AIMD atoms slide outside of unit cell with NVT ensemble

[keya.joy]

I am running an AIMD simulation for an ion moving inside bilayer graphene and when I visualize the trajectory in JMOL and VMD, the sheets appear to slide apart even though the simulation is with NVT ensemble. The fractional coordinates in the XDATCAR file are outside of the [0, 1] range at times and I was wondering if anybody experienced this. This phenomenon happens at different temperatures and different supercell sizes.
What does it mean for the coordinates to be outside [0, 1] and could that be an acceptable result?

[ahampel]

Hi keya.joy,

Coordinates outside [0,1] themselves are normal in MD and simply indicate atoms crossing periodic boundaries. VASP will not automatically change the coordinates back to be inside the unit-cell to indicate where the atom originally came from. This can be sometimes be hard to track in VMD or Jmol. I think for VMD for an example there is a plugin tool called pbc that supports wrapping the atoms back into the unit cell.

However, if the actual interlayer distance between the graphene sheets increases during the simulation, then this is a physical effect and not related to the coordinate representation. In that case it is worth checking whether dispersion interactions are included, since graphene bilayers require vdW corrections to remain bound. Maybe you can clarify what you mean by "slide apart"?

Best,
Alex

[keya.joy]

Thank you for the response! I have DFT-D3 corrections on and the interlayer distance doesn't change much, but they slide apart horizontally so I will check out the pcb tool. I attached images that help describing the sliding that I am seeing.

[ahampel]

Hi,

I am not very experienced with MD but I think this could be a visualization problem. Can you try to use pbc wrap -all ( I think this is the option name) . Maybe you can also check if the lattice vectors change during the MD run (I think you said you run NVT with cell fixed, but just to double check)? This could be a hint to whether something is going wrong. Let me know how it looks with wrapping enabled.

Best,
Alex