Correlation of pure component Gibbs energy. Using UNIFAC group contribution

Abstract. The viscosity of primary and secondary organic aerosol (SOA) has important implications for the processing of aqueous organic aerosol phases in the atmosphere, their involvement in climate forcing, and transboundary pollution. Here we introduce a new thermodynamics-based group-contribution model, which is capable of accurately predicting the dynamic viscosity of a 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. The mixture viscosity modelling framework builds on the thermodynamic activity coefficient model AIOMFAC (Aerosol Inorganic–Organic Mixtures Functional groups Activity Coefficients) for predictions of liquid mixture non-ideality, including liquid–liquid phase separation, and the calorimetric glass transition temperature model by DeRieux et al. (2018) for pure-component viscosity values of organic components. Comparing this new model with simplified modelling approaches reveals that the group-contribution method is the most accurate in predicting mixture viscosity, although accurate pure-component viscosity predictions (and associated experimental data) are key and one of the main sources of uncertainties in current models, including the model presented here. Nonetheless, we find excellent agreement between the viscosity predictions and measurements for systems in which mixture constituents have a molar mass below 350 g mol−1. As such, we demonstrate the validity of the model in quantifying mixture viscosity for aqueous binary mixtures (glycerol, citric acid, sucrose, and trehalose), aqueous multicomponent mixtures (citric acid + sucrose and a mixture of nine dicarboxylic acids), and aqueous SOA surrogate mixtures derived from the oxidation of α-pinene, toluene, or isoprene. We also use the model to assess the expected change in SOA particle viscosity during idealized adiabatic air parcel transport from the surface to higher altitudes within the troposphere. This work demonstrates the capability and flexibility of our model in predicting the viscosity for organic mixtures of varying degrees of complexity and its applicability for modelling SOA viscosity over a wide range of temperatures and relative humidities.

Download Full-text

Pure-component vapor pressures using UNIFAC group contribution

Industrial & Engineering Chemistry Fundamentals ◽

10.1021/i100003a010 ◽

1981 ◽

Vol 20 (3) ◽

pp. 239-246 ◽

Cited By ~ 47

Author(s):

Torben Jensen ◽

Aage Fredenslund ◽

Peter Rasmussen

Keyword(s):

Pure Component ◽

Group Contribution ◽

Vapor Pressures

Download Full-text

Extension of the UNIFAC group-contribution method for the prediction of pure-component vapor pressures

Industrial & Engineering Chemistry Process Design and Development ◽

10.1021/i200022a015 ◽

1983 ◽

Vol 22 (3) ◽

pp. 433-436 ◽

Cited By ~ 16

Author(s):

Osnat Ben Yair ◽

Aage Fredenslund

Keyword(s):

Pure Component ◽

Group Contribution ◽

Vapor Pressures ◽

Group Contribution Method

Download Full-text

Comparison of different methods for estimating activity coefficients in binary mixtures of hydrocarbons

Collection of Czechoslovak Chemical Communications ◽

10.1135/cccc19811541 ◽

1981 ◽

Vol 46 (7) ◽

pp. 1541-1548

Author(s):

Vladimír Dohnal

Keyword(s):

Activity Coefficient ◽

Binary Mixtures ◽

Activity Coefficients ◽

Regular Solution Model ◽

Pure Component ◽

Group Contribution ◽

Coefficient Estimation ◽

The Mean ◽

Group Contribution Methods ◽

The One

The accuracy and reliability of a number of different methods for predicting activity coefficients in binary solutions of hydrocarbons was tested. Various modifications of the regular solution model and of the one-parameter Wilson equation and various group-contribution methods were applied to a set of 53 binary mixtures of hydrocarbons of different types. The agreement of the calculated and experimental dependence of activity coefficients on composition was considered. On using the best methods requiring the knowledge of pure component properties only, it is necessary to expect on the average an error of 7% in the value of activity coefficient. When using the group-contribution methods, which employ condensed information on related systems for the prediction, the mean error in the activity coefficient estimation lies about 4%.

Download Full-text