Ice nucleation active site-independent stochasticity in heterogeneous ice nucleation

Clouds are central elements of the hydrological cycle and climate. They are responsible for precipitation and strongly impact global and regional temperature by reflecting incoming solar radiation, and absorbing heat emitted from Earth. The amount of ice contained in clouds determines much of their properties and their impact on climate. Cloud ice is mostly formed by heterogeneous nucleation (HIN), catalyzed by insoluble airborne particles (such as dust, biological particles and soot). Despite its importance, the description of HIN in climate models is vastly incomplete and remains one of the major challenges that impedes further progress in estimates of anthropogenic climate change.

 

One important open question is whether HIN should be modelled as a deterministic (time independent) or as a stochastic (time dependent) process. The typical sigmoidal shape of freezing curves - i.e., the fraction of frozen samples as a function of time - suggests the latter to be the correct approach. Current stochastic models cast the entire stochastic behavior on the differences between the IN activity of different ice nucleation active sites (INAS) in the sample. Recent studies however indicate that a large part of stochasticity may be independent from INAS variability.

 

The current work uses large scale molecular simulations freezing droplet ensembles on pure graphitic surfaces to explore the molecular scale origins of INAS independent stochasticity. We find that the interplay of three main factors: the size fluctuations of the pre-critical ice embryo, capillary wave fluctuations of the non-frozen droplet surface and the extent of stacking disorder - i.e.: cubic to hexagonal polymorph ratio - in the frozen droplet manifests as observable stochasticity even if the properties of the IN surface are identical. We conclude by quantifying the extent of INAS-independent stochasticity as a range of contact angles in the framework of the Adsorption Nucleation Theory. The resulting representation is then used to provide a stochastic adsorption-based parameterization of deposition freezing on soot particles.