Particles in the atmosphere grow or evaporate depending on the local saturation ratio, and under some conditions the local radiation balance. The simulation of supersaturation for growth of aerosol droplets and ice crystals has been achieved by different techniques. In principle, expansion, mixing and combined thermal and vapor diffusion produce supersaturation over a wide range from a few tenths to a few hundred percent. In practice, specific techniques suit some ranges rather than others. The Wilson chamber rapid expansion is more appropriate for droplet homogeneous nucleation at high supersaturation (Schmitt 1992); the annular cylinder diffusion chamber for ice nucleation near water saturation (Rogers 1993). Such an instrument responds to silver iodide nuclei with a few seconds time response (Rogers 1994). The dynamic thermal vapor diffusion chamber is ideal for the growth of ice from the vapor, under supersaturations relevant to the atmosphere. This chamber is also suited for simulation of fall velocity, by incorporating it into a horizontal wind tunnel and independently adjusting the flow to approximate the terminal speed as the crystal grows (Hallett and Knight, 1994). The advantage of all diffusion chambers is that they are steady state, with temperature, supersaturation and air velocity subject to independent control. Static chambers have invariant temperature and supersaturation profiles; dynamic chambers give changing conditions moving away from the entrance port, but which are invariant at a given location.
A large expansion chamber is ideal for studying gas--aerosol
conversion. Experiments in the Calspan 600 m
expansion
chamber examined conversion of SO
to sulfate and demonstrated
increased mass Cloud Condensation Nuclei (CCN) during absorption
in cloud droplets produced during expansion (Hoppel et al 1994).
This system has an advantage in as far as trace gases can be
readily added, changing acidity as well as composition, and the
cloud can be readily recycled by compression. Such a large
system is evidently not appropriate for surface deposition
studies, where particle loss to a surface with diffusive flux is
important. Such results are shown by Chen et al (1992) although
the large aspect ratio of their chamber gives less than ideal
geometry. Growth of aerosol under non steady state conditions
and approach to saturation was examined in a wetted wall vertical
laminar flow reactor with particles growing during vertical
ascent and showed agreement with simple theory (Li et al 1992).
This simulates aerosols in the human lung. Coating the chamber
with solutions of known concentration could in principle simulate
lower relative humidities, although local concentration changes
in such solutions films may lead to some uncertainty. Direct
measurement of small (few %) supersaturation can be made using
the technique of Gerber (1991) by measurement of droplet growth
on nuclei deposited on a hydrophobic surface. This instrument
demonstrates the occurrence of super-saturations up to 0.5% in
radiation fogs, which has implication for the nature of the
mixing process.
A new technique for producing small supersaturation and transient subsaturation conditions uses an expansion chamber containing a water solution of tetraethylene glycol. The water vapor pressure is equivalent to 2.5% relative humidity and can be increased during expansion to values up to 160% relative humidity. The presence of the glycol has minimum effect on nucleation (Schmitt and Hagen, 1994).
Subsaturation studies for ice particles is important for break up and melt studies. This can be readily achieved by growing crystals in a thermal vapor diffusion chamber, and then evaporating in reverse flow in undersaturated air produced by heating air, ice saturated at a lower temperature. Simulation is achieved by ventilation adjusted to near terminal velocity (Dong et al 1994). This chamber has great versatility in controlling environmental conditions; its disadvantage lies in the necessity for growth on a filament support.
A question of terminology is relevant. Fukuta (1994) points out that the term ``thermal diffusion chamber'' (originally described by Schaefer for ice growth in the Project Cirrus study (Schaefer, 1952) implies a gradient of temperature and vapor, and proposes ``thermal (gradient) vapor diffusion chamber.'' Since two quantities diffuse, perhaps ``thermal vapor (gradient) diffusion chamber'' might be more appropriate. When incorporated into a wind tunnel, the term ``dynamic'' is added; this implies momentum diffusion, although the boundary conditions for velocity (v = 0 top, bottom) differ from temperature and vapor density (top high, bottom low). This is to be contrasted with the ``thermal vapor saturation chamber,'' (as defined by Fukuta) used for ice nuclei activation in a filter paper, where the vapor density is assumed constant throughout the chamber. This was first used for ice growth and epitaxy studies some 30 years ago. Its function depends on a minimum growth and negligible vapor depletion at the lower boundary. In principle also, this could be incorporated into a wind tunnel to replenish the moist air, when the term ``dynamic'' would be appropriate.
The conventional mixing chamber, where a small volume of nuclei are rapidly mixed with a much larger volume of supercooled cloud, has been improved to give more uniform ice particles in free growth. Song and Lamb (1994a) established a rotation inflow such that crystals were centrifugally prevented from contacting a line air sink in a vertical cylindrical system, having air input from the bottom, with crystals falling out to be collected below. Arnott, et al (1995) employ a different technique with a continuously formed and seeded supercooled cloud overlying colder saturated air below, into which crystals fall.
A more controlled mixing environment has been developed by Strum and Toor (1992) where a warm saturated jet mixes with a colder environment to produce a supersaturation at the edges of the jet. Drop sizes across the jet are measured by a phase Doppler particle analyzer. Maximum supersaturation of a few 100% can lead to cloud forming nuclei either in the original jet or in the surroundings. The actual concentration of droplets appears related to the time-supersaturation history. This experiment, in principle, can be made to simulate less extreme processes at cloud edges where mixing mechanisms, as individual entities give quite specific results. Concentrations of droplet numbers in dilution result in a relation between mean volume radius and concentration (Telford et al, 1993). This is perhaps where the theoretical insights of non-steady state condensation process may have some relevance (Fukuta, 1992).
An elegant technique has been developed for remotely sensing
the temperature of an acoustically suspended evaporating drop,
diameter 200 
m to 3 mm (Rubel and Seaver, 1994). This uses
fluorescence of trace europium in solution as Eu
.
Measurements show discontinuity in drop evaporation rate and
temperature as it evaporates, related to the formation of a
complete octadecanol (a low vapor pressure alcohol) surface
layer. Effects are consistent with modelled rate processes of
the wet bulb and show promise for examining the role of trace
organic materials on the behavior of haze droplets with changing
humidity.