Modeling soot particles as stable radicals: a chemical kinetic study on formation and oxidation. Part II. Soot oxidation in flow reactors and laminar flames

Abstract

Recent experimental and theoretical studies have shed light on the persistent radical behavior of large polycyclic aromatic hydrocarbons (PAHs) and soot particles. The first detailed soot model developed based on these findings is here proposed. While its good performances in predicting soot formation and growth when soot oxidation is negligible are presented in the companion paper of this work (Part I), the model is here validated against 45 cases of particle size distribution functions (PSDF) and soot volume fraction under oxidizing conditions. Notably, we show that by considering stable radical particles and aggregates it is possible to reproduce soot oxidation by O2 in plug flow reactors under very diluted conditions, whereas existing models could not predict particle consumption in the absence of particle surface activating species. Therefore, this work paves the way for a different approach to soot modeling that could help to better interpret soot reactivity. Oxidation of soot produced from different fuels (ethylene, n-heptane, and toluene, i.e. leading to different particle graphitization) at different temperatures (950–1073 K) and O2 concentrations (1000–10000 ppm) is successfully modeled adopting reference oxidation rates of gas-phase resonantly stabilized PAH radicals, featuring analogous reactivity compared to soot species, and by also accounting for different particle hydrogenation levels. Good model performances are also obtained in both rich and lean premixed ethylene flames (Tmax∼1480–1510 K) and in two sets of counterflow ethylene soot formation/oxidation flames (SFO, Tmax∼2560–2825 K), where OH radical becomes the dominant oxidizer. Finally, model limitations emerging in some of the cases considered are discussed through a sensitivity analysis and with references to other oxidation mechanisms proposed in the literature in order to highlight possible future improvement of the model predictive capability