Have a look at the following:
Molecular-dynamics simulations of steady-state growth of ion-deposited
tetrahedral amorphous carbon films
H. U. Jäger and K. Albe
Citation: J. Appl. Phys. 88, 1129 (2000); doi: 10.1063/1.373787
The trick that Jager and Albe suggest is to increase the cutoff distance
in the Tersoff potential. They suggest R and S values of 1.95 and 2.25
respectively.
Good luck!
John
Hi John,
Thanks for the tip, i have tried it and it increases the sp3 ratio.
However, I still could not get the RDF graph to “complete” as well.
Below is the link for the RDF graph generated based on the modified Tersoff:
https://www.dropbox.com/s/hw8ju6o3acna4aw/modTersoff.png
The graph ends at the beginning of the second peak, instead of completing it.
My initial thoughts are that it’s a problem of equilibration, but increasing the final equilibration time does not have any visible effect.
It seems that for both standard and modified Tersoff potential in the simulations i’ve done, the RDF graph does not extend fully, but I could not find any factors that might be responsible for it.
Wonder if anyone in the LAMMPS community have successfully generated a valid complete RDF graph from either Tersoff or modified Tersoff potential?
Rgds
Damien
Have a look at the following:
Molecular-dynamics simulations of steady-state growth of ion-deposited
tetrahedral amorphous carbon films
H. U. Jäger and K. Albe
Citation: J. Appl. Phys. 88, 1129 (2000); doi: 10.1063/1.373787
The trick that Jager and Albe suggest is to increase the cutoff distance
in the Tersoff potential. They suggest R and S values of 1.95 and 2.25
respectively.
Good luck!
John
Hi Damien,
Your question was regarding a Tersoff potential. In my opinion,
the Albe potentials are very high quality and the best you are
going to get for a pure Tersoff potential.
You mentioned before that you got good results from with REBO.
Why not just use that?
You might also consider the approximations contained in these
potentials. Tersoff potentials are first nearest neighbor potentials.
So if you are looking for second neighbors in your RDF in a
disordered material you likely won’t see them.
REBO is also first nearest neighbor potential, but it includes second
neighbors in the overbinding and dihedral terms of the bond order
function. This might be why you get a little better result from your
REBO model.
There are many other excellent potentials already implemented in
LAMMPS that might be useful to you. Have a look at ReaxFF, LCBPOI,
and AIREBO. These are all excellent potentials that can treat amorphous
carbon. The list of potentials not implemented is much longer, but
unless you are willing to do a lot of implementation work thats as
good as it gets.
The literature on carbons is 30 years old and theres a lot of very
outstanding papers. Perhaps you can contact a group directly if there
is a method that might best match your problem.
Best,
John