Mineral Dust and its Microphysical Interactions with Clouds - Mineral Dust: A Key Player in the Earth System - page 294

convection; we focus here on ice clouds which form in situ. In the absence of ice-
nucleating particles, haze particles become sufficiently dilute and sufficiently cold
that ice nucleates homogeneously within them at an RH below water saturation
(Koop et al. 2000 ). Hence, in the upper troposphere, ice crystals tend to form
well below water saturation. If heterogeneous ice nuclei are present, nucleation
can occur at much lower ice supersaturation. Mineral dust particles free of any
coatings are thought to initiate ice formation through deposition nucleation at low
supersaturations (RH i 110-120 %). Recently, it has been suggested that what is

observed as the direct deposition of ice onto a solid material is in fact a result of
condensation of water into cracks or pores which then freezes (Marcolli 2013 ).
When mineral dusts become coated with chemically inert (or reactive) soluble
materials, the pathway of nucleation changes and nucleation tends to occur at a
higher supersaturation. Instead of nucleation occurring on a mineral surface in
contact with air, nucleation occurs in an aqueous solution which Hoose and Möhler
( 2012 ) term “immersion freezing of solution droplets.”
The complexity and range of nucleation pathways which are possible in the
atmosphere means that it is critical to discuss IN concentrations in connection to a
particular pathway under a set of specific conditions pertinent for a particular cloud
type. As mentioned earlier we have no underpinning fundamental understanding of
ice nucleation, and in fact, different materials may nucleate ice for different reasons.
However, there have been some advances, for example, Knopf and Alpert ( 2013 )
suggest a means of describing immersion mode nucleation in both solution droplets
and pure water using a water activity-based approach. In the following paragraphs,
we review several methods for quantifying ice nucleation by mineral dust particles.
In the past, it was common to quote the conditions under which an observable
amount of ice formed in an experiment (e.g., Pruppacher and Klett 1997 ). These
threshold ice nucleation conditions (temperature or RH) are experiment-specific
values and difficult to translate to the atmosphere. For example, reported freezing
temperatures in recent work on feldspar mineral dust immersed in water droplets
ranged from 25 ı C depending on the surface area available for nucleation

(Atkinson et al. 2013 ). Since the surface area available and the sensitivity of
ice particle detection vary between different instruments, the threshold freezing
temperature will be different for different instruments. Hence, it is necessary
to parameterize the freezing probability in a form which is then transferable to
atmospheric conditions. This task remains a challenge, but a number of parame-
terizations/models/descriptions have been put forward.
Nucleation is a probabilistic, stochastic, process and the probability of nucleation
increases with a greater surface area of IN and for longer times (Kashchiev 2000 ;
Mullin 2001 ). For an idealized uniform ice-nucleating material where each particle
of the same size has the same probability of nucleating ice as the next particle, the
probability of droplets remaining liquid is

5to

N 2
N 1 D

exp . J het st/ (12.4)