The nucleation of new phases is a crucial part of processes of interest to chemistry, physics, geology, pharmacy, meteorogy, and materials science and engineering, yet nucleation is among the few areas where the discrepancy between theoretical predictions and experimental determinations reaches many orders of magnitude. This problem is even more severe for complex nucleating systems, such as nucleation of crystals from solution, where even the reproducibility of the determinations is often at issue. I will discuss recent findings, suggesting that the nucleation of crystals of many colloid and protein, and molecular and ionic small-molecule materials in solution occurs via a two-step mechanism: (1) the solute molecules assemble into disordered and fluid droplets of mesoscopic sizes, and (2) ordered crystalline nuclei form within these dense liquid droplets. I will summarize the experimental findings supporting this mechanism in various systems. I will then discuss a recent phenomenological theory which provides a rate law for the kinetics of nucleation via the two-step mechanism, and tests of the theory predictions using experimental data. I will present recent independent evidence of the existence of the fluid droplets which serve as precursors to the nuclei. I will discuss how this mechanism explains the role of foreign substrates in acceleration of the nucleation rate and the selection of the crystal polymorph emerging from the supersaturated solution. The low volume fraction of the dense liquid precursor explains the discrepancies between nucleation theory and experiment, while its variability underlies inconsistencies of the experimental determinations of the nucleation rate.

Proc. Natl. Acad. Sci. USA 2000, 97, 6277; Crystal Growth and Design 2004, 4, 671-685; J. Chem. Phys. 2005, 122, 174905; J. Amer. Chem. Soc. 2005, J. Chem. Phys. 2005, 123, 014904; 127, 3433-3438; Biophys. J. 2007, 92, 267; J. Phys. Chem. B 2007, 111, 3106