All the models considered at the previous chapter explain the mechanism of melting within the bulk material. These theories consider melting process as a breakdown of the crystal lattice which occurs uniformly throughout the solid at the melting point, i.e. melting is required to be homogeneous. However, there is ample evidences that melting is actually heterogeneous, and that it involves nucleation of the liquid phase at some preferred sites of the solid (the free surface, grain boundaries, large dislocations and disclinations, etc), followed by subsequent growth of the liquid phase. The most likely site from where the solids start to melt is the free surface of the solid. A premelting of the surface offers an elegant solution to an intriguing puzzle of melting, namely why a liquid could be cooled below the freezing point (``supercooling''), but a solid can not be heated above the melting point (``superheating''). On the basis of classical nucleation theory one expects, upon melting and freezing, that these hysteresis effects should occur. The absence of ``superheating'' of solids, is indicative of the general absence of an energy barrier for the nucleation of a melt, yet such a barrier does exist for solidification. Apparently the formation of a liquid surface layer at temperatures below the melting temperature eliminates the need for liquid nucleation at melting transition, thus no metastability effects exist.
Tammann  was the first to point out that surface may play an important rôle in initiating melting. The concept that melting may start at the surface can be inferred from an empirical criterion formulated by Lindemann . Relying on the criterion one may argue that the outermost atomic layer of the crystal should disorder far below the bulk melting point. The loosely bounded surface atoms have a reduced number of neighbors, and therefore have a higher vibrational amplitude than in the bulk. Consequently, at the surface the Lindemann criterion is satisfied at a lower temperature. Surface melting involves a formation of a thin disordered layer at some relatively high temperature. Surface melting considered here has nothing to with the trivial melting caused by temperature gradient, for example, melting of an ice cube floating in a glass of water. In this case the surface melts simply because the outside is hotter than the inside.
While numerous attempts to detect surface-initiated melting phenomena have been made, it is only recently that surface melting has been observed directly on a microscopic level by employing atomically clean, well characterized surfaces. The first direct observations were made using Rutherford backscattering , in conjunction with shadowing and blocking. Since then, other techniques have been employed such as calorimetry, electron, neutron and X-ray diffraction, microscopy, ellipsometry, and helium scattering, and even visual inspection with the naked eye. Most experiments have been carried out near equilibrium. But even when a solid is heated suddenly, melting still tends to be initiated at the surface . Under some conditions it is possible to force the solid to break down internally, but then the melting will tend to begin at internal interfaces such as grain boundaries.
Theoretical aspects of surface melting have been studied by phenomenological Landau theories [27-30], lattice models , and density functional theory . A more detailed review of those methods is given below. Computer simulations play a leading part in the studying of the phenomena in a microscopic level. The story of surface melting is not yet complete. Still to be explored is the evolution of layer structure at the surface region in detail.