Monodisperse magnetic nanoparticles have generated huge interest in applied research within the last years. The control of their monodispersity and surface properties leads to a variety of nanotechnological applications. The use as non-volatile data storage media and sensor applications are in the focus of industry. Moreover, magnetic nanoparticles are close to be employed in tumor therapy, bio-labelling or contrast agents in magnetic imaging. To an increasing extend these are studied in fundamental research as well, since nanoparticles with diameters of a few nm bridge the gap between atomic and solid state physics. In particular, the intrinsic magnetic properties of fine particles such as the magnetocrystalline anisotropy, the saturation magnetization and the Curie temperature are affected by the reduction of their size. The subject of this thesis is the investigation of magnetic Co/CoO core-shell nanoparticles and metallic Co nanoparticles with diameters in the range of 9-14 nm. The nanoparticles have been prepared by means of organometallic synthesis, and they exhibit a high degree of monodispersity (σ < 10%). The colloidal Co/CoO nanoparticles consist of a fcc ferromagnetic Co core covered with a naturally formed antiferromagnetic CoO shell. The main purpose of this study is the direct correlation of structure and magnetism in the particles. For better understanding of the influence of the antiferromagnetic shell and the Co/CoO interface on the magnetism of the particles pure metallic Co nanoparticles were studied. These have been repared from Co/CoO particles by reduction of the oxidic shell with a reactive plasma treatment. A erromagnetic/antiferromagnetic exchange coupled system shows an additional unidirectional contribution to the total magnetic anisotropy energy which can be measured by the so-called exchange bias. A general description of exchange anisotropy at ferromagnetic/antiferromagnetic interfaces is still lacking, since exchange bias strongly depends on the interface conditions. To address the interface magnetism explicitly, techniques have to be applied which give direct access to the interface. For direct comparison to metallic nanoparticles it is desirable to have control on the oxide shell without any loss of Co atoms. The system Co/CoO with typical layer thicknesses of 2-3 nm of the antiferromagnet CoO is a prototype for exchange bias. As shown in this thesis, the nanoparticles present a very strong exchange bias of µ₀H_EB = 0.4 T at 10 K compared to Co/CoO thin films. The microscopic quantities which govern the exchange bias are (i) the magnetic moments of the core and the shell Co atoms, (ii) the number of contributing interface moments, and (iii) the magnetic anisotropies of Co and CoO. The magnetic moment and its orbital contribution are quantitatively measured to understand the microscopic mechanisms which determine the exchange bias and the magnetic anisotropy energy. Two techniques which provide the necessary information with submonolayer sensitivity are Ferromagnetic Resonance (FMR) and X-ray Magnetic Circular Dichroism (XMCD). Explicitly, the ratio of orbital-to-spin magnetic moment can be measured. FMR probes the ferromagnetic core, only, while XMCD in the total electron yield mode is surface sensitive and - as demonstrated in this thesis - can be employed to extract the contribution of buried interfaces. The unique combination of both techniques yields a fcc bulk-like Co magnetic moment of a ferromagnetically ordered Co core (≤5 nm) and uncompensated Co²⁺ moments at the Co/CoO interface carrying a large orbital moment. XMCD studies at two different oxide shell thicknesses show that the interface moments are coupled parallel to the core. The effective magnetic anisotropy energy of Co/CoO particles has been determined by FMR to K^eff = 9 µeV/atom at T = 15 K which is much smaller than the hcp bulk value of 65 µeV/atom at T = 0 K and surprisingly matches the fcc bulk Co value of 8.5 µeV/atom. By frequency-dependent FMR in ambient conditions in a 18 months period of time after sample preparation, it has been measured that directly after the deposition of the particles on a substrate a 2-2.5 nm thick CoO shell forms within hours. After three weeks the shell grows to about 3 nm. From this point of time on the CoO shell acts as a self-passivating layer for each individual particle. The oxidation slows down to a few atoms per day. This experiment demonstrates that small structural modifications can be addressed by FMR investigations. In order to verify and to compare the results of Co/CoO particles to pure metallic particles, an in-situ reactive plasma etching process has been employed. By successive oxygen and hydrogen plasma exposure it was possible (i) to remove the organic ligands and (ii) to reduce the entire CoO to metallic Co. This method allowed the controlled modification of the surface without any agglomeration or movement of the particles on the substrate for coverages up to one monolayer. Slightly larger coverages admitted to partial sintering of particles in a top layer while particles in the bottom layer remained fixed to the substrate. The removal of the organic ligands led to the formation of a double layer system which shows exchange coupling between particles perpendicular to the substrate. Using detailed structural and morphological studies as input parameters for theory, Landau-Lifshitz-Gilbert type of simulations have been performed to simulate element-specific hysteresis loops. The calculations have shown that (i) the metallic particles exhibit a low effective magnetic anisotropy energy density K^eff = 1.5 µeV/atom at T = 15 K which is less than 20% of the fcc bulk Co anisotropy constant. (ii) The exchange coupling strength suggests that in medium only four atomic pairs of Co atoms are exchange coupled at the interface between top and bottom layer particles. The magnetic moment of the metallic particles has been determined to be 1.56 µ_B/atom by means of XMCD investigations on monolayer samples. The reduction compared to the bulk magnetic moment of 1.72 µ_B/atom has been assigned to the hydrogen-load of particles during the hydrogen plasma exposure. Annealing at T = 950 K successfully dissociates cobalt hydride and forms pure Co nanoparticles and a total magnetic moment of 1.83 µ_B/atom has been found. The enhancement compared to the bulk magnetic moment can be explained by the contribution of about 10% surface atoms in reduced symmetry which is manifested in an enhanced orbital moment. In future studies the surface anisotropy can be measured in pure metallic nanoparticles. Compared to ultrathin films the nanoparticle approach has the big advantage that the contact area to the substrate is very small and thus the influence of the substrate/ferromagnet interface is negligible. To correlate the magnetic properties to the structure of the nanoparticles detailed transmission electron microscopy (TEM) investigations have been performed. High-resolution TEM has shown that the Co/CoO particles consist of multiply twinned fcc Co core and a multigrain fcc CoO shell. The thickness of the CoO shell has been determined to be 2-2.5 nm by energy-filtered TEM. The passivating CoO shell encases a metallic Co core of 5-9 nm depending on the total particle diameter. With these results the low effective magnetic anisotropy energy density K^eff = 1.5 µ_B/atom found for metallic Co nanoparticles can be explained by the multiply twinned structure of the ferromagnet. Individual grains with randomly oriented anisotropy axes reduce K^eff remarkably and produce an effective uniaxial anisotropy. In the case of Co/CoO nanoparticles the oxide shell increases K^eff which results in a magnetic hardening.