PBC in A/B directions still working for molecule between electrodes?

[naohjohn]

I'm calculating the transmission spectrum of Au/Molecule/Au systems.

Since the molecule is large (hundreds of atoms  ), I make a supercell quite “crowded”  in A/B directions to save time.
 
After relaxation in VASP(it's faster in convergence?) , some H atoms in the molecule go across the boundary in A or B direction and VASP make a translation of them to make them within the supercell (still in bond with a C atom and far away from others), just like the molecule cut by the bounday.

ATK gives extremely large transmission in that case.

However, if  I translate these separate H atoms back to form a complete molecule in VNL window, the transmission is reasonable(not sure, no similar publications  ).

So how does this happen? Does the periodic boundary condition still working? Or I need to increase A/B length to prevent this from happening?

Thank you for any advice.

[Anders Blom]

The PBC should take of this, provided you do have PBC in the calculation, and not Multigrid with Dirichlet or Neumann. If you are concerned you can email the structure privately, we keep all customer data confidential.