Insufficient oxidation capacity in the photooxidation of aromatic VOCs results in underestimation of SOA yield

), influence atmospheric chemistry, climate, and public health. However, their formation processes remain insufficiently represented in atmospheric models due to their complexity. Here, we investigate SOA formation from representative aromatic hydrocarbons (toluene, m-xylene, and 1,3,5-trimethylbenzene) under varying precursor concentrations using controlled smog chamber experiments. Leveraging high-resolution mass spectrometry and machine learning approaches, we demonstrate that oxidation capacity (characterized by the cumulative OH exposure), modulated by precursor concentration, is the primary driver of SOA yield. A key and novel finding is that at lower precursor concentrations (∼10 ppb), which are more representative of real atmospheric conditions, the enhanced OH exposure per molecule promotes more efficient multigenerational oxidation. This process favors the production of low-volatility organic compounds, thereby substantially increasing SOA formation compared to that at higher precursor concentrations. These findings, validated by the machine learning analysis, provide mechanistic insights into SOA evolution and underscore the need for improved representation of multigenerational oxidation in atmospheric models to enhance air quality and climate predictions.