hello
Can you give me some advice about Zn pseudopotential generation?
Zn: 1s2 2s2 2p6 3s2 3p6 3d10 4s2
4s2 as valence
I have used different datasets . But the result is always not good.
This is one of my input files:
Zn 30
LDA-PW scalarrelativistic loggrid 2100
4 3 3 0 0
4 0 2
0 0 0
c
c
c
v
c
c
c
0
2.3
y
3.0
n
Vanderbilt
1 0 ultrasoft
2.0
2.3
2
default
0
Zn PAW datasets
-
Alain_Jacques
- Posts: 279
- Joined: Sat Aug 15, 2009 9:34 pm
- Location: Université catholique de Louvain - Belgium
Re: Zn PAW datasets
Hello Tiger,
just a few ideas...
1. no need to have such a large grid - your pseudo will be dog slow. Start with 1200-1500, try to go down to 600-900.
2. if you look at the orbital energies, you'll find that if 4s is valence, 3d should be valence too.
3. maximum l for basis and projector functions should be 2
4. add at least one s projector, two p projectors, one d projector - more if you need accuracy
So I would start playing with
or similarly (by adding an empty 4p valence, rearranging the projectors and reducing rpaw)
This is untested, unfinished, un ... all what you can think about. Look at the valence electrons energies comparison in the "Zn" file when changing parameters. Plot the wfn files to spot large (>15) coefficients and cranky waves. Plot logderiv files to spot ghosts; adjust reference energies for unbound states and matching radii.
Kind regards,
Alain
just a few ideas...
1. no need to have such a large grid - your pseudo will be dog slow. Start with 1200-1500, try to go down to 600-900.
2. if you look at the orbital energies, you'll find that if 4s is valence, 3d should be valence too.
3. maximum l for basis and projector functions should be 2
4. add at least one s projector, two p projectors, one d projector - more if you need accuracy
So I would start playing with
Code: Select all
Zn 30
LDA-PW scalarrelativistic loggrid 1200
4 3 3 0 0
0 0 0
c
c
c
v
c
c
v
2
2.3
y
1.
n
y
2.
y
3.
n
y
4.
n
vanderbilt
3 0.0 ultrasoft
2.25
2.25
2.25
2.25
2.25
2.25
0or similarly (by adding an empty 4p valence, rearranging the projectors and reducing rpaw)
Code: Select all
Zn 30
LDA-PW scalarrelativistic loggrid 1200
4 4 3 0 0
4 1 0
0 0 0
c
c
c
v
c
c
v
v
2
2.0
y
1.
n
y
2.
n
y
3.
n
vanderbilt
3 0.0 ultrasoft
1.95
1.95
1.95
1.95
1.95
1.95
0This is untested, unfinished, un ... all what you can think about. Look at the valence electrons energies comparison in the "Zn" file when changing parameters. Plot the wfn files to spot large (>15) coefficients and cranky waves. Plot logderiv files to spot ghosts; adjust reference energies for unbound states and matching radii.
Kind regards,
Alain
Re: Zn PAW datasets
Dear Alain_Jacques
Thanks for your good advice.
It works.
I will try to test it and finish it .
Kind regards
Tiger
Thanks for your good advice.
It works.
I will try to test it and finish it .
Kind regards
Tiger
Re: Zn PAW datasets
Hi,
I found Zn atomic data file Zn.gga.atompaw and Zn.lda.atompaw under PAW atomic data (ABINIT table) recently.
However, after downloading I got a HTML file insteated of a atompaw file. I downloaded Ce.gga.atompaw
before. There is no trouble.
Regards
Zhifan Chen
Clark Atlanta University
USA
I found Zn atomic data file Zn.gga.atompaw and Zn.lda.atompaw under PAW atomic data (ABINIT table) recently.
However, after downloading I got a HTML file insteated of a atompaw file. I downloaded Ce.gga.atompaw
before. There is no trouble.
Regards
Zhifan Chen
Clark Atlanta University
USA