playing with dielectric constant

Hi lammps-users,
I am trying an experiment where dielectric permittivity effects are provided simultaneously from two sources: (1) molecular solvent and (2) background dielectric constant. Let’s say, for the sake of discussion, I have a system of salt water. However, the “water” is a weakened solvent (i.e., the dipole is not as strong as normal water). I would like the “effective permittivity” to still be something like water (epsilon_r = ~80). So the idea is if the solvent intrinsically provides a dielectric constant of, say epsilon_molecule = 10, I want to “boost” the overall dielectric constant up to 80 by using the dielectric command (maybe “dielectric 8”?).
However, the hypothetical value 10 was determined previously in LAMMPS (pure solvent system with a background electric field), given the command: “dielectric 1.0”. Now in my “salt water” system, adjusting the LAMMPS dielectric constant for the boost will naturally change the molecular dielectric constant that I had measured.
In actuality, I believe what I’m looking for is a way to selectively reduce only ion-ion Coulombics by a certain factor, but leave ion-solvent and solvent-solvent interactions at full strength (dielectric 1.0). To my knowledge, this is not implemented in LAMMPS. Does anyone have any suggestions for how to do this? Keep in mind the long-range Coulombics must be handled consistently as well.
I understand this setup is rather unorthodox, so please let me know if it requires further explanation. Thanks!

Jonathan

Hi lammps-users,
   I am trying an experiment where dielectric permittivity effects are
provided simultaneously from two sources: (1) molecular solvent and (2)
background dielectric constant. Let’s say, for the sake of discussion, I
have a system of salt water. However, the “water” is a weakened solvent
(i.e., the dipole is not as strong as normal water). I would like the
“effective permittivity” to still be something like water (epsilon_r = ~80).
So the idea is if the solvent intrinsically provides a dielectric constant
of, say epsilon_molecule = 10, I want to “boost” the overall dielectric
constant up to 80 by using the dielectric command (maybe “dielectric 8”?).
   However, the hypothetical value 10 was determined previously in LAMMPS
(pure solvent system with a background electric field), given the command:
“dielectric 1.0”. Now in my “salt water” system, adjusting the LAMMPS
dielectric constant for the boost will naturally change the molecular
dielectric constant that I had measured.
   In actuality, I believe what I’m looking for is a way to selectively
reduce only ion-ion Coulombics by a certain factor, but leave ion-solvent
and solvent-solvent interactions at full strength (dielectric 1.0). To my
knowledge, this is not implemented in LAMMPS. Does anyone have any
suggestions for how to do this? Keep in mind the long-range Coulombics must
be handled consistently as well.
   I understand this setup is rather unorthodox, so please let me know if it
requires further explanation. Thanks!

i don't think that modifying the dielectric constant as such would be
a good way to model the system that you are describing. i believe the
key is in the phrase "selectively weaken ion-ion coulombics". from my
experiences with dielectric properties, i don't think that this is
going to be a step function, but rather a smooth function, so it could
be modeled by using hybrid/overlay. please have a look at the
coul/diel pair style that is meant to be used in a similar fashion for
implicit solvent simulations. there are some references, too.

http://lammps.sandia.gov/doc/pair_coul_diel.html

axel.

Interesting, thanks Axel. I wasn't aware of this pair style. I can probably use this for what I'm trying to do.

Jonathan

Hi Axel,
   A quick follow-up question. If I use "coul/diel" for one of my pair interactions (one out of several, for example), should the epsilon value be consistent with the value I give to the "dielectric" command? I hope that LAMMPS will not doubly-apply the dielectric constant in that case (specifically for the PPPM solver). Thanks!

Jonathan

Hi Axel,
   A quick follow-up question. If I use "coul/diel" for one of my pair interactions (one out of several, for example), should the epsilon value be consistent with the value I give to the "dielectric" command? I hope that LAMMPS will not doubly-apply the dielectric constant in that case (specifically for the PPPM solver). Thanks!

please carefully read the papers that are referenced in the docs. the
dielectric constant you provide in the pair style is independent of
the global dielectric constant (which is essentially a global scaling
factor for all coulomb interactions and you could achieve the exact
same effect by just setting the per atom charges to correspondingly
scaled values, which is what i would personally prefer to do to avoid
any consistency issues).
this sets the correction for how coulomb interactions need to be
screened differently depending on their distance from each other.

axel.