Dear all
I am trying to simulate Carbon nanotube using Tersoff potential, its running fine but now I want to add LJ potential along with Tersoff potential for all carbon atoms to add van der Waals interaction in system. My part of my input file
units real
atom_style atomic
pair_style hybrid/overlay tersoff lj/cut 10.29
pair_coeff * * tersoff …/C.tersoff C
pair_coeff * * lj/cut 0.0556417 3.430
compute 1 all pair tersoff
compute 2 all pair lj/cut
thermo_style custom step temp c_1 c_2 epair evdwl ecoul ebond pe ke etotal press vol
But its considering only LJ potential, when I’m computing c_1 (tersoff pair potential) its coming zero. But I want summation of two interactions as total interaction.I know its over-writing, but how to define the two potentials together. is it possible?
Dear all
I am trying to simulate Carbon nanotube using Tersoff potential, its
running fine but now I want to add LJ potential along with Tersoff
potential for all carbon atoms to add van der Waals interaction in system.
My part of my input file
units real
atom_style atomic
pair_style hybrid/overlay tersoff lj/cut 10.29
pair_coeff * * tersoff ../C.tersoff C
pair_coeff * * lj/cut 0.0556417 3.430
compute 1 all pair tersoff
compute 2 all pair lj/cut
thermo_style custom step temp c_1 c_2 epair evdwl ecoul ebond pe ke
etotal press vol
But its considering only LJ potential, when I'm computing c_1 (tersoff
pair potential) its coming zero. But I want summation of two interactions
as total interaction.I know its over-writing, but how to define the two
potentials together. is it possible?
there are two issues at hand here.
1) i cannot reproduce what you are claiming. when i set up a test
simulation in the way you describe, i do get both the energy contributed by
tersoff and by lj/cut. are you using the current version of LAMMPS? if you
run the input with tersoff only, do you get any energy contributions, i.e.
are the atoms close enough to interact via tersoff?
2) your model is bogus! you are double counting the close interactions.
while tersoff does not contain long-range interactions, lj/cut also has
(strong) short range interactions, that are put on top of tersoff. that
*has* to be bogus. you would have to use a more complex model to handle
this kind of setup, where you use switching functions to transition from
one interaction to the other. please have a look at AIREBO, for an example
how this can be done. it most certainly won't work via hybrid/overlay.
axel.
Thanks Axel for your response,
Coming to your 1st point, when I ran using only Tersoff, the pair-wise interaction energy was not zero, when I’m switching to both Tersoff and lj/cut, lj/cut energy is coming, but Tersoff energy is becoming zero, and I’ve also visualized atoms are not hold by bond anymore, structure is totally deforming, whereas with only Tersoff nanotube structure was as it should be. There was a huge difference in Tersoff(~ -100000) and Lj(~ -60) energy. I’m using LAMMPS 5 Sep 14.
Actually I want to use COMB3 potential in future. For comparing results I want to run Tersoff+LJ+coulomb(I know its not directly comparable). As far as I know Tersoff potential does not considered van der Waals interaction, when I will study adsorption(physisorption due to van der Waals interaction) on nanotube it will require LJ potential so I need Tersoff + LJ.
And one more thing I want to know, does COMB3 potential include LJ interaction?
Thanks Axel for your response,
Coming to your 1st point, when I ran using only Tersoff, the pair-wise
interaction energy was not zero, when I'm switching to both Tersoff and
lj/cut, lj/cut energy is coming, but Tersoff energy is becoming zero, and
I've also visualized atoms are not hold by bond anymore, structure is
totally deforming, whereas with only Tersoff nanotube structure was as it
should be. There was a huge difference in Tersoff(~ -100000) and Lj(~ -60)
energy. I'm using LAMMPS 5 Sep 14.
that sounds like you are observing what i mentioned: that kind of model is
bogus. also, *always* double check with the last patch level of LAMMPS
before reporting a "problem". nobody wants to track down and correct a
problem that has already been found and corrected.
Actually I want to use COMB3 potential in future. For comparing results I
want to run Tersoff+LJ+coulomb(I know its not directly comparable). As far
as I know Tersoff
what is the point of this comparison? you already know that you are going
to compare apples to oranges? this seems to be quite a waste of time to me.
potential does not considered van der Waals interaction, when I will study
adsorption(physisorption due to van der Waals interaction) on nanotube it
will require LJ potential so I need Tersoff + LJ.
as i mentioned before, you cannot just add an unmodified LJ term to an
unmodified tersoff. that is flat out wrong. from your description it is
also not clear why you need tersoff. and what the coulomb support is going
to add to the description. in any case, this all seems to be a problem of a
very badly designed model and not a question of LAMMPS. LAMMPS follows the
GI-GO principle (garbage in, garbage out) and no "bugfix" will make a badly
designed simulation produce correct results. like i already mentioned, look
at AIREBO and how it is designed and implemented. that is the way how you
can make this work. you *do* need some kind of switching function that
smoothly transitions from using one kind of interaction for short range to
another type of interaction for long range. to do this well with a
many-body potential is non-trivial. what you are doing is definitely bogus.
And one more thing I want to know, does COMB3 potential include LJ
interaction?
i don't know. how about reading the papers that describe it? too obvious??
axel.