The ABINIT Users Mailing List ( CLOSED )

["Lukas Thulin" <lthulin@nanoptek.com>]

Thanks Martin, I'll give that a try.
Luke

-----Original Message-----
From: Martin Stankovski [mailto:martin.stankovski@uclouvain.be] 
Sent: Thursday, September 03, 2009 2:12 AM
To: forum@abinit.org
Subject: RE: [abinit-forum] supercell band structures

Dear Lukas,

I see that you are doing a PAW calculation. I have recently experienced a
similar thing, and it went away when I specified "istwfk *1" in all my
ground-state input files, telling abinit not to use additional symmetry
shortcuts in the wavefunction description for special points. I suspect
there might be a small bug in the symmetry-handling routines for PAW which
stop the cycle from converging to very high precision for some states.

Happy abiniting,
Martin :)

-----Original Message-----
From: Lukas Thulin [mailto:lthulin@nanoptek.com] 
Sent: den 2 september 2009 20:41
To: forum@abinit.org
Subject: RE: [abinit-forum] supercell band structures

Actually, when I look more carefully, all the bands converge to better than
tolwfr given that nbdbuf defaults to two.  For some reason at the Gamma
point one very deep (5th) valance band eigenvalue is giving me that 6.8e-11
value that is preventing the iscf cycle from stopping.  This problem doesn't
occur in the bulk case. 

-----Original Message-----
From: Lukas Thulin [mailto:lthulin@nanoptek.com] 
Sent: Wednesday, September 02, 2009 1:54 PM
To: forum@abinit.org
Subject: RE: [abinit-forum] supercell band structures

Thanks for the response.

-----Original Message-----
From: matthieu verstraete [mailto:matthieu.jean.verstraete@gmail.com] 
Sent: Wednesday, September 02, 2009 11:44 AM
To: forum@abinit.org
Subject: Re: [abinit-forum] supercell band structures

>> 1)      Despite high k-sampling (8,8,1), highly converged (toldfde=1e-9)
>> scf run, high energy cutoffs, etc. I can’t converge the iscf run to
>> tolwfr=1e-12.  Even for nstep=500 the calculation stops at
tolwfr=6.8e-11.
>> Is this ok, or should I expect better convergence?#
are you using nbdbuf to neglect the last few bands in the convergence
criterion? Check the log file for the "res" fields: your bands may all
be converged very well (you can get down to 1.e-20 and more in most
cases, but vacuum is difficult) except the last few. Are you using
iscf 7 or 17?

I see, in general the last two bands are poorly converged and the rest are
on the order of 1e-13.
I'm using iscf 17.

btw toldfe 1.e-9 and 8x8 k-points is not _that_ tight convergence. It
depends what you look at, that's the real criterion for convergence.

Right, I suppose I was basing it on the band plot not changing as I
increased sampling, etc.

>>
>> 2)      Even without any vacuum the simple tetragonal supercell bands
>> calculation doesn’t quite agree with the bulk (body centered tetragonal)
>> bands.  There is some difference especially around the gamma point and
some
>> additional bands (blue in the plot) are found in the supercell case.
In the supercell case you have more atoms per unit cell, and have
folded back the bulk bands into the slab BZ. I trust you have made
sure the path for the band structures is the same in absolute
coordinates: for the usual reduced coordinates, the k-points you plot
would be very different in the 2 unit cells (bulk and slab).

Just to verify, I'm plotting along a line from Gamma to X.  In Cartesian X
is (1/2a,1/2a,0).
For bulk (bct) case, basis vectors are (a,0,0), (0,a,0), (a/2,a/2,c/2)
giving reciprocal vectors (1/a,0,-1/c), (0,1/a,-1/c), and (0,0,2/c), that
makes X (1/2,1/2,1/2).
For supercell (simple tetragonal) with basis vectors (1/a,0,0), (0,1/a,0),
and (0,0,2/xc) giving reciprocal vectors (1/a,0,0), (0,1/a,0), and
(0,0,2/xc) that makes X (1/2,1/2,0).


>>
>> 3)      As I increase the layers of vacuum I get some occupied states
>> (green in plot) within the gap.  I don’t understand why these states
appear,
>
> Surface states, probably. neat. I would also like to here if anyone has
any
> comments about points 1) and 2). Cheers,
correct. You can plot the wf with cut3d to check. Often they are not
strictly localized on the surface, but also hybridize with bulk
states. This is all physical and gives neat experiments too.

Sounds interesting, I guess I wasn't expecting surface states when plotting
bands along a direction perpendicular to the surface.

Matthieu


>
> adam
>
>>
>> and for the work function calculation, should I ignore these for the HOMO
>> energy value.
>>
>>
>>
>> Thanks for any suggestions,
>>
>> Luke Thulin
>>
>> Nanoptek Corp
>>
>>
>>
>> I am using Abinit v5.8.4 with the following input files:
>>
>>
>>
>> # Crystalline anatase titania
>>
>> #Geometry optimization parameters, none for this case.
>> ionmov 0
>>
>> #Use to get a vacuum potential reference
>> prt1dm 1
>>
>> #Definition of the unit cell;
>> acell 3.802 3.802 9.68411/2 angstrom    # a,a,c/2
>>
>> #Orient the vacuum above the (001) plane.
>> rprim   1.0            0.0              0.0
>>         0.0            1.0              0.0
>>         0.0            0.0              6.5
>> chkprim 0   # This input variable allows to use non-primitive unit cells.
>>
>> #Definition of the atom types
>> ntypat 2          # There are two types of atoms: Ti and O
>> znucl 22 8
>>
>> #Definition of the atoms
>> natom 18           # There are 18 atoms in our supercell.
>> typat 1 1 2 2 2 2 1 1 2 2 2 2 1 1 2 2 2 2
>> xangst
>>    0.000  0.000  0.0000000000E+00   #layer 1
>>    0.000  1.901  2.4210275000E+00
>>    0.000  0.000 -2.0020593599E+00
>>    0.000  0.000  2.0020593599E+00
>>    0.000  1.901  4.1896814012E-01
>>    0.000  1.901  4.4230868599E+00
>>    1.901  1.901  4.8420550000E+00   #layer 2
>>    1.901  3.802  7.2630825000E+00
>>    1.901  1.901  2.8399956400E+00
>>    1.901  1.901  6.8441143600E+00
>>    1.901  3.802  5.2610231400E+00
>>    1.901  3.802  9.2651418600E+00
>>    0.000  0.000  9.6841100000E+00   #layer 3
>>    0.000  1.901  1.2105137500E+01
>>    0.000  0.000  7.6820506400E+00
>>    0.000  0.000  1.1686169360E+01
>>    0.000  1.901  1.0103078140E+01
>>    0.000  1.901  1.4107196860E+01
>>
>> #Definition of the planewave basis set
>> ecut  24.0         # Maximal kinetic energy cut-off, in Hartree
>> pawecutdg 60.0
>>
>> #Definition of the k-point grid;
>> kptopt 1
>> kptrlatt  8  0  0
>>           0  8  0
>>           0  0  1
>> shiftk 0.5 0.5 0.0
>>
>> #Definition of the SCF procedure
>> iscf 17
>> nstep 150         # Maximal number of SCF cycles
>> toldfe 1.0d-9     # Will stop when, twice in a row, the difference
>>                   # between two consecutive evaluations of energy
>>                   # differ by less than toldfe (in Hartree)
>> #SCF preconditioner
>> iprcel 45         # Inhomogeneous systems work better with RPA method.
>> diemix 0.7        # Default for PAW
>>
>>
>>
>>
>>
>> Band structure calculation input file:
>>
>> # Crystalline anatase titania
>> #
>> # Computation of the band structure.
>> # Abinit automatically looks for density from a previous run, file
>> #Slabi_DEN, proceeds with non-SCF band structure calculation.
>>
>> #The band structure
>> iscf    -2     #Perform a non-scf calculation
>> kptopt  -1     #Use one segment and rely on kptbounds and ndivk
>> nband   90     #The number of bands
>> ndivk   22     #The number of samples for each segment
>>
>> kptbounds    0.0  0.0  0.0  #Gamma
>>              1/2  1/2  0.0  #X
>>
>> tolwfr  1.0d-12       # Tolerance of the wavefunction squared residual
>> enunit  1             # Will output the eigenenergies in eV
>> prteig  1             # Will outupt the eigenenergies to a separate file
>>
>> #Definition of the unit cell;
>> acell 3.802 3.802 9.68411/2 angstrom    # a,a,c/2
>>
>> #Alter rprim, oriented with vacuum above the (001) plane.
>> rprim   1.0            0.0              0.0
>>         0.0            1.0              0.0
>>         0.0            0.0              6.5
>> chkprim 0   # This input variable allows to use non-primitive unit cells.
>>
>> #Definition of the atom types
>> ntypat 2          # There are two types of atoms
>> znucl 22 8
>>
>> #Definition of the atoms
>> natom 18           # There are 18 atoms in our supercell
>> typat 1 1 2 2 2 2 1 1 2 2 2 2 1 1 2 2 2 2
>> xangst
>>    0.000  0.000  0.0000000000E+00   #layer 1
>>    0.000  1.901  2.4210275000E+00
>>    0.000  0.000 -2.0020593599E+00
>>    0.000  0.000  2.0020593599E+00
>>    0.000  1.901  4.1896814012E-01
>>    0.000  1.901  4.4230868599E+00
>>    1.901  1.901  4.8420550000E+00   #layer 2
>>    1.901  3.802  7.2630825000E+00
>>    1.901  1.901  2.8399956400E+00
>>    1.901  1.901  6.8441143600E+00
>>    1.901  3.802  5.2610231400E+00
>>    1.901  3.802  9.2651418600E+00
>>    0.000  0.000  9.6841100000E+00   #layer 3
>>    0.000  1.901  1.2105137500E+01
>>    0.000  0.000  7.6820506400E+00
>>    0.000  0.000  1.1686169360E+01
>>    0.000  1.901  1.0103078140E+01
>>    0.000  1.901  1.4107196860E+01
>>
>> #Definition of the planewave basis set
>> ecut  24.0         # Maximal kinetic energy cut-off, in Hartree
>> pawecutdg 60.0
>>
>> nstep 250         # Maximal number of non-SCF cycles
>>
>>
>>
>>
>>
>>
>



-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

Universite de Liège
Institut de Physique, Bat. B5
Allée du 6 aout, 17
B- 4000 Sart Tilman, Liège
Belgium

Phone : +32 4 366 37 50
Fax   : +32 4 366 36 29

Mail : matthieu.jean.verstraete@gmail.com




No virus found in this incoming message.
Checked by AVG - www.avg.com 
Version: 8.5.409 / Virus Database: 270.13.75/2340 - Release Date: 09/01/09
20:03:00