This thesis is aimed at establishing modalities of time-resolved photoelectron spectroscopy (tr-PES) conducted at a free-electron laser (FEL) source and at a high harmonic generation (HHG) source for imaging the motion of atoms, charge and energy at photoexcited hybrid organic/inorganic interfaces. Transfer of charge and energy across interfaces lies at the heart of surface science and device physics and involves a complex interplay between the motion of electrons and atoms. At hybrid organic/inorganic interfaces involving planar molecules, such as pentacene and copper(II)-phthalocyanine (CuPc), atomic motions in out-of-plane direction are particularly apparent. Such hybrid interfaces are of importance to, e.g., next-generation functional devices, smart catalytic surfaces and molecular machines. In this work, two hybrid interfaces – pentacene atop Ag(110) and copper(II)-phthalocyanine (CuPc) atop titanium disulfide (1T-TiSe2) – are characterized by means of modalities of tr-PES. The experiments were conducted at a HHG source and at the FEL source FLASH at Deutsches Elektronen-Synchrotron DESY (Hamburg, Germany). Both sources provide photon pulses with temporal widths of ∼ 100 fs and thus allow for resolving the non-equilibrium dynamics at hybrid interfaces involving both electronic and atomic motion on their intrinsic time scales. While the photon energy at this HHG source is limited to the UV-range, photon energies can be tuned from the UV-range to the soft x-ray-range at FLASH. With this increased energy range, not only macroscopic electronic information can be accessed from the sample’s valence and conduction states, but also site-specific structural and chemical information encoded in the core-level signatures becomes accessible. Here, the combined information from the valence band and core-level dynamics is obtained by performing time- and angle-resolved photoelectron spectroscopy (tr-ARPES) in the UV-range and subsequently performing time-resolved x-ray photoelectron spectroscopy (tr-XPS) and time-resolved photoelectron diffraction (tr-XPD) in the soft x-ray regime in the same experimental setup. The sample’s bandstructure in energy-momentum space and time is captured by a time-of-flight momentum microscope with femtosecond temporal and sub-Ångström spatial resolutions. In the investigated systems, out-of-equilibrium dynamics are traced that are connected to the transfer of charge and energy across the hybrid interfaces. While energetic shifts and complementary population dynamics are observed for molecular and substrate states, the shapes of involved molecular orbitals change in energy-momentum space on a subpicosecond time scale. In combination with theory support, these changes are attributed to iiiatomic reorganizations at the interface and transient molecular structures are reconstructed with sub-Ångström precision. Unique to the material combination of CuPc/TiSe2, a structural rearrangement on the macroscopic scale is traced simultaneously: ∼ 60 % of the molecules undergo a concerted, unidirectional in-plane rotation. This surprising observation and its origin are detailed in this thesis and connected to a particularly efficient charge transfer across the CuPc/TiSe2 interface, resulting in a charging of ∼ 45 % of CuPc molecules.

Paclitaxel (PTX) is one of the leading drugs against breast and ovarian cancer. Due to its low solubility, treatment of the patients with this drug requires a very well-suited combination with a soluble pharmaceutical excipient to increase the bioavailability and reduce the strong side ef-fects. One efficient way to achieve this in the future could be the incorporation of PTX into pol-ymeric micelles composed of poly(2-oxazoline) based triblock copolymers (POL) which ena-bles PTX loadings of up to 50 wt.%. However, structural information at an atomic level and thus the knowledge of interaction sites within these promising but complex PTX-POL formula-tions were not yet available. Such results could support the future development of improved excipients for PTX and suitable excipients for other pharmaceutical drugs. Therefore, a solid-state MAS NMR investigation of these amorphous formulations with different POL-PTX com-positions was performed in this thesis as this gives insights of the local structure at an atomic level in its solid state. NMR in solution showed very broad 13C signals of PTX for this system due to the reduced mobility of the incorporated drug which exclude this as an analytical meth-od. In a first study, crystalline PTX was structurally characterized by solid-state NMR as no com-plete 13C spectrum assignment and no 1H NMR data existed for the solid state. In addition, the asymmetric unit of the PTX crystal structure consists of two molecules (Z'=2) that can only be investigated in its solid state. As crystalline PTX in total has about 100 different 13C and 1H chemical shifts with very small differences due to Z’=2, and furthermore, its unit cell consisting of more than 900 atoms, accompanying GIPAW (CASTEP) calculations were required for NMR signal assignments. These calculations were performed using the first three available purely hydrous and anhydrous PTX structures, which were determined by XRD and published by Vel-la-Zarb et al. in 2013. Within this thesis, is was discovered that two investigated batches of commercially available PTX from the same supplier both contained an identical and so far un-known PTX phase that was elucidated by PXRD as well as solid-state NMR data. One of the two batches consists of an additional phase that was shown to be very similar to a known hy-drated phase published in 2013.[1] By heating the batch with the mixture of the two phases un-der vacuum, it is transformed completely to the new dry phase occurring in both PTX batches. Since the drying conditions to obtain anhydrous PTX in-situ on the PXRD setup described by Vella-Zarb et. al.[1] were much softer than ours, we identify our dry phase as a relaxed version of their published anhydrate structure. The PXRD data of the new anhydrate phase was trans-ferred into a new structural model, which currently undergoes geometry optimization. Based on solid-state NMR data at MAS spinning frequencies up to 100 kHz, a 13C and a partial 1H signal assignment for the new anhydrous structure were achieved. These results provided sufficient structural information for further investigations of the micellar POL-PTX system. In a second study, the applicability and benefit of two-dimensional solid-state 14N-1H HMQC MAS NMR spectra for the characterization of amorphous POL-PTX formulations was investi-gated. The mentioned technique has never been applied to a system of similar complexity be-fore and was chosen because around 84% of the small-molecule drugs contain at least one nitrogen atom. In addition, the number of nitrogen atoms in both POL and PTX is much smaller than the number of carbons or hydrogens, which significantly reduces the spectral complexity. 14N has a natural abundance of 99.6% but leads to quadrupolar broadening due to its nuclear spin quantum number I = 1. While this is usually undesirable due to broadening in the resulting 1D 14N NMR spectra, this effect is explicitly used in the 2D 14N-1H HMQC MAS experiment. The indirect 14N measurement can avoid the broadening while maintaining the advantage of the high natural abundance and making use of the much more dispersed signals due to the additional quadrupolar shifts as compared to 15N. This measurement method could be successfully applied to the complex amorphous POL-PTX mixtures. With increasing PTX loading of the formulations, additional peaks arise as spatial proximities of the amide nitrogens of POL to NH or OH groups of PTX. In addition, the 14N quadrupolar shift of these amide nitrogens decreases with increasing PTX content indicating a more symmetric nitrogen environment. The latter can be explained by a transformation of the trigonal planar coordination of the tertiary amide nitrogen atoms in pure POL towards a more tetrahedral environment upon PTX loading induced by the formation of hydrogen bonds with NH/OH groups of PTX. In the third and last project, the results of the two abovementioned studies were used and ex-tended by solid state 13C and two-dimensional 1H-13C as well as 1H-1H MAS NMR data with the aim to derive a structural model of the POL-PTX formulations at an atomic level. The knowledge of the NMR signal assignments for crystalline PTX was transferred to amorphous PTX (present in the micelles of the formulations). The 13C solid-state NMR signals were evalu-ated concerning changes in chemical shifts and full widths of half maximum (FWHM) for the different PTX loadings. In this way, the required information about possible interaction sites at an atomic level becomes available. Due to the complexity of these systems, such proximities often cannot be assigned to special atoms, but more to groups of atoms, as the individual de-velopments of line widths and line shifts are mutually dependent. An advantageous aspect for this analysis was that pure POL already forms unloaded micelles. The evaluation of the data showed that the terminal phenyl groups of PTX seem to be most involved in the interaction by the establishment of the micelle for lowest drug loading and that they are likely to react to the change in the amount of PTX molecules as well. For the incorporation of PTX in the micelles, the following model could be obtained: For lowest drug loading, PTX is mainly located in the inner part of the micelles. Upon further increasing of the loading, it progressively extends to-ward the micellar shell. This could be well shown by the increasing interactions of the hydro-phobic butyl chain of POL and PTX, proceeding in the direction of the polymer backbone with rising drug load. Furthermore, due to the size of PTX and the hydrodynamic radius of the mi-celles, even at the lowest loading, the PTX molecules partially reach the core-shell interface of the micelle. Upon increasing the drug loading, the surface coverage with PTX clusters increas-es based on the obtained model approach. The latter result is supported by DLS and SANS data of this system. The abovementioned results of the 14N-1H HMQC MAS investigation of the POL-PTX formulations support the outlined model. As an outlook, the currently running geometry optimization and subsequently scheduled calcu-lation of the chemical shieldings of the newly obtained anhydrous PTX crystal structure can further improve the solid-state NMR characterization through determination of further spatial proximities among protons using the existing 2D 1H(DQ)-1H(SQ) solid-state MAS NMR spec-trum at 100 kHz rotor spinning frequency. The 2D 14N-1H HMQC MAS NMR experiments were shown to have great potential as a technique for the analysis of other disordered and amor-phous drug delivery systems as well. The results of this thesis should be subsequently applied to other micellar systems with varying pharmaceutical excipients or active ingredients with the goal of systematically achieving higher drug loadings (e.g., for the investigated PTX, the similar drug docetaxel or even different natural products). Additionally, it is planned to transfer the knowledge to another complex polymer system containing poly(amino acids) which offers hy-drogen bonding donor sites for additional intermolecular interactions. Currently, the POL-PTX system is investigated by further SANS studies that may provide another puzzle piece to the model as complementary measurement method in the future. In addition, the use of MD simu-lations might be considered in the future. This would allow a computerized linking of the differ-ent pieces of information with the aim to determine the most likely model.

In the present work we investigated various charge transfer processes, as they appear in the versatile world of organic semiconductors by probing the spin states of the corresponding charge carrier species via electron paramagnetic resonance (EPR) spectroscopy. All studied material systems are carbon-based compounds, either belonging to the group of polymers, fullerenes, or single-wall carbon nanotubes (SWNTs). In the first instance, we addressed the change of the open circuit voltage (Voc) with the fullerene blend stoichiometry in fullerene-based solar cells for organic photovoltaics (OPV). The voltage depends strongly on the energy separation between the lowest unoccupied molecular orbital (LUMO) of the donor and the highest occupied molecular orbital (HOMO) of the acceptor. By exploiting the Gaussian distribution of the charge carriers in a two-level system, and thus also their spins in the EPR experiment, it could be shown that the LUMOs get closer by a few to a few hundred meV when going from pure fullerene materials to a fullerene mixture. The reason for this strong energetic effect is likely the formation of a fullerene alloy. Further, we investigated the chemical doping mechanism of SWNTs with a (6,5)-chirality and their behaviour under optical excitation. In order to determine the unintentional (pre)-doping of SWNTs, EPR spectra of the raw material as well as after different purification steps were recorded. This facilitated the determination of nanotube defects and atmospheric p-doping as the causes of the measured EPR signals. In order to deliberately transfer additional charge carriers to the nanotubes, we added the redox-active substance AuCl3 where we determined an associated doping-yield of (1.5±0.2)%. In addition, a statistical occupation model was developed which can be used to simulate the distribution of EPR active, i.e. unpaired and localised charge carriers on the nanotubes. Finally, we investigated the charge transfer behaviour of (6,5)-SWNTs together with the polymer P3HT and the fullerene PC60BM after optical excitation.

Die vorliegende Dissertation widmete sich der Aufklärung der Photodissoziationsdynamik der drei Xylyl-Radikale ortho-, meta- und para-Xylyl sowie des Benzyl-Radikals mit Hilfe des Velocity-Map-Imagings. Diese reaktiven Intermediate sind insbesondere im Bereich der Verbrennungschemie von hoher Relevanz, da sie die primären Zerfallsprodukte der Xylole und des Toluols darstellen, welche als Antiklopfmittel in Ottokraftstoffen Verwendung finden.Dementsprechend ist eine Betrachtung des weiteren Zerfalls dieser resonanz-stabilisierten Radikale, insbesondere unter dem Gesichtspunkt der Rußbildung, von entscheidender Bedeutung. Für alle drei Xylyl-Radikale konnte eine selektive pyrolytische Generierung aus den entsprechenden 2-(Methylphenyl)ethylnitriten realisiert werden. Die isomerspezifische Identifikation erfolgte mit Hilfe von REMPI-Spektroskopie der jeweiligen D0 -> D3-Übergänge. Nachfolgend wurde die Photodissoziation aller drei Xylyl-Isomere nach Anregung des D3-Zustandes bei ca. 310 nm und nach Anregung der D-Bande bei 250 nm untersucht. Das „einfachste” Experiment stellte in diesem Zusammenhang die Photodissoziation des para-Xylyl-Radikals dar. Es konnte die von Hemberger et al. in thermischen Zerfallsexperimenten beobachtete Reaktion p-Xylyl -> p-Xylylen + H verifiziert werden. Die VMI-Experimente lieferten die Kennwerte <fT>(309.6nm) = 33 % und <fT>(250nm) = 19 % unter Erhalt isotroper Images für beide Anregungswellenlängen. Die dazugehörigen Dissoziationsratenkonstanten wurden zu kH(309.6nm) ≈ 10^8 s-1 und kH(250nm) ≈ 5*10^7 s-1 bestimmt. Es ist verblüffend, dass die Photodissoziation scheinbar bei der höheren Anregungswellenlänge von 309.6 nm (und somit bei geringerer Anregungsenergie) schneller verläuft als bei 250 nm. Darüber hinaus ist es nicht möglich, die beobachteten Raten mittels des statistischen Modells der RRKM-Theorie zu beschreiben. Des Weiteren konnten auch die Translationsenergieverteilungen nicht mit dem „Quack-Fit” für statistische Dissoziationen angefittet werden. Bei der Photodissoziation des para-Xylyl-Radikals liegt eine Dissoziation nach Rückkehr in den rovibronisch hochangeregten elektronischen Grundzustand infolge der Photoanregung vor. Hierbei thermalisiert die innere Energie im elektronischen Grundzustand vor der Dissoziation scheinbar nur teilweise, sodass keine vollständige statistische Verteilung dieser innerhalb des para-Xylyls gegeben ist. Da dies eine Grundvoraussetzung der gängigen statistischen Modelle darstellt, ist es nicht verwunderlich, dass keine quantitative Reproduktion der experimentellen Ergebnisse durch Anwendung dieser Modelle ermöglicht wird. Bei entsprechenden Experimenten zum ortho-Isomer konnten diese statistischen Modelle ebenfalls nicht zur quantitativen Beschreibung der Dissoziation verwendet werden. Abermals wurde mit kH(311.1nm) ≈ 10^8 s-1 und kH(250nm) ≈ 5*10^7 s-1 eine schnellere Dissoziation bei geringerer Anregungsenergie festgestellt. Dies erscheint demnach charakteristisch für die Xylyl-Radikale. Innerhalb der VMI-Experimente wurden isotrope Verteilungen erhalten, deren Fragmenttranslationsenergieverteilung nach Anregung des D3-Niveaus bei 311.1 nm jedoch nicht durch die von Hemberger et al. beschriebene Hauptreaktion o-Xylyl -> o-Xylylen + H erklärt werden konnte. Eine Fragmentation nach o-Xylyl -> Benzocyclobuten + H konnte auf diesem Weg als Hauptdissoziationspfad identifiziert werden. Innerhalb der Studien von Hemberger et al. ist eine Reaktion zu Benzocyclobuten bei Anregung mit 311.1 nm energetisch nicht zugänglich. Mittels quantenchemischer Rechnung konnte jedoch ein bislang unbekannter, energetisch zugänglicher Reaktionspfad zur Bildung von Benzocyclobuten unter simultaner Ringschlussreaktion und Wasserstofffragmentation identifiziert und charakterisiert werden. Die Kennwerte der Photodissoziationsreaktion des ortho-Xylyls konnten hierdurch zu <fT>(311.1nm) = 30 % und <fT>(250nm) = 16 % bestimmt werden. Wie bereits im Fall des para-Isomers liegt die Vermutung nahe, dass es sich um eine Dissoziation aus dem rovibronisch hoch-angeregten elektronischen Grundzustand handelt, welcher nicht vollständig vor der Fragmentation thermalisiert. Im Rahmen der Experimente zum letzten der drei Xylyl-Isomere, dem meta-Xylyl-Radikal, konnte mit VMI eine Fragmentation nach m-Xylyl -> m-Xylylen + H als Hauptdissoziationpfad ausgeschlossen werden. Innerhalb der Experimente nach Anregung des D3-Niveaus um 310 nm konnten mit para-Xylylen und Benzocyclobuten zwei Reaktionsprodukte festgestellt werden, welche die erhaltene Translationsenergieverteilung erklären könnten, wobei die entsprechende maximale Überschussenergie einer Fragmentation zu para-Xylylen den Nullabfall der Verteilung geringfügig besser widerspiegelt. Die mittlere Fragmenttranslationsenergie liegt mit <fT>(p-Xylylen) = 29 % respektive <fT>(Bcb) = 25 % leicht unterhalb der entsprechenden Kennwerte der para- beziehungsweise ortho-Xylyl Experimente. Durch die nötige, der Dissoziation vorausgehende Isomerisierung scheint ein höherer Thermalisierungsgrad der Schwingungs- und Rotationsenergie innerhalb des elektronischen Grundzustands erreicht zu werden, aus welchem die geringen <fT>-Werte resultieren könnten. Der Effekt verminderter <fT>-Werte wurde in den Experimenten bei 250 nm nicht gefunden (<fT>(p-Xylylen) = 19 % respektive <fT>(Bcb) = 17 %). Vergleicht man an dieser Stelle die <ET>- anstelle der <fT>-Werte (<ET>(para) = 0.41 eV, <ET>(ortho) = 0.38 eV, <ET>(meta) = 0.41 eV), stellt man fest, dass <ET>(meta) = <ET>(para) gilt und somit ein weiteres Indiz dafür gefunden wurde, dass eine Umlagerung zu para-Xylyl mit anschließender Fragmentation zu para-Xylylen möglicherweise gegenüber jener zum ortho-Isomer mit nachfolgender Bcb-Bildung bevorzugt ist. Dies würde darüber hinaus im Einklang mit den Studien von Hemberger et al. stehen, in welchen beim thermischen Zerfall des meta-Xylyls para-Xylylen als alleiniges Fragmentationsprodukt gefunden wurde. Eine Betrachtung der Umlagerung mittels RRKM wies jedoch keinen bevorzugten Isomerisierungspfad aus. Schlussendlich lässt sich aufgrund der ermittelten Ratenkonstanten (kH(310nm) ≈ 10^8 s-1, kH(250nm) ≈ 4*10^7 s-1) sowie den <fT>-Werten vermuten, dass die Isomerisierung langsamer als die Dissoziation bei 310 nm verläuft, jedoch zumindest auf einer ähnlichen Zeitskala wie die entsprechende Dissoziation nach Anregung bei 250 nm. Eine zweifelsfreie Interpretation der meta-Xylyl Experimente gestaltet sich jedoch als schwierig. Innerhalb der Studien zur Photodissoziation des Benzyl-Radikals konnten literaturbekannte Daten zur Fragmentation nach Anregung um 250 nm in guter Übereinstimmung reproduziert werden. Die experimentellen Daten zur Untersuchung der Photodissoziation nach Anregung des D3-Niveaus konnten jedoch nicht eindeutig interpretiert werden. Die literaturbekannte Lage des D3-Niveaus bei 305.3 nm konnte mittels REMPI-Spektroskopie reproduziert werden und anschließende 1H-Photofragmentspektren zeigten, dass eine Anregung des D3-Niveaus zur Bildung von Wasserstofffragmenten führt. Die beobachteten 1H-Fragmente zeigten jedoch eine deutlich zu hohe Überschussenergie für eine Einphotonenabsorption, sodass diese Mehrphotonenabsorptionen zugeordnet werden müssen. Es lässt sich vermuten, dass die Wasserstofffragmente aus einer Anregung eines „superexcited states” oberhalb des Ionisationspotentials, wahrscheinlich durch Zweiphotonenabsorption, stammen. Dieser „superexcited state” zeigt scheinbar keine (vollständige) Autoionisation und führt nachfolgend zumindest teilweise zur Fragmentation des Benzyl-Radikals. In der Folge liegt die Vermutung nahe, dass die Energien eines einzelnen 305 nm-Photons nicht zur Initiierung einer Photodissoziation des Benzyl-Radikals ausreichend ist oder aber, dass diese Photodissoziation zu langsam ist, um sie in einem VMI-Experiment zu beobachten. Potential für weitere Experimente zur Photodissoziation des Benzyl-Radikals nach Anregung des D3-Niveaus wird an dieser Stelle nicht gesehen.

The understanding of excimer formation in organic materials is of fundamental importance, since excimers profoundly influence their functional performance in applications such as light-harvesting, photovoltaics or organic electronics. We present a joint experimental and theoretical study of the ultrafast dynamics of excimer formation in the pyrene dimer in a supersonic jet, which is the archetype of an excimer forming system. We perform simulations of the nonadiabatic photodynamics in the frame of TDDFT that reveal two distinct excimer formation pathways in the gas-phase dimer. The first pathway involves local excited state relaxation close to the initial Franck–Condon geometry that is characterized by a strong excitation of the stacking coordinate exhibiting damped oscillations with a period of 350 fs that persist for several picoseconds. The second excimer forming pathway involves large amplitude oscillations along the parallel shift coordinate with a period of ≈900 fs that after intramolecular vibrational energy redistribution leads to the formation of a perfectly stacked dimer. The electronic relaxation within the excitonic manifold is mediated by the presence of intermolecular conical intersections formed between fully delocalized excitonic states. Such conical intersections may generally arise in stacked π-conjugated aggregates due to the interplay between the long-range and short-range electronic coupling. The simulations are supported by picosecond photoionization experiments in a supersonic jet that provide a time-constant for the excimer formation of around 6–7 ps, in good agreement with theory. Finally, in order to explore how the crystal environment influences the excimer formation dynamics we perform large scale QM/MM nonadiabatic dynamics simulations on a pyrene crystal in the framework of the long-range corrected tight-binding TDDFT. In contrast to the isolated dimer, the excimer formation in the crystal follows a single reaction pathway in which the initially excited parallel slip motion is strongly damped by the interaction with the surrounding molecules leading to the slow excimer stabilization on a picosecond time scale.