We present the results of a new crystal structure prediction of ROY performed with the GRACE program, including the first wide search for Z′ = 2 structures. In addition, we present calculations of lattice-vibrational contributions to the polymorphs' relative stability. Because of excessive electron delocalization, the polymorphs of ROY constitute a surprisingly challenging system for density functional approximations, raising troubling questions regarding the performance of some of the most popular computational methods. Despite this, lattice dynamics calculations performed very well in reproducing the experimental cyano-group stretching frequencies, allowing us to propose a plausible candidate structure for the RPL polymorph, identified by comparing experimental Raman frequencies with calculated phonon modes.

Here, we report a general strategy for rapidly cultivating single crystals from melt microdroplets within tens of minutes at the microgram scale.

The rate of crystal nucleation has been measured in four glass-forming molecular liquids: D-sorbitol, D-arabitol, D-xylitol, and glycerol. These polyalcohols have similar rates of crystal growth when compared at the same temperature relative to T_g (the glass transition temperature), peaking near 1.4 T_g, while the nucleation rates J are vastly different. In D-sorbitol and D-arabitol, J reaches a maximum of ∼10⁸ m⁻³ s⁻¹ near 1.1 T_g, whereas J < 10⁻² m⁻³ s⁻¹ in D-xylitol and <1 m⁻³ s⁻¹ in glycerol based on no nucleation in large samples after long waits. This confirms the fundamentally different mechanisms for nucleation and growth. Near T_g, both nucleation and growth slow down with a similar temperature dependence, suggesting a similar kinetic barrier for the two processes. This temperature dependence is significantly weaker than that of viscosity η, approximately following η^−0.75. This indicates that viscosity is a poor representative of the kinetic barrier for nucleation, and a better choice is the crystal growth rate. Under the latter assumption, the classical nucleation theory (CNT) describes our data reasonably well, yielding σ = 0.013 J/m² for D-sorbitol and 0.026 J/m² for D-arabitol, where σ is the critical nucleus/liquid interfacial free energy. There is no strong indication that the CNT fails as the length scale for corporative rearrangement exceeds the size of the critical nucleus, as recently suggested for lithium disilicate.