HOMO-LUMO gap vs band gap from PDOS

[Arup Sarkar]

Hello all CP2K users,

I have a very basic query related to HOMO-LUMO gap calculation in CP2K (9.1). I am trying to calculate the HOMO-LUMO gap and also partial density-of-state (PDOS) of a nanocluster. First I optimized the geometry using periodic boundary conditions by putting my large nanocluster inside a large box so that no interaction occurs with its periodic image. Then, I am calculating the ENERGY with HOMO-LUMO settings using the box size and POISSON solver XYZ or Periodic. 

I am getting the same HOMO-LUMO gap even I use non-periodic boundary conditions with POISSON solver wavelet with the same input settings. Now, when I am trying to calculate the energy gap or more precisely the band gap from PDOS calculation (of course using periodic boundary conditions at gamma point) with smearing (and also without smearing) and with OT (and also without OT),  the band gap is less than the HOMO-LUMO gap from the previous calculations. 

Now it is known that, band gap from band or DOS calculation is usually considered smaller than the cluster HOMO-LUMO gap but this is true if my HOMO-LUMO calculation is performed in non-periodic mode. Why the periodic HOMO-LUMO and non-periodic HOMO-LUMO gap matches in CP2K ? Is it converting my calculations automatically to a non-periodic calculation even I mention POISSON solver periodic during the HOMO-LUMO energy calculations ?

  With regards,

Dr. Arup Sarkar

Postdoctoral Researcher

Max Planck Institute for Polymer Research