On the evolution of sub- and super-saturated water uptake of secondary organic aerosol in chamber experiments from mixed precursors

To better understand the chemical controls of sub- and super-saturated aerosol water uptake, we designed and conducted a series of chamber experiments to investigate the evolution of secondary organic aerosol (SOA) particle physicochemical properties during photo-oxidation of single and mixed biogenic (α-pinene, isoprene) and anthropogenic (o-cresol) volatile organic compounds (VOCs) in the presence of ammonium sulfate seeds. During the 6h experiments, the cloud condensation nuclei (CCN) activity at super-saturation of water (0.1%-0.5%), hygroscopic growth factor at 90% relative humidity (RH), and non-refractory PM1 chemical composition were recorded concurrently. Attempts to use the hygroscopicity parameter κ to reconcile water uptake ability below and above water saturation from various VOC precursor systems were made, aiming to predict the CCN activity from the sub-saturated hygroscopicity. The thermodynamic model AIOMFAC (aerosol inorganic-organic mixtures functional groups activity coefficients) was used to simulate κ values of model compound mixtures to compare with the observation and to isolate the controlling factors of water uptake at different RHs. The sub- and super-saturated water uptake (in terms of both κHTDMA and κCCN) were mainly controlled by the SOA mass fraction, which depended on the SOA production rate of the precursors, and the SOA composition played a second-order role. For the reconciliation of κHTDMA and κCCN, the κHTDMA/κCCN ratio increased with the SOA mass fraction and this was observed in all investigated single and mixed VOC systems, independent of initial VOC concentrations and sources. For all VOC systems, the mean κHTDMA of aerosol particles was ∼25% lower than the κCCN at the beginning of the experiments with inorganic seeds. With the increase of condensed SOA on inorganic seed particles throughout the experiments, the discrepancy of κHTDMA and κCCN became weaker (down to ∼0%) and finally the mean κHTDMA was ∼60% higher than κCCN on average when the SOA mass fraction approached ∼0.8. As indicated by AIOMFAC model simulations, non-ideality alone cannot fully explain the κ discrepancy at high SOA mass fraction (0.8). A good agreement in κCCN between model and observation was achieved by doubling the molecular weight of the model compounds or by reducing the dry particle size in the CCN counter. This indicates that the evaporation of semi-volatile organics in the CCN counter together with non-ideality could have led to the observed κ discrepancy. As a result, the predicted CCN number concentrations from the κHTDMA and particle number size distribution were ∼10% lower than CCN counter measurement on average at the beginning, and further even turned to an overestimation of ∼20% on average when the SOA mass fraction was ∼0.8. This chemical composition-dependent performances of the κ-Köhler approach on CCN prediction can introduce a variable uncertainty in predicting cloud droplet numbers from the sub-saturated water uptake, the influence of which on models still needs to be investigated.