectories of ions in the analyser depend on the ion energy. As sketched in the figure, ions with different energies will hit the detector at different places, which makes it possible to dis-

Variabie ~ ^L apertures ï

Trajectories - - îonslelectrons

Sample i _.-/ ₁₄₅ •

...

'

Figure 3.3: Schematic picture of the LEIS apparatus ERISS

tinguish between different energies. In this way a 100 e V energy range of the spectrum ( called window) is taken simultaneously. The total spectrum is taken by measuring the entire energy range window by window. The ion dose needed for a proper spectrum is therefore reduced relative to machines with so-called Cylindrical Mirror Analysers (CMAs), which only record small energy windows.

For further enhancement of the sensitivity, the detector signal is summed over the azimuthal angle. The beam damage is reduced even more by moving the sample holder under the beam, to spread the ion dose over a 3 x 3 mm² surface. In this way reasonable statistics can be reached with a dose of only 10¹⁴ ions/cm²• Since the sputter rate of 3 keV He ions on poly- mers is 0.1-0.2 [27], at maximum 2·10¹³ atoms/cm² are sputtered. Since this is only about 1%

of a monolayer, sputteringduringa single measurement is not significant.

Master thesis Eindhoven University ofTechnology page 19

4. Experimental set-up

4.1. PolyLED production facility

The polyLED production set-up used in this research is designed for making polyLEDs under controlled conditions. A schematic picture of the set-up is displayed in figure 4.1.

sluice

UVazone eh amber

glove box

~ 1-1:20 < 1 ppm

(:!!;il

s~n=re/

currentllight characterisation

port for suitcase

sluice

transfer rod

UHV evaparatien

1Ff-.f=!=4~ chamber

Ca Au Al

Figure 4.1: Schematic picture of the set-up for polyLED production and characteri- sation, after UV ozone cleaning samples can be introduced into the glove box (02 and

H20 < 1 ppm) without contact to air. Transfer of in situ spin-coated polymer layers to

the UHV evaporation chamber is also done without air contact.

The central part of the set-up is a glove box filled with nitrogen. In order to remove oxygen and water vapour, the nitrogen is continuously circulated and filtered. In this way, both 02

and H20 levels are kept below 1 ppm during normal operation. Solvents are removed by an active carbon filter. The glove box window is equipped with a yellow shield to proteet sam- ples inside from light with a shorter wavelength than yellow.

At the left si de of the glove box, a UV ozone chamber is mounted for ITO cleaning. When the chamber is filled with air and the UV lamp is switched on, oxygen is converted into ozone (0₃₎ which reacts with organic species. In this way the ITO anode of a polyLED can be cleaned from organic species.

After the treatment, the chamber is pumped down and filled with nitrogen from the glove box.

By opening the valve, the cleaned samplescan be introduced into the glove box without con- tact to air.

Inside the glove box, the PPV polymer layer is spin-coated on the ITO anode. A droplet of solution is applied, subsequently the substrate is rotated (spinned). The sample is fixed on the rotating parts of the spin-coater with the aid of a vacuum.

Degradation of polymer LEDs: the role of oxygen and heating

At the right side of the glove box, a UHV evaporation chamber is attached, where catbodes are deposited. The chamber is pumped down with the aid of a turbomolecular pump. The base pressure is 1·10-⁹ mbar (1·10-⁷ Pa), during evaporation of calcium or aluminium, pressure is in the order of 10-⁷ mbar (10-⁵ Pa). The evaporation chamber is connected with the glove box via a transfer chamber, also equipped with a turbomolecular pump. In this way samples can be introduced into the evaporation chamber without breaking the vacuum of the main chamber.

In order to study effects of oxygen on polyLED performance, oxygen can be introduced into the transfer chamber to oxidise samples. An ordinary halogen lamp, which is shielded to pre- vent UV emission, is also placed in the chamber, for photo-oxidising samples.

Samples are introduced in the system in a sample holder with six positions, so six samples are prepared in a single batch.

The evaporation chamber has three positions for evaporators. Calcium (Ca) and aluminium (Al) are deposited with the aid of commercial Riber ABN 135 L effusion cells, whereas for gold (Au) deposition a home-built evaporator is used. In order to achieve a deposition rate of 5 Á/s for Ca and 1-2 Á/s for Al, the crucibles of Ca and Al are heated by resistive heater bas- kets to 510

oe

oe,

The gold evaporator is not operated at a specific temperature, but the heater current is set at 47 A. The gold evaporator is not equipped with a shutter.

Layer thicknesses are monitored with an Intellemetrics IL 150 quartz crystal thickness moni- tor.

In the evaporation chamber in situ current and light versus voltage characterisation is possi- ble. Light is measured with photodiodes (Burr Brown OPT 301). A NTC resistor pressed on the backside of the glass substrate of the polyLED measures temperatures. These NTC resis- tors are part of the sample bolder, so the sample temperature is measured during characterisa- tion in UHV and in the glove box. The system does not allow light measurement on all the six positions, because one position has a glass fibre for fluorescence thermometry measurements instead of a photodiode. Current can however be measured at this position. In order to obtain reliable measurements, only two different treatments (three polyLEDs per treatment) per batch of six polyLEDs are applied to compare effects of different treatments.

The characterisation can be done manually or fully automated, using a measuring system run- ning under the operating system VxWorks. The electrooie equipment for measuring current, brightness and temperature is home-built. In the glove box device characterisation is also pos- sible. Here, light can be measured on all positions.

Transport of samples from the glove boxtoother set-ups (for using analysis techniques) under dry nitrogen atmosphere is possible by attaching a vacuum suitcase to the glove box.

4.2. Polymer structure and solution preparation

The chemica} structures of the polymers, mainly used in this research are drawn in figure 4.2.

The orange-emitting NRS PPV in figure 4.2a is a random copolymer of OC1C₁₀-PPV (I) and OCw-phenyl-PPV (11), figure 4.2b shows the yellow-emitting meta-OCw-biphenyl-PPV.

The copolymer NRS PPV belongs to the class of non red-shifting (NRS) polymers. Polymers like OC 1Cw-PPV show a temperature-dependent ordering, resulting in a red-shift of the light

page 22 Master thesis Eindhoven University of Technology