[Wien] Freestanding Pb slab calculation

[Peter Blaha]

Have you looked how the generated structure looks like ? (xcrysden)

It is not correct.

I'm not the expert with structgen and thus cannot tell you, where the error is.

In any case, an fcc-111 surface is very easy (although one cannot use
directly the supercell program).

Have a look into example_struct_files/Gold_23l.struct
This is an fcc 111 structure. All you need to do is to adjust the
lattice parameters and eventually delete some of the atoms (if you want just
6 layers instead of 23).

Remember the stacking: A-B-C .... !

PS. In general it is NOT a good idea to neglect warnings/errors/suggestions
from nn or sgroup, unless "you know why you are doing it".

A 6-layer slab has inversion symmetry and only 6 atoms/cell !
It runs in a couple of minutes, not 19 h.





Am 07.09.2010 16:00, schrieb Bartosz Slomski:
> Dear users,
>
> I am trying to calculate the band structure of a free standing Pb slab
> of 6 ML thickness and 12 ML of vacuum for different lattice parameter.
> Starting with a =4.95 Å the SCF cylce converges, but my band structure
> along M-G-M looks strange, thats why I need help in order to fix the
> problem.
>
> Let me brievely review the steps I have done so far to obtain a band
> structure.
>
> 1.) Generation of slab geometry
> a) Pb unit cell, with lattice a=4.95Å, lattice type:F
> b) slab generation using octave
> i) s=loadstruct("Pb495.struct");
> ii) s1 = makeconventional(s);
> iii) s2 = makesurface(s1, [1 1 1],1,34,67) (film thickness = 34 bohr
> = 6 ML, vacuum thickness = 68 bohr = 12 ML)
> iv) savestruct(s2,"Pb_slab.struct","P1");
> After copying the Pb_slab.struct to Pb495.struct "Structgen" shows a
> spacegroup: 1P1_ , 22 inequivalent atoms, 1 symmetry operation
>
> 2.) Initialization of the struct
> i) nn-bondlength factor 2 (default)-->here I get a warning: WARNING:
> Mult not equal. PLEASE CHECK outputnn-file
> ii) I ignored the warning and did not used the new struct_nn file
> iii) After x sgroup (0.001u 0.001s 0:00.00 0.0% 0+0k 0+0io 0pf+0w)
> I get the warning:
> warning: !!! Number of inequivalent atoms has changed. !!! Old
> value= 22 New value= 12 warning: !!!
> Bravais lattice has changed. sgroup found: 156 (P 3 m 1) Note that
> shift vectors for this space group are defined
> which I also ignored
> iv) x symmetry gives: alpha(3) .lt. 89.8; reset to 90.1 0.003u 0.002s
> 0:00.00 0.0% 0+0k 0+0io 0pf+0w
> So far I did not accept a new struct file
> v) x lstart gives with -6.0 Ry a core electron leak of 0.010, and 7.0
> also, but at 8.0 Ry I got no warning.
> the case.inst file
> vi) for the first SCF run I lowered the R*kmax value to 5.0 instead of
> 7.0 and choosed TEMP as Fermi method with eval = 0.005 (as proposed by
> robert)
> vii) in x kgen I set the grid to 4x4x1 and got the message:
> 1 symmetry operations without inversion
> inversion added (non-spinpolarized non-so calculation)
> viii) the x dstart is always automatically marked as "interactively
> complex selected" although it was said that inversion was added.
> Anyway I started x dstart -c and got: 14.215u 3.626s 0:18.03 98.8%
> 0+0k 0+0io 0pf+0w
>
> ix) after choosing "no-spin polarization" I start a SCF cycle with
> following criteria:
> Energy = 0.0001 Ry and Charge = 0.001 e
>
> After 19h of calculation and 30 cycles the convergency is reached to
> Energy = 0.00015 and charge -0.0009, I save the lapw using save_lapw.
>
> In files/error files I got a message: Error in LAPW1, but only there,
> because the dayfile says nothing about errors:
>
>  > stop
>
> ec cc and fc_conv 1 1 1
> :CHARGE convergence: 1 0.001 -.0009186
> :ENERGY convergence: 1 0.0001 .0000157900000000
>  > mixer (14:16:04) 3.840u 1.244s 0:11.53 44.0% 0+0k 0+131960io
> 0pf+0w
>  > lcore (14:15:53) 0.672u 0.204s 0:02.08 41.8% 0+0k 0+26048io
> 0pf+0w
>  > lapw2 -c (14:14:06) 97.114u 7.196s 1:47.02 97.4% 0+0k
> 0+135112io 0pf+0w
>  > lapw1 -c (13:40:11) 2016.750u 6.508s 33:55.25 99.4% 0+0k
> 0+1144152io 0pf+0w
>  > lapw0 (13:37:00) 184.023u 1.468s 3:10.82 97.2% 0+0k
> 0+85080io 0pf+0w
>
>
> 3) Then I start to calculate the bandstructure creating a cut along
> M-G-M with 40 k-points: x lapw1 -band -c
>
>
> I have almost no experience in performing DFT, except some which I got
> in the workshop in Nantes. But there everything was fine :) Now I run
> into several problems, as you can see.
>
> Perhabs an experience user will see immediately that soemthing bad happens.
>
> Because I still run lapw1 -band -c I will provide the calculated band
> structure later (if somebody is interested). For now I need a red line,
> what files I should check or any other hint.
>
> Thanks and best regards,
>
> Bartosz
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------