Classical nucleation theory of immersion freezing: Sensitivity of contact angle schemes to thermodynamic and kinetic parameters

• Published in: Atmospheric chemistry and physics Vol. 2016; no. 1; p. 1
• Main Authors: Ickes, L, Welti, A, Lohmann, U
• Format: Journal Article
• Language: English
• Published: Copernicus GmbH 18.01.2016
• Subjects: Comparative analysis; Stochastic processes; Thermodynamics
• ISSN: 1680-7316; 1680-7324

Heterogeneous ice formation by immersion freezing in mixed-phase clouds can be parameterized in general circulation models (GCMs) by Classical Nucleation Theory (CNT). CNT parameterization schemes describe freezing as a stochastic process including the properties of insoluble aerosol particles, so called ice nuclei, in the droplets. There are different ways how to describe the properties of aerosol particles (i.e. contact angle schemes), which are compiled and tested in this paper. The goal of this study is to find a parameterization scheme for GCMs to describe immersion freezing with the ability to shift and adjust the slope of the freezing curve compared to homogeneous freezing to match experimental data. The results of using CNT are very sensitive to unconstrained kinetic and thermodynamic parameters in the case of homogeneous freezing leading to uncertainties in calculated nucleation rates J.sub.hom of several orders of magnitude. Here we investigate how sensitive the outcome of a parameter estimation for contact angle schemes from experimental data is to kinetic and thermodynamic parameters. We show that additional free parameter can mask the uncertainty of J.sub.imm due to thermodynamic and kinetic parameters. Different CNT formulations are fitted to an extensive immersion freezing dataset as a function of particle diameter (d), temperature T and time t for different mineral dust types, namely kaolinite, illite, montmorillonite, microcline (K-feldspar) and Arizona test dust. It is investigated how accurate different CNT formulations (with the estimated fit parameters) reproduce the measured freezing curves, especially the time and particle size dependence of the freezing process. The results are compared to a simplified deterministic freezing scheme. It is evaluated in this context which CNT based parameterization scheme to represent particle properties is a good choice to describe immersion freezing in a GCM.