Surface melting away from equilibrium

Anisotropy in surface-initiated melting of the triclinic molecular crystal 1,3,5-triamino-2,4,6-trinitrobenzene: A molecular dynamics study

Surface-initiated melting of 1,3,5-triamino-2,4,6-trinitrobenzene (TATB), a triclinic molecular crystal, was investigated using molecular dynamics simulations. Simulations were performed for the three principal crystallographic planes exposed to vacuum, with the normal vectors to the planes given by b × c, c × a, and a × b (where a, b, and c define the edge vectors of the unit cell), denoted as (100), (010), and (001), respectively. The best estimate of the normal melting temperature for TATB is 851 ± 5 K. The nature and extent of disordering of the crystal-vacuum interface depend on the exposed crystallographic face, with the (001) face exhibiting incomplete melting and superheating. This is attributed to the anisotropy of the inter-molecular hydrogen bonding and the propensity of the crystal to form stacking faults in directions approximately perpendicular to the (100) and (010) faces. For all three crystal orientations, formation of molecular vacancies in the lattice at the crystal-vacuum (or crystal-quasi-liquid layer) interface precedes the complete loss of order at the interface.

Ice premelting during differential scanning calorimetry

Premelting at the surface of ice crystals is caused by factors such as temperature, radius of curvature, and solute composition. When polycrystalline ice samples are warmed from well below the equilibrium melting point, surface melting may begin at temperatures as low as -15 degrees C. However, it has been reported (Bronshteyn and Steponkus, 1993. Biophys. J. 65:1853-1865) that when polycrystalline ice was warmed in a differential scanning calorimetry (DSC) pan, melting began at about -50 degrees C, this extreme behavior being attributed to short-range forces. We show that there is no driving force for such premelting, and that for pure water samples in DSC pans curvature effects will cause premelting typically at just a few degrees below the equilibrium melting point. We also show that the rate of warming affects the slope of the DSC baseline and that this might be incorrectly interpreted as an endotherm. The work has consequences for DSC operators who use water as a standard in systems where subfreezing runs are important.