- Email: [email protected]
Theories of Crystal Surface Melting and Non-Melting
99
THEORIES OF CRYSTAL SURFACE MELTING AND NON-MELTING 1 E. Tosa tt 1·a ,b , A• Trayanova,c , A. C. LeVl.a,d , F. Erco 1essl.a ,an d P. Carneva l,e
a International School for Advanced Studies, 1-34100 b International Centre for Theoretical Physics, P.O. Italy c Institute of Physical Chemistry, Bulgarian Academy Bulgaria d Department of Physics, University of Genova, Italy e IEM Rome Scientific Center and ECSEC, v. Giorgione
Trieste, Italy Box 586, 1-34100 Trieste, of Sciences, Sofia, 159, 00100 Rome, Italy
When the temperature of a crystal is raised to approach its bulk melting value Tr1, some low-index crystal surfaces break up and develop a thin quasiliquid layer of increasing thickness (surface melting), whilst others do not (surface non-melting). Although empiricially very well known, this phenomenon has proved surprisingly elusive theoretically, and a satisfactory description has yet to be found. We have developed theoretical approaches to crystal surface melting at three different levels of increasing microscopic detail. At the first level, we have a theory of the Ginzburg-Landau, or Cahn-Hilliard type. This theory is based on two order parameters; density and "crystallinity". Many features, including occurrence of non-melting versus melting, facedependence, and the effect of long-range forces are easy to envisage in this picture 1 At the next microscopic level, we have a statistical-mechanical description of the solid-vapor interface, adapted on a lattice. Using mean-field and free-volume approximations, the appearance and growth of the quasi-liquid layer, and the corresponding resolution of the surface force energy were followed in detail, particularly for the Lennard-Jones (100) and (110) faces 2. Lastly, molecular dynamics (r1D) computer simulations were applied to study the problem. While the large temperature fluctuations due to limited size are usually fatal to studies of the quasi-liquid in cases of melting, r1D appears to be quite suitable for studying cases of non-melting surfaces. By way of example, we conducted a detailed study of Au(III), both reconstructed and unreconstructed. The former remains fully stable up to and above TM. The
100
latter presents a case of "blocked" surface melting.
Here, the first two
outermost layers break up at about 100 K below T but this seed of liquid M, layer does not further propagate into the crystal as T is approached. M 1. A. C. Levi and E. Tosatti, to be publ ished. 2. ,n.. Trayanov and E. Tosatti, to be published. 3. F. Ercolessi, P. Carnevali and E. Tosatti, Phys , Rev. B (to aooear).
THEORIES OF CRYSTAL SURFACE MELTING AND NON-MELTING 1 E. Tosa tt 1·a ,b , A• Trayanova,c , A. C. LeVl.a,d , F. Erco 1essl.a ,an d P. Carneva l,e
a International School for Advanced Studies, 1-34100 b International Centre for Theoretical Physics, P.O. Italy c Institute of Physical Chemistry, Bulgarian Academy Bulgaria d Department of Physics, University of Genova, Italy e IEM Rome Scientific Center and ECSEC, v. Giorgione
Trieste, Italy Box 586, 1-34100 Trieste, of Sciences, Sofia, 159, 00100 Rome, Italy
When the temperature of a crystal is raised to approach its bulk melting value Tr1, some low-index crystal surfaces break up and develop a thin quasiliquid layer of increasing thickness (surface melting), whilst others do not (surface non-melting). Although empiricially very well known, this phenomenon has proved surprisingly elusive theoretically, and a satisfactory description has yet to be found. We have developed theoretical approaches to crystal surface melting at three different levels of increasing microscopic detail. At the first level, we have a theory of the Ginzburg-Landau, or Cahn-Hilliard type. This theory is based on two order parameters; density and "crystallinity". Many features, including occurrence of non-melting versus melting, facedependence, and the effect of long-range forces are easy to envisage in this picture 1 At the next microscopic level, we have a statistical-mechanical description of the solid-vapor interface, adapted on a lattice. Using mean-field and free-volume approximations, the appearance and growth of the quasi-liquid layer, and the corresponding resolution of the surface force energy were followed in detail, particularly for the Lennard-Jones (100) and (110) faces 2. Lastly, molecular dynamics (r1D) computer simulations were applied to study the problem. While the large temperature fluctuations due to limited size are usually fatal to studies of the quasi-liquid in cases of melting, r1D appears to be quite suitable for studying cases of non-melting surfaces. By way of example, we conducted a detailed study of Au(III), both reconstructed and unreconstructed. The former remains fully stable up to and above TM. The
100
latter presents a case of "blocked" surface melting.
Here, the first two
outermost layers break up at about 100 K below T but this seed of liquid M, layer does not further propagate into the crystal as T is approached. M 1. A. C. Levi and E. Tosatti, to be publ ished. 2. ,n.. Trayanov and E. Tosatti, to be published. 3. F. Ercolessi, P. Carnevali and E. Tosatti, Phys , Rev. B (to aooear).