Effects of the tbu-pyrene-tetraone deposition on the work function of silver

Transcription

1 Effects of the tbu-pyrene-tetraone deposition on the work function of silver BACHELORARBEIT zur Erlangung des akademischen Grades Bachelor of Science (B. Sc.) im Fach Physik eingereicht an der Mathematisch-Naturwissenschaftlichen Fakultät I Institut für Physik Humboldt-Universität zu Berlin von Frau Helena Stange geboren am in Hamburg Betreuung: 1. Prof. Dr. Norbert Koch 2. Priv.-Doz. Dr. Stefan Kirstein

2 eingereicht am: 29. Oktober 2010 ii

3 Contents 1 Introduction Organic electronics Charge carrier injection barriers Comparison of PyTon and tbu-pyton Theoretical Background Work function of metals Composition of the work function Influence of an interface dipole Hole injection barrier Tuning of the hole injection barrier Preliminary considerations based on the Helmoltz equation Experimental Methods Photoelectron spectroscopy Particularities of PES for organic materials Ultra high vacuum technique Experimental Setup Experimental setup Sample preparation UPS measurements Results Deposition of tbu-pyton on Ag(111) Work function Valence region Deposition of α-npd on top of a tbu-pyton film Hole injection barrier Charging Discussion Charge transfer Maximum increase of the work function Downshift of the work function Push back effect Conformation effects Combination of flat-on and edge-on molecular orientations Reorientation of the molecules from a flat-on to a edge-on orientation i

4 6.3.5 Step edge decoration Further evolution of the work function Hole injection barrier reduction Summary 30 A 33 A.1 STM images of PyTon ii

5 List of Figures 1.1 3D image of PyTon and tbu-pyton [4] Chemical structre: a) PyTon and b) tbu-pyton Work function of a metal close to the surface and at the limit of infinte distance Illustration of the influence of interface dipoles induced by electron acceptor or donor molecules on the work function Schematic energy level diagram for the interface of an adsorbed organic layer and a metal. a) Vaccum-level alignment for a metal with a low work function metal and b) a high work function metal for the Schottky-Mott limit of no interface dipole. c) Pinning of the HOMO with a molecular level P at E F. h and e designate the hole and the electron injection barrier Tuning of the hole injection barrier by a (sub-)monolayer of an electron acceptor molecule. a) Modification of the work function. b) Reduction of the hole injection barrier of the hole transport layer. Figure taken from [2] Simplified energy level diagram a) Vacuum level alignment b) Reduction of h via submonolayer of electron acceptor molecule c)level pinning. The interface dipole is denoted ID and the hole transport layer HTL Inelastic mean free path as a function of the kinetic energy [7] Principle of a PES system Schematic energy level diagram for the sample and the analyzer. Illustration based on [14] Principle of the UPS study of an organic/metal interface. Illustration based on [11] Evaporation source with tbu-pyton Schematic drawing of the two interconnected ultra high vacuum chambers SECOs of the a) first and b) second series of measurements Modification of the work function Φ as function of nominal tbu- PyTon film thickness UPS survey of the second series of measurements Molecular states in the UPS surveys a) Zoom to the Fermi edge b) Emerging peak close to the Fermi edge a) SECOs of the third series of measurements. b) UPS surveys of the third series of measurements iii

6 5.7 Determination of the HOMO level of the α-npd layer STM image of tbu-pyton on Ag(111) [9] Model of contributions of push back and dipole formation. a) Effects presented as seperate functions. b) Sum of the effects as one function depicted together with the data of figure A.1 STM image of PyTon taken by Hendrik Glowatzki [9] iv

7 Chapter 1 Introduction 1.1 Organic electronics The field of organic electronics is defined by the use of carbon-based polymers or small molecules as materials for the conducting part of electronic devices. Organic electronic devices are characterised by low-priced raw materials, their simple and low-cost production methods as well as potential flexiblity (beneficial for e.g. rollable displays). An unique feature is the possibility to tune properties of the device by engineering the functional groups of the molecules. These numerous advantages have lead to an increase in research efforts, despite of the low conductivity of organic materials compared to metal or silicium based circuits. Examples of organic electronic devices are organic photovoltaic cells (OPVCs), organic light emitting diodes (OLEDs) and organic field-effect transistors (OFETs). Many organic electronic devices consist of multiple layers with specific functions. For instance in mondern multilayer OLEDs there is a hole transport layer with good hole conduction properties on top of the anode, an electron transport layer on top of the cathode and a recombination layer in between [13]. Therefore an improvement in understanding of the multitude of organic/organic and metal/organic interfaces is of huge importance for the progress in organic electronics. 1.2 Charge carrier injection barriers Reducing charge carrier injection barriers at the interface of the conducting electrode and an organic layer is crucial for the improvement of performance of all organic electronic devices, independent of their specific task. In the case of the hole injection barrier h at the anode one of the concepts being investigated is the introduction of a monolayer - or even thinner film - of a strong electron acceptor molecule between the metal electrode and the organic layer. The electron acceptor molecules induce dipoles at the interface, resulting in an increase of the work function [15]. A linear dependence of the h decrease of a subsequently deposited hole transport layer on the work function increase has been observed for some molecules [16] [15]. The extent to which the work function increases depends directly on the norm of the dipole moment p perpendicular to the surface. This norm is in turn the product of the elementary charge q and the distance d between the charges. Consequently one way to increase the dipoles magnitude is to enlarge this distance. 1

8 2 1.3 Comparison of PyTon and tbu-pyton The aim of this work is to verify if the workfunction can indeed be increased by this approach. Therefore the molecule pyrene-tetraone (PyTon), which has already been known for its electron accepting character [5], was modified by adding bulky butyl groups at its extremes to expand the distance of the electron accepting centre to the metal surface. This new molecule is called tbu-pyrene-tetraone (tbu-pyton). It was produced by Ralph Rieger, Max Planck Institut für Polymerforschung, group Prof. Klaus Müllen. A 3D image (figure 1.1) shows the differences in geometry of the two molecules. Their chemical structure is given in figure 1.2. In this work the effects of the tbu-pyton deposition on the work function of a Ag(111) crystal are examined and compared to the results obtained for PyTon in [5]. Additionally a layer of N,N -diphenyl-n,n -bis(1-naphthyl)-1,1 -biphenyl-4,4 -diamine (α-npd), a typical hole transport layer material, is deposited on top of the tbu-pyton film to analyse the modification of the hole injection barrier. Figure 1.1: 3D image of PyTon and tbu-pyton [4] Having introduced the aim of this work in this first chapter, the second chapter gives more detailed information about the theoretical background. Additionally a first estimation of the magnitude of the expected effect on the work function by the deposition of tbu-pyton is made. The third chapter deals with the fundamentals of photoelectron spectroscopy, which was used in this study. In the fourth chapter the experimental setup is described. In the next chapter the results of the measurements are given. These are discussed and compared to the values of PyTon in chapter six. The last chapter highlights the results of this work in a brief summary. Figure 1.2: Chemical structre: a) PyTon and b) tbu-pyton

9 Chapter 2 Theoretical Background 2.1 Work function of metals The work function of a metal is defined as the potential difference between the Fermi level (E F ) and the vacuum energy level (E vac ). In the case of metals the Fermi level is identical to the highest occupied electronic state and constant within the crystal. Figure 2.1: Work function of a metal close to the surface and at the limit of infinte distance Composition of the work function Two factors contribute to the work function of a metal. One is the difference in energy of the metallic crystal caused by the loss of an electron. This part of the work function is isotropic. The other factor is the energy needed to traverse the dipole layer at the surface. 3

10 4 At the surface of metallic crystals, an intrinsic dipole layer exists. This is due to the fact that the charge distribution of the electron cloud of a surface atom differs from the one of a bulk atom because of missing adjacent atoms. Part of the electron cloud is spilling out to the vaccuum because there is no repulsion by the next atomic layer. This outspilling part of the electron cloud causes a surface dipole by producing a negative charge close to the surface. At the same time there is another surface effect, the so called smoothing effect, which has a contrarious effect and will be explained later in chapter 6 [30]. However, for metals the overall surface dipole normally has a work function increasing effect. The importance of the position of neighbouring atoms for the shape of the electron cloud leads to the phenomenon of the surface dipoles magnitude being dependent on which surface of the crystal is regarded. Consequently E vac differs for different surfaces of the same metal [14]. At large distances the influence of the dipole layer on the work function decreases with r 2 like the field of a regular dipole. This can be described as decline of the vaccum level towards a converged value E vac as depicted in figure 2.1. Accordingly the surface work function Φ S and the average work function Φ are determined Influence of an interface dipole If a strong electron donor or acceptor molecule is deposited on the surface of a metal, an additional interface dipole is created by charge transfer. In case of an electron donor molecule the interface dipole is charged oppositely to the intrinsic surface dipole, so the total dipole is lowered in magnitude, leading to a decrease of the work function. In contrast, if an electron acceptor molecule is used, the interface dipole amplifies the surface dipole and the work function increases [2]. Figure 2.2 illustrates the two cases. Independently of its electron accepting or donating character an adsorbant always has a decreasing push back effect on the surface dipole that needs to be considered for the total dipole. This means that the part of the electron cloud of the metal which spills out into the vacuum is pushed back into the metal by the repulsive interaction with the electron cloud of the molecule. Compared to the situation without adsorbant the area above the surface is now charged less negative, leading to a lowered vacuum energy level [11] and consquently to a decrease of the work function. The additional potential that an electron has to overcome to leave the metal is described quantitatively by the Helmholtz equation 2.1, where q stands for the elementary charge, N for the surface dipole density, p for the perpendicular dipole moment, ϵ 0 for the vacuum permitivity and ϵ r for the relative dielectric constant [13]. The perpendicular dipole moment p is given by p = q d, with d being the distance between the charges. q N p Φ = (2.1) ϵ 0 ϵ r

11 5 Figure 2.2: Illustration of the influence of interface dipoles induced by electron acceptor or donor molecules on the work function 2.2 Hole injection barrier Generally the hole injection barrier at the interface of a metal and an organic material is the potential difference between E F of the metal and the highest occupied molecular level (HOMO) of the organic molecule. Analogously the electron injection barrier is defined as the potential difference between E F and the lowest unoccupied molecular level (LUMO). Figure 2.3 a) illustrates the situation for the Schottky- Mott limit of no interface dipole. If the work function is increased (for example by using another metal), the hole injection barrier is reduced (figure 2.3 b) ). However, a limit is set to this relation when the work function is growing in a way that the hole injection barrier would virtually disappear. Without a hole injection barrier a charge transfer will occur to reconstitute thermodynamic equilibrium resulting in positive polarons at the interface. Thereby an oppositely orientated new interface dipole is induced, inhibiting further charge transfer. A molecular polaron level is pinned at the Fermi level. The other energy levels are also pinned, with the HOMO level situated slightly under the critical point of no hole injection barrier (figure 2.3 c). The magnitude of the hole injection barrier is now independent of a further increase of the work function Tuning of the hole injection barrier As has been mentioned in the introduction, there is great interest in controling the hole injection barrier between the metal electrode and the hole transport layer. As described above, electron acceptor molecules on a metal surface induce an interface dipole increasing the work function. More molecules with contact to the surface induce additional dipoles. Therefore the work function can be modified linearly by the deposition of submonolayer amounts of an organic electron acceptor molecule on the metal surface until a coverage with a complete monolayer is reached [15] [16]: Figure 2.4 a). A second layer of a different organic molecule serving as hole transport material is then deposited on top of the modified substrate. Under the condition that charge transfer between the two organic layers can be excluded, the

12 6 Figure 2.3: Schematic energy level diagram for the interface of an adsorbed organic layer and a metal. a) Vaccum-level alignment for a metal with a low work function metal and b) a high work function metal for the Schottky-Mott limit of no interface dipole. c) Pinning of the HOMO with a molecular level P at E F. h and e designate the hole and the electron injection barrier. vaccum energy levels align and a reduction of the hole injection barrier in the first organic layer induces a corresponding change in the hole transport layer [16]. This mechanism is shown in figure 2.4 b). Figure 2.4: Tuning of the hole injection barrier by a (sub-)monolayer of an electron acceptor molecule. a) Modification of the work function. b) Reduction of the hole injection barrier of the hole transport layer. Figure taken from [2] A simplified energy level diagram of the process is given in figure 2.5. The so called push back effect is not included since it only attenuates the impact of the interface dipole. The ideal vacuum energy level alignment of the organic hole transport layer and the metal substrate is shown infigure 2.5 a). The modification of the work function and the hole injection barrier by the insertion of a monolayer of an electron acceptor molecule is depicted in figure 2.5 b). Noteworthy to point out this method is not able reduce the hole injection barrier to zero, since it only works as long as the regime of level pinning is not reached. Figure 2.5 c) illustrates how the oppositely orientated interface dipole induced by positive polarons conserves a minimal hole injection barrier.

13 7 Figure 2.5: Simplified energy level diagram a) Vacuum level alignment b) Reduction of h via submonolayer of electron acceptor molecule c)level pinning. The interface dipole is denoted ID and the hole transport layer HTL. 2.3 Preliminary considerations based on the Helmoltz equation The Helmoltz equation (equ. 2.1) can be used to make some preliminary calculations in order to estimate the possible magnitude of the work function change by the deposition of PyTon and tbu-pyton on silver. The distance d is estimated by the distance of the electron accepting centre of the molecule to the metal surface. For PyTon this distance is known by x-ray standing wave (XSW) measurements to be 2.46 Å. [26]. The density of the molecule coverage can be estimated from scanning tunneling microscope (STM) data [9] (see image in the appendix A.1) to be 0.97 molecules per nm 2. Since there is only one electron accepting region in the molecule, the dipole density equals the molecule density. As the relative dielectric constant ϵ r is not known neither for PyTon nor for tbu-pyton, it is estimated in both cases with ϵ r = 3 which is a typical value for organic material [10]. Using these values for the Helmholtz equation leads to a theoretical change in the work function of Φ = 1.11eV for PyTon. In the case of tbu-pyton the distance d is increased by d Å compared to PyTon. This estimation is founded on the geometry of the butyl groups analysed by XSW [26]. Probably the density N decreases for tbu-pyton compared to PyTon, because of the bulky butyl groups. However, since there is no data available for tbu-pyton, the previous value for N is used as an approximation. The resulting value for the change in the work function is Φ = 1.48eV, depending on the value for d = 2.51 to 2.53 Å. Accordingly the difference between the two cases is Φ Φ 0.37eV at maximum. The comparison of the calculated value for PyTon of Φ = 1.11eV with the experimental result of Φ = 0.35eV [5] shows that the Helmoltz equation alone is not a good model because it overestimates Φ by about three times the experimental result. This is not surprising since the model leaves unconsiderated a whole bunch of other effects possibly taking place at the interface due to the interaction of the metal and the molecule. However, knowing this quantitative relation one can estimate that the expected change might be of the scale of a third of the calculated value of 0.3 ev, that is 0.1 ev.

14 Chapter 3 Experimental Methods 3.1 Photoelectron spectroscopy Photoelectron spectroscopy (PES) is a widely used technique to investigate the binding energies of the occupied states of a solid. The method is based on the external photoelectric effect. The sample is exposed to a beam of monochromatic photons, which cause the emission of electrons in case the photon energy exceeds a certain boundary value given by the characteristic work function Φ S of the solid. Depending on the wavelength of the applied photons PES is referred to as UPS (ultraviolet photoelectron spectroscopy, 10nm to 400nm) or XPS (x-ray photoelectron spectroscopy, 0.01nm to 10 nm). The former is mostly used for analysing the valence region, while the latter - because of its higher photon energy - is applied to probe the core levels. In any case PES is a surface sensitive method due to the small inelastic mean free path of the electrons in condensed matter. The principle of PES analysis only works if a significant number of electrons can leave the sample without loosing energy by interaction with other particles. From the universal mean free path curve (fig. 3.1) one obtains a mean free path λ of a few Å for photon energies used in UPS (10 ev to 100 ev) and some tens of Å for XPS (100 ev to 100 kev). The surface sensitivity is one of the reasons why PES measurements require ultra high vacuum conditions in order to avoid surface contamination with foreign atoms. Generally PES only works in ultra high vacuum because the electrons have to reach the analyzer and the detector without interaction with other particles. The kinetic energy of the electrons leaving the sample depends on their binding energy E B relative to the Fermi level E F and the work function of the sample: E kin = hν E B Φ S (3.1) An electrostatic analyzer can be used to measure this energy. The experimental setup used for this study was equipped with a hemispherical analyzer. It consists of two parallel conducting hemispherical plates at different potentials, creating an electric field between them. Only electrons with a certain energy, the pass energy, follow a trajectory that allows them to pass the analyzer and reach the detector. At the analyzer s entrance the electrons are accelerated or decelerated variably by an electron lense system. In this way electrons of different kinetic energies can be detected. The principle is illustrated in figure

15 9 Figure 3.1: Inelastic mean free path as a function of the kinetic energy [7] Figure 3.2: Principle of a PES system However the kinetic energy of the electrons measured by the analyzer Ekin mes is not equivalent to equation 3.1. In figure 3.3 the variation of E vac between the sample and the analyzer is shown. For Ekin mes of the sample Φ S only the difference between the work function and the analyzer Φ A is relevant. In order to obtain Ekin mes this contact potential Φ C = Φ A Φ S needs to be substracted from 3.1. E mes kin = hν E B Φ Φ C = hν E B Φ A (3.2) Figure 3.4 illustrates how UPS spectra can be used to analyse an organic/metal interface. In figure 3.4 a) a schematic energy level diagram, the photoemission process and the UPS spectrum for a pristine metal surface are shown. The electrons with the highest kinetic energy are the ones which had no binding energy relative to the Fermi level E F of the metal. Hence the Fermi edge is defined by E kin,f = hν Φ A. The electrons with the lowest kinetic energy at E vac are the ones with E kin = 0 relative to the sample and Ekin mes = Φ C relative to the analyzer. Accordingly an

16 10 Figure 3.3: Schematic energy level diagram for the sample and the analyzer. Illustration based on [14] expression for the work function of the sample is given by Φ S = hν (E max kin E min kin ) = hν (E kin,f E vac ) (3.3) Figure 3.4 b) gives the equivalent picture for a thin organic layer deposited on top of the metal surface. With increasing layer thickness the vast majority of the registered photons originate from the organic layer because of the small mean free path discussed above. The most energetic electrons steem from the HOMO of the organic layer. The vacuum level might be shifted by Φ due to the interaction at the organic/metal interface. Apart from the work function various paramters can be deduced from the spectra. The ionization energy of the organic material is given by E ion = hν (E kin,homo E org vac) (3.4) The hole injection barrier h is considered to be the difference between the highest occupied molecular orbital (HOMO) and E kin,f h = E kin,f E kin,homo (3.5) The lower energetic end of the spectra is constituted by the secondary electron cut-off (SECO). It is caused by photoelectrons inelastically scattered within the sample. In order to measure the complete SECO a negative potential of a few volts has to be applied to the sample because only electrons with E kin < Φ A Φ S can enter the analyzer, as can bee seen in figure 3.3. Hereby an upward shift of all features in the spectra is induced which needs to be substracted to determine the true kinetic energy. However, for many parameters only the relative position of the features is important.

17 11 Figure 3.4: Principle of the UPS study of an organic/metal interface. Illustration based on [11] 3.2 Particularities of PES for organic materials When an electron is emitted through exitation by a photon, the sample is charged due to the ionization of the atom. The substrate is grounded so charge neutrality can be re-established easily for conducting samples. However, many organic molecules have a low conductivity which is why an accumulation of positive charges at the sample surface can occur. This leads to a decrease of the work function, because of a lowered vacuum potential E vac close to the surface. The thicker the molecular layer is, the more important this effect becomes because the charge carriers have a longer way through the poor conducting organic material. One way to try to reduce the positve charge is the illumination with visible light, creating additional charge carriers at the interface of the organic layer and the metal [17]. On the other hand reducing the intensity of the incident photon beam also leads to a smaller charging effect, because less electrons leave the sample. Another problem is that a high intensity of the incident photon beam may cause photodegradation of the molecules of an organic sample, for example by photodissociation. Therefore lower intensities are used for the analysis of organic molecules than for metals. 3.3 Ultra high vacuum technique As explained above PES analysis requires ultra high vacuum conditions, among other reasons because of its surface sensitivity. Additionally ultra high vacuum is necessary for sample preparation as well, for example to assure a clean surface for evaporation. For PES a ultra high vacuum of about 1x10 7 mbar or lower needs to be guaranteed. At this pressure a monolayer of foreign atoms on a sample is established within approxametely 25 seconds if every molecule reaching the surface sticks to it (Calculation based on the formula for monolayer formation time and data for air molecules at room temperature given in [27]). The low pressures of ultra high

18 12 vacuum are achieved by the combination of several pumps. A mechanical prevacuum pump creates a prevacuum of about 10 3 mbar. It usually works with oil as lubricant, which needs to be prevent from entering the UHV chamber in any case. A turbomolecular pump reduces the vacuum up to 10 7 mbar. Turbomolecular pumps need to be backed up by a prevacuum pump, because they can not start working at atmosphere pressure. For further improvement the whole apparatus needs to be baked out in order to desorb water and other substances from the walls and remove it from the chamber. Now ion getter and titanium sublimation pumps can reduce the vacuum to mbar. Only certain materials are suitable for the construction and use in ultra high vacuum chambers. Their vapor pressure should be very low, even at the higher temperatures necessary for example for the baking procedure and sample heating. This requirement is fulfilled very well by metals like molybdenum, tantalum and tungsten. For the construction of vacuum chambers stainless steel and oxide ceramics are used which also have a low vacuum pressure. Grease has a high vacuum pressure and is not appropriate for UHV. Therefore attention has to be paid to prevent it from entering the UHV chamber as mentioned above.

19 Chapter 4 Experimental Setup 4.1 Experimental setup For sample preparation and measurements an interconnected system of two ultra high vacuum chambers was used. Therefore the sample could be transferred from the sample preparation chamber to the analysis chamber without breaking ultra high vacuum conditions. The base pressure was 4x10 9 mbar for the preparation chamber and 2x10 10 mbar for the analysis chamber. The sample preparation chamber was equipped with a sputter gun and a quarz crystal microbalance. A load lock allowed sample change and four different evaporation sources could be used. A photo of the evaporation source used for this work is shown in figure 4.1. The tbu-pyton molecules are distinguishable as a tiny yellow spot on the tantalum pin hole source used as container. In the upper part of the image the quartz crystal microbalance, the probe holder and the manipulator can be seen. Figure 4.1: Evaporation source with tbu-pyton 13

20 14 The sample heating system allowed the annealing of the sample. In the analysis chamber XPS und UPS measurements could be done with a hemispherical electrostatic energy analyzer. The position of the x-ray tube was variable, so photon beam intensity could be modified. A Mg and an Al anode were disposable. However for this work only the Mg anode was used. For UPS measurements a helium lamp was used as photon source. Additionally the analysis chamber included a sample storage device, a STM and LEED system. The latter two however, were not used for this work. A schematic drawing of the experimental setup is depicted in figure 4.2. Figure 4.2: Schematic drawing of the two interconnected ultra high vacuum chambers

21 Sample preparation An Ag(111) crystal was used as a substrate for all measurements. In order to obtain a clean surface the crystal was sputtered with Ar-ions for at least half an hour. Afterwards the crystal was annealed up to 450 C. This procedure was repeated twice and then the crystal surface was checked for contamination via XPS. This was done by a XPS survey scan from a kinetic energy of 50 ev to 1255 ev. Additionally the regions where the peaks of the 1s states of typical contaminants, such as oxygen, nitrogen and carbon, could be expected were scanned more precisely. Once the cleanliness of the surface was assured, the evaporation source was prepared, heating it until a evaporation rate of a bit less than 1 Ångstroem per minute was achieved. Before the first measurement was done, the evaporation source was outgased to avoid contamination with solvents. The evaporation rate was measured by a quartz crystal microbalance, which was installed in the preparation chamber at the position, on top of the evaporation source, where later the sample was placed. Even though the quartz crystal microbalance displays results in Ångstroem, it rather indicates the mass deposited than the real layer thickness, because it is not known if the growth of the molecular film is of a layer-by-layer type [14]. When a stable rate was observed, the sample was transferred to the preparation chamber. There it was exposed to the molecular beam for a time corresponding to the required deposition according to the rate measured previously. It is rather difficult to estimate the error for the deposition amount. Fluctuations of about 10 percent were observed even when the evaporation rate was relatively stable, so this could be considered as lower limit for the error. 4.3 UPS measurements After the deposition of the molecular layer the sample was transferred to the analysis chamber. The position of the sample was optimized by modifying it until maximum intensity for a previously selected peak of the silver spectrum was reached in the fixed energy modus. For every prepared film an UPS survey was done from a kinetic energy of 10 ev to 22 ev. The Fermi edge was localised at a kinetic energy of ev. Then a negative potential of -10 V was applied to the sample, so the SECO could be measured via UPS. Within one series of measurement the next layer was deposited on the sample directly after the analysis. The pressure in the analysis chamber was kept at the same level, within ± mbar, during a series of measurement. The analysis was done with the SpecsLab2 software. The error of all energy values determined from the spectra is estimated to be ±0.05eV

22 Chapter 5 Results Three series of measurements were taken. The first one was a survey in which an overview of the spectral changes was established. Therefore it began with the deposition of a very thin layer of 1 Å of tbu-pyton on the Ag(111) crystal, which was then doubled to 2 Å. The layer thicknesses were doubled until a film thickness of 32 Å was reached. For the second series of measurements layers from 1 to 9 Å with a step size of 1 Å were analysed, since previously the most prominent spectral changes had been observed in this region. Additionaly one layer of 40 Å was analysed in order to find out about the variation of the work function for thick layers. During the third series of measurements a 5 Å layer of tbu-pyton was deposited on the Ag(111) crystal. At this film thickness maximum increase in the work function was reached beforehand. This high work function template was then covered by 50 Å of α-npd, used as a prototype of a hole transport material. In all series of measurements UPS survey and SECO measurements were taken at every deposition step. 5.1 Deposition of tbu-pyton on Ag(111) Work function The SECO was measured for every deposition step as described in section 3. In the first series of measurements the SECO shifted slightly to lower kinetic energies compared to the value of pristine Ag(111) for small tbu-pyton deposition amounts. For increasing deposition amounts it reached higher values with a maximum at 8 Å nominal film thickness (figure 5.1 a)). This trend was confirmed by the second series of measurements, where the SECO with the highest kinetic energy was measured for layers of 5 and 6 Å (deposition steps for which no analysis was made in the first series of measurements). In figure 5.1 b) an overview of the SECO measurements of the second series is depicted. The dashed lines mark the maximum and minimum SECO values measured, as well as the value for pristine Ag(111). The small deviations of the different series of measurements are partly due to the error in deposition amount. Additionally slightly different charging effects may occur because during the first series the helium lamp for UPS was turned off and on for every analysis, while it was operating continuously during the second series of measurements. In both series the SECO shifted again towards lower kinetic energies after the maximum. From equation 3.3 it can be derived that the described behaviour of the SECOs corresponds to an inital decrease of the work function, followed by an increase and a final decrease. The change of the work function Φ in relation to the value for 16

23 17 pristine Ag(111) is shown in figure 5.2, confirming the consistency of the data taken in the different series of measurements. The maxium increase of the work function of Φ = 0.43eV was observed for a nominal film thickness of 5 Å. Figure 5.1: SECOs of the a) first and b) second series of measurements 0,5 0,4 0,3 0,2 Measurement series 1 Measurement series 2 Measurement series 3 Measurement series 2 reduced beam intensity ev] 0,1 0,0-0,1-0, coverage [A] Figure 5.2: Modification of the work function Φ as function of nominal tbu-pyton film thickness

24 Valence region An overview of the UPS valence band survey spectra taken in the second series of measurements is plotted in figure 5.3. The three prominent peaks of the pristine Ag(111) surface correspond to the 4d band of this metal. With increasing tbu- PyTon deposition, the silver states loose intensity while molecular features become apparent. At a tbu-pyton layer thickness of 4 Å the silver peaks can not be distinguished anymore. Instead new peaks can be obvserved, which are broader and correspond to molecular states. Two of them are marked in figure 5.4 by the dashed lines. The survey of the pristine Ag(111) surface has been reduced in intensity in order to allow its inclusion in the figure. Intensity [arb. units] E B [ev] Figure 5.3: UPS survey of the second series of measurements In figure 5.5 a) a zoom directly on the Fermi edge is given. For the pristine Ag(111) surface and thin tbu-pyton layers (1 and 2 Å) a sharp peak was observed at the Fermi edge. Due to its position this peak can be identified most probably as the surface state of silver [22]. The fact that this peak could be seen indicates a very clean surface. To the same degree as the silver features diminished, the intensity of the Fermi edge decreased and the surface state disappeared. A second zoom (figure 5.5 b)) into the valence band close to the Fermi edge shows that for increasing deposition of tbu-pyton a new peak, marked by a dashed line, emerged. It got more pronounced the thicker the tbu-pyton layer was. An exeption is the 40 Å layer, where the peak is not discernable. The intensity of the 0 Å and 1 Å layer was extenuated in order to include them in the figure and make the peak visible at the same time.

25 19 x 0,1 0 Intensity [arb. units] E B [ev] Figure 5.4: Molecular states in the UPS surveys Figure 5.5: a) Zoom to the Fermi edge b) Emerging peak close to the Fermi edge

26 5.2 Deposition of α-npd on top of a tbu-pyton film In the third series of measurements a tbu-pyton layer of 5 Å was evaporated on the cleaned Ag(111) crystal. As can be seen in figure 5.2, the work function increase was even a little higher than in the previous measurements. This points to a deposition amount slightly closer to the optimal monolayer coverage. A 50 Å layer of α-npd was deposited on top of this film. The results of the SECO measurements for this template are depicted in figure 5.6 a). The α-npd layer lowered the work function, 20 Figure 5.6: a) SECOs of the third series of measurements. b) UPS surveys of the third series of measurements however it was still higher than for pristine Ag(111). The UPS valence band surveys (figure 5.6 b)) showed the typical features of clean Ag(111), including the surface state for the substrate, and a similiar evolution as in figure 5.3 for the tbu-pyton covered sample of the same layer thickness. In case of the additional α-npd layer it can be seen that new peaks, corresponding probably to the filled molecular states of α-npd, replaced the ones of tbu-pyton Hole injection barrier As mentioned in section 3, the hole injection barrier is defined as the distance between the HOMO level and the Fermi edge. Figure 5.7 shows a zoom into the region of the α-npd HOMO level. The position of the HOMO peak is determined to be at a binding energy of 075 ev. This yields an offset which equals the hole injection barrier h = 0.75eV. The shape of the HOMO level in figure 5.7 resembles the HOMO level observed for α-npd on Au in [19], fortifying the assumption of the peaks in the lowest graph of figure 5.6 b) being the molecular states of α-npd. 5.3 Charging During the second series of measurements some SECO measurements were repeated with a lowered helium lamp intensity or changed visible light conditions in order to

27 21 Intensity [arb. units] npd tbu-pyton ,0 1,5 1,0 0,5 0,0-0,5 E B [ev] Figure 5.7: Determination of the HOMO level of the α-npd layer evaluate charging effects of the organic layer. For small deposition amounts of tbu- PyTon like 7 Å or 8 Å additional illumination with visible light and shading of the chamber s windows did not produce any change in the postion of the SECO. In case of the thicker layer of 40 Å shading decreased the work function a little by 0.02 ev. Curiously additional visible light sources provoked an identical change in the work function. In any case both effects lie within the error of measurement. In contrast to this for a reduced intensity of the incident photon beam a significant increase of the work function by 0.07eV at maxiumum was observed. Further reduction of the beam intensity was limited by the construction of the He-light source. Therefore it is possible that 0.07 is not the absolute maximum. The results of the SECO measurements with reduced intensity are included in figure 5.2.

28 Chapter 6 Discussion The aim of this study was to determine if the work function increase caused by the deposition of PyTon could be augmented by enlarging its distance to the metal surface via the addition of butyl groups, converting PyTon to tbu-pyton. A necessary condition for the work function increase by the deposition of an organic molecule - as described in chapter 2 - is a charge transfer from the metal to the molecule. In the first section of this chapter the evidence given by experimental results for such a charge transfer is discussed. Then the maximum increase of the Ag(111) work function by the deposition of tbu-pyton is compared to the results for PyTon in order to see if the initial approach is valid. In the third section an unexpected evolution of the work function for small deposition amounts is discussed. Therefore several imaginable explanations are considered. Afterwards the evolution of the work function for higher deposition amounts is discussed in section four. Finally in the fifth section the effect of the tbu-pyton deposition on the hole injection barrier of a subsequentely deposited hole transport material (α-npd) is regarded. 6.1 Charge transfer Figure 5.5 b) shows a newly emerging peak in the valence band close to the Fermi edge for growing tbu-pyton layer thickness. PyTon is an electron acceptor molecule [5]. The similiar structure of the molecules makes the assumption plausible that this is the case for tbu-pyton as well. The deposition of a strong electron acceptor molecule on a metallic surface usually leads to a charge transfer from the metal towards the molecule. Examples for this phenomenon are tetracyanoquinodimethane (TCAQ) [15] and tetrafluoro-tetracyanoquinodimethane (F4-TCNQ) [16]. This charge transfer tends to increase the work function and normally new states, so called charge transfer (CT) states appear in the energy gap. These usually correspond to a level derived from the former LUMO of the molecule and metal bands. This level is located just below the Fermi level [15] [14]. In agreement with this analysis the peak visible in figure 5.5 b) is a strong indication for a charge transfer from the metal to the molecule. The charge transfer state disappears for the 40 Å layer, because the spectrum is than dominated by the molecules of the organic bulk. Due to the charge transer, the interaction between tbu-pyton and the silver surface can be classified as chemisorptive. The charge transfer causes the dipole which induces the desired increase of the work function. 22

29 Maximum increase of the work function The maximum increase of the work function reached with the deposition of tbu- PyTon on Ag(111) is Φ = 0.43eV. Compared to the previous result for the work function increase by the deposition of PyTon on Ag(111) of Φ = 0.35eV [5] this is an augmentation of 0.08 ev. This value is in good agreement with the preleminary estimation made in section 2.3. The result could be taken as a hint that the approach to increase the charge transfer dipole by expanding the distance of the electron accepting centre of PyTon to the surface actually works. However, this can not be taken for granted, since we do not know exactly which effect causes the work function modification. It is not assured that the tbu-pyton molecules adsorb on the surface in the same flat-on manner as the PyTon molecules, which is an important prerequisite for the approach. From the STM images for tbu-pyton on Ag(111) in figure 6.1 taken by Hendrik Glowatzki [9], a flat-on configuration cannot be deduced with absolute certitude. Apart from this, the sample was prepared in a different way compared to the ones used for the UPS measurements. The monolayer was achieved by depositing multilayers first and then desorbing all but the monolayer by annealing. This can have a significant influence on the molecular conformation. Thus it can not be assured that figure 6.1 actually shows a configuration corresponding to the samples prepared in this work. Figure 6.1: STM image of tbu-pyton on Ag(111) [9] 6.3 Downshift of the work function The results of the measurements in section 5 (figure 5.2) show an unexpected downshift of the work function for small depositions of tbu-pyton below 4 Å, before the increase of the work function starts. We cannot determine with absolute certitude what is the cause of this phenonemon, nonetheless several possible explanations are considered in this chapter. Although discussed seperately, the effects do not exclude each other and might even all contribute their part.

30 24 The measurements done for PyTon in [5] do not include data for the range of 1 to 2 Å. For 3 Å coverage of PyTon an increase of the work function was measured. However, since PyTon has a smaller molecular weight it should create more dipoles than tbu-pyton for the same nominal coverage. So it is unknown if PyTon causes a similar decrease of the work function for a comparable number of molecules on the surface. That is why the speculations on the cause of the downshift do not concentrate on the butyl-groups which differentiate the two molecules Push back effect As explained in the second chapter, the formation of interface dipoles is supposed to increase the work function to a maximum value for a complete monolayer if a charge transfer from the metal to the molecule takes place. However, when depositing molecules on clean metal surfaces, the so called push back effect described in section 2.1.2, which causes a work function lowering must be considered [12]. The sum of the two effects (charge transfer and push back ) might deliver an explanation for the down and upward shift of the work function as observed in figure 5.2. A model based on the two following assumptions can be used to deduce quantitative estimations for the magnitude of the two effects necessary to explain the observed shift. Firstly, we suppose the increasing impact of the created charge transfer dipoles on the work function rises linearly until a monolayer is completed and the maximum value is reached according to the Helmholtz equation (see equation 2.1). This is reasonable because every newly deposited molecule adds an interface dipole. When the surface is completely filled no new interface dipole can be created. Secondly, the push back effect is getting stronger linearly as more molecules are deposited on the surface until the maximum push back possible is reached. Normally the push back effect should have its maximum for a monolayer as well. However, one could assume that for some reason the maximum push back is reached already for smaller deposition amounts. In this case the deposition amount with the maximum decrease in the change of the work function can be identified as the one where the maximum push back is reached. This model is illustrated in figure 6.2. On the Figure 6.2: Model of contributions of push back and dipole formation. a) Effects presented as seperate functions. b) Sum of the effects as one function depicted together with the data of figure 5.2

31 25 left side (figure 6.2 a) the effects are represented seperately. Their sum on the right side (figure 6.2 b)) reproduces the first part of the curve in figure 5.2 quite well. The slopes of the curves in figure 6.2 a) are calculated from the results in chapter 5 using the equations y 1 = a x 1 + b x 1 (6.1) y 2 = a x 2 + b x 1 (6.2) Being y 1 = 0, 125eV, y 2 = 0.43eV, x 1 = 3Å and x 2 = 5Å one obtains a = (y 2 y 1 )/(x 2 x 1 ) 0.28eV/Å and b = 0.32eV/Å. According to this analysis the two effects can be regarded seperately. This yields for a maximum pushback effect of Φ = b x 1 = 1.02eV and a theoretical maximum positive shift of the work function due to interface dipole formation of Φ = a x 2 = 1.39eV for tbu-pyton on Ag(111). In order to check if these results are reasonable they are compared with values found beforehand. For the push back effect among the highest values reported, we find -0.7 ev for tetratetra-contane (TTC) on Cu(111) [2] and -0.5 for Xe on Cu(111) [1]. Thus the above calculated value for the push back of ev is quite high. For the interface dipole induced by charge transfer density functional theory (DFT) modeling delivered values of +0.7 ev for hexaazatriphenylene-hexacarbonitrile (HATCN) and +1.7 ev for F4-TCNQ on Ag(111) [2]. Though the latter value does not consider an oppositely directed intramolecular dipole. Compared to these results the calculated value of ev seems to be possible, though also high. The linear evolution of the push back effect can be justified in the same way as for the interface dipoles. Every additional molecule contributes to the repulsion of the electronic cloud at its adsorbtion site. Coherently the maximum push back would be reached for a full monolayer. This means at the same time as the maximum interface dipole effect and not before, contradicting the assumption made for the model used above. One scenario to explain why the push back effect might reach its maximum before completion of the monolayer would be a coverage-dependent geometrical change of the unit cell size occupied by one molecule. Assuming the unit cell size shrinks with increasing deposition, additional molecules could still be added to the first layer to a certain extend, even after a complete coverage of the surface with the large unit cell. Since the surface is already covered, these molecules would not increase the push back effect, but still they would contribute additional interface dipoles. A shrinking unit cell with increasing coverage has been observed for copperphthalocyanine (CuPc) on Ag(111) [21]. The space occupied by one molecule varied from 217 Å 2 for a coverage of 0.89 monolayers to 192 Å 2 for a full monolayer, corresponding to a decrease of about 12 percent. The equivalent increase in deposited mass is 14 percent. In order to explain the behaviour of the work function in this study, the unit cell of tbu-pyton would need to shrink by about 40 percent from a nominal coverage of 3 Å to 5 Å, which corresponds to an increase in mass of 67 percent. A 40 percent decrease in size of a unit cell appears to be a rather improbable scenario.

32 26 Apart from this, one could argue that the shrinking of the copper-phthalocyanine unit cell described in [21], which is attributed by the authors to intermolecular repulsion, takes places before the a full monolayer is completed. However, this is not necessarily the case as can be seen when explaining the method used by the authors of [21] for coverage determination. They measured the coverage by XPS using the C1 peak integral. For the calibration of the method the coverage with the smallest unit cell available was used. Consequently when the authors of [21] write about a 0.89 monolayer coverage they use a misleading notation. They deal with a coverage that has 89 percent of the mass of the monolayer used for calibration, however since the unit cell of the 0.89 layer is bigger it could constitute a monolayer as well. Nevertheless, if repulsion between the molecules is the reason for the shrinking process, as supposed in [21], it is a legitimate question whether it can be energetically favourable for the molecules to squeeze themselves into a nearly completed monolayer instead of starting a second layer at a larger distance. To sum it up, there are some aspects contradicting the model outlined in this section being the only explanation for the work function decrease at small deposition amounts of tbu-pyton, although it could contribute partly Conformation effects Chemisorption of organic molecules is defined by a strong interaction with the metal surface, including charge transfer. Given this case, many molecules (like for example 3,4,9,10-perylene tetracarboxylic dianhydride (PTCDA) [8]) undergo a modification of their conformation at the adsorption on the surface due to the interaction with the metal. Sometimes this adsorption-induced distortion causes an intramolecular dipole which also has an influence on the work function [18]. The intramolecular dipole is caused by an inhomogenous distribution of electron density within the molecule due to some atoms with higher electronegativity compared to others. PyTon on Ag(111) has been found in a bend conformation for which DFT modeling predicts an intramolecular dipole which lowers the work function by Φ mol = 0.44eV [5]. This is to say the intramolecular dipole is oriented oppositely to the dipole caused by the charge transfer between the molecule and the metal surface. Due to the huge similarity of the two molecules, one could expect tbu-pyton to adopt a similar bend conformation with the same effect. One possible explanation of the observed down- and upward shift of the work function in the sub-monolayer regime could be that with increasing molecular density the interaction between neighbouring molecules needs to be taken into account, like it is for example the case for the intramolecular repulsion of CuPc [21]. This interaction might lead to a further change in the molecular conformation, weakening the intramolecular dipole Φ mol. If the intramolecular dipole gets weaker the sum of charge transfer dipole and the intramolecular dipole influence could be negative for a small nominal layer thickness of tbu-pyton and turn positive with increasing deposition amounts. From the data obtained for tbu-pyton in this study the alternative described in this section can neither be confirmed nor rejected. Maybe x-ray standing wave measurementes, like done by the authors of [18], at different film thicknesses could clarify the situation.

33 Combination of flat-on and edge-on molecular orientations Another idea on what is behind the reversion in the evolution of the work function is to consider a possible combination of flat-on and edge-on molecular orientations. A flat-on orientation describes a situation where the aromatic rings are found parallel to the substrate surface, while a so called edge-on orientation corresponds to upright molecules. For tbu-pyton one would consider a flat-on orientation the most probable configuration, since it allows the electron accepting centre of the molecule (the former PyTon) to be as close a possible to the surface. Generally a flat-on conformation is known to be the normal case as long as substrate-molecule interactions are stronger than intermolecular interactions [2]. However some aromatic molecules have been found to exhibit a combination of flat-on and edge-on orientations. Up to the first half of the monolayer p-hexaphenyl and p-quaterphenyl adsorb in a flaton orientation. Then the following molecules place themselves between the flat-on molecules in an edge-on orientation [29] [25] [24] [23]. A similiar behaviour, though not for the first but for the second monolayer, was observed for quaterrylene [6]. We could assume that the tbu-pyton molecules also adsorb first in a flat-on and later, starting at about half a monolayer, in an edge-on orientation. In this case it could be possible that the flat-on adsorbed molecules, due to various effects (push back, intramolecular dipole), tend to decrease the work function despite the interface dipole they induce. In contrast, the edge-on adsorbed molecules could have an overall increasing effect on the work function because their electron accepting centre is further away from the surface, a fact that should increase the interface dipole moment. They also should have a modified molecular conformation, which might reduce the intramolecular dipole. This is for example the case for HATCN, another organic electron acceptor molecule, on Ag(111), which has a larger intramolecular dipole in the flat-on than in the edge-on conformation [3]. Similarely to the previous section, more information is needed in order to confirm or exclude this explanation. Maybe more STM images of a sample prepared in the same way as for this study could help Reorientation of the molecules from a flat-on to a edge-on orientation Instead of a different adsorbtion orientation of the molecules from the second half of the monolayer one could also imagine a reorientation of the molecules as a cause for the described behaviour of the work function. For HATCN on Ag(111) a somehow analogous evolution of the work function has been observed. At small deposition amounts of less than 3 Å the work function, though it does not decrease, stays essentially constant before starting to increase strongly. It has been shown that this evolution follows a gradual reorientation of the flat-on molecules to an edge-on orientation [3]. This is explained by the strong bonding of the outer functional CN groups of HATCN to the metal. This bonding is stronger in the edge-on orientation. Nevertheless the flat-on orientation is energetically more favourable until a monolayer is completed, because more CN groups are in contact with the surface. However, once the surface is covered this number can be increased by the reorientation of the molecules to the edge-on orientation [3].

34 28 Notwithstanding the butyl-groups, which are the outer functional groups of tbu- PyTon, have a lower electron density than the CN groups. So probably their bonding to the surface is not as strong as the one of the CN groups of HATCN. Therefore it can be regarded as rather improbable that a similar reorientation takes place for tbu-pyton, though one should not exclude this option completely Step edge decoration The metal crystal used as a substrate for tbu-pyton is Ag(111). The (111)-surface is not completely smooth, but presents so called step edges where an atomic terrace ends. An increase in density of step edges of a (110) tungsten crystal has been shown to reduce the metals work function [20]. This can be explained by the smoothing effect of the surface of the metal s electronic cloud, introduced by Smoluchowski: Every atom is surrounded by a Wigner-Seitz cell, including all points which are closer to this atom than to any other of the crystal lattice. Within the bulk of the crystal, all atoms have the same radial charge distribution in the Wigner-Seitz cell. However at the surface the charge distribution is disturbed because of the missing next atomic layer. It is energetically more convenient for the electrons to remain in the depressions rather than in the elevations formed by the Wigner-Seitz cells of the surface atoms. The consequence is a dipole layer formation with a decreasing effect on the work function [30]. The smoothing effect is less pronounced for densly packed than for open structures like step edges [20]. One could imagine that the tbu-pyton molecules might have a preference to adsorb on the step edges before covering the residual terraces. At least this behaviour has been seen for other organic molecules. The organic semiconductor rubrene is an example of a molecule that decorates the step edges of an Ag(111) crystal in the first place [28]. If this is the case for tbu-pyton, the work function increasing effect of the molecules might not be able to compensate the smoothing effect at the step edges. This way an overall decrease of the work function could be possible for small deposition amounts. With an increasing nominal layer thickness the terraces would be covered as well, leading to a reversion in the trend by increasing the work function, because at this point the smoothing effect looses importance for the average work function. STM images of the sample for different small deposition amounts would be useful to see if step edge decoration takes place, which is a necessary condition for this scenario. 6.4 Further evolution of the work function After the maximum increase, the change of the work function decreases again, compared to its original value. The most probable explanation for this behaviour is charging due to an ionization of the atoms in the sample as described in section 3.2. This is supported by the results given in section 5.3. An evidence of the influence of charging on the observed decrease in the work function is the attenuation of the decreasing trend by the measurement done with reduced photon beam intensity that can be seen in 5.2. Even if the former level of work function increase is not reached it can be assumed that this is only due to the impossibility of further intensity reduction of the helium lamp. The down shift of the work function upon the deposition

35 29 of α-npd in the third series of measurements 5.6 can also be attributed to charging effects. 6.5 Hole injection barrier reduction The desire to reduce the hole injection barrier h from the metal electrode to the hole transport layer is the motivation for efforts to increase the metal s work function. Therefore it is important to check if an observed work function increase actually translates into a reduction of h. The h of α-npd on pristine silver is 1.4 ev [2]. For α-npd on a tbu-pyton covered Ag(111) crystal a h of 0.75 ev was measured (section 5.2.1), thus a reduction of 0.65 ev was achieved. Unfortunately no reference value for PyTon is available, so it can not be determined if the h reduction is bigger with tbu-pyton. The following calculation shall visualize the consequences of the achieved barrier reduction for the current through a metal/orgnic interface in an organic electronic device. The contact between a metal and a semiconductor is called a Schottky diode. The current density j S through an ideal Schottky diode depends exponentially on the charge injection barrier [2]. j S = A T 2 exp( h k B T ) exp( eu ext βk B T 1) (6.3) In this so called Schockley formula A = 4πm (k 2 B/h 3 ) denotes the Richardson constant with m being the reduced mass of the charge carriers, T the absolut temperature, U ext the applied voltage and β an ideality factor of the diode. A reduction of h by x results thus in a current density j S given by j S = j S exp( x k B T ) (6.4) At room temperature k B T is approxemately 25meV. For an ideal Schottky contact a hole injection barrier decrease of 0.65eV, as achieved in this study with tbu- PyTon for an Ag(111)/α-NPD interface, would thus lead to a theoretical increase of the current density by eleven orders of magnitude. Such an increase can not be expected in practice. However, it gives an impression of how already a relatively small h reduction can increase the current density considerably.

36 Chapter 7 Summary There is a great interest in organic electronis to reduce the hole injection barrier at the interface of organic materials and metal electrodes in order to improve device performance. One way to achieve this objective is to increase the work function of the metal by the deposition of an electron acceptor molecule between the organic material and the metal surface. The Helmholtz equation states that the work function increase due to the interface dipole created by the molecule depends on its distance to the surface. The aim of this work was to determine if the work function increase reached for a given system (PyTon on Ag(111)) could be augmented by enlarging the molecules distance to the metal surface. This was achieved by modifying the molecule with bulky substituents: tbu-pyton, which consist of a PyTon molecule enhanced by two butyl groups, was chosen for the comparative measurements. Analysing the dependence of the work function on the tbu-pyton film thickness clarified the optimal layer thickness. Additionally it was examined how the work function increase affects the hole injection barrier from the Ag(111) crystal into α-npd, which served as a prototypical hole transport layer. An increase of the work function by 0.43eV as a consequence of the deposition of tbu-pyton was measured. This is a slight improvement of 0.08eV, compared to the value for PyTon [5]. The magnitude of the effect is in good agreement with an estimation based on the Helmoltz equation. These two observations can be regarded as evidence in favour of the validity of the original approach. Nevertheless, further investigation on the conformation and orientation of the tbu-pyton on the Ag(111) surface would be necessary to clarify if this is actually the case. A reduction of the hole injection barrier from Ag(111) into α-npd by 0.65eV was measured. Using the Shockley formula it was calculated that for an ideal Schottky contact between a metal and an organic semiconductor such a reduction signifies a theoretical increase of the current density by eleven orders of magnitude. Even though such an increase is not realistic in practice, it shows that already a relatively small reduction of the hole injection barrier can improve the current density significantly. For small deposition amounts of tbu-pyton an unexpected decrease of the work function was observed. It is unknown if PyTon on Ag(111) has a similar effect, because there are no data for these layer thicknesses. Five different approaches to explain this phenomenon were discussed in detail. A combination of the push back effect and the influence of the interface dipoles was examined as a possible cause, as well as a density dependent change of the intramolecular dipole. Furthermore a 30

37 31 combination of flat-on and edge-on orientated molecules and a reorientation from a flat-on towards an edge-on orientation were considered. Additionally the possibility of step edge decoration leading to a work function decrease due to the smoothing effect described by Smoluchowski [30] was discussed. Further investigations of tbu-pyton layers on Ag(111) with other anlaysis methods, as for example STM, XSW or reflection absorption infrared spectroscopy (RAIRS) at different film thicknesses might help to decide if one of these ideas includes what actually happens.

38 Acknowledgement First I would like to thank Prof. Norbert Koch for the possibility to do my bachelor thesis in his research group and the advice for the data interpretation. Special thanks go to Dr. Benjamin Bröcker for the supervision. Beginning with the patient introduction to the work with the UHV chamber, the assistence during the measurements to the discussion of the results his support was outstanding. I am very grateful to Stefanie Winkler for elucidating discussions and countless helpful comments. Also, I thank Johannes Frisch for answering small and big questions and help with handeling the Origin program, as well as Raphael Schlesinger for assistence in the preparation of measurements. In general I am thankful to the SMS group for the friendly working atmosphere. For providing the STM images and their explanation I would like to thank Dr. Hendrik Glowatzki. I would like to thank Héctor for the help with the computer, his patience and understanding. Additionally I would like to thank my family for their support in my studies and especially my brother Moritz for the logistic help while handing in this work. 32

39 Appendix A A.1 STM images of PyTon Figure A.1: STM image of PyTon taken by Hendrik Glowatzki [9] 33