[lammps-users] question on periodic boundary condition


I have a question on the periodic boundary conditions. I was under the impression that when using these BCs any size box should behave as a bulk material. One of the criteria used to determine when a crystal has melted is the Lindemann criterion, e.g melting occurs when the RMSD of atoms exceed 10% of the lattice constant. However, I have noticed that when the box is defined smaller than 40 x 40 x 20 nm the RMSD exceeds 10% of the lattice constant a temperatures much lower than the bulk melting point. Why would this happen if the simulated material is a bulk and the periodic boundary conditions make it such that end effects are negligible? I was expecting that one could define a box of any size with periodic boundary conditions and get the same results.

Also, I have some problems with the timestep used for the EAM potential. When I define a free surface the recommended value of 0.001 ps works fine. However, the same timestep for the case with periodic boundaries on all sides produces large oscillations on the RMSD when the temperature increases. I have found that 0.005 ps eliminates these oscillations, but I would like to know why this is the case.


Jaime Sanchez

Hi, Jaime. Your premise is wrong. Finite system size can affect things
much, even with periodic boundary conditions.
Finite size does affect the rms
displacement in a crystal. The reason is that any mode with wavelength
larger than the periodic side length is frozen out. This can affect
things like melting. For some details of the effect of finite size on
melting, see "Effect of Finite System Size on Thermal Fluctuations -
Implications for Melting" by M. O. Robbins, G. S. Grest, and K. Kremer,
Physical Review B volume 42 page 5579, 1990.