It is perhaps with good reason that one of the more
controversial areas of laboratory studies is that of particle
electrification. Particle processes involving charge separation
are far from equilibrium for both ice phase processes and for
ionic effects. Laboratory simulation of ice effects is
particularly difficult as not only does particle impact and
bounce occur, but the relative state of the surface of the two
particles as determined by growth, evaporation, droplet freezing,
defect structure and surface chemistry are all of potential
importance. Yet cloud electrification is a very real phenomena,
and ice particle interaction plays an important role. At the
center of arguments is the extent to which any laboratory
experiment simulates the thunderstorm cloud processes. This
requires simulation of ambient conditions---as for all the other
topics discussed above---but it also requires simulation of
impact on a riming surface. Aircraft studies in Florida (Willis
et al 1994) show that magnitude of electric field is low with all
supercooled cloud yet increases rapidly with initiation of the
ice phase, and is linked with specific vertical shear regions of
a convective cloud where the all supercooled water in the updraft
interfaces with the all ice in the downdraft. There are
large gradients of ice concentration, liquid water content, and
surface temperature of particles in the horizontal over a few
hundred meters, and over a range of temperatures. Simulation
requires reproducing all of the parameters, besides the particle
bounce which results in the charge transfer. Attempts to
simplify the concepts have differentiated processes taking place
during evaporation or growth alone. Laboratory experiments of
Dong and Hallett (1992, see above) found positive charging of
growing ice at temperatures below --4
C, with negative
charging above this temperature; the charging rate was greater
(X 10) for growth of ice crystals from the vapor on fresh rime.
Evaporating crystals charged negatively, reaching a maximum at a
fall velocity of 2 m s
. This was related to differential
mobility of ions in the air and of charged defects in the ice
lattice. Brooks and Saunders (1994) used a laboratory technique
of bounce in an electric field to show that inductive effects
could only just be of importance; a result consistent with an
extensive numerical model using laboratory derived criteria
(Ziegler et al, 1991).
Accretion experiments involved measurement of charge on a
rotating system riming in a supercooled cloud, seeded to produce
ice, (Saunders et al 1991) together with similar measurements in
a wind tunnel. This gives a reversal of sign depending on
temperature and liquid water content in the range 1/4 to
1 g m
. Williams et al (1991) argue that the state of the
riming surface differs in growth and evaporation. The collision
process separates charge even if riming is not occurring
(Williams et al, 1994). A key issue of the debate has centered
on measurement of liquid water and ice content of the laboratory
produced clouds, and comparison with earlier studies (Saunders,
1993, Williams and Zhang 1993, and Saunders, 1994). A major
uncertainty is the local conditions of rime growth, and
particularly the temperature and vapor gradients around freezing
droplets on the graupel surface as particles bounce under a wide
range of surface conditions.
It is of interest that the complex nature of the riming
surface has implications for retention or exclusion of gases
during freezing, an important aspect of SO
--SO
conversion
by H
O
, where discrepancies exist between laboratory and
field studies (Snider et al, 1992). Baker and Dash (1994), in a
theoretical study, postulate a transfer of surface charge layer
from one crystal to another crystal with different curvature,
temperature and surface layer thickness. This mechanism is very
difficult in concept and difficult to investigate experimentally
in view of the variability of surface conditions. An experiment
on frost electrification (Rydok and Williams 1991) showed that a
precooled ice sphere frosted and became charged positively when
exposed in a warm, moist environment. There is no attempt here
to simulate atmospheric processes directly as thermal gradients,
moisture gradients, and local droplet nucleation and capture are
exaggerated. Nevertheless, careful analysis of the details of
such a complex process, as in the riming case discussed above,
may be capable of yielding insight into what is a very complex
physical and possibly chemical problem.
A further complication for growth of particles by
accretion is that single (1 cm) particles may go in and out of
``wet'' growth (with surface temperature of 0
C)
depending on local liquid water content fluctuations which, as
shown by a theoretical study, may lead to a hysteresis because
significant surface roughness will inhibit wet growth at slower
fall speed and at lower density, (Johnson and Rasmussen, 1992).
Should wet growth begin, a much lower liquid water content will
be required to return to dry growth. A continued discussion on
the nature and thickness of supercooled water films on
hailstones, growing completely wet, observed in laboratory
studies shows that there are still gaps in our understanding of
this complex process (Lozowski, 1991, List, 1991). The thickness
of the film may be determined by an equilibrium between accretion
rate in shear flow and heat transport---the question under debate
is the relative importance of the terms. The inherent instability
of an ice dendrite growing from an underlying ice surface into
a supercooled liquid above in quasi-steady state resulting
from shear flow and the arrival of supercooled droplets.
This is difficult both to examine and model; there seems
to be room for an innovative laboratory study
of this interface. The experimental techniques are readily
criticized in as far as measurements of key quantities are often
imprecise; liquid water content is important, yet instantaneous
measurements in the laboratory in a mixed phase cloud which
differs from place to place provides a major challenge. Add to
this uncertainties concerning the impurity and ice defect
structure and the true nature of the problem becomes evident.
The simplicity of the diffusion chamber approach is to be
contrasted with the full simulation approach of the riming wind
tunnel or rotation experiment. Riming surface numerical
simulation is a brave attempt but fraught with uncertainties with
so many unknown physical and chemical processes on the scale of
an individual droplet. Nevertheless progress is significant.