Sources of vdw pseudopotentials

Asked by Tony on 2017-03-16

Can anyone share some well tested vdw pseudopotentials for atoms such as H, C, Si,...?
Thank you in advance.

Question information

English Edit question
Siesta Edit question
No assignee Edit question
Solved by:
Last query:
Last reply:

This question was reopened

  • 2017-03-17 by Tony
Tony (tony2017) said : #1

Well, I handled myself with Atom

Tony (tony2017) said : #2

When I generated vdw pseudo-potential for C atom, I observed in the OUT file some warning such as:

n = 1 l = 0 s = 0.0
        a extr 1.730
        r extr 0.174
        r zero
        r 90/99 % 0.488 0.795

 n = 2 l = 0 s = 0.0
        a extr 0.352 -0.823
        r extr 0.138 1.208
        r zero 0.377
        r 90/99 % 2.511 3.841

 n = 2 l = 1 s = 0.0
        a extr 0.710
        r extr 1.194
        r zero
        r 90/99 % 3.144 5.250

 n = 3 l = 2 s = 0.0
        a extr 0.205
        r extr 111.620
        r zero
        r 90/99 % 118.006 119.491
        WARNING: This orbital is not bound!

 n = 4 l = 3 s = 0.0
        a extr 0.223
        r extr 111.595
        r zero
        r 90/99 % 118.006 119.491
        WARNING: This orbital is not bound!

I got similar warnings when generating the GGA pseudo-potential. I ran an example for Si in the directory "Tutorial/PS_Generation" of Atom, I also observed such warning in the OUT file.

Do I have to worry bout this?

Tony (tony2017) said : #3

An other question, I have seen that some in put files with prefix "pe", that is not "pg"-pseudopotential generation. What does "pe" stand for? I compared the OUT files generated by these two prefixes they result in different total energies.

Tony (tony2017) said : #4

I am reporting another issue: vdw pseudopotentials do not work with transiesta:

I used the generated vdw pseudopotentials for C and H atoms to make a test with transiesta (calculated for bilayer graphene ribbons). it has showed an error:

GHOST: No ghost state for L = 0
GHOST: WARNING: Ghost state for L = 1
KBgen: WARNING: Ghost states have been detected
KBgen: WARNING: Some parameter should be changed in the
KBgen: WARNING: pseudopotential generation procedure.
Stopping Program from Node: 0
Stopping Program from Node: 0

and calculations stopped.

It is worth noting that I cheched that vdw of C atoms using with Siesta calculation can lead to the band structure of bilayer graphene without any problems. And I noted that there were no Ghost states found in Siesta calculations:

GHOST: No ghost state for L = 0
GHOST: No ghost state for L = 1
GHOST: No ghost state for L = 2

So why does transiesta report this kind of error?

Tony (tony2017) said : #7

Since no ones has suggested solution for this problem, I had to investigated myself.

Checking in detail I have seen that the error of Ghost state detected comes from the pseudopotential of H atoms, not C atoms. Thus my previous comparison with calculations using C atoms is not relevant.

I did a search and found a topic reported the same issue: https://<email address hidden>/msg06022.html

Although this topic did not show a completed solution, I have eventually figured out the way:

1. Generating a new vdw PP for H atoms with some modification in the inp file:
the input *.inp file of H atoms that I downloaded from the Siesta site has the cutoff radii for different shells in values:
1.33000 1.33000 0.37000 1.33000

Since the PP of C atoms did not show any problem, I tried to learn the setup in *.inp file for C atom and I have observed that the 4h radius should be larger than the others. So I set up a modification in the *.inp of H atom, for instance:
 1.33000 1.33000 0.37000 1.700000
Then I generated again the vdw PP with Atom program.

2. In the main fdf file, I also set (without this setup we still see the error):

%block PS.lmax
   C.vdw 3
   H.vdw 3
%endblock PS.lmax

Combining these 2 solutions, I have successfully removed the error of Ghost state detected.

Hope this information is useful for who has the same issue.

Yelda Kadioglu (yelda) said : #8

Hi, i generated relativistic PP and tried to do SOC calculation with siesta-4.1b3 version but i got an error about ghost state for l=0, i tried block PS.lmax as yours but it didnt solve the problem. I tried to change radius but i dont know what number i write to relevant part. How do you you decide on ? thanks