Top Page > Browsing
Initial Charge Densities
Date: 2016/11/08 17:07
Name: Clarke

Dear Openmx experts,

I am new user of Openmx. In the manual under‘Atom.SpeciesAndCoordinates’tag, it is written that "The sixth and seventh columns give the number of initial charges for up and down spin states of each atom, respectively. The sum of up and down charges must be the number of valence electrons for the atomic element."

For Carbon atom, the valance electrons are four. So we defined 2 and 2 in the sixth and seventh column when we use pseudopotential C5.0-s2p1. It works well, as it is also given in the ‘work directory’ in negf_examples. So the distribution of initial charges for spin up and spin down states must be same for the elements belonging in Carbon group (IV).

I am dealing with 'Tin' with pseudopotential Sn7.0-s2p2d3f1, which comes in the column of element 'Carbon' having four valence electrons. As my chemistry knowledge is very less, the 'Tin' also has four valence electrons. I am defining the initial charge densities of each 'Tin' atom by giving 2.5 and 1.5 or 2.0 and 2.0, which is, of course, the sum of the valence electrons. But when I ran the program it gives the following error:

<SetPara_DFT> PAOs of species Sn were normally found.
<SetPara_DFT> PAOs of species H were normally found.
<SetPara_DFT> VPSs of species Sn were normally found.
Sn_CA13.vps is j-dependent.
In case of scf.SpinOrbit.Coupling=off,
j-dependent pseudo potentials are averaged by j-degeneracy,
which corresponds to a scalar relativistic treatment.
<SetPara_DFT> VPSs of species H were normally found.
H_CA13.vps is j-dependent.
In case of scf.SpinOrbit.Coupling=off,
j-dependent pseudo potentials are averaged by j-degeneracy,
which corresponds to a scalar relativistic treatment.
Invalid values for the initial densities of atom 1 (valid sum: 14.000)
Invalid values for the initial densities of atom 2 (valid sum: 14.000)
Invalid values for the initial densities of atom 3 (valid sum: 14.000)
Invalid values for the initial densities of atom 4 (valid sum: 14.000)
Invalid values for the initial densities of atom 5 (valid sum: 14.000)
.
.
Invalid values for the initial densities of atom 20 (valid sum: 14.000)
Invalid values for the initial densities of atom 21 (valid sum: 14.000)
Invalid values for the initial densities of atom 22 (valid sum: 14.000)
Invalid values for the initial densities of atom 23 (valid sum: 14.000)
Invalid values for the initial densities of atom 24 (valid sum: 14.000)

It says the sum of the initial charge densities must be equal to 14. I changed it and gave the sum of the initial charge densities equal to 14 (7 for up and 7 for down), it works well.

My question is that the sum of the charge densities must be equal to the number of the valence electrons (as mentioned in the manual), however 'Tin' has only four valence electrons, not 14. Then why it is giving the above error when I am giving the sum of the charge densities is equal to 4 (same as in case of carbon because carbon also has 4 valance electrons as Tin). Am I making serious mistake in defining the initial charge densities of Tin atoms?

Your comments and suggestions in this regard would be highly appreciated.

Thank you.
メンテ
Page: [1]

Re: Initial Charge Densities ( No.1 )
Date: 2016/11/08 19:36
Name: Artem Pulkin

Hi,

Do not worry: these are only initial occupation numbers. They are only important for magnetic systems where several alternative magnetic configurations are possible.

Now, your particular question is about the pseudopotential approximation. There, we define a frozen core and valence electrons for each atom. The particular Tin pseudopotential you are talking about has 36 frozen electrons and 14 valence. You may increase the balance towards 46 frozen electrons and 4 valence. This will simplify your calculations considerably, however, since only 4 electrons participate in bonding the properties of your solid will be described poorly.

So, in short, valence electrons are not defined strictly. When you look into the periodic table they just correspond to the outermost atomic shell, however, given a pseudopotential approximation one typically includes several more "semicore" electrons into the valence shell.

Keywords: pseudopotential, valence, semicore

Artem
メンテ
Re: Initial Charge Densities ( No.2 )
Date: 2016/11/09 17:45
Name: Clarke

Dear Artem,

Thank you for your reply and giving proper suggestion. I have completed my calculations successfully. I plotted transmission vs energy-chemical potential of the left lead (E-mu_L). It is observed that the Fermi level is not at zero but it is shifted to around 0.75 eV. I have gone through forum posts, it is found that Openmx does not calculate Fermi level instead it calculates the chemical potential. The calculation are done at default value of temperature i.e. at T = 300 K.

So my question is:

1) the obtained results are correct or incorrect, and how to defend it if the Fermi level is not at zero

2) Is there any way to set the Fermi level at zero.

I have taken a look on the results reported in the literature using Openmx that almost everybody plotting transmission vs E-mu, and they are showing Fermi level at zero, but I did not understand the setting of Fermi level at zero.

Your help would be highly appreciated.

Thanking you.

Clarke
メンテ
Re: Initial Charge Densities ( No.3 )
Date: 2016/11/09 19:40
Name: Artem Pulkin

1) I do not know. In earlier version of OpenMX everything related to transmissions was relative to the Fermi level. You may always run examples provided with the distribution and check if it is the case.

2) Maybe just subtract its value?
メンテ
Re: Initial Charge Densities ( No.4 )
Date: 2016/11/10 00:08
Name: Clarke

Dear Artem,

Thank you for your reply. Prof. Ozaki mentioned the following:


What OpenMX gives is not Fermi level (EF), but chemical potential(mu) for a given
temperature (T). The mu is found in OpenMX so that the total number of electrons
can be conserved at T. Also, we have a relation:

lim T->0 mu = EF

at the forum discussion http://www.openmx-square.org/forum/patio.cgi?mode=view&no=975

Acording to the above discussion,

If the calculations are done at the finite temperature the Fermi level may not be at Zero. In this case can we say that the Fermi level is not as zero because the calculations are done at finite temperature that may shift the Fermi level from zero.


2) May be just subtract its value?

All the researchers are just subtracting its value to match the Fermi level at zero.

Your help would be highly appreciated because all my results depends on this statement. Please help me in this regard.

Thanking you.

Clarke


メンテ
Re: Initial Charge Densities ( No.5 )
Date: 2016/11/11 13:09
Name: Clarke

Dear Artem,

I checked with graphene nanoribbon. The Fermi level is coming near at zero but exactly at zero. My question is that if I am dealing with 'Tin'then why the Fermi level is not at zero.

Please help in this regard.

Thanking you.
メンテ
Re: Initial Charge Densities ( No.6 )
Date: 2016/11/11 19:16
Name: Artem Pulkin

"My question is that if I am dealing with 'Tin'then why the Fermi level is not at zero."

Look, you may set your energy origin at arbitrary value without losing physical sense. The Fermi level is set with respect to THE ORIGIN, thus, generally, the Fermi level is non-zero. When you plot a band structure for the publication you typically want to set your Fermi level to zero since it is the only "special" energy on your plot. So people take the calculated band structure energies and subtract the Fermi level from them setting THE ORIGIN to the Fermi level. Then, they state that the Fermi level on the plot at zero.

For example, when simulating 2D structures, your energy origin is close to the vacuum level. The Fermi level is always below it, thus, you always find it negative (for example, E_F = -5 eV). In 3D structures it is different (there is a reason for that): it can be both positive and negative. I.e. ARBITRARY.

Finally, a few words about the chemical potential. While there is an exact relation between the Fermi level and the chemical potential they are almost equal. As far as I remember the difference between them is less than the order of magnitude of temperature 300K = 25 meV. However, the error in the calculated Fermi level or chemical potential (whatever you call it) is typically larger: your chosen k-point grid is just too small to capture the 25 meV difference. The chemical potential of band gap materials, for example, should always be in the conduction energy region. However one always finds it at some arbitrary levels inside the band gap. So while it is the chemical potential by concept, it smells, tastes and looks like the Fermi level.

Hope I was clear enough. Whether you have more questions or not I suggest you to follow some solid state physics courses.

Artem
メンテ
Re: Initial Charge Densities ( No.7 )
Date: 2016/11/14 00:19
Name: Clarke

Dear Artem,

Thanking you for your reply.

First two paragraphs are OK to me no problem. My main concern is

>>However, the error in the calculated Fermi level or chemical potential (whatever you call it) is typically larger: your chosen k-point grid is just too small to capture the 25 meV difference.

In this, you talk about k-point grid:

Is it:
1) scf k-point grid of leads

or
2) scf.Kgrid (in NEGF file)
or
3) NEGF.scf.Kgrid
or
4) NEGF.tran.Kgrid

One more thing about NEGF.gate.voltage (Is it not working when we work with system like left lead-central region-right lead) because I am dealing with this type of system and applying gate voltage under zero and finite bias voltages, but I am not seeing any difference between the results with and wtithout gate voltages. However, the gate voltage mainly effect the central region, in my case the central region is very thin, may be not more than 8-10 atoms.

Your suggestions would be highly appreciated.

Thanking you.
メンテ
Re: Initial Charge Densities ( No.8 )
Date: 2016/11/14 21:44
Name: Artem Pulkin

"In this, you talk about k-point grid:"

The Fermi level is calculated prior to any Greens function calculation so you should be concerned about the first option: scf.kgrid of your leads.

Regarding the rest: if you want to have answers please ask questions.
メンテ
Re: Initial Charge Densities ( No.9 )
Date: 2016/11/15 15:29
Name: Clarke

Dear Artem,

Thank you very much for your help. Now I am doing some calculation on graphene nanoribbon with spinorbit coupling but the results are not differnt looks same. The results are not affected by SOC because it is weak in graphene, is it the case. For spin orbit coupling I used the following input file

#
# File Name
#

System.CurrrentDirectory ./ # default=./
System.Name lead-l-6zgnrsoc
level.of.stdout 1 # default=1 (1-3)
level.of.fileout 1 # default=1 (0-2)


#
# Definition of Atomic Species
#

Species.Number 2
<Definition.of.Atomic.Species
C C5.0-s2p1 C_CA13
H H5.0-s2 H_CA13
Definition.of.Atomic.Species>


NEGF.output_hks true
NEGF.filename.hks lead-l-6zgnrsoc.hks


#
# Atoms
#

Atoms.Number 28
Atoms.SpeciesAndCoordinates.Unit Ang # Ang|AU
<Atoms.SpeciesAndCoordinates
1 C 0.000000 0.000000 0.000000 2.500000 1.500000
2 C 0.000000 2.836000 0.000000 2.500000 1.500000
3 C 2.456048 0.000000 0.000000 2.500000 1.500000
4 C 1.228024 0.709000 0.000000 2.500000 1.500000
5 C 1.228024 2.127000 0.000000 2.500000 1.500000
........
25 H 0.000000 -1.090000 0.000000 0.500000 0.500000
26 H 2.456048 -1.090000 0.000000 0.500000 0.500000
27 H 0.000000 12.434000 0.000000 0.500000 0.500000
28 H 2.456048 12.434000 0.000000 0.500000 0.500000
Atoms.SpeciesAndCoordinates>
Atoms.UnitVectors.Unit Ang # Ang|AU
<Atoms.UnitVectors
4.902196 0.000000 0.000000
0.000000 23.524000 0.000000
0.000000 0.000000 10.000000
Atoms.UnitVectors>


#
# SCF or Electronic System
#


scf.XcType LSDA-CA # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization NC # On|Off|NC
scf.SpinOrbit.Coupling on # On|Off
scf.ElectronicTemperature 300.0 # default=300 (K)
scf.energycutoff 200.0 # default=150 (Ry)
#scf.Ngrid 140 140 100 # about=200 (Ry)
scf.maxIter 1000 # default=40
scf.EigenvalueSolver Band # DC|GDC|Cluster|Band
scf.lapack.dste dstevx # dstegr|dstedc|dstevx, default=dstegr
scf.Kgrid 201 1 1 # means n1 x n2 x n3
scf.Mixing.Type rmm-diisk # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
scf.Kerker.factor 10.0 # default=1
scf.Init.Mixing.Weight 0.0001 # default=0.30
scf.Min.Mixing.Weight 0.0001 # default=0.001
scf.Max.Mixing.Weight 0.105 # default=0.40
scf.Mixing.History 150 # default=5
scf.Mixing.StartPulay 10 # default=6
scf.Mixing.EveryPulay 1 # default=6
scf.criterion 1.0e-6 # default=1.0e-6 (Hartree)

#
# MD or Geometry Optimization
#



メンテ

Page: [1]

Title (must) Move the thread to the top
Your Name (must)
E-Mail (must)
URL
Password (used in modification of the submitted text)
Comment (must)

   Save Cookie