Hello professor Steve, Axel and all other lammps users
I have a question regarding calculation of hydrogen bonding in lammps using
: pair_style hbond/dreiding/lj command
i have read the documentation carefully for many times. using this command
for hbond in lammps, we need two cutoff distances (global inner spline and
global cutoff for donor-acceptor interactions)
(Interesting. I did not notice this earlier. Why do the docs call it
a "spline cutoff"?)
and one global angle cutoff
for acceptor-hydrogen-donor interactions.
As all we know, hydrogen bond is usually defined via geometric criteria for
example: the A-D distance <3.5 angstrom, and A-H-D angle >120.
I would like to know how we can apply this definition in above lammps
command, are these two geometric values the same as global cutoff for A-D
interactions and global angle cutoff for A-H-D interactions? if so, what is
the value of global inner spline cutoff?
Any help is highly appreciated,
thanks in advance
Are you asking why we recommend using a 9.0 - 11.0 cutoff distance
(when hydrogen bonds are < 3.5 Angstroms)?
Actually, in the the original Mayo JPC 1990 paper they did not use any
cutoff at all. (This is equivalent to an infinite cutoff.) However
on page 8903 of that paper (footnote, left-hand column), they
recommend using a 9.0 Angstrom cutoff. That's why I use 9.0 Angstroms
for the inner cutoff. They recommended it in their paper.
Journal of Physical Chemistry, Vol. 94, No. 26, 1990, 8897
The outer cutoff (11.0 in the example online) does not matter as long
as it is at least 0.5 Angstroms greater than the inner cutoff. The
inner and outer cutoff distances are usually very similar to each
other.
--- discussion ---
You are correct that hydrogen-bond distances are usually much smaller:
on the order of 2.5-4.0 Angstroms. Beyond this, the interaction
between the donor and acceptor atoms weakens rapidly. But you should
not confuse this with the cutoff distance(s).
Keep in mind that the "Morse(r)" and "LJ(r)" functions never decay to
zero at any radius. (See web page link:)
http://lammps.sandia.gov/doc/pair_hbond_dreiding.html
If we did not use cutoffs, then LAMMPS would have to spend time
evaluating the Morse(r) or LJ(r) functions for every possible pair of
donor and accpetor atoms in the entire system.
As you know, there is no reason to wait for LAMMPS to spending time
evaluating the Morse(r) function when the distance between the donor
and acceptor atoms is extremely large. So the cutoff distance(s) you
are asking about are supposed to be "extremely large". Beyond the
outer cutoff distance, the interaction energy between the donor and
acceptor atom is negligible. This does not occur at 3.5 Angstroms (at
least, not if you are using pair_style hbond/dreiding).
You have probably seen this before if you have run other simulations
in the past. For example, the cutoff distance(s) used in the
van-der-Walls (Lennard-Jones) interaction is usually much larger than
"sigma". (Recall that "sigma" is the sum of the atom radii, which is
also near the separation distance where they are most stable).
I hope my explanation was clear.
Cheers!
Andrew