Studies on the freezing of pure liquids III. Homogeneous nucleation in molten alkali halides

Abstract

Current methods of interpreting critical supercooling are discussed. In the method of Turnbull it is assumed that the experimental observation time matches the reciprocal of the steady-state nucleation frequency. This assumption is shown to be incompatible, for certain types of experiment, with the nucleation time lag predicted by the theory of part II. An alternative criterion of freezing which avoids this difficulty is that under the conditions of a threshold experiment the observation time is comparable with the nucleation time lag. This criterion may be used in conjunction with the theory of part II to interpret critical supercoolings without making separate assumptions about nucleation frequencies. Threshold nucleation rates, sizes of crystal nuclei, and liquid-solid interfacial free energies for alkali halides are calculated from the supercooling data of part I. Data for nuclei are examined in relation to the properties of the liquid and solid compounds in bulk form. Certain outstanding problems in the interpretation of results obtained in this and other studies are discussed with respect to present nucleation theory.

The crystal-fluid interfacial free energy and nucleation rate of NaCl from different simulation methods

“…The agreement with both the simulation of Ref. 32 and the experiment [34][35][36] is confirmed by the fit. Moreover, the fit works well even at very deep supercoolings, where we have estimated the rate by means of brute force simulations (black squares in Fig.…”

“…The seeding method also allows estimating free energy barriers and nucleation rates for several temperatures. Using classical nucleation theory, we obtain a fit for our nucleation rate data that are consistent with experiments [34][35][36] and that extrapolate well to high supercoolings, where we have evaluated the rate with brute force simulations.…”

“…Contrary to γ cf , the nucleation rate can be directly measured experimentally. [34][35][36] By means of our seeding simulations, CNT, and calculations of the attachment rate of particles to the critical cluster, we estimate the nucleation rate and find a good agreement between numerical simulations and experiments.…”

“…The fit shows a good agreement with experimental data. [34][35][36] We have also used CNT to obtain a nucleation rate curve for US and MI. Since γ cf does not depend on temperature, one can take the average γ cf calculated at coexistence with these techniques and use CNT as we did to fit the seeding data to obtain the cyan curve shown in Fig.…”

“…In Fig. 11(b), we also include the only available experimental data for the nucleation of NaCl from its melt [34][35][36] (blue triangle). Remarkably, the experimental data are fully consistent with the predictions of the model.…”