Predictive Models for Gas-Particle Partitioning and Viscosity of Atmospheric Aerosols

Atmospheric aerosol features like liquid-liquid phase separation, high viscosity and glass formation have gained interest due to their impacts on heterogeneous chemistry, hygroscopicity, the equilibration time scale, aerosol mass concentrations, ice nucleation and aerosol-cloud effects. We have developed a thermodynamic model framework for coupled gas-particle partitioning, solid-liquid and liquid-liquid equilibrium with the Aerosol Inorganic-Organic Mixtures Functional groups Activity Coefficients (AIOMFAC) model at its core. For well-characterized aerosol systems, this framework enables predictions for a wide variety of process-level applications, usually targeting conditions encountered in Earth's troposphere. Furthermore, the framework may also apply to particle compositions and temperature ranges relevant for conditions in other planetary atmospheres.

In air quality and chemistry-climate models, detailed aerosol chemical composition information is frequently lacking; concurrently, keeping computational costs of process modules low is critical. In this context, we have developed a reduced-complexity model, the Binary Aerosol Thermodynamics (BAT) model, based on AIOMFAC-generated data. BAT can be driven with either low- or high-fidelity input information, such as average molar mass and oxygen-to-carbon ratio of aerosol components. Coupled to a volatility basis set, this model can predict water uptake, miscibility and cloud formation potential of organic aerosols.

Beyond equilibrium, gas-particle partitioning can be time-sensitive in the presence of viscous, semi-solid to glassy particle phases expected to exist at low temperatures and/or low aerosol water contents. We introduce a new thermodynamics-based group-contribution model, which is capable of accurately predicting the dynamic viscosity of an aqueous organic mixture over several orders of magnitude (∼10^-3 to 10¹² Pa s) as a function of temperature and mixture composition, accounting for the effect of relative humidity on aerosol water content. Comparing this new model with simplified modeling approaches reveals that the developed group-contribution method is the most accurate in predicting mixture viscosity, although accurate pure-component viscosity predictions (or experimental data) are key.