In this work, the production of nanoparticles in two different gas-phase synthesis processes was studied by means of numerical simulations. Complementarily, the applied modelling approaches were validated against experimental data, and new modelling approaches were investigated. The
desired products in both examined processes were iron oxide (Fe₂O₃) particles, produced either from the gaseous precursor iron pentacarbonyl (FeCO₅) in flame synthesis or from the liquid precursor iron (III) nitrate (Fe(NO₃)₃) in spray flame pyrolysis. Besides their different precursors, the processes can be distinguished by their flow type: laminar flow in the flame reactor and turbulent flow in the spray flame reactor.

In the first part of the work, highly-resolved simulations of two laminar, iron-oxide particle forming, hydrogen/oxygen flames in a low-pressure environment were performed. The flames were characterized by their different orientations, one burning from the top down (down-firing
flame, DFF) and the other, as is commonly done, from the bottom up (up-firing flame, UFF). Substantial variations between the temperature-time profile of the DFF and UFF and a good agreement between modeled and experimentally measured particle sizes were found. Moreover, a early particle formation zone was detected in the experiments, and the corresponding reaction kinetics model from simulations reproduced this observation for iron clusters. Finally, simulations reproduced a systematic shift of the results. This shift could be attributed to the influence of the
probing system in the experiment The influence of the probing system on the flame structure was further investigated in the subsequent study. A methane-oxygen flame, operated at low pressure, was investigated experimentally (by another working group) and numerically. The measurements were performed with two different probing nozzles, which differed in their material (quartz/metal) and their orifice diameter (90 µm/550 µm). Experiments and simulations showed good agreement, and several factors, that should be considered when evaluating the sample effect, could be derived (e.g., sample temperature and suction effects).
The second part of the work investigated nanoparticle synthesis in a turbulent flow. For this purpose, the turbulent spray flame ’SpraySyn’ was simulated using large eddy simulations (LES). In the scope of the research group SPP1980, the resulting particle size distribution could be validated with experimentally measured data for the first time and agreed well. A sub-filter model for the coagulation source term was implemented and tested based on the validated data. The sub-filter model did not show a significant influence on the results. As the final part of this work, a Lagrangian-transported filtered probability density function (FDF) method with a sectional particle model was implemented in the LES code and validated against a generic test case. Initial simulations of the ”SpraySyn” flame were carried out using the FDF method, in which a significant impact could not be observed compared to the conventional description of particle growth. It is important to note that the reason for the lack of influence at this stage remains unclear. Further investigation is required to gain a deeper understanding of the underlying factors. It is important to emphasize that this observation is based on preliminary results that still require more extensive validation.