Dear SIESTA users, I need to calculate Si unreconstructed and unrelaxed surface formation energy very accurately. Previously I used another atomic orbitals based code, AIMPRO and found that it gives big error. The relation I used is
Esurf=(Etot-uN)/2 Esurf - surface energy Etot - total energy of the slab u - energy per one Si atom in Si bulk N - number of atoms in the slab For Si(100) surface in AIMPRO it does about 159 meV/A2, whereas VASP gives 149 meV/A2. (LDA, both calculations converged) In my opinion this is because there is not enough basis functions in vacuum between slabs, so the electron density distribution cannot be approximated correctly and this makes total energy of the slab higher. I did some tests placing "ghost" atoms above Si surface and they indeed lower surface energy considerably. My question is if there is a way to calculate surface formation energy accurately in atomic orbitals based code like SIESTA or AIMPRO? These "ghost" atoms are confusing me, I don't know how to test convergence with respect to these atoms. The cutoff energy in plane waves based code is much more straightforward. Kind regards Ruslan