New Results on Ice Growth from the Vapor; Ice multiplication without riming

Ice crystals have been grown from the vapor at substantially uniform and constant temperature and supersaturation, in a closed chamber in a constant temperature bath, with the vapor source a layer of supercooled water solution in the bottom of the chamber, and no applied ventilation. The equilibrium freezing point of the supercooled solution was varied from 0 to about -0.5C. Most of the 96 runs were between -4.8 and -5.5C, with a few between -0.5 and -1.5. Plate growth can occur at all of the temperatures used, and generally starts as rounded discs that grow into hexagonal, faceted plates. Extremely anisotropic growth does occur below water supersaturation. The growth habits can be plates, sheaths or needles, and evidently are critically dependent upon how the crystals are started, but as of this writing no reproducible way of predetermining the habit that forms has been found. Needle growth at -5C occurs in two types: exaggerated c-axis growth at the corner of a sheath, and a faster form: smoothly tapered, sharp needles growing at a slight angle to the c-axis with no detectable basal facet at the tip. Maximum growth velocity of these, measured so far, is 5 microns per second. These two types seem distinct at this point, but there may be a continuum of growth habits in between. New crystals fairly commonly nucleate spontaneously at the tips of these fast-growing needles, initiating needle growth there at a new orientation. (This is most surprising). The supercooled puddle controlling the vapor pressure freezes when a crystal grows into it, of course, but it fairly often freezes spontaneously in the growth experiments at -5C with no visible contact, or even close approach of a crystal to it. (The growing crystals are about 2.5 cm above it). Freezing never happens in controls without a growing crystal, and must represent an ice multiplication mechanism, without riming but with no visible cause. When growth down the wall occurs it is very slow (about 0.1 microns per second), is easily detectable, and is not involved. Perhaps the explanation of this involves extremely thin whiskers, but whatever is going on may be the mechanism of the Hallett-Mossop process.