Most realistic, oh-lattice surface simulations are done canonically co
nserving particles. For some applications, however, such as studying t
he thermal behavior of rare gas solid surfaces, these constitute bad w
orking conditions. Surface layer occupancies are believed to change wi
th temperature, particularly at preroughening, and naturally call for
a grand canonical approach, where particle number is controlled by a c
hemical potential. We report preliminary results of novel realistic gr
and canonical Monte Carlo simulations of the Lennard-Jones fcc(111) su
rface, believed to represent a quantitative model of Ar (111). The res
ults are successful and highly informative for temperatures up to roug
hly 0.8 T-m, where clear precursor signals of preroughening are found.
At higher temperatures, convergence to equilibrium is hampered by lar
ge fluctuations. (C) 1998 Elsevier Science B.V. All rights reserved.