Toluene photo-oxidation and secondary organic aerosol formation: EUROCHAMP-2020 multi-chamber experiments

Atmospheric simulation chambers (ASCs) are one of the most advanced tools for the experimental investigation of the oxidation of volatile organic compounds (VOCs) and the subsequent secondary organic aerosol (SOA) formation. Toluene is one of the most prevalent anthropogenic VOCs. Its photo-oxidation yields a wide range of products in the gas phase and a significant amount of SOA. Some of the remaining uncertainties about toluene atmospheric chemistry are possibly linked with chamber artifacts. In this study, several atmospheric simulation chambers, characterized by a great diversity (size, shape, material of walls, light source, instrumentation, measurement techniques, etc.), performed several toluene photo-oxidation experiments under different pre-set conditions (levels of toluene, NOx, and relative humidity, presence, or lack of seeds). A model based on the Master Chemical Mechanism (MCM) and a SOA production module were used to facilitate the synthesis of the results. The results of the multiple-chamber toluene experiments suggest that a combination of facilities can provide a better picture of the overall behavior and that significant gaps remain in our understanding of the system, especially in the later oxidation stages. For cresol, a first-generation product, the observed gas-phase yields, ranging from 3% to 8% under low-NOx conditions, were consistent with model predictions. In contrast, the measured gas-phase yields of benzaldehyde (8-16%%) were higher than the predicted (3-5%) yields, highlighting uncertainties in the H-abstraction pathway of the toluene reaction with hydroxyl radicals (OH). Glyoxal and methylglyoxal yields varied between facilities, with the model often failing to capture their temporal profiles. Additionally, the MCM-based model struggled to reproduce concentrations of oxygenated products (e.g., C7H8O2 and C7H8O3), suggesting shortcomings in simulating later oxidation stages. Most notably, the model consistently underpredicted SOA mass across experiments, pointing to critical gaps in the representation of SOA-forming pathways in the currently used version of the MCM.