How to calculate internal energy of an irregular structure

Message

fanghz · #1 Post by **fanghz** » Thu Oct 16, 2008 9:41 am

Dear Admin,

If I want to calculate internal energy of an irregular polyhedron (such as defected Icosahedron or Triangular prism), how can I construct the three cell vectors in the POSCAR file ? Can I use a cubic supercell and then relax the volume to accommodate the structure?

Many thanks for your answers !
<span class='smallblacktext'>[ Edited ]</span>

#2 Post by **admin** » Thu Oct 16, 2008 10:21 am

please note that vasp is based on translational invariance and check whether your structures can be space-filling, if you intend to study a bulk system.
--) For bulk structures, The starting geometry including the choice of the Bravais matrix should be reasonable, I cannot imagine how you accomplish to put e.g. a triangular prism into a cubic box and expect equally reasonable forces and stress tensors on all atoms from such an input structure.
For free clusters, there is no problem at all, please choose a large cubic box and a reasonable input geometry for the cluster.

fanghz · #3 Post by **fanghz** » Thu Oct 16, 2008 10:52 am

Many thanks! Indeed, I'm interested in the free clusters, not the bulk structure.
Do you mean that if I choose I a cubic box large enough, the box size will have no influence on the internal energy and I don't have to relax it ?

#4 Post by **admin** » Mon Oct 20, 2008 11:01 am

yes, you are correct. please simply choose a cubic box with sufficient vacuum on each side, i.e. neighboring clusters in adjacent unit cells (periodic boundary conditions!) should be separated by at least 8-10 Ã…. If the cluster has a dipole (or multipole) moment, pelase don't forget to use dipole corections (IDIPOL=4; LDIPOL=.True.)
Do NOT relax the box size and shape itself, just the coordinates of the atoms in the cluster (i.e. ISIF=0, 1 or 2)
remind that you need an 1x1x1 k-mesh only for a free cluster!

fanghz · #5 Post by **fanghz** » Mon Jan 12, 2009 1:54 am

Dear Admin,

I have checked the results and found the intel energy is related with the box size (the box size has been set large enough). I don't know which energy I should use. I want to calculate the energy of an icosahedral clusters and compare it with other type clusters. Please help me to solve this problem.

Many thanks!