Breaking of solvent molecules

[helvetico]

Dear all

I am doing MD simulations of metal surfaces with solvents, mainly H2O. I noticed that the ions (atoms) are written back to the unit cell when leaving it and that therefore the H2O molecules are broken when e.g. the H atom is leaving the cell entering it on the other side resulting in wrong bond lengths.

How can I avoid this breaking? When I want to calculate bond frequencies (via VACF) of water on the surface and compare them to bulk water I will get wrong results because the bond increases from e.g. 1A to 1A+(x coordinate of cell).

Secondly but less important: it would be nice when the molecules are not written back to the cell. It would be possible to calculate diffusion coefficients (via MSD) of the solvents and solutes depending on an applied charge. Any way to circumvent this?