The prediction of atmospheric ice formation represents one of the grand challenges in atmospheric sciences. This may in part due to the fact that (i) different ice nucleation pathways exist that lead to ice crystal formation; (ii) only a small fraction of the aerosol particles act as ice-nucleating particles (INPs), on the order of 10^-1 to 10^-5; (iii) the different ice nucleation activities of various particle-types that are likely the result of the individually different physicochemical particle properties such as composition and morphology.

The Intergovernmental Panel on Climate Change (2013) ascertains that cloud adjustments due to aerosols contribute to the largest uncertainties to the global radiative forcing and as such climate. This latest assessment contributes a cooling effect to those adjustments. However, it is important to note that the formation of ice crystals present in mixed-phase clouds and cirrus clouds are not accounted for in this estimated. For example, it is suggested that cirrus clouds have a predominantly warming effect on a global scale.

Prediction of ice crystal number concentrations from INPs is challenging and an ongoing field of research. It is complicated by the insufficient understanding how aerosol particles act as INPs.

The right hand side shows a schematic overview of different ice nucleation pathways, exemplary given as a function of temperature, relative humidity, and supersaturation with respect to ice (RHice and Sice, respectively). (A) Homogeneous ice nucleation, (B) immersion freezing, (C) deliquescence and water uptake followed by immersion freezing, (D) deposition ice nucleation, (E) contact ice nucleation, (F) inside-out freezing. Symbol forms and
sizes not to scale. (Knopf et al., 2018). The existence of multiple pathways to form ice are one of the reasons why it is challenging to accurately predict INPs and number of ice crystals.

In our group we use custom-built ice nucleation cells that allow simulation of tropospheric temperatures and humidity coupled to optical microscopy or scanning electron microscopy to determine under which thermodynamic conditions aerosol particles act as INPs.

The experimental data serve for theoretical interpretation and derivation of parameterizations that can be applied in cloud resolving and climate models. We recently have shown that immersion freezing can be well described by a stochastic freezing process as assumed in classical nucleation theory (Knopf et al. 2020).

Our group focuses on homogeneous ice nucleation and heterogeneous ice nucleation by immersion freezing and deposition ice nucleation. We employ well-specified laboratory generated particle systems to test our understanding of ice formation but also use field-collected (ground, ship, airborne) authentic particles as samples for INPs. Since the majority of ambient particles are associated with organic matter, particular focus is placed on how organic compounds alter the ice nucleation capability of the aerosol (Knopf et al., 2018).