Synchrotron radiation study of free and adsorbed organic molecules

Transcription

1 Synchrotron radiation study of free and adsorbed organic molecules Licentiate Thesis Teng ZHANG December 2016

2 Abstract In this licentiate thesis, organic molecules, namely Cobalt Phthalocyanine (CoPc) and Biphenylene, have been studied by means of synchrotron radiation-based spectroscopic methods (Photoemission Spectroscopy (PES) and X-ray Absorption Spectroscopy (XAS) in combination with Density Functional Theory (DFT) calculations. Paper I is a combined experimental and theoretical investigation of electronic structure of CoPc. addressing the atomic character of the Highest Occupied Molecular Orbital (HOMO) and the electronic configuration of the molecular ground state. Both these aspects are still under discussion since different experimental and theoretical studies have given controversial results. Previous works have indicated the CoPc ground state to either be described by the 2 A 1g or 2 E g, or by a mix of the two electronic configurations. Regrading the debated the atomic character of the HOMO of CoPc, it has been suggested to be either metal 3d-like and localized on the central Co atom or originating in the organic ligand of the molecule. In this thesis the valence photoemission results for CoPc in gas phase and as adsorbed films on Au(111) together with the DFT simulations, consistently indicate that the HOMO is derived only by the organic ligand, with mainly contribution from the carbon atoms with no metal character. Moreover, the good agreement between the experimental and theoretical results, confirms that the ground state of CoPc is correctly described by the 2 A 1g configuration. In Paper II, PES and XAS have been used to investigate the occupied and empty density of states of biphenylene films of different thicknesses, deposited onto a Cu(111) crystal. The results have been compared to previous gas phase spectra and single molecule Density Functional Theory (DFT) calculations to get insights into the possible modification of the molecular electronic structure in the film induced by the adsorption on a surface. Furthermore, XAS measurements allowed the characterizion of the variation of the molecular arrangement with the film thickness and helped to clarify the substrate-molecule interaction. c Teng ZHANG 2016

3 List of papers This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I II Exploring the electronic structure of CoPc by photoemission and absorption spectroscopy T. Zhang, I. Brumboiu, J. Lüder, C. Grazioli, V. Lanzilotto, E. Giangrisostomi, R. Ovsyannikov, Y. Sassa, I. Bidermane, M. Stupar, M. de Simone, M. Coreno, B. Ressel, M. Pedio, P. Rudolf, B. Brena and C. Puglia Electronic structure investigations of biphenylene films R. Totani,C. Grazioli, T. Zhang, I. Bidermane, J. Lüder, M. de Simone, M. Coreno, B. Brena, L. Lozzi and C. Puglia Submitted to J. Phys. Chem.

4 Comments on my own participation The studies presented in the licentiate thesis are a result of teamwork. The work cannot be achieved without the contribution and collaboration of all the authors, expert either in experiment or theory, listing in the two papers. In Paper I, I have been the main responsible for performing the experiments, analyzing the data and writing the manuscript. In Paper II, I have participated in the experiment and discussions, as well as partly involved in writing the manuscripts.

5 Contents 1 Introduction Molecules Cobalt Phthalocyanine (CoPc) Biphenylene Techniques Photoelectron spectroscopy (PES) X-ray Photoelectron Spectroscopy (XPS) Satellites in PES Ultraviolet Photoelectron Spectroscopy (UPS) Angle-Resolved PhotoEmission Spectroscopy (ARPES) and Angle-Resolved Time-Of-Flight (ARTOF) spectrometer Near Edge X-ray Absorption Fine Structure Spectroscopy (NEXAFS) Experimental set-up The GasPhase Photoemission beamline at ELETTRA LowDosePES endstation at PM4 beamline at BESSY II The I311 beamline at MAX IV laboratory Sample preparation Gas phase measurements Adsorption measurements Substrate preparations Film preparations Summary of papers Cobalt Phthalocyanine (CoPc) and Biphenylene The electronic structure of CoPc XAS of Co L-edge of CoPc CoPc Valence levels CoPc film on Au (111) surface Comparison of valence levels in Pc s measured in gas phase Biphenylene on Cu (111) C 1s core level photoemission

6 6.3.2 Valence photoemission NEXAFS Conclusion Acknowledgement References

7 1. Introduction Natural materials have been fundamental resources for the human society from ancient era. However, the development of the new technologies within energy and information, both essential driving forces of nowadays world, demands for novel materials. Along with the development of our technological society, the call for materials is not anymore restricted to what is found in nature, but to more complex products, obtained by processing at different levels the natural materials by chemical or physical methods, or to purely man-made materials. Nowadays we need novel materials for a wide range of human activities. We want to improve the waterproof and breathable performance to something better than Gore-Tex for our out-door jacket, the thermal insulation ability and durability for space shuttle thermal protection system, the efficiency for solar cells, the biocompatibility and lifetime of material for artificial organs for biomedical applications, magnetic properties of materials for data storage, the electronic performance of transistors and thus to get faster and stable CPUs, the sensing capability of materials for sensors devices being as the most important components of Internet of Things (IoT) [1], just to name a few. As a result of being the carrier of a variety of important properties, the research of novel materials is always highly promoted. To maximise the interaction property of a material as well as to minimise the size of a device, functional materials are now processed and studied down to nanoscale. At such scale, the properties of a material are often completely different from the normal size material. For example, the surface heterogeneous property would be more critical for a nanomaterial while homogeneous property is more significant for a bulk material. Surface and interface properties of a material are crucial to applications such as heterogeneous catalysis, semiconductor device fabrication, solar cells, fuel cells, molecular devices built by bottom-up self-assembly processes. As a preliminary step towards making such ideal tiny devices, studies of thin films of interesting molecules with special potential, grown in carefully controlled way, are done. We usually make such molecular films of different thicknesses in-situ (from few Ångström to some nonometer) in Ultra High Vacuum (UHV) chambers with a pressure lower than 10 8 mbar. Then the electronic properties of the films are studied by PhotoElectron Spectroscopy (PES) and X-ray Absorption Spectroscopy (XAS) with synchrotron radiation light, which is an effective powerful light source dedicated to the front-line material research of great value in both industry and science. To exclude the influence of the substrate on the electronic structure of the molecules, gas phase measurements are sometimes performed 1

8 helping to identify the intrinsic electronic property of the isolated molecule and moreover to enlighten the modification induced by the adsorption on surfaces and in films. In this thesis, two molecules have been studied: Cobalt Phthalocyanine (CoPc) and biphenylene. Phthalocyanines (Pc s) are blue-colour compounds widely used in industry since their first synthesis by chance in 1928 [2]. However Pc s might have been accidentally prepared long before such date [3]. Pc s exhibit fascinating physical and chemical properties that make them important in molecular functional materials. Pc s have been subject of a huge number of studies because of their multiple applications such as dyes, catalysts and coatings. At present they are one of the most studied organic materials for possible applications in nanodevices and spintronics [4]. However, the amount of research investigations conducted on Pc s in gas phase compared to the Pc s films is quite limited [5] [6]. Recently, studies have shown the potential of using transition metal Pc s (with a transition metal atom in the molecular centre) and in particular CoPc as single-molecule magnets [7]. The magnetic and electronic properties of CoPc are determined by the occupation of the 3d orbitals of the transition metal in the center of the molecules [8]. However, until now, neither experimental results or density functional theory (DFT) computational calculations have been able to give a clear description of the ground state configuration of CoPc as well as the d-state contributions to the electronic valence structure [9] [10] [11] [12]. In Paper I, a comparison between our gas phase experimental results and new DTF simulations confirms that the CoPc ground state is correctly described by the 2 A 1g electronic configuration. The atomic contributions to the Highest Occupied Molecular Orbital (HOMO) of CoPc have been addressed by performing different photoemission spectroscopy experiments on CoPc both in gas phase and thin film samples. Our results clearly show that the HOMO is formed only by the organic ligand, with mainly contribution from the carbon atoms and slightly from the nitrogens. Biphenylene (C 12 H 8 ) is a cyclic hydrocarbon composed of two benzene rings connected by a cyclobutadiene ring. It was the first π-electronic hydrocarbon systems discovered to show evidence of antiaromaticity [13]. It has special properties due to presence of a formal cyclobutadiene ring which makes it interesting from both fundamental and technological point of views [14]. For instance, polyphthalidylidene biphenylene thin films can achieve anomalously high conductivity. Moreover, being the initial precursor of a 2- D porous graphite-like molecular network, named biphenylene carbon, it is expected to play a major role as a novel organic material [15]. Previous gas phase measurements as well as DFT calculations gave a detailed description of the characteristic contributions by the non-equivalent carbon atoms (C α, C β, and C γ ) present in the molecule to the C 1s X-ray photoemission spectrum [16]. In Paper II, we present a study by means of PES as 2

9 well as Near-Edge X-ray Absorption Spectroscopy (NEXAFS) of biphenylene films of different thicknesses adsorbed on Cu (111) surface. The substratemolecule interaction of biphenylene with the the Cu (111) surface could be addressed by comparing our experimental and theoretical results with the previous gas phase study. The molecular orientation changing from lying to standing on the surface with increasing coverages, was found by NEXAFS. The biphenylene was found weakly adsorbed on Cu (111) surface and the films preserving a molecular like character. 3

10 2. Molecules 2.1 Cobalt Phthalocyanine (CoPc) The Phthalocyanines (Pc s), discovered accidentally in 1907 by Braun and Tcherniac, are intensive blue-green colored macrocyclic compound. They are similar to porphyrins which are very important in nature for taking part in different processes and reactions in living organisms, as they are the active site in many different enzymes, hemoglobin, and in chlorophyll just to mention a few. Unlike porphyrins, found in nature, Pc s are synthetic organic planar aromatic macrocycles formed by four isoindole units linked together by four bridging nitrogen atoms. In Figure 2.1, a metal-free Pc, H 2 Pc (with a molecular formula C 32 H 18 N 8 ) (a) is shown beside a metal Pc (MPc) (b), obtained by replacing the two hydrogen atoms at the center of the metal-free Pc by a metal atom, as Co in Figure 2.1 (b) for example. The Pc central ring can host almost all types of metal atoms and a variety of substituents on the ligands can be also used to modify the Pc s, tuning their electronic and optical properties. (a) H 2 Pc (b) CoPc (c) D 4h ligand field splitting E b 1g NH NH a 1g e g b 2g Figure 2.1. Molecular structure of metal-free phthalocyanine (a) and cobalt phthalocyanine (b). The schematic representation of the d-level splitting in a D 4h ligand field and a simple picture of one of the possible Co 3d electronic structures, in accordance to the ground state we calculated by DFT, is shown in (c). Nowadays Pc s are commonly used as pigment in the industry being a dye of very high thermal stability and low cost. However a variety of different applications, in optolelctronics (OLEDs for example) and solar cells, in catalysis [17] [18] [19], in fuel cells [20] and in sensing materials (organic field effect transistors used as gas sensors [21]), have also been widely implemented. Pc s 4

11 are perfect candidates for these kinds of applications due to their photochromic properties, thermal and chemical stability, together with their attracting molecular structure that makes them forming supramolecular self-assemblied thin films. In films and bulk materials, the Pc s form molecular stackings where the orientation of the molecules results in different crystalline phases (for example α and β phases). Moreover, the stacking orientation with respect to the surface often varies with the thickness of the deposited film leading to slightly different electronic structures/property of the material [4]. Recent studies have also shown the possibility of using Transition Metal Phthalocyanines (TMPc s), which have a transition metal in the molecular centre (like Co, Fe and Mn), as single-molecule magnets [7]. In this thesis, CoPc has been studied. The molecular structure is shown in Figure 2.1 (b). It is known that the TMPc s are planar organic molecules characterized by a delocalized electron distribution in the organic ligands that form bivalent bonds to the central transition metal atom via the central nitrogen atoms. The electronic structure of TMPc is determined by the bonding configuration of the 3d-states of the central transition metal with the organic ring [6] [22]. The molecular symmetry of flat TMPc belongs to the D 4h point group. According to the ligand field theory [23], placing a transition metal in a D 4h symmetry environment (as in TMPc s) causes the splitting of the degenerate 3d states into b 1g (d x 2 y 2), a 1g(d z 2), e g (d zx,d yz ) and b 2g (d xy ) levels, being e g a doublet, as shown in Figure 2.1 (c). The 3d 7 open shell configuration of the CoPc results in a total spin S = 1/2, with in total 3 holes in the 3d orbitals and the b 1g (d x 2 y2) orbital completely empty. There are debates on the TMPc s thin film morphology when adsorbed on surfaces and their related electronic-magnetic properties [24]. In particular, the electronic configuration of the ground state of CoPc is still unclear. Until now DFT computational simulations have not been able to give a unique description of the d-state contributions to the electronic valence structure due to the complexity to describe the hybridization of the Co 3d states with the Pc molecular orbitals (MO) [12]. Experimental and theoretical results of CoPc thin films showed the electronic ground state configuration to be a mix of the 2 A 1g and the 2 E g first excited state, at slightly higher energy [9] [10] [11]. Other studies have instead proposed the 2 E g configuration with a 3d-hole in the e g state, as description of the ground state of CoPc [8] [25]. The character of the Highest Occupied Molecular Orbital (HOMO) of CoPc molecules is also still under debate, suggested to be either metal 3d-like and localized on the central Co atom [25] [26], or originated in the organic ligand of the molecule [27]. In this thesis, the atomic character of the HOMO of CoPc has been addressed by performing photoemission spectroscopy experiments measured on both CoPc gas phase and film samples. By varying the photon energies used in the investigations, we could disentangle the atomic contribution to different 5

12 valence photoemission features and in particular to the HOMO peak, taking into account the diverse photoemission cross-sections of Co, C and N. Together with the ab-initial calculation, our results clearly show that the HOMO is only formed by the organic ligand, with mainly contributions from the carbon atoms and slightly from the nitrogens. Moreover, a comparison between our experimental results and new DFT simulations performed by our colleagues, confirms that the molecular ground state is correctly described by only 2 A 1g electronic configuration. 2.2 Biphenylene Biphenylene, first synthesized by Lothrop in 1941 [28], is a pale yellowish powder with melting temperature of about 110 C. As shown in Figure 2.2, it is a cyclic hydrocarbon with chemical formula C 12 H 8, formed by two benzene rings and a cyclobutadiene ring in between. J. Waser et al. first calculated the crystal structure of biphenylene in 1944 proposing a planar structure for the biphenylene molecule [29]. It was one of the first hydrocarbon systems discovered to show evidence of anti-aromaticity due to its electronic structure with 4nπ-electrons [30]. Figure 2.2. Structure of biphenylene molecules, formed by two benzene rings and a cyclobutadiene ring in between. However, although antiaromatic, the long distance between the two benzyne groups weakens the bonding of the cyclobutadiene ring giving biphenylene an intermediate character between anti-aromatic and aromatic molecule [16]. This can explain why biphenylene is a stable molecule [31] but however more reactive than benzene [14]. The intermediate aromatic and anti-aromatic character, uncommon photophysical properties [32] and anomalously high conductivity shown in thin polyphthalidylidene biphenylene films [33], make biphenylene a very interesting molecule, target of many experimental and theoretical studies [34] [33] [35] [36] [37]. Novel two dimensional (2-D) functional materials, like graphene or boronnitride, are considered to be prospective candidates for the next-generation electronic and optical devices [38] [39]. In this connection, biphenylene becomes interesting being the precursor of a 2-D porous graphene-like molecular network called biphenylene carbon. This new material would be characterized by delocalized band with good dispersion and gap separation [15]. Bipheny- 6

13 lene sheets, originating from biphenylene dimers [40], have been hypothesized and their one-dimensional derivatives, like ribbons and tubes, have been theoretically investigated [41]. Our previous results were obtained using a combined experimental and theoretical analysis of the core, valence, and unoccupied electronic states of biphenylene in gas phase [16]. However, to make use of the functionality of the molecules in more technological interesting system (like in biphenylene carbon), a study of biphenylene adsorbed in films of different thicknesses on different substrates has been carried out for tracing the molecular growing process with biphenylene as building blocks. The most promising surface to grow this kind of covalently-bonded carbon-materials is to use coinage metal surfaces, i.e. Au, Ag, Cu, Ni, etc since this kind of 2-D structure can be obtained by funtionalizing the biphenylene rings with bromine. In fact, previous studies have shown coinage surfaces to be able to promote a 2-D molecular ordering when the building blocks are funtionalized by halide atoms [42] [43]. In this thesis, Cu (111) crystal has been used to grow biphenylene films of different thicknesses. A comparison with the previous gas phase results allowed to characterize the electronic structure modifications induced by the molecules-substrate interactions for low molecular coverages, and the molecular arrangement within the film of different thicknesses. 7

14 3. Techniques 3.1 Photoelectron spectroscopy (PES) PhotoelElectron Spectroscopy (PES) is based on the photoelectric effect, a phenomenon first detected by Heinrich Hertz in 1887 and also known as the Hertz Effect [44]. According to the photoelectric effect, electrons are emitted from their original electronic orbitals to the vacuum level when the material interacts with photons of enough high energy for being ionized. A schematic description of the process of various PES techniques are presented in Figure 3.1, namely core and valence level photoemission spectroscopy. For core level photoemission X-ray photons are usually used and for this reason the technique is also called X-ray Photoelectron Spectroscopy (XPS). On the other hand, Ultraviolet Photoelectron Spectroscopy (UPS) indicates photoemission investigations of valence levels using UltraViolet (UV) light. Vacuum e - e - hν hν Ground State XPS UPS Figure 3.1. A schematic illustration describing the process examined by the PES technique, namely XPS and UPS. During the XPS process, electrons as deep as from the core level are emitted after the excitation by the incoming photon hν. In a UPS experiment, electrons from the valence level are emitted. A typical experimental setup of PES is shown in Figure 3.2. After the photoemission process, the kinetic energy of the photoelectrons, E k, is measured by the electron analyser. Then, considering the energy conservation, knowing the excitation energy (hν) and measuring the kinetic energy E k of the photoemitted electrons, the binding energies can be determined according to the formula: E B = hν E k φ spec, (3.1) 8

15 where φ spec is the analyzer work function, which is an intrinsic property of the analyser and its setting parameters. (From the photon source) hν z e - (To the analyser) y electron analyser θ φ x sample Figure 3.2. An experimental setup of a Photoemission Spectroscopy (PES) experiment. A photon impinges on a sample and an electron is excited, escaping into vacuum and finally detected by an electron analyzer. Polar (θ) and azimuthal (ϕ) angles of the photoelectron with respect to the normal to the surface (z-axis) and to the plane (x-axis), respectively, are displayed. When studying surfaces and adsorbates, the sample and the spectrometer are usually in electrical contact so that their Fermi levels get in good alignment. For these experiments, the binding energies are referred to the Fermi level since the Fermi edge can be used as an internal calibration point. In gas phase PES measurements, the binding energies are calibrated referring to the vacuum level. In this case, a reference gas, with spectroscopic lines of known binding energy (with reference to vacuum level), is introduced and measured simultaneously with the sample in order to calibrate the binding energy. Due to, among other things, the different references used for the binding energy scales, comparing film and gas phase measurements requires a shift to align the corresponding spectral features. Each chemical elements has its own, unique set of binding energies, whereby PES can be used as a powerful method for chemical analysis. For this reason, PES is also called Electron Spectroscopy for Chemical Analysis (ESCA) [45] [46]. Usually PES requires Ultra-High Vacuum (UHV) conditions to prolong the mean free path of the outcoming photoelectrons [47]. PES obviously requires a sufficiently monochromatized photon source in order to give accurate information. The electron kinetic energies (E k ) are measured either by deflecting the emitted electrons with a Lorentz force using magnetic or electrostatic fields (e.g. in a hemispherical analyser, as the SES200 used in our gas phase experiments) or by applying a drift tube and measuring the time-of-flight of the electrons, i.e. velocities (for example the ARTOF spectrometer used in some of the experiments reported in this thesis). 9

16 3.1.1 X-ray Photoelectron Spectroscopy (XPS) As already mentioned, X-ray Photoelectron Spectroscopy (XPS), as shown in Figure 3.1, also known as Electron Spectroscopy for Chemical Analysis (ESCA), is a quantitative technique using high energy photons for the excitation (typically in the range from 200 ev to 10 kev). The technique was developed by Professor Kai Siegbahn from Uppsala and his collaborators. The technique was quickly applied in the field of surface science, and it is still considered one of the most powerful techniques for material characterization. It can often be considered as being essentially nondestructive to most materials, including organic molecules and polymers. Kai Siegbahn was awarded the Nobel Prize in 1981, for his contribution to the development of electron spectroscopy. XPS measures the binding energies of the atoms in the sample and it is an element-specific method. However, the conventional position of the binding energy of a specific element and the line profile of a spectral feature can vary as a result of the chemical environment of the atom or molecule. For core levels we can observe a chemical shift giving more detailed information about the bonding and the interaction of the atom with the surroundings. Through a careful analysis of the chemical shift, we can study changes in the electronic structure of a system. For an adsorbate we can, for instance, characterize the interaction and bonding with the surface or with other atoms/molecules. The shifts may also be sensitive to the orientation of the adsorbates and to their adsorption sites Satellites in PES The photoemission process is fast and it can often be treated within the so called sudden approximation. The systems we study are multi-electron systems, where interactions between electrons will give rise to satellite features in addition to a main peak. In simple picture, the satellites may be viewed as excitations in the core hole final state. For excitation to bound states we use the term shake-up and for excitation to continuum states we use the term shake-off. The shake-up features carry information of the multi-electron system. Detailed information such as the atomic chemical environment, HOMO-LUMO gap for adsorbed molecular systems and solids can be obtained by studying the shake-up s Ultraviolet Photoelectron Spectroscopy (UPS) Ultraviolet Photoelectron Spectroscopy (UPS) is a technique developed by David W. Turner at 1960 s for valence photoelectron studies of free molecules in the gas phase. 10

17 UPS uses photon energies ranging from few ev to hundred ev, i.e. low photon energy which allows a detailed investigation of the valence orbitals as shown in Figur 3.1. In the thesis, UPS has been used to study both molecular vapor (gas phase) and solid films Angle-Resolved PhotoEmission Spectroscopy (ARPES) and Angle-Resolved Time-Of-Flight (ARTOF) spectrometer As described above, the normal PES detects the kinetic energy of the photoemitted electrons. In Angle-Resolved PhotoEmission Spectroscopy (ARPES), a technique of refinement of the ordinary PES, the emission angles of the emitted electron, θ and ϕ, are also registered (see Figure 3.2). Thus, since electrons with different momenta escape at different angles from the surface of the sample, ARPES measurements allow to determine the electron momentum, as it can be calculated from the emission angles of electrons with respect to the sample surface. The relationship between Ek vacuum and the momentum k vacuum of the photoemitted electrons is: Ek vacuum = (kvacuum ) 2 2m The parallel component of the k-vector is conserved: (3.2) hk sample = hk vacuum (3.3) With the basic geometry of an ARPES measurement shown in Figure 3.2, we get: hk sample = hk vacuum = 2mE k sinθ (3.4) In summary, ARPES is a powerful tool to get simultaneously information about the energy and the momentum of an electron emitted from the sample surface after the photoionization process. Specifically, by locating the crossings of the bands with the Fermi level, the Fermi surfaces can be obtained. ARPES is one of the most direct methods of studying the electronic structure of surfaces and solids. Angle-Resolved Time-Of-Flight (ARTOF) spectrometer is a new type of analyzer for ARPES. An ARTOF spectrometer measures at the same time energy and momentum of the electrons with high transmission and high resolution. The construction of the new ARTOF spectrometer dates back to 2004 [48]. The fundamental principle of the ARTOF is to use time-of-flight measurement to determine the electron velocities and thus the electron kinetic energies. An advanced electron lens system makes it possible to simultaneously 11

18 determine the emission angles of the electrons. Using the time-of-flight concept, ARTOF needs pulsed sources for the excitation with maximum repetition rate below 3 MHz. Thus ARTOF needs a synchrotron radiation facility of a special kind or a pulsed laser source. In the thesis, all the experiments carried out with ARTOF analyzer were done at the LowDosePES endstation at PM4 beamline at the BESSY II synchrotron located in Berlin, Germany. 3.2 Near Edge X-ray Absorption Fine Structure Spectroscopy (NEXAFS) Unlike PES that probes occupied electronic structure, NEXAFS investigates the unoccupied atomic/molecular valence electronic states in the presence of a core-hole [49]. NEXAFS is also an element-specific technique. A NEXAFS spectrum is measured by irradiating the sample with a tunable light source to excite core electrons into unoccupied valence levels. One usually records NEXAFS spectra by monitoring the amount of decay processes, e.g. through emission of a photon (fluorescent decay) or emission of Auger electrons. For low-z elements such as N or C the Auger decay is more probable while for high-z element it is more common to have fluorescence decay. In general, a NEXAFS spectrum reflects the dependence of the photoabsorption cross section vs the photon energy near the absorption edge, i.e. from just below the certain ionization threshold up to around 50 ev above it [50]. The X-ray absorption cross section can be calculated from Fermi s Golden Rule where we expand the Hamiltonian as follows: σ x f e p i 2 δ ( E f E i hν ) ρ f (E), (3.5) where σ x is the X-ray absorption cross section, e the unit electric field vector, p the dipole transition operator and ρ f (E) is the density of final states. From Equation 3.5 we see that the transition intensities depend on the orientation of the electric field vector relative to the orientation of the molecule. In this way NEXAFS makes it possible to monitor / determine real space orientation of the molecules adsorbed on a surface. The intensity will be proportional to the direction of the final state orbital according to [49]: I e f p i 1s 2 cos 2 δ (3.6) with δ being the angle between the electric field vector E of the incoming light, and the direction of the final state orbital. This describes how the intensity of a resonance can change by varying the angle of incidence of the light. NEXAFS is often recorded in the following modes: partial, total or Auger electron yield by multi-channel plate or by electron spectrometer. During the measurement for Auger electron yield by an electron analyzer, we usually choose a fixed kinetic window for the Auger electrons of the target decay channel. 12

19 4. Experimental set-up 4.1 The GasPhase Photoemission beamline at ELETTRA GasPhase is a beamline at the Elettra synchrotron facility in Trieste, Italy, characterized by high resolution and high photon flux and devoted to research on gaseous systems. It is an undulator-based beamline with a monochromator consisting of a plane mirror and five spherical gratings. The beam line provides a photon energy range from 13 ev up to 900 ev and a photon spot size of around 200 µm at the target. The base pressure of the experimental chamber is in the 10 9 mbar range. However, when measuring a gas phase sample, the pressure in the chamber can reach 10 5 mbar but, thanks to the differential pumping system, the pressure in the analyser is kept within a safe working range. In order to get a better base pressure, a cryo-trap, cooled by liquid nitrogen, has been used during the experiments. During our characterizations of the CoPc molecules, the experimental chamber was equipped with a home-built effusive molecular oven nozzle [51] and a VG Scienta SES200 spectrometer mounted at the magic angle (54.7 ) with respect to the beam of the linearly polarized incident light. 4.2 LowDosePES endstation at PM4 beamline at BESSY II The LowDosePES is a permanent experimental station located at the PM4 dipole beamline at BESSY II synchrotron in Berlin, Germany. It is equipped with a Angle-Resolved Time-Of-Flight (ARTOF) spectrometer as well as with a VG Scienta SES100 hemispherical analyser. For excitation the ARTOF needs a pulsed source with a maximum repetition rate of few MHz. The BESSY II ring works in three modes, Standard Fill Pattern (Multi Bunch Hybrid), Single Bunch Operation and Low-alpha Multi Bunch Hybrid Mode. In Single Bunch Operation only one bucket with a bunch current of 13 ma is populated with electrons. The photon pulse train derived from this bunch has a frequency of 1.25 MHz. Single bunch operation is the most suitable operation mode for ARTOF experiments. However, only two weeks of single bunch operation twice a year are provided. For the other 2 hybrid operational modes, LowDosePES uses a chopper [52], i.e. a rotating wheel to achieve pseudo single bunch conditions. This wheel is installed at 13

20 the intermediate focus of the beamline, allowing only one pulse per turn out of the 400 possible X-ray flashes at BESSY II storage ring, to fulfil the repetition rate of the pulsed source that ARTOF needs. In our experiments, the endstation consisted of a fast-entry load-lock chamber, a radial distribution chambers transfer system (UFO), a lateral preparation chamber (Racoon), a vertical preparation chamber (Top chamber) equipped with LEED and a main chamber for spectroscopic experiments. The system is equipped with a manipulator with x, y modules, a z translation, a q rotary platform and with Janis cryostat which can reach a temperature down to 4 K with liquid Helium cooling. This endstation is still under upgrading and a new laser source has just been included to the system [53]. During our experiments, the base pressure for the Racoon chamber after baking was in low 10 9 mbar range. The base pressure in main chamber was also in low 10 9 mbar range at room temperature or in low range with helium cooling during the experiment. The thermal evaporator used in our experiments was brought from Uppsala and mounted to the Racoon chamber during the measurements. The Racoon chamber was used to prepare the clean Au (111) surface by sputtering and annealing cycles as well as to deposit CoPc films of different thicknesses with the help of the evaporator. 4.3 The I311 beamline at MAX IV laboratory Biphenylene films were characterized at the photoemission branch of beamline I311 at MAX IV laboratory, the Swedish synchrotron radiation facility, in Lund [54]. With the permanent shut down of the MAX I, II and III accelerators on the 13th of December, 2015 in preparation for the new MAX IV facility, I-311 is now history. I-311 was an undulator-based soft X-ray beamline ( ev), equipped with a modified Zeiss SX-700 monochromator. The spectroscopy end-station was equipped with a hemispherical electron energy analyzer (Scienta SES200) for high resolution PES and NEXAFS. The manipulator in the analysis chamber was equipped with resistive heating and a cold finger which allowed to reach a temperature of about 70 K on the sample cooling with liquid nitrogen. A heating band wrapped around a glass tube containing the biphenylene powder was used to deposit the molecules onto the surface by sublimation. The pressure in the analysis chamber and in the preparation chamber, during the experiments, was kept in the low mbar range. 14

21 5. Sample preparation 5.1 Gas phase measurements The gas phase measurements were performed at the GasPhase photoemission beamline at the Elettra synchrotron facility, Italy. The gas phase photoemission experiment in general needs a constant gas flow into the analysis chamber. This is a task which is not straight forward for some big molecules such as phthalocyanines due to the difficulty to keep a steady molecular vapour flux at a temperature ranging from 320 C to 435 C for days, when acquiring data. As a consequence, compared to the abundant Photoemission Spectroscopy (PES) and Near Edge X-ray Absorption Fine Structure (NEXAFS) studies of CoPc thin films, only very few gas phase measurements on CoPc have been carried out. However, such experiments are important since they contribute significantly to the understanding of the electronic structure of the molecules and also for identifying the modification induced by the adsorption when the molecules are deposited on surfaces. In our case, when studying CoPc, this kind of investigations could give insights into the molecular ground state configuration of CoPc isolated molecules. In the case of biphenylene on Cu(111), such studies enabled us to identify the influence of the molecular-substrate interaction to the molecular electronic structure when adsorbed at low coverage or as thin films. As mentioned, the experimental chamber at the GasPhase beam line was equipped with a home built effusive molecular oven nozzle [51] to heat and evaporate the CoPc molecules. Depending on the molecules, it is possible to choose a proper nozzle diameter to obtain, for example, the preferable molecular evaporation rate. It is always difficult to evaporate molecules with high molecular weight, as CoPc s, due to the need to reach high temperatures. Moreover, phthalocyanines are often clogging the nozzle and it is necessary to replace it. We find that the best evaporation condition for CoPc was to use the largest size of the nozzle, i.e. removing the nozzle cap, resulting in an aperture size of about 1 cm in diameter. This leads to a reduced sample density but we still gained in experimental time by preventing the nozzle from clogging, avoiding time-consuming changes of CoPc sample and necessary cleaning procedures. The purity of the CoPc powder used in our experiments, purchased from Sigma-Aldrich, was 97 %. To further purify the sample from water and CO 2 as well as molecular residues from the synthesis, namely phthalonitrile, the CoPc powder was heated by gradually raising the temperature up to 315 C in the experimental chamber. The cleaning procedure takes at least 8 hours before the 15

22 experiments. We had to warm up the oven carefully, avoiding big temperature steps and preventing the cooling of the molecular flux which would condense at the nozzle. The cleanness of CoPc was checked by valence PES until no peaks due to contaminants, such as H 2 O, CO 2 or phthalonitrile, could be seen. To have a constant flux, the count rate was also checked and compared each time by measuring the HOMO peak. To improve the vacuum in the chamber, a cryo-trap, cooled by liquid nitrogen, was used during the experiment. 5.2 Adsorption measurements Substrate preparations Au (111) The Au (111) single crystal, used in CoPc film experiments (Paper I in this thesis), was bought from MaTecK. The crystal was mounted on a molybdenum Omicron standard sample plate, used at LowDosePES beam line at BESSY II. The crystal was fixed by a tantalum wire (diameter 0.3 mm) which was spot welded to the sample plate. The Au (111) surface was prepared with standard Ar + sputtering and flashing (up to 500 C) cycles. Usually two cycles were needed, one cycle with the sample facing the sputter gun and the other cycle at 45 with respect to the sputter gun. For the sputtering, the parameters P Argon = mbar, 1 kv, 10 ma emission current, 20 minutes were used. The temperature for the flashing was read from both a pyrometer and a thermocouple located near the sample plate. The base pressure in the preparation chamber (Racoon) was in low 10 9 mbar range. The cleanness of the Au (111) ) surface was checked with the ArTOF spectrometer, until the Au surface state was seen in the band dispersion spectra. Cu (111) The Cu (111) was used as substrate for biphenylene films (Paper II in this thesis) at I311 in Lund. The surface was prepared with standard procedure of Ar + sputtering (1 kv, 20 minutes) and annealing (547 C, 20 minutes) to clean and reconstruct the Cu (111) crystal surface. The surface cleanness was checked with PES until no signal from contaminations was detected Film preparations Cobalt Phthalocyanine (CoPc) CoPc ( Sigma-Aldrich, 97 %) was evaporated through a thermal evaporator as shown in Figure 5.1. The evaporator set-up with defined filament type and 16

23 Figure 5.1. The evaporator used for CoPc deposition. A filament made from a tantalum wire (diameter 0.5 mm) was wrapped around the glass crucible (6 rounds). A molybdenum sheet was wrapped and inserted into the crucible in order to keep the temperature stable. A thermocouple wire was dipped into the crucible between the crucible and the molybdenum sheet. Finally, the CoPc powder was filled into the crucible. We have found that the evaporation rate was quite reproducible with this setting, i.e. even if we changed the filament and/or molecules, we could get almost same evaporation temperature at same filament current. wrapping as well as the crucible size, ensures the reproduciblility of the evaporator through current vs. temperature characterizations. CoPc films of different thicknesses were grown on Au (111) substrate. The sample thicknesses were estimated by comparing with results from previous studies [55]. The CoPc monolayer can be made either by just deposition or by desorbing the multilayer sample at a flashing temperature of 411 C. The pressure in the Racoon chamber during deposition was in the high 10 9 mbar range. Biphenylene The biphenylene powder (99% purity) was bought from Sigma-Aldrich. A glass tube wrapped with a heating band around was used to heat up the molecule. The glass tube was connected to the preparation chamber through a leak valve and pumped separately through a dedicated pumping system. Before the deposition, the biphenylene was outgassed at the temperature around 100 C. The same temperature was used to evaporate and deposit biphenylene on the freshly prepared Cu (111) crystal. It is known that at 17

24 room temperature biphenylene will only form a monolayer. We needed to cool the Cu (111) crystal down to -183 C with liquid nitrogen for increasing the sticking coefficient of biphenyelene and getting molecular films of different thicknesses. To get a monolayer, an annealing temperature of -53 C was used to remove thick biphenylene layers. At this temperature, the biphenylene multilayers desorb leaving only a coverage of about 1 monolayer on the surface. The pressure in the preparation chamber during deposition was in the low 10 8 mbar range. 18

25 6. Summary of papers 6.1 Cobalt Phthalocyanine (CoPc) and Biphenylene Recent studies have shown the possibilities of using Transition Metal Phthalocyanines (TMPc s), like Cobalt Phthalocyanine (CoPc), as single-molecule magnets [7]. CoPc is a planar organic molecule with magnetic property determined by the bonding configuration of the 3d-states of the central transition metal with the organic ring. In Paper I, we address two still open questions about the electronic structure of the CoPc molecule, namely the atomic character of the HOMO and the electronic configuration of the ground state. This paper combines gas phase and thin film experimental investigations by means of different X-ray spectroscopies as well as Density Functional Theory (DFT) and Multiplet Ligand Field Theory (MLFT) simulations. These allow us to identify the organic ligand character of the HOMO, the Co character for HOMO-1 and confirm that the molecular ground state is well described by the 2 A 1g configuration. In Paper II we investigate the electronic structure and the molecular orientation of biphenylene adsorbed on a single crystal Cu(111) surface. Also for this study the combination of experimental characterizations of biphenylene in gas phase and as thin films along with theoretical simulations have been shown to be a very powerful strategy in giving an accurate description of the system. 6.2 The electronic structure of CoPc Here we present and discuss the most relevant results which are reported in Paper I XAS of Co L-edge of CoPc The first gas phase experimental Co L 2,3 edge XAS spectrum is shown in Figure 6.1 (a) in comparison with the angle resolved MLFT simulation. The L 3 edge is formed by two main subgroups of unoccupied molecular orbitals, π* (A) at lower photon energies and σ* (C1, C2) at higher energies. The L 2,3 edges are separated by about 15 ev (same as the ionization energies of Co 2p 3/2 and 2p 1/2 for the Co metal) with L 3 at lower photon energy than L 2. The L 2 and L 3 main features have nearly the same width of about 7.2 ev. 19

26 pyr (a) A C1 C2 Co L 2,3 edge XAS Intensity (a.u.) Exp. B D E F A' C1' C2' Total MLFT OPL MLFT IPL MLFT Photon Energy [ev] 800 (b) (c) A Co L 2,3 edge XAS Intensity (a.u.) B C1 Exp Photon Energy [ev] Figure 6.1. (a) Co L edge NEXAFS experimental result (shown in red) in comparison with the angular resolved Multiplet Ligand Field Theory (MLFT) simulations (shown in black) where the in-plane (IPL, green) and the out-of-plane (OPL, purple) contributions are resolved. (b) The zoom-in of the experimental Co L edge NEXAFS. (c) Atomic orbital- and spin-resolved DFT calculations of the occupied and unoccupied valence density of states of CoPc. 20

27 According to the MLFT (Figure 6.1 (a)) and the DFT simulation (Figure 6.1 (c)), the main peak A at ev is an out-of-plane feature, and is associated with a Co 2p 3/2 transition into the a 1g (d z 2) level, while the C1 (780.2 ev) and C2 (781.4 ev) correspond to excitations into in-plane orbitals associated with Co 2p 3/2 into b 1g (d x 2 y2). The small shoulder B at ev (Figure 6.1 (b)) can be associated to transitions into the e 1g (d zx ) orbital. Similar to the L 3 edge, the A, C1 and C2 peaks of L 2 are related to the transition from Co 2p 1/2 to a 1g spin-down, b 1g spin-up and b 1g spin-down states, respectively. In comparison to the L 3 peaks, these features are much broader, however the energy differences between A and C1, C1 and C2 are roughly the same as observed already for the L 3 edge. According to the ligand field theory, the b 1g state provides the main contribution to the σ bonding, having its lobes directed towards the nitrogen ligand. Our DFT calculations also predict that the b 1g state is the highest in energy and remains completely empty (Figure 6.2 (c)). The other states, a 1g, e g and b 2g are close in energy and their population in the CoPc ground state, as already mentioned, is still under debate [56]. As described in Section 6.2.2, the ground state of CoPc in our DFT calculation is 2 A 1g, which corresponds to the electronic configuration with the e g state completely filled. However, in the experimental XA spectrum the e g state is still visible (feature B in Figure 6.1 (b)) even though with very low intensity, appearing at the same relative energy position as found in previous study on CoPc thin films [56]. An unoccupied molecular orbital containing small contributions from the e g state of Co is revealed by our DFT calculation at 0.6 ev above LUMO in the VB spectrum (Figure 6.1 (c)). This molecular orbital also contains contributions from the p z states of the nitrogen atoms of the aza bridge (N aza ) as well as of pyrrole (N pyr ). It is very likely that the B feature in the Co L 3 -edge XAS spectrum originates from transitions from the Co 2p 3/2 to this hybrid unoccupied state. The reason for the poor resolution of the e g feature in the gas phase measurements is the impossibility to disentangle orbital of different symmetries for free molecules, as instead possible for ordered films thanks to XAS dichroism CoPc Valence levels The discussions of the electronic structure of CoPc concern the atomic character of the HOMO but also the electronic configuration which would correspond to the molecular ground state. Our recent DFT calculations predict the 2 A 1g configuration (in Figure 6.2 (c)) to well describe the molecular ground state. This is confirmed by the very good agreement between the experimental photoemission spectrum (taken at 50 ev photon energy) and the simulation, presented in Figure 6.2 (b). 21

28 (a) 150 ev Intensity (a.u.) 100 ev 80 ev 50 ev Binding Energy [ev] (b) (c) Intensity (a.u.) H G F E' E N D C' C B A B' A' Co 3d C Exp. Cal. E D 4h ligand field splitting 2 ( A 1g) b 1g a 1g e g b 2g Binding Energy [ev] Figure 6.2. (a) Experimental valence level of CoPc in gas phase measured with 50 ev (red), 80 ev (black), 100 ev (blue) and 150 ev (green) photon energies. (b) Experimental valence levels of gas phase CoPc measured with 50 ev photon energy (black) shown in comparison with the simulation considering the molecular ground state described by the ( 2 A 1g ) configuration (in (c)). The different valence features are labeled with letters from A to H. The calculated C 2p (green filled), N 2p (blue) and Co 3d (orange filled) contributions are also shown. 22

29 From Figure 6.2 (a), we clearly discern the HOMO at about 6.38 ev which decreases in intensity with increasing photon energy. Taking advantage of the different photoemission cross sections that C, and N atoms have with respect to the Co metal [57], and knowing that the intensity of photoemission peaks with carbon or nitrogen character will be reduced by increasing the photon energy (as expected for lighter elements), we performed measurements at different photon energies. This is a valid strategy for disentangling the atomic character of the different valence state as already shown in previous studies [5] [6] [58]. Looking at the results, we can conclude that the HOMO is mostly due to intensity contributions from the organic ligand of the molecule, i.e. C and N. The other features have instead some more Co metal character. The different intensity ratio variation with changing photon energy of the peaks E and E indicates that they have different contributions from Co atom. Figure 6.1 (b) shows DFT results considering the 2 A 1g as the molecular ground state electron configuration. The DFT results predict the HOMO feature at 6.12 ev, which is in good agreement with the measured binding energy at 6.38 ev. The calculations also indicate that the HOMO fully originates from the C 2p states related to the a 1u molecular orbital, coming from the C-C hybridization of π type orbitals, in accordance with previous studies [26] [27]. The DFT results distinguish the other peaks in the valence spectra (from B to H) as related to molecular states formed by C, N and Co atomic orbitals with features B and B having important Co 3d contributions. These theoretical results are confirmed by the gas phase photon energy dependent spectra shown in Figure 6.2 (a). The experimental results are also in great agreement with the simulation shown in Figure 6.1 (c). An analysis of the Co 3d levels, shows that B and B are occupied e g states alongside the spin-down b 2g component. The spinup b 2g state contributes to slightly higher binding energies, giving rise to the experimentally observed intensity increase of peak C. Finally, the singly occupied a 1g state contributes to the E feature of the experimental spectrum which noticeably increases in intensity as the photon energy is increased. According to the calculation, peak E fully originates in 2p ligand states. This appears to be the case in the experimental spectra as well, since peak E decreases in intensity relative to E when higher photon energies are used CoPc film on Au (111) surface In order to carefully address the character of the HOMO and to identify if any metal character could be hidden underneath the HOMO intensity, we also performed a study about CoPc films grown at different coverages on a Au (111) surface. 23

30 CoPc film vs vapor Figure 6.3 shows the comparison between the valence photoemission results of CoPc in gas phase, measured with 36 ev photon energy, and of a thin film, grown by thermal evaporation on Au (111), taken with 40 ev photon energy. The spectrum of the film is shifted and aligned to the gas phase spectrum with HOMO at 6.38 ev. The differences with respect to the cross-section at these photon energies can be neglected, making it possible to compare the two spectra directly. Apart from a reasonable overall broadening the spectrum of the film show quite similar appearances with the gas phase results, indicating that the CoPc film mantains a molecular-like character, as already observed for other metal Pc s films [5]. The molecular character of the CoPc film makes it possible to study the HOMO origin through an investigation of CoPc adsorbed on a surface. CoPc valence Gas Phase Film Intensity (a.u.) hv=36 ev hv=40 ev Binding Energy [ev] Figure 6.3. Comparison between the valence photoemission spectra of gas phase CoPc, taken with 36 ev photon energy (red) and of a CoPc film absorbed on Au (111), measured with 40 ev photon energy (black). CoPc adsorption on Au (111) In Figure 6.4 we show valence PE spectra taken with an ARTOF spectrometer of molecular depositions of increasing thicknesses. The figure includes the results of clean Au (111) substrate together with SubMonoLayer (SubML), MonoLayer (ML) and thin film coverages of Co phthalocyanines. The spectra show a new peak (marked as IS in Figure 6.4) at about 0.55 ev binding energy already at very low molecular coverage ( SubML) while the HOMO is clear visible at 1.15 ev for higher molecular deposition. The new peak at 0.55 ev is the so-called Interface State (IS) ascribed, in previous studies, to a new state 24

31 of fully metallic character [56] [55], originated by the interaction between the Co 3d out-of plane orbitals and the delocalized states of the Au (111) surface. hv = 80 ev HOMO Intensity (a.u.) IS Film ML SubML Au(111) Binding Energy [ev] Figure 6.4. Valence photoemission spectra of CoPc deposited on Au(111) at the given molecular thicknesses, taken with a photon energy of 80 ev. The binding energy is calibrated to the Fermi level. The coverages from bottom to top are increasing from clean substrate (Au (111)), submonolayer (SubML), monolayer (ML) to film (of a thickness of about 2 nm), respectively. CoPc HOMO character study We performed a photon energy dependent valence photoemission investigation for addressing the atomic character of the CoPc HOMO following the intensity variation of the HOMO and of the Interface State, which is known to be purely metallic. The diversity of the photoionization cross-sections for Co 3d and C atoms, would result in a variation of the Co 3d / C 2p ration from 9 (55 ev) to 21 (80 ev). Then we expected a different intensity variation of the HOMO and the IS peaks in case the HOMO would be just of organic ligand origin. (a) (b) (c) (d) 57 ev 80 ev ML FIlm Intensity (a.u.) Film Intensity (a.u.) Film Intensity (a.u.) 57 ev Intensity (a.u.) 57 ev ML ML 80 ev 80 ev Binding Energy [ev] Binding Energy [ev] Binding Energy [ev] Binding Energy [ev] Figure 6.5. Spectra of the valence states of a monolayer (ML) and film (about 2 nm) of CoPc, taken with photon energies of 57 ev (a) and 80 ev (b). In (c) and (d), the photon energy dependent spectra of ML and film samples are shown. All spectra are normalized and calibrated at the Fermi edge of the substrate. 25

32 The comparison of valence photoemission spectra of a film of CoPc (about 2 nm thick) and a monolayer sample measured with 57 ev and 80 ev photon energies are shown in Figure 6.5 (a) and (b), respectively. The figures evidence the evolution of the HOMO peak (at about 1.15 ev) for increasing molecular coverage and being more intensive in the spectra taken at lower photon energy (57 ev) as an effect of the photoionization cross section. As expected the intensity of interface state (0.55 ev) does not increase with the thickness of the film, since it is due to the interaction of the first molecular layer with the Au substrate. The intensity of this peak neither changes so much with the different photon energies used, having Co 3d states quite similar photoionization cross sections between 55 ev and 80 ev. Figures 6.5 (c) and (d), comparing the spectra of samples of same molecular thickness taken with different photon energies, evidence still more clearly the dependence of the peak intensities on the photoionization cross section. These results, confirm that HOMO is due to the organic ligand of the molecule with no metallic contribution, while the interface state, can be assigned to the a purely metallic state, in agreement with the previous studies.[26] [55] [56] Comparison of valence levels in Pc s measured in gas phase In Figure 6.6 we compare the valence spectrum of CoPc with the valence results of FePc, MnPc and H 2 Pc from previous studies [6]. All these Pc s have a common features with mainly C 2p character appearing at around 6.4 ev and similar broad structures in the binding energy region between 7.7 ev and 9.5 ev. However, the three metal Pc s present additional features appearing at different energies with Co, Fe and Mn character respectively. Only the HOMO of MnPc has mainly metallic character. Figure 6.6. The comparison of CoPc (green), FePc (black), MnPc (red) and H 2 Pc (blue) measured in gas phase. For all the Pc s the features at round 6.5 ev have mainly C 2p character. Only for MnPc the HOMO has metallic character whereas for FePc and CoPc the HOMO has organic ligand character. 26