As the requirements for steel producers in terms of purity are increasing, the aim of this work was to elaborate a deeper understanding of these non-metallic inclusions (NMI). Not only the phase composition but also their morphology was examined more closely. The samples to be examined were polished steel samples with exposed inclusions. The chemism of these particles was either in the system CaO-Al2O3-MgO or in the system CaO-Al2O3-MgO-SiO2, depending on the steel grade in which the inclusion was present. It should also be noted that zirconia were also present in some inclusions. To clarify the morphology of the NMI and its dependence on the cooling rate, two different approaches were followed. On the one hand, the samples obtained were prepared in such a way that experiments with the "high temperature laser scanning confocal microscope" (HT-LSCM) were possible. The sample was heated to such an extent that the inclusions melted and were subsequently cooled at a defined rate. Three different cooling rates were applied: 25, 250 and 1000°C/min. By means of a microscope, the melting and solidification of the particles could be followed and observed in-situ. Furthermore, samples of synthetic raw materials with a chemical composition similar of the NMI were prepared using a simultaneous thermal analysis apparatus (STA). On the one hand, these samples served to determine the transformation temperature of the mineral phases, but more importantly, they fulfilled the purpose of producing sufficiently large samples with a cooling rate of 25 °C/min. To clarify the phase composition of the inclusions before and after the temperature treatment, as well as the STA samples, a scanning electron microscope with associated energy dispersive X-ray spectroscopy (EDX) was used. To analyse the microstructure after the HT-LSCM test below the surface, the specimens were ground and polished. X-ray diffraction analysis (XRD) was also performed on the STA samples. In order to identify the phases present, especially the finely intergrown matrix, the obtained results of XRD and EDX analysis were evaluated by thermochemical calculations. To validate the identified phases, phase diagrams of the current system were subsequently used. MA spinel (less frequently pure periclase crystals) was identified as a typical magnesium-containing phase. For the calcium aluminates, CA is the most common phase, whereas C12A7, C2AS and C3MA2 often occur as a gusset phase. In the case of a content of zirconia, which most probably enters the steel through erosion of refractory functional products (casting tube and shadow tube) or through the sliding sand, the phases CaZrO3 and Ca13Zr2A12O35 could be detected. The analysis of the HT-LSCM showed that the microstructure generated by a cooling rate of 1000°C/min was amorphous, while the samples with cooling rates of 25 and 250°C/min solidified crystalline. When comparing the microstructures of the STA samples and the HT-LSCM samples of the same cooling rate, differences were found. These differences can be attributed to the segregation of the liquid due to sedimentation of crystalline magnesium-containing phases during temperature treatment in the HT-LSCM. With the experiments in the STA apparatus, the phase composition and morphology of the original inclusions could be reproduced. Furthermore, the reproduction of dendritic periclase was achieved and the cause for this formation could be traced back to the chemical composition of the NMI.