Thanks! Now I know where to go.
El 11-01-2014 15:21, N H escribió: > Hi Christian ... > > Not running surface calculations since the end of my PhD ... but I have a guess. > > If i got it correctly, you are freenzing two layer on one extreme of your slab. If that is true you have a surface farly out of the equilibrium. That wehere your negative freq. are probably comming. > > It also breaks the symetry of your system and might hamper electronic convergence of surface states. So the system just goes to any "metallic" local minimum where it can spread the unrelaxed dangling bonds around. > > In your place I would make sure that I have two symmetric surfaces in your slab and let only inner atoms frozen. > > ( If your frozen layers are already on the middle of your slab, just forget what I said). > > Cheers > On Jan 11, 2014 9:26 PM, "CHRISTIAN VIALES MONTERO" <[email protected]> wrote: > >> Dear SIESTA users, while working with a Silicon surface (7 layers, 2 constrained to bulk geometries (according to previous calculations), Si(100)-(2x1) reconstruction, 2 atoms per layer, with 20.9703 Ang of vacuum and 8.1350 Ang of slab height, BLYP/DZP with BLYP/TM2 PP (tested for transferability and similarity), in a 7×12×1 Monkhorst Pack kGrid and a 3×6×1 supercell, with DM.Tolerance=1.d-5, DM.Energy.Tolerance=1.d-7 eV and MD.MaxForceTol=0.00045 eV/Ang (to coincide with previous B3LYP5/6-311G* and PM7 calculations on non-periodic equivalent systems) and set to PAO.EnergyShift=70 meV and MeshCutoff=200 Ry (as indicated by a total-energy-per-atom convergence analysis)) I ran into two problems: 1) I found 4 vibrational bands between 0 and -200 cm-1; and 2) there is no gap between the 28th and 29th (14 atoms*4 electrons/2 electrons per band=there should be 28 valence bands) electronic bands (or nearby, checked also by plotting the total DOS). >> >> Since the force tolerance is that small, might the imaginary frequecies originate from constraining the lower two layers? >> >> In that case, what would you recommend to do next? >> >> The fact that the lower layer is not reconstructed could explain the lack of a surface band gap? >> >> By capping the lower layer with hydrogen atoms (and thereby transforming the system to some sort of nanoribbon) or working with a symmetric slab (probably a 14 layer system (with the central 2 fixed?)) could the problem be solved? >> >> Thanks to all in advance.
