Polycrystals are often expected to start melting at their free surface or at the interface between grains. These grain boundaries are said to melt when they are invaded by a thick liquid film at the bulk melting temperature Tm. Premelting is sometimes predicted, with liquid-like layers appearing between grains at temperatures below Tm.
We review this topic, and recall a general argument which predicts that, although systems with short-range forces might show grain boundary melting and premelting, grain boundaries can only be wet incompletely by the liquid at $\Tm$ in systems with long-range Van der Waals forces. We describe experiments where the two limits are observed, on colloidal and helium crystals, respectively. |