ELECTRON-DIFFRACTION STUDIES OF THE KINETICS OF PHASE-CHANGES IN CLUSTERS .4. FREEZING OF AMMONIA

Clusters of liquid ammonia produced by the condensation of vapor in supersonic flow were monitored by electron diffraction, in flight, while they froze. Freezing began when clusters had cooled to 120 K by evaporation and froze to single crystals of the normal cubic phase, space group P2(1)3. Even though initially undercooled by 75 deg, they would have warmed to the melting point well before the completion of freezing if the process in the clusters had been strictly adiabatic. Yet they froze completely, and their temperature was 128 K within a dozen microseconds after nucleation. The kinetics of evaporative cooling of the liquid and of the freezing clusters are treated, and the amount of ablation suffered during solidification is calculated. The rate of formation of solid critical nuclei in the undercooled liquid ranged from 1.2 x 10(30) to 1.5 X 10(30) m(-3) s(-1) in different stream tubes of the jet. The interfacial free energy of the liquid-solid boundary was derived from the nucleation rate by applying the Turnbull-Fisher nucleation theory. Its value, sigma(sl) = 23 mJ/m(2), is consistent with Turnbull's empirical relation correlating sigma(sl) with the heat of fusion. Inferences of the temperature dependence of the nucleation rate make it unlikely that the rate of freezing of ammonia increases sufficiently at deep undercooling to be accessible to simulation by molecular dynamics techniques at currently available computational speeds. Several new aspects of cluster formation and behavior were seen in this research. In contrast to all previous observations of phase changes by this technique, where clusters were generated from dilute vapors seeded into a carrier gas, the ammonia microdrops were condensed from the neat vapor. The present transformation was also the first in which neither of the phases was plastically crystalline, verifying that unusual molecular mobility in the solid phase is not a sine qua non for transitions to occur on a microsecond time scale. Cluster diameters began at 100 Angstrom in the center line of the jet but fell off by 25% toward the periphery. Skimming at different angles enabled a delicate selection of particle size. Freezing of the smaller clusters began earlier in the flow and at slightly higher nucleation rates, for reasons that are discussed.