Hallo Peter, you are right, the pseudo gaps should vanish after the change of symmetry, I was just wondering why the transition seemed to be not continouus and did not think too much about the k-points (I am on the road and can't use our cluster to check 100^3 or more)
Maybe Pablo can test with smaller changes and much more k-points how the transition and the band splittings develop with the parameters. PS.: I just do not remember that I ever was asked to shift the atom positions after x symmetry. For test reasons I tried several times calculations using P1, in particular for hexagonal structures with only few atoms, or because I forgot something when starting with a cif file in P1. PSS: the k-meshers were identical as I used them long time ago, where I found a solution of the optimisation close to z=1/4 after I used "wrong" start structural parameters in a calculation and finished it more or less for fun to see the energy but not DOS, etc.. Ciao Gerhard DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy: "I think the problem, to be quite honest with you, is that you have never actually known what the question is." ==================================== Dr. Gerhard H. Fecher Institut of Physics Johannes Gutenberg - University 55099 Mainz ________________________________________ Von: Wien [wien-boun...@zeus.theochem.tuwien.ac.at] im Auftrag von Peter Blaha [peter.bl...@tuwien.ac.at] Gesendet: Freitag, 8. Juli 2022 13:07 An: wien@zeus.theochem.tuwien.ac.at Betreff: Re: [Wien] Bi almost cubic Plot the bands and you will "understand" the minimum at EF in the 2-atom structure when you consider how the tetrahedron method works (i.e. by connecting bands according to energy (and not character). Due to the finite k-mesh you get a lot of "avoided" crossings and a pseudogap in the DOS. Try a mesh of 100**3 or even 200**3 k-points for the DOS (probably you still get some artefacts at EF, but they should be "smaller". It is a nice example of Fermi surface nesting and a resulting Peierls distortion. PS: I don't think anything has changed in sgroup. It finds the identical SG for both structures, but with c/2. And in addition, it shifts the origin, so that the Bi atom sits at the origin. PPS: The 2 k-meshes should NOT be identical for the 1 and 2 atom cells, but as close as possible be of similar k-point density. Am 08.07.2022 um 12:22 schrieb Fecher, Gerhard: > Hallo Peter, > > b) I did that for all tested structures P1 and the one after sgroup/symmetry > > the Fermi energy is slightly different (0.4092709547 or 0.4104773792) if one > compares the different structures, > and, if one plots the density of states as Pablo did they are also clearly > different > In the original structure with 2 atoms in the cell, one has a very clear > minimum at the Fermi energy, > in the reduced structure with only one atom and c/2, one has a high density > of states at the Fermi energy. > I performed both calculations with 25x25x25 initial k-mesh and otherwise also > identical parameters > > For 1 atom c/2 one has 2 half filled bands crossing the Fermi energy==> > clearly metallic > For 2 atoms c, the bands are nearly filled up to Ef or empty (within 0.01 > electrons) at the used parameters ==> clearly semimetallic > > I do not remember that sgroup asked to change the structure (warning: !!! > Unit cell has been reduced. sgroup found: 166 (R -3 m) ) > in the possibly very old version that I used in the past for my calculations > with x=1/4 > > Ciao > Gerhard > > DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy: > "I think the problem, to be quite honest with you, > is that you have never actually known what the question is." > > ==================================== > Dr. Gerhard H. Fecher > Institut of Physics > Johannes Gutenberg - University > 55099 Mainz > ________________________________________ > Von: Wien [wien-boun...@zeus.theochem.tuwien.ac.at] im Auftrag von Peter > Blaha [peter.bl...@tuwien.ac.at] > Gesendet: Freitag, 8. Juli 2022 11:50 > An: wien@zeus.theochem.tuwien.ac.at > Betreff: Re: [Wien] Bi almost cubic > > This is the effect of: > > > a) a complicated "folding" of the BZ together with a shift of origin. > > b) Plot the band structures (use xcrysden in the hexagonal setting to get > "identical k-mehes". > > c) you will see that some bands at some k-points agree, others do not and > they are very complicated to trace by backfolding. > > d) In any case it should become obvious what the DOS at EF looks so > different. It is most likely a k-mesh problem (you should use enormous > unshifted meshes), but still, the Tetrahedron method has no problems for the > DOS at EF in the small one-atom cell since all bands go through EF in a > straight line and you get a metal. > > For the doubled cell, there are many "pseudo gaps" and the Tetrahedron method > will make a completely different interpolation for the bands and if your > k-mesh is not VERY dense, give you a semimetal (or "gap"). > > > Otherwise, the results are "identical" as they should, but you have to be > careful with the interpretation. > > > When A is changed to 0.25; > ---------------- > R LATTICE,NONEQUIV.ATOMS: 1 166_R-3m > MODE OF CALC=RELA unit=ang > 8.591340 8.591340 22.415740 90.000000 90.000000120.000000 > ATOM 1: X=0.25000000 Y=0.25000000 Z=0.25000000 > ATOM 1:X= 0.75000000 Y=0.75000000 Z=0.75000000 > Bi NPT= 781 R0=0.00000500 RMT= 2.5000 Z: 83.000 > ------------------- > sgroup gives this warning; > > warning: !!! Unit cell has been reduced. > sgroup found: 166 (R -3 m) > > and the cell is reduced to; > -------------------- > R LATTICE,NONEQUIV.ATOMS: 1 166 R-3m > MODE OF CALC=RELA unit=ang > 8.591340 8.591340 11.207870 90.000000 90.000000120.000000 > ATOM 1: X=0.00000000 Y=0.00000000 Z=0.00000000 > Bi1 NPT= 781 R0=0.00000500 RMT= 2.5000 Z: 83.0 > --------------------- > which is "semicubic" with an angle; > > Angle is 87.539 deg > and only one Bi atom in the cell, now in the corners > > And in this reduced Bi structure the "gap" at Ef in DOS disappears. > > What I see is that with A=0.249 (0.25-0.001) and A=0.2499 (0.25-0.0001) > the DOS have a "gap" and they are quite symilar, but with the addition of the > 0.0001 (A=0.25) the cel is reduced and the "gap" disappears. > > I hope that this answers your questions and becomes clear what I am trying to > show. > Saludos > > Pablo > ________________________________ > > > > > _______________________________________________ > Wien mailing list > Wien@zeus.theochem.tuwien.ac.at<mailto:Wien@zeus.theochem.tuwien.ac.at> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien > SEARCH the MAILING-LIST at: > http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html > > > -- > ----------------------------------------------------------------------- > Peter Blaha, Inst. f. Materials Chemistry, TU Vienna, A-1060 Vienna > Phone: +43-158801165300 > Email: peter.bl...@tuwien.ac.at<mailto:peter.bl...@tuwien.ac.at> > WWW: http://www.imc.tuwien.ac.at WIEN2k: http://www.wien2k.at > ------------------------------------------------------------------------- > > _______________________________________________ > Wien mailing list > Wien@zeus.theochem.tuwien.ac.at > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien > SEARCH the MAILING-LIST at: > http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html -- ----------------------------------------------------------------------- Peter Blaha, Inst. f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-158801165300 Email: peter.bl...@tuwien.ac.at WWW: http://www.imc.tuwien.ac.at WIEN2k: http://www.wien2k.at ------------------------------------------------------------------------- _______________________________________________ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html _______________________________________________ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html