Re: [Wien] metal supercell band folding dielectric function

[MingWenmei]

Thank you, Professor Pblaha.
What you said is true only for semiconductors, because they have finite band 
gap separating "valence bands" from "conduction bands" and this will not change 
even with band folding from larger supercell calculations. 
However, for metals, because around Fermi energy the bands are partially 
occupied, then some states from the SAME band in primary unit cell calculation 
might fold back to bands with DIFFERENT band index.  
The simplest example is say we have a metal with one single band of parabolic 
dispersion when do an "unit cell" calculation. When we do dielectric function 
calculation, we only have the INTRA band contribution, that is, Drude-like 
dielectric function, which is only determined by plasma frequency and damping 
parameter.
This means all states are from one single band. However when we do a cell of 2 
times the unit cell, then basically we will have TWO bands due to band folding. 
The INTRA band contribution won't change because the plasma frequency is the 
same. However in this case we will have INTER band transition contributing to 
the dielectric function. This is something extra compared to ONE band 
calculation. It means the dielectric function will depend on the cell you are 
choosing. Physically this is WRONG.
Thank you for your suggestion,
Wenmei



> Date: Sat, 30 Mar 2013 08:27:39 +0100
> From: pbl...@theochem.tuwien.ac.at
> To: wien@zeus.theochem.tuwien.ac.at
> Subject: Re: [Wien] metal supercell band folding dielectric function
> 
> a) Do an optics calculation for the same 32 atom Au-supercell, but without 
> dopant,
> just "label" one Au atom by "Au1", so that the symmetry is identical to your 
> Cd-doped
> calculation. This should give identical eps2 than the small Au calculation 
> (beware of
> k-mesh effects !!!)
> b) Use bandstructure with "fat-bands" and partial DOS for analysis of your 
> doped system.
> You can also run optics with matrix-elements restricted to eg. only Cd or 
> only Au
> (missing crossterms). See UG for details.
> 
> Am 30.03.2013 00:27, schrieb MingWenmei:
> > Dear all,
> >
> > I am calculating the interband dielectric function of Au with low 
> > concentration of Cd dopant (say 1/32 atom concentration). Very strikingly 
> > in low energy (< 1eV) range,
> > there appears a high peak (~a few hundred). However for FCC pure gold, the 
> > interband dielectric function is very small (around zero) below 1eV.
> >
> > So my question is physically why the dielectric function changes this 
> > dramatically? I did the band structure calculation. I found within small 
> > energy range (<1eV) of Fermi
> > energy for the doped system, there are some extra bands. I began to realize 
> > it is this small energy difference of the bands above Fermi energy that 
> > causes the high peak of
> > dielectric function.
> >
> > These extra bands may be from doped Cd and also may be from the 
> > band-folding of host Au in the supercell calculation. The contribution from 
> > supercell itself to the
> > dielectric peak seems not to make sense physically. For if band-folding can 
> > affect the value of dielectric function, then even for pure FCC Au we may 
> > get significantly
> > different dielectric functions by choosing "unit cell" of different size 
> > (say 1x1x1, 4x4x4 times primary cell )
> >
> > Could anyone give some comment to understand this?
> >
> > Thank you a lot,
> >
> > Wenmei
> >
> >
> >
> >
> >
> >
> > ___
> > Wien mailing list
> > Wien@zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> 
> -- 
> -
> Peter Blaha
> Inst. Materials Chemistry, TU Vienna
> Getreidemarkt 9, A-1060 Vienna, Austria
> Tel: +43-1-5880115671
> Fax: +43-1-5880115698
> email: pbl...@theochem.tuwien.ac.at
> -
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
  ___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

[Wien] metal supercell band folding dielectric function

[MingWenmei]

Dear all,
I am calculating the interband dielectric function of Au with low concentration 
of Cd dopant (say 1/32 atom concentration). Very strikingly in low energy (< 
1eV) range, there appears a high peak (~a few hundred). However for FCC pure 
gold, the interband dielectric function is very small (around zero) below 1eV. 
So my question is physically why the dielectric function changes this 
dramatically? I did the band structure calculation. I found within small energy 
range (<1eV) of Fermi energy for the doped system, there are some extra bands. 
I began to realize it is this small energy difference of the bands above Fermi 
energy that causes the high peak of dielectric function. 
These extra bands may be from doped Cd and also may be from the band-folding of 
host Au in the supercell calculation. The contribution from supercell itself to 
the dielectric peak seems not to make sense physically. For if band-folding can 
affect the value of dielectric function, then even for pure FCC Au we may get 
significantly different dielectric functions by choosing "unit cell" of 
different size (say 1x1x1, 4x4x4 times primary cell )
Could anyone give some comment to understand this? 
Thank you a lot,
Wenmei



  ___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

[Wien] MBJLDA divergence charge

[MingWenmei]

Dear all,
I was doing the total energy calculation of (0001) surface of In2Se3 using 
MBJLDA potential.  Before I switched to MBJLDA everything went well with 
conventional LDA/GGA+SO, then once I started using MBJLDA on top of previous 
GGA charge density, the charge density started to diverge gradually(if mixing 
factor is small, the rate of diverging is slower, otherwise it diverges in 4-5 
steps) .   I checked previous posts and only found one relevant issue. But in 
that post no solutions were posted.
for MBJLDA I used (1) PARTT as the mixing method;(2) copy bulk In2Se3.grr to be 
surface_In2Se3.grr.  also delete surface_In2Se3.in0_grr

in cycle 2ETEST: 142.07565580   CTEST: .0196913in cycle 3ETEST: 
141.983307605000   CTEST: .0362174in cycle 4ETEST: .017627965000   
CTEST: .0506755in cycle 5ETEST: .017781875000   CTEST: .0632420in cycle 
6ETEST: .01830710   CTEST: .0741105in cycle 7ETEST: 
.01510429   CTEST: .0833564in cycle 8ETEST: .00956088   
CTEST: .0913016in cycle 9ETEST: .00426698   CTEST: .0980860in cycle 
10ETEST: .000849745000   CTEST: .1002863in cycle 11ETEST: 
.016063705000   CTEST: .1565063in cycle 12ETEST: .673544965000   
CTEST: .1325724in cycle 13ETEST: .33292142   CTEST: 1.0285099 LAPW0 
ENDSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT 
- ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - 
ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - ErrorSELECT - Error
>   stop error
I really appreciate your help if you can give me some suggestions. Thanks
Wenmei

  
-- next part --
An HTML attachment was scrubbed...
URL: 

[Wien] [SPAM?] RE: symmetry converts "hexagonal structure" to "orthorhombic structure"

[MingWenmei]

Dear Prof. Phlaha,
Thanks so much. inplance lattices should be of equal length. I did not find 
this minor difference before.  After this everything works.
Wenmei

> Date: Mon, 30 Apr 2012 08:55:43 +0200
> From: pblaha at theochem.tuwien.ac.at
> To: wien at zeus.theochem.tuwien.ac.at
> Subject: Re: [Wien] symmetry converts "hexagonal structure" to "orthorhombic 
> structure"
> 
> You created somehow a supercell, but you have a strange b-lattice parameter:
> 
>  7.59395   7.59396 155.16510  90.0  90.0 120.0
> 
> Is it intended that a and b differ by 0.1  
> 
> If you make them identical, everything works.
> 
> If you want them different, make them "clearly" different, since the symmetry 
> programs
> have difficulty to recognize if there is a symmetry or not for this input.
> 
> PS: Hexagonal and orthorhombic lattices are quite close to each other. Try to 
> span a
> "hexagonal lattice" with another unit cell which has angles of 90 degree !!!
> 
> Am 29.04.2012 18:37, schrieb MingWenmei:
> > Dear all,
> >
> > I met a serious problem in the initialization of my structure. It is a 
> > hexagonal supercell. Before x symmetry everything goes well: x group 
> > generates a hexagonal supercell with 6
> > symmetry matrices. But x symmetry always crashes down with the following 
> > error message. Also I found from outputs that the symmetry gives a "PGLSYM: 
> > THE CRYSTAL SYSTEM IS
> > ORTHORHOMBIC", not even a hexagonal structure. How is it possible for 
> > "symmetry" to convert a hexagonal structure to an orthorhombic structure?
> >
> >  > symmetry (10:18:34) 2 missing
> > 0.003u 0.012s 0:00.03 33.3% 0+0k 0+0io 0pf+0w
> > -- ERROR --
> > ERROR: (multiplicity of atom 1 )*(number of pointgroup-operations)
> > ERROR: is NOT = (number of spacegroup-operations)
> > ERROR: MULT: &! nbsp; 1 ISYM: 6 NSYM 2
> > ERROR: Check your struct file with x sgroup
> > -- ERROR --
> > -- ERROR --
> > ERROR: (multiplicity of atom 2 )*(number of pointgroup-operations)
> > ERROR: is NOT = (number of spacegroup-operations)
> > ERROR: MULT: 1 ISYM: 6 NSYM 2
> > ERROR: Check your struct file with x sgroup
> > -- ERROR --
> >
> > I attached the "structure" and "structure_sgroup", Please look a look if 
> > you can help.
> > My wien2k is 11.1 version
> >
> > Thanks so much,
> > Wenmei
> >
> >
> > ___
> > 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: +43-1-58801-165300 FAX: +43-1-58801-165982
> Email: blaha at theochem.tuwien.ac.atWWW: 
> http://info.tuwien.ac.at/theochem/
> --
> ___
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
  
-- next part --
An HTML attachment was scrubbed...
URL: 
<http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20120501/4d4fda06/attachment.htm>

[Wien] symmetry converts "hexagonal structure" to "orthorhombic structure"

[MingWenmei]

Dear all,
I met a serious problem in the initialization of my structure.  It is a 
hexagonal supercell. Before x symmetry everything goes well: x group generates 
a hexagonal supercell with 6 symmetry matrices. But x symmetry always crashes 
down with the following error message. Also I found from outputs that the 
symmetry gives a  "PGLSYM: THE CRYSTAL SYSTEM IS ORTHORHOMBIC", not even a 
hexagonal structure.  How is it possible for "symmetry" to convert a hexagonal 
structure to an orthorhombic structure?
>   symmetry(10:18:34) 2 missing 0.003u 0.012s 0:00.03 33.3% 0+0k 0+0io 
> 0pf+0w -- ERROR -- ERROR: (multiplicity of atom   
> 1 )*(number of pointgroup-operations) ERROR: is NOT = (number of 
> spacegroup-operations) ERROR: MULT:   1  ISYM:   6  NSYM  
>  2 ERROR: Check your struct file withx sgroup  -- ERROR 
> -- -- ERROR -- ERROR: (multiplicity 
> of atom   2 )*(number of pointgroup-operations) ERROR: is NOT = 
> (number of spacegroup-operations) ERROR: MULT:   1  ISYM:   6 
>  NSYM   2 ERROR: Check your struct file withx sgroup  -- 
> ERROR --
I attached the "structure" and "structure_sgroup", Please look a look if you 
can help. My wien2k is 11.1 version
Thanks so much,Wenmei 
-- next part --
An HTML attachment was scrubbed...
URL: 

-- next part --
A non-text attachment was scrubbed...
Name: testql.struct
Type: application/octet-stream
Size: 2106 bytes
Desc: not available
URL: 

-- next part --
A non-text attachment was scrubbed...
Name: testql.struct_sgroup
Type: application/octet-stream
Size: 9617 bytes
Desc: not available
URL: 

[Wien] x symmetry 2 missing

[MingWenmei]

Sorry I forgot to attach the two files. Now they are here





Dear all,
I'm trying to do a calculation of slab structure, but in the initialization 
"symmetry" is always stuck there with the warning "symmetry (00:58:14) 2 
missing". As I looked into the source code  this warning is from 
"SRC_symmetry/dirdef.f. Basically it terminated when trying to determine the 
local rotation matrix and symmetry.   So in struct_st it only has the first 4 
four atoms with the remaining 28 omiited.
I searched for previous similar cases like 
http://zeus.theochem.tuwien.ac.at/pipermail/wien/2011-July/014986.html. I 
replaced the rstruc.f and recompiled, but problem is still here without any 
effect.
Please take a look at the attached for Struct and Struct_st files I got. Any 
help will be greatly appreciated. 
Thanks so much.Wenmei   
  
-- next part --
An HTML attachment was scrubbed...
URL: 

-- next part --
A non-text attachment was scrubbed...
Name: test.struct_st
Type: application/octet-stream
Size: 1596 bytes
Desc: not available
URL: 

-- next part --
A non-text attachment was scrubbed...
Name: test.struct
Type: application/octet-stream
Size: 9803 bytes
Desc: not available
URL: 

[Wien] x symmetry 2 missing

[MingWenmei]

Dear all,
I'm trying to do a calculation of slab structure, but in the initialization 
"symmetry" is always stuck there with the warning "symmetry (00:58:14) 2 
missing". As I looked into the source code  this warning is from 
"SRC_symmetry/dirdef.f. Basically it terminated when trying to determine the 
local rotation matrix and symmetry.   So in struct_st it only has the first 4 
four atoms with the remaining 28 omiited.
I searched for previous similar cases like 
http://zeus.theochem.tuwien.ac.at/pipermail/wien/2011-July/014986.html. I 
replaced the rstruc.f and recompiled, but problem is still here without any 
effect.
Please take a look at the attached for Struct and Struct_st files I got. Any 
help will be greatly appreciated. 
Thanks so much.Wenmei 
-- next part --
An HTML attachment was scrubbed...
URL: 

[Wien] "R0 too large"----"SELECT - Error"

[MingWenmei]

Dear all, 

Recently I started using Wien2k_11_1 for the band-structure of HgS, even I 
generated the structure by w2web i was always reminded R0 for Hg is too large: 
the R0 automatically generated from w2web is 0.0001 for Hg. so I gradually 
deduced R0 to 0.05 at which the "R0 too large" complain disappeared. 
However starting from the scf calculation I cannot even successfully get 
through lapw1:   
LAPW0 END
SELECT - Error
SELECT - Error
SELECT - Error
SELECT - Error
I tried to adjust the linear energy of semi-core state, but I did not work at 
all.

Then I changed R0 for Hg to 0.1 in case.struct and did the initialization 
again, the this time the scf calculation goes very smooth, but "R0 too large" 
complain was still there. I thus become concerned with the reliability of my 
calculation with "R0 too large" complain even eventually I am able to get the 
converged result.
 
I checked these input files especially case.in1 for both the situations: they 
are exactly the same. Linear Energies are identical.

The last version of wien2k I had is Wien2k_09. Since I was never complained 
about "R0 too large", so I did not pay any attention to the magnitude of R0 and 
everything seems OK. I am thinking that such R0 inspection for lstart might be 
latest feature for wien2k.

Any suggestions are highly appreciated and thanks a lot.

Ming 
  
-- next part --
An HTML attachment was scrubbed...
URL: 

[Wien] Projected bulk bandstructure

[MingWenmei]

Hi Pblaha,
Thank you so much for you immediate reply.I want to put together the slab band 
structure and corresponding bulk band structure, and then compare the two. If 
there appears some extra bands other than the well-defined bulk bands, we may 
perceive these extra bands are closely related to the surface structure from 
the slab.Then my question is: how I do the 2D-projected bulk band structure, 
since in wien2k the available band structure plot is basically 3-D plot(in the 
sense you should specify the k-points in (kx,ky,kz) form).
Thanks for sharing some experience.Wenmei Ming

> Date: Mon, 4 Apr 2011 08:33:25 +0200
> From: pblaha at theochem.tuwien.ac.at
> To: wien at zeus.theochem.tuwien.ac.at
> Subject: Re: [Wien] Projected bulk bandstructure
> 
> You would use the "fat band" representation and specify the surface atoms
> and s,p or d-character.
> 
> This will show you bands which have large character at the surface.
> 
> As next step you repeat that for other atoms. The partial charges of a
> "surface state" decay rapidly into the bulk and the corresponding band
> should not have much (any) character of the central planes, while a
> "bulk" state usually is big every where or has some oszillations.
> 
> Finally you may inspect the suspect surface state directly in the
> case.qtl file and check the decainig charges into the bulk (or even
> plot them as function of layers).
> 
> 
> 
> Am 04.04.2011 03:48, schrieb MingWenmei:
> > Dear all,
> > 
> > In order to get the surface state band structure from a slab calculation, I 
> > want to plot the bulk projected band structure for comparison. For example 
> > the surface of interest is 
> > (110), a very tedious way to get bulk projected band structure is first to 
> > discretize (001) direction with a quite small interval, then do a bunch of 
> > bulk band structure 
> > calculations with the k-points in the first step, and finally overlap the 
> > band-structures obtained from steps with the shared k-point indices 
> > parallel to the surface.
> > 
> > I am wondering there is! some other easier way for me to do that, or some 
> > program already implemented in Wien2k for this purpose?
> > 
> > Thanks so much.
> > Wenmei Ming
> > 
> > 
> > 
> > ___
> > 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.atWWW: 
> http://info.tuwien.ac.at/theochem/
> --
> ___
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
  
-- next part --
An HTML attachment was scrubbed...
URL: 
<http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110405/bf296569/attachment.htm>

[Wien] Projected bulk bandstructure

[MingWenmei]

Dear all,
In order to get the surface state band structure from a slab calculation, I 
want to plot the bulk projected band structure  for comparison. For example the 
surface of interest is (110), a very tedious way to get bulk projected band 
structure is first to discretize (001) direction with a quite small interval, 
then do a bunch of bulk band structure calculations with the k-points in the 
first step, and finally overlap the band-structures obtained from steps with 
the shared k-point indices parallel to the surface.
I am wondering there is some other easier way for me to do that, or some 
program already implemented in Wien2k for this purpose?
Thanks so much.Wenmei Ming
-- next part --
An HTML attachment was scrubbed...
URL: 

[Wien] eigen-spin treatment in core states calculation

[MingWenmei]

Dear all, 

 

In the calculation of core states, Wien2k treats the spin-up electron and 
spin-down electron respectively 

from their full relativistic Dirac equations. However, in such calculations the 
spin(S) is not good quantum number

of the system, but the total angular momentum(J=l+S). Actually, in the 
/SRC_lcore/insld.f the spherical potential file

"case.vspup" is read for solution the "spin-up" Dirac equation and "case.vspdn" 
for solution of "spin-dn" Driac equation. 

 

Anyone can give me some tips to understand such treatment of independent 
"spin-up" and "spin-down" electrons 

with the framework of Dirac equation? In full relativistc situation, the 
spin-up and spin-down states are always mixed together

such that the eigen-spin states actually do not exist.

 

 

Thanks so much.

Wenmei Ming

 

Institute of Physics, 

Chinese Academy of Sciences

Beijing 100190

China

_
Messenger??Messenger???
http://im.live.cn/safe/
-- next part --
An HTML attachment was scrubbed...
URL: