The failure of the Classical Nucleation Theory at low temperatures resolved

A relevant observation about crystal nucleation kinetics in glass-forming substances has been a matter of intense debate for several decades. The Classical Nucleation Theory (CNT) allegedly fails to describe the temperature dependence of the homogeneous crystal nucleation rates below the temperature of maximum nucleation rate. This failure was reported for several glass-forming substances and is known as nucleation "breakdown". Some reasonable explanations for this apparent break have been advanced in the literature, however, the simplest hypothesis has never been tested: that this break is a byproduct of nucleation datasets that have not reached the steady-state regime. In this work, we tested this hypothesis by thoroughly analyzing new and published nucleation data for supercooled Li2Si2O5, BaSi2O5, Na4CaSi3O9, and Na2Ca2Si3O9 liquids, using only datasets for which steady-state conditions (likely) have been reached. For that purpose, we used three restraining conditions: i) Nucleation and diffusion data were measured in the same glass batch, and each batch was individually analyzed; ii) only nucleation rate data that passed a steady-state regime test were used in the analysis; iii) the uncertainty and regression confidence bands were computed and considered. With this strategy, we proved that the alleged nucleation break is indeed an experimental artifact! This result ends a four decade-old dilemma and corroborates the use of CNT for analyses of crystal nucleation rates.