Abstract

Heterogeneous catalytic conversions account for approximately 80% of the processes in conventional refineries.1 Achieving a detailed, comprehensive analysis of the full compositional spectrum, including the reactivity of representative molecules in these feeds, is critical for understanding reaction mechanisms and optimizing catalyst design.2 This doctoral thesis addresses key analytical challenges in integrating advanced techniques into traditional workflows to study compositional changes and reaction networks in unconventional feedstocks, proposing methods for translation of qualitative molecular information into a quantitative description.

The study begins with the hydrocracking of vacuum gas oil (VGO) blended with oxygenated plastics, such as PMMA and PET. A multi-technique approach, combining gas chromatography, bulk elemental analysis, and Fourier-transform ion cyclotron resonance mass spectrometry (FT-ICR MS), elucidates heteroatom removal mechanisms in waste-refinery blends. The results point-out to a direct relationship between the polymer structure (linear vs. aromatic) with the enhancing of heteroatom removal efficiency.3,4

Next, the thesis explores blends of VGO and pyrolysis waxes from HDPE, introducing a method to transform qualitative FT-ICR MS data into a quantitative description. This chapter focus on exposing the feasibility on integrating data from different techniques to calculate a mass-fraction based distribution, while using results from routine analysis like simulated distillation to define compositional space limits.

Finally, the thesis extends this quantitative approach to bio-oils, addressing challenges such as the absence of commercial internal standards or model compounds. It evaluates the influence of water content and ionization source on method applicability to bio-oils with varying hydrodeoxygenation (HDO) levels. The results highlight the relevance of extending the acquired knowledge in complex petroleum samples and their limitations to oxygen-rich biomass-derived oils.