Crystal Nucleation from a Hard-Sphere Liquid—The Discrepancy Between Experiments and Simulations

Understanding crystal nucleation from the melt is central to the study of glass formation. The hard-sphere system has been particularly useful here, because it allows direct experimental observation of the nucleation events in colloidal suspensions. On the theoretical side, there are detailed determinations of the equations of state of the two phases, calculations of the interfacial energies, as well direct simulations of the nucleation events. At high liquid densities, there is fair agreement between the nucleation frequencies measured in experiments and simulations. At lower liquid densities, however, abundant nucleation is observed where simulations predicts there should be none.<br/><br/>This talk will be a critical review of the possible origins of this discrepancy: (i) Establishing homogeneous nucleation conditions by suppressing heterogeneous nucleation at the sample cell walls; (ii) Determining the (small) degree of softness and the size dispersion of the polymeric spheres; (iii) Determining the effects of (ii) on the equations of state and the phase diagram; (iv) Evaluating and comparing the different crystal identification algorithms, especially for the small crystal sizes; and (v) Evaluating the methods for extracting crystal nucleation rates.<br/><br/>Central to this study is the concept of the crystal-liquid interfacial energy. Some experimental values can be extracted from the equilibrium size distribution of crystalline fluctuations in the liquid. Under non-equilibrium conditions, we determine the critical nucleus size, at which a crystal is equally likely to grow as to shrink. Classical nucleation theory, combined with the appropriate equations of state, then also gives values for the interfacial energy. These will be compared with earlier determinations on macroscopic interfaces.