Dear Lammps Users and Developers,
I am going to write Tersoff potential file for an InGaN system. As written in the lammps document for a binary system, the file must have 8 entries and for a 3 elements, 27 entries is required. For my InGaN system, I want to use the parameters mentioned in the following paper ( table 2 in page 4). However, only the parameters for pair interactions, i.e., InN and GaN is reported. I need to have 27 lines in my file but I am not sure how I can calculate other interactions including 3 body interactions such as In-In-G, In-Ga-N and etc.
There are many papers which reported new Tersoff parameter for different system/alloys using fitting Tersoff to DFT data. But they just reported two elements interactions which I don’t know how to write them in LAMMPS format (please see the attached paper for example). How I can convert the two body interaction to the lammps format (3 body interaction)?
D. Powell, M. A. Migliorato, and A. G. Cullis, 2007, "Optimized Tersoff potential parameters for tetrahedrally bonded III-V semiconductors, "Phys. Rev. B 75, 115202.
http://journals.aps.org/prb/abstract/10.1103/PhysRevB.75.115202
Thanks
Hossein
1.pdf (136 KB)
Hi Hossein,
In fact, there are no true three dimensional (that is, depending on the nature of atoms i, j and k) parameters in the Tersoff potential. I suggest that you check for symmetries in a three-elements Tersoff potential file, eg in /potentials/BNC.tersoff
Arthur
Dear Lammps Users and Developers,
I am going to write Tersoff potential file for an InGaN system. As written
in the lammps document for a binary system, the file must have 8 entries and
for a 3 elements, 27 entries is required. For my InGaN system, I want to use
the parameters mentioned in the following paper ( table 2 in page 4).
However, only the parameters for pair interactions, i.e., InN and GaN is
reported. I need to have 27 lines in my file but I am not sure how I can
calculate other interactions including 3 body interactions such as In-In-G,
In-Ga-N and etc.
There are many papers which reported new Tersoff parameter for different
system/alloys using fitting Tersoff to DFT data. But they just reported two
elements interactions which I don't know how to write them in LAMMPS format
(please see the attached paper for example). How I can convert the two body
interaction to the lammps format (3 body interaction)?
you are mixing up the terms multi-component system and manybody
interactions. the potentials you refer to all *do* have 3 body
interaction terms, only they are parameterized for binary systems.
determining the parameters for the interactions between three
different elements requires an additional parameterization step. if
two of the elements are similar, those can sometimes be inferred from
mixing rules, combining the parameters from two binary interaction
parameter sets.
axel.
Hi Hossein,
In fact, there are no true three dimensional (that is, depending on the
nature of atoms i, j and k) parameters in the Tersoff potential.
Sorry but Tersoff do have a couple of terms that depends on atoms i, j, and
k so it does have "true" three dimensional parameters.
Ray
There are “terms”, like the angular function g(theta), which depend on the nature of atoms i, j and k, but there are no “parameters” depending on the nature of the three atoms. See equations (1a) to (1e) in , all the parameters either depend on the central atom i, or the bond, ij or ik. Now, the format used in lammps allows to use three dimensional parameters instead of one- or two-dimensional ones. This gives much more flexibility, but is different from what is described in Tersoff’s papers. Arthur
Thank you all for responding to this email.
I am sorry but I haven’t figured out how to calculate the parameters for mulch-component systems yet. Could you please explain a little bit more?
For a ternary alloy (ABC), in Tersoff file I know X1 = {AAB, AAC, BBA, BBC, CCA, CCB}, and X2= {ABC, ABA, … etc} are 3 body interactions and X3= {ABB, ACC, BAA, BCC, CAA, CBB} are 2 body interactions. If I have just the parameters for AB and AC, How I can calculate ABC or ABB or etc parameters. I mean, I think the parameters for AB is the parameters for pair interaction between atom A and atom B, how I can modify the parameters to include the third atom B in this iteration, i.e., ABB or ABA or ABC.
Thanks
Hossein
Thank you all for responding to this email.
I am sorry but I haven't figured out how to calculate the parameters for
mulch-component systems yet. Could you please explain a little bit more?
For a ternary alloy (ABC), in Tersoff file I know X1 = {AAB, AAC, BBA, BBC,
CCA, CCB}, and X2= {ABC, ABA, .. etc} are 3 body interactions and X3= {ABB,
ACC, BAA, BCC, CAA, CBB} are 2 body interactions. If I have just the
parameters for AB and AC, How I can calculate ABC or ABB or etc parameters.
I mean, I think the parameters for AB is the parameters for pair interaction
between atom A and atom B, how I can modify the parameters to include the
third atom B in this iteration, i.e., ABB or ABA or ABC.
at this point it seems to me, that you have to very carefully re-read
the pair_style tersoff documentation. it does explain in detail the
functional form that LAMMPS uses. it also explains that this
functional form is a superset of several functional forms used by
Tersoff and other authors in the literature and how these can be
mapped to the generalized form that LAMMPS uses. it also explains, how
you can derive missing mixed terms from existing binary
parameterizations using the provided mixing rules, but that *requires*
that the original parameters correspond to a specific variety of the
Tersoff potential (called Tersoff_2). all the necessary references are
given as well. so what you need to do is to first verify to which of
the available variants the paper you have corresponds (best by
converting/entering the data for multiple of those binary
parameterizations and then trying to reproduce properties from that
reference publication). once that is successful, you could continue to
see, if those mixing rules can be applied and the missing parameters
for the ternary compound can be derived that way. otherwise, you will
have to either find a different publication or learn how to do the
parameterization yourself.
axel.