Dear LAMMPS users
I have a simulation which contains methane molecules. What kind of potential should I use to simulate covalent bonds? I also want this bonds to be breakable (e.g. in high temperatures) and then to be restored somehow to make other hydrocarbon compounds.
Many thanks for any advice
Konstantin
Dear LAMMPS users
I have a simulation which contains methane molecules. What kind of
potential should I use to simulate covalent bonds? I also want this
bonds to be breakable (e.g. in high temperatures) and then to be
restored somehow to make other hydrocarbon compounds.
konstantin,
that sound very much like you would need to look for
an ab initio MD code. i don't think lammps would be
a good choice here, as classical potentials usually are
not transferable across temperature ranges, not to mention
the few of them support breaking and forming bonds.
cheers,
axel.
Axel,
I believe AIREBO ["A reactive potential for hydrocarbons with
intermolecular interactions", J. Chem. Phys. 112, 6472 (2000)]. It has
been implemented in LAMMPS (I have not used LAMMPS for it though) and
is an improvement to the Brenner Potential for hydrocarbons.
I also think the recently added ReaxFF potentials in LAMMPS can
simulate covalent Bond breaking and recombination. I am not sure if it
has parameters for simulating hydrocarbons?
Any reason for discounting the above two potentials?
Best Regards,
Manoj
Axel,
I believe AIREBO ["A reactive potential for hydrocarbons with
intermolecular interactions", J. Chem. Phys. 112, 6472 (2000)]. It has
been implemented in LAMMPS (I have not used LAMMPS for it though) and
is an improvement to the Brenner Potential for hydrocarbons.I also think the recently added ReaxFF potentials in LAMMPS can
simulate covalent Bond breaking and recombination. I am not sure if it
has parameters for simulating hydrocarbons?Any reason for discounting the above two potentials?
manoj,
well, it depends on what you are looking for. you have to
use any of these potentials for kind of systems they are
parameterized for, and that usually limits the kind of
reactions that you can see or the temperature range that
you can use. it worries me occasionally, how many people
pay little attention to these boundary conditions and just
run simulations with whatever looks convenient.
the OP mentioned high temperatures and reactions and that
spells for me (intermediate) formation of radicals, complex
chemical reactions, formation of h2 and unsaturated bonds
etc. pp. if my guess is correct, then you _have_ to use
an ab initio MD code (although then you have to deal with
corrections for the frequently incomplete handling of
non-bonded dispersion interactions).
it is my impression that tersoff-type potentials are generally
used for systems where the type of bond doesn't change (or change
much), e.g. crack propagation in solids, stability of CNTs.
i happily stand corrected, if my impression is wrong.
i have been involved a lot with DFT based AIMD codes and
particularly QM/MM interfaces, so i am likely to be a bit
biased towards that, but probably also bit more sensitive
to where classical potentials have limitations and AIMD
has its strengths (with the immensely increased computational
cost of AIMD methods over classical potentials, one has to
always think carefully about when to use it or not).
it is probably best to ask the OP to provide more details
about the system in questions and then have a closer look
at what would work or what would not. i wouldn't want to
completely exclude either approach.
cheers,
axel.
Axel,
I agree with your comments on usage of classical potentials. Ab-Initio
is still a black art (box) for me though I tune in to the Quantum
Espresso mail-list. I wanted an expert's view on the matter. I also
felt that the exact problem was not fully stated and your answer could
drive away the uninitiated from AIREBO / ReaxFF.
One question remains 
-> What is OP (my guess - Other Person)?
Manoj
Axel,
manoj,
I agree with your comments on usage of classical potentials. Ab-Initio
is still a black art (box) for me though I tune in to the Quantum
Espresso mail-list. I wanted an expert's view on the matter. I also
asking an "expert" is not always helpful. remember the "if your only
tool is a hammer, everything looks like a nail" proverb. if your
expert is more comfortable with classical MD, you are likely to get
a recommendation leaning towards that and vice versa for somebody with
lots of experience in both. i have been spending significant time
using either method (but not using many body potentials, hence my
reservations 
) and it is not always easy to make a choice between
those methods (considering that you can do AIMD on various "levels
of theory", e.g. tight-binding, semi-empirical, hartree-fock, DFT
and hybrid QM/MM).
felt that the exact problem was not fully stated and your answer could
drive away the uninitiated from AIREBO / ReaxFF.
that argument works both ways: if you are ignorant about the
limitations of these potentials you can make just as bad a choice.
in my opinion, good force field parametrization is as much
a black art as, say, pseudopotential generation for plane
wave based DFT calculation.
because of the black art status, many people shy away from
the effort of really looking into it to properly understand it.
and thus they are - again, in my personal opinion - taking a
needlessly high risk of doing complete nonsensical calculations.
One question remains
OP = original poster.
i'm getting old and lazy and forget about that fact,
that usenet slang may not be understood so well.
sorry about that.
cheers,
axel.