COMB charges for Si and O in the input file with coords to read

Hi,
I am trying to use the COMB potential for creating a Si substrate and have divided the substrate with 512 atoms into fixed and moving atom groups. The fixed atoms do not have any charge but am not understanding what value of charge to use for the moving Si atoms.

I referred to the lammps comb example form HfO2 and found a charge value of 3.6 for Hf with an atomic number 72 and -1.8 for O used in the data files. Could somebody please advise how this values were calculated.

Also, I am using the fix qeq/comb to calculate the equilibrium charge on each atom which is calculated by the electronegativity equation. Below is the file generated.

Charge equilibration on step 5000
iteration: 0, enegtot -0.56024, enegmax 0.0562825, fq deviation: 0.0104876
iteration: 1, enegtot -0.560245, enegmax 0.056132, fq deviation: 0.0104515
iteration: 2, enegtot -0.560255, enegmax 0.0558465, fq deviation: 0.0103836
iteration: 3, enegtot -0.560269, enegmax 0.0554407, fq deviation: 0.0102881
iteration: 4, enegtot -0.560286, enegmax 0.0549286, fq deviation: 0.0101678
iteration: 5, enegtot -0.560306, enegmax 0.0543232, fq deviation: 0.0100288
iteration: 6, enegtot -0.560328, enegmax 0.0536369, fq deviation: 0.00987297
iteration: 7, enegtot -0.560352, enegmax 0.0528813, fq deviation: 0.00970749
iteration: 8, enegtot -0.560378, enegmax 0.0520669, fq deviation: 0.00953108
iteration: 9, enegtot -0.560404, enegmax 0.0512039, fq deviation: 0.00934512




iteration: 69, enegtot -0.560682, enegmax 0.0135017, fq deviation: 0.00304473
Charges converged in 70 iterations to 0.0029946495 tolerance

I understand that the equilibrated value of the charges is 0.0029946495 but is this what is to be used as charge in the data file for Si in my new iteration?

Regards,
Saketh.

Hi, could anybody please reply to the query below…

Ray can comment.

I think the answer is that is does not matter what you set the charges
to in a 2nd run. If you use COMB, it will reset the charges for
you as it proceeds.

Steve

Hi, could anybody please reply to the query below..
________________________________
From: [email protected]...
To: [email protected]
Date: Tue, 8 Nov 2011 17:37:48 +0000
Subject: [lammps-users] COMB charges for Si and O in the input file with
coords to read

Hi,
I am trying to use the COMB potential for creating a Si substrate and have
divided the substrate with 512 atoms into fixed and moving atom groups. The
fixed atoms do not have any charge but am not understanding what value of
charge to use for the moving Si atoms.
I referred to the lammps comb example for HfO2 and found a charge value of
3.6 for Hf with an atomic number 72 and -1.8 for O used in the data files.
Could somebody please advise how this values were calculated.

if you do run charge equilibration, those are initial guesses.

Also, I am using the fix qeq/comb to calculate the equilibrium charge on
each atom which is calculated by the electronegativity equation. Below is
the file generated.
Charge equilibration on step 5000
iteration: 0, enegtot -0.56024, enegmax 0.0562825, fq deviation: 0.0104876
iteration: 1, enegtot -0.560245, enegmax 0.056132, fq deviation: 0.0104515

[...]

Charges converged in 70 iterations to 0.0029946495 tolerance
I understand that the equilibrated value of the charges is 0.0029946495 but

this is nonsense. the charges are equilibrated up to that _tolerance_,
i.e. accuracy. if you want to see the individual charges you either have
to write them out to a custom dump or a restart file and convert the
latter with restart2data.

is this what is to be used as charge in the data file for Si in my new
iteration?

the qeq fix updates the internal per atom charges, so they
will be in the restart file you can continue with the equilibrated
charges.

cheers,
   axel.

Hi Saketh,

The 3.6 and -1.8 for Hf and O in the example script are initial charges
for the HfO2 compound. COMB potential, based on Qeq principle, will
equilibrate the charges to find a lowest energy state based on your input
coordinates. The input charges are just initial guesses that are close to
the equilibrium so that Qeq takes less time/iterations.

The 0.00299 from the Qeq output is the deviation in electronegativities
(X), or the charge force, among each atom. When X is almost equal for all
of the atoms (in this case difference < 0.003) then charge values don't
change anymore and reach an equilibrium charge state. This difference is
the "precision" in the fix qeq/comb command. If you use the provided
in.comb.HfO2 example script, the last two thermo output columns are the
equilibrium charges for Hf and O.

In your case, you can set all charge to zero and let COMB find equilibrium
charges, or you can provide your initial guesses for your fixed and moving
atoms.

Ray Shan