Potential
Posted: Sun Dec 02, 2007 6:15 am
Dear users,
I'm trying to reproduce the band offset calculations of Van de Walle and Martin (PRB "34", 5621 (1986)) for the pseudomorphic Si-Ge interface . The supercell consists of two 8-atom crystallographic cells of Si and Ge (16 atoms) with the interface normal to the z-direction. I'm using CA-LDA USPPs with a 400 eV cutoff, 40 irreducible k-points (8x8x4 M-P grid) -- a bit of an overkill. I re-compiled the code with the option in main.F to generate the total potential (including the XC part), and produced the LOCPOT shown in the following link:
http://129.219.38.204/SiGe/SiGe-interface-sup.jpg
To ensure the proper comparison I adopted the atomic positions and cell geometry used by van de Walle and Martin (see citation given above). I confirmed that their ansatz is very close to equilibrium, resulting in residual forces < 0.01 eV/Angs.
I'm perplexed by several things. First, the potential step (~ 2.7 eV) is about three times the magnitude reported by van de Walle and Martin (~ 0.85 eV). Furthermore, the "polarity" of the VASP total potential on the Si side of the slab appears reversed with respect to that on the Ge side (the figure in link). How is the non-locality of the PS contributions to LOCPOT defined? Is there a way to plot only the l=1 component?
Any/all comments would be greatly appreciated!
Andrew Chizmeshya
I'm trying to reproduce the band offset calculations of Van de Walle and Martin (PRB "34", 5621 (1986)) for the pseudomorphic Si-Ge interface . The supercell consists of two 8-atom crystallographic cells of Si and Ge (16 atoms) with the interface normal to the z-direction. I'm using CA-LDA USPPs with a 400 eV cutoff, 40 irreducible k-points (8x8x4 M-P grid) -- a bit of an overkill. I re-compiled the code with the option in main.F to generate the total potential (including the XC part), and produced the LOCPOT shown in the following link:
http://129.219.38.204/SiGe/SiGe-interface-sup.jpg
To ensure the proper comparison I adopted the atomic positions and cell geometry used by van de Walle and Martin (see citation given above). I confirmed that their ansatz is very close to equilibrium, resulting in residual forces < 0.01 eV/Angs.
I'm perplexed by several things. First, the potential step (~ 2.7 eV) is about three times the magnitude reported by van de Walle and Martin (~ 0.85 eV). Furthermore, the "polarity" of the VASP total potential on the Si side of the slab appears reversed with respect to that on the Ge side (the figure in link). How is the non-locality of the PS contributions to LOCPOT defined? Is there a way to plot only the l=1 component?
Any/all comments would be greatly appreciated!
Andrew Chizmeshya