An experimental setup for probing ultrafast dynamics at the diffraction limit was developed, characterized and demonstrated in the scope of the thesis, aiming for optical investigations while simultaneously approaching the physical limits on the length and timescale. An overview of this experimental setup was given in Chapter 2, as well as the considerations that led to the selection of the individual components. Broadband laser pulses with a length of 9.3 fs, close to the transform limit of 7.6 fs, were focused in a NA = 1.4 immersion oil objective, to the diffraction limit of below 300 nm (FWHM). The spatial focus shape was characterized with off-resonance gold nanorod scatterers scanned through the focal volume. For further insights into the functionality and limitations of the pulse shaper, its calibration procedure was reviewed. The deviations between designed and experimental pulse shapes were attributed to pulse-shaper artifacts, including voltage-dependent inter-layer as well as intra-layer LCD-pixel crosstalk, Fabry-Pérot-type reflections in the LCD layers, and space-time coupling. A pixel-dependent correction was experimentally carried out, which can be seen as an extension of the initial calibration to all possible voltage combinations of the two LCD layers. The capabilities of the experimental setup were demonstrated in two types of experiments, targeting the nonlinearity of gold (Chapter 3) as well as two-dimensional spectroscopy at micro-structured surfaces (Chapter 4). Investigating thin films, an upper bound for the absolute value for the imaginary part of the nonlinear refractive index of gold could be set to |n′′ 2 (Au)| < 0.6·10−16 m2/W, together with |n′ 2 (Au)| < 1.2·10−16 m2/W as an upper bound for the absolute value of the real part. Finite-difference time-domain simulations on y-shaped gold nanostructures indicated that a phase change of ∆Φ ≥ 0.07 rad between two plasmonic modes would induce a sufficient change in the spatial contrast of emission to the far-field to be visible in the experiment. As the latter could not be observed, this value of ∆Φ was determined as the upper bound for the experimentally induced phase change. An upper bound of 52 GW/cm2 was found for the damage threshold. In Chapter 4, a novel method for nonlinear spectroscopy on surfaces was presented. Termed coherent two-dimensional fluorescence micro-spectroscopy, it is capable of exploring ultrafast dynamics in nanostructures and molecular systems at the diffraction limit. Two-dimensional spectra of spatially isolated hotspots in structured thin films of fluorinated zinc phthalocyanine (F16ZnPc) dye were taken with a 27-step phase-cycling scheme. Observed artifacts in the 2D maps were identified as a consequence from deviations between the desired and the experimental pulse shapes. The optimization procedures described in Chapter 2 successfully suppressed the deviations to a level where the separation from the nonlinear sample response was feasible. The experimental setup and methods developed and presented in the scope of this thesis demonstrate its flexibility and capability to study microscopic systems on surfaces. The systems exemplarily shown are consisting of metal-organic dyes and metallic nanostructures, represent samples currently under research in the growing fields of organic semiconductors and plasmonics.

Time-resolved optical spectroscopy has become an important tool to investigate the dynamics of quantum mechanical processes in matter. In typical applications, a first “pump” pulse excites the system under investigation from the thermal equilibrium to an excited state, and a second variable time-delayed “probe” pulse then maps the dynamics of the excited system. Although advanced nonlinear techniques have been developed to investigate, e.g., coherent quantum effects, all of these techniques are limited in their spatial resolution. The laser focus diameter has a lower bound given by Abbe’s diffraction limit, which is roughly half the optical excitation wavelength—corresponding to about 400nm in the presented experiments. In the time-resolved experiments that have been suggested so far, averaging over the sample volume within this focus cannot be avoided. In this thesis, two approaches were developed to overcome the diffraction limit in optical spectroscopy and to enable the investigation of coherent processes on the nanoscale. In the first approach, analytic solutions were found to calculate optimal polarizationshaped laser pulses that provide optical near-field pump–probe pulse sequences in the vicinity of a nanostructure. These near-field pulse sequences were designed to allow excitation of a quantum system at one specific position at a certain time and probing at a different position at a later time. In the second approach, the concept of coherent two-dimensional (2D) spectroscopy, which has had great impact on the investigation of coherent quantum effects in recent years, was combined with photoemission electron microscopy, which yields a spatial resolution well below the optical diffraction limit. Using the analytic solutions, optical near fields were investigated in terms of spectroscopic applications. Near fields that are excited with polarization-shaped femtosecond laser pulses in the vicinity of appropriate nanostructures feature two properties that are especially interesting in the view of spectroscopic applications: On the one hand, control of the spatial distribution of the optical fields is achieved on the order of nanometers. On the other hand, the temporal evolution of these fields can be adjusted on the order of femtoseconds. In this thesis, solutions were found to calculate the optimal polarizationshaped laser pulses that control the near field in a general manner. The main idea to achieve this deterministic control was to disentangle the spatial and temporal near-field control. First, the spatial distribution of the optical near field was controlled by assigning the correct state of polarization for each frequency within the polarization-shaped laser pulse independently. The remaining total phase—not employed for spatial control—was then used for temporal near-field compression, which, in experimental applications, would lead to an enhancement of the nonlinear signal at the respective location. In contrast to the use of optical near fields, where pump–probe sequences themselves are localized below the diffraction limit and the detection does not have to provide the spatial resolution, a different approach was suggested in this thesis to gain spectroscopic information on the nanoscale. The new method was termed “Coherent two-dimensional (2D) nanoscopy” and transfers the concept of “conventional” coherent 2D spectroscopy to photoemission electron microscopy. The pulse sequences used for the investigation of quantum systems in this method are still limited by diffraction. However, the new key concept is to detect locally generated photoelectrons instead of optical signals. This yields a spatial resolution that is well below the optical diffraction limit. In “conventional” 2D spectroscopy a triple-pulse sequence initiates a four wave mixing process that creates a coherence. In a quantum mechanical process, this coherence is converted into a population by emission of an electric field, which is measured in the experiment. Contrarily, in the developed 2D nanoscopy, four-wave mixing is initiated by a quadruple-pulse sequence, which leaves the quantum system in an electronic population. This electronic population carries coherent information about the investigated quantum system and can be mapped with a spatial resolution down to a few nanometers given by the spatial resolution of the photoemission electron microscope. Hence, 2D nanoscopy can be considered a generalization of time-resolved photoemission experiments. In the future, it may be of similar beneficial value for the field of photoemission research as “conventional” 2D spectroscopy has proven to be for optical spectroscopy and nuclear magnetic resonance experiments. In a first experimental implementation of coherent 2D nanoscopy coherent processes on a corrugated silver surface were measured and unexpected long coherence lifetimes could be determined.

Im Mittelpunkt der vorliegenden Arbeit stand das Studium der langwelligen Magneto-Kollektivmoden quasi-zweidimensionaler Elektronengase (Q2DEG) in GaAs. Diese Anregungen, die sich in Zyklotronresonanzen und Magneto-Intersubband-Resonanzen untergliedern, wurden mittels der Ferninfrarot-Fourierspektroskopie in einem Magnetfeldregime 0 T &#8804; B &#8804; 17 T untersucht. Die Zyklotronresonanz wurde über einen sehr weiten und umfassenden Dichtebereich von 1x10^11 cm^-2 bis 1.2x10^12 cm^-2 im Temperaturintervall 0.3 K < T < 80 K vermessen. Dabei kamen grundlegend unterschiedliche Proben-Strukturen mit Elektronenbeweglichkeiten im Bereich 5x10^5 cm^2/Vs bis 7x10^6 cm^2/Vs zum Einsatz, die unter unterschiedlichen Optimierungsgesichtspunkten hergestellt wurden. Mit den verfügbaren Proben und Parametern konnten mittels der Zyklotronresonanz die Regimes des Integralen (IQHE) und des Fraktionalen Quanten-Hall-Effektes (FQHE) abgedeckt und die bei hohen Temperaturen dominierenden Polaron-Renormierungen grundlegend charakterisiert werden. Zur Analyse und Interpretation der experimentellen Daten wurden theoretische Modelle zur mehrkomponentigen Zyklotronresonanz unte r den Aspekten der Polaron-Renormierung, der Leitungsband-Nichtparabolizität, der Streuung an Störstellen, der Abschirmung, sowie der Elektron-Elektron-Wechselwirkung und den mit ihr zusammenhängenden Grundzuständen entwickelt und mit diesen numerische Modell- und Anpassungsrechnungen durchgeführt. Die Magneto-Intersubband-Resonanzen wurden im Regime des IQHE experimentell untersucht. Dabei wurde die Gitterkopplertechnik zur Ankopplung des Lichtfeldes an diese Anregungen eingesetzt. Zum Verständnis und zur Interpretation der Messergebnisse wurden die selbstkonsistenten Gleichungen zur Berechnung der Magneto-Landau-Subband-Struktur und der dazu kompatiblen Dichteantwort im Rahmen der Hartree-Fock- (HFA) bzw. der zeitabhängigen Hartree-Fock-Näherung (TDHFA) aufgestellt und der numerische Lösungsweg dargelegt. Anhand von Anpassungsrechnungen wurde daraufhin die Magnetfeldabhängigkeit der Intersubband-Resonanzen analysiert.