Reaxff coupled to an electric field

Hi all,

It seems to me that in current reaxff implementation, the charge
equilibration process doesn't consider the polarization caused by an
external electric field. I'm wondering if this will be incorporated in
LAMMPS in future?

Thanks!

Hai

Hi all,

It seems to me that in current reaxff implementation, the charge
equilibration process doesn't consider the polarization caused by an
external electric field. I'm wondering if this will be incorporated in
LAMMPS in future?

this is most likely, if you would contribute
the code yourself. :wink:
many of the features in LAMMPS only exist
because somebody needed it and implemented
it. particularly for more "exotic" features.

cheers,
    axel.

I'll let Aidan and Ray comment on this idea
and what work would be involved.

Steve

I think it would not be hard to add a uniform electric field term to the
charge equilibration energy. As usual, it would be helpful to define a
representative calculation, including expected inputs and outputs.

Aidan

aidan,

I think it would not be hard to add a uniform electric field term to the
charge equilibration energy. As usual, it would be helpful to define a
representative calculation, including expected inputs and outputs.

i am overall a bit worried by the somewhat restricted
realism of this approach.

several of my colleagues in the ICMS work on voltage gated ion
channels (= proteins embedded in lipid bilayers that can change
conformation based on the drop of the electrostatic potential), and
using a uniform electric field significantly messes up the results.
instead they use an enforced charge separation (in their case
simply by inserting a vacuum in the water/electrolyte and just
swapping anions and cations between the two water/electrolyte
segments until the desired electrostatic potential drop is reached).

i wonder if something similar would be meaningful with reaxff?
particularly, since there is no long-range electrostatics...

just a thought,
      axel.

To add polarization, one can implement a "compute" that calculates the polarization vector based on the fluctuating charge, fluctuating dipole model proposed by HA Sterne et. al., JPCB 1999.

In their model, the vector is calculated from the electric field generated by the atomic charges and the neighboring induced dipoles. In this "compute", one can also add the effect of the external electric field.

This compute should require some work.

Best,
Ray

Hi Axel,

If the material is spatially uniform on some lengthscale greater than
atomic, the exteranl field will also be uniform on that lengthscale, and
so the uniform field assumption will probably be adequate. I agree that
for ion channels, this is not the case.

I also agree that imposing a potential difference by adding and removing
charges at both ends is more physically realistic than a uniform field. It
requires no modification of the charge equilibration scheme. Instead the
burden is shifted to constructing the two charged reservoirs. It is
important to note that the charge equilibration scheme in LAMMPS/ReaxFF
enforces charge conservation globally on the entire cell, which
corresponds to infinite charge conductivity. So charge separation can only
occur due to separation of chemical environments. There is a modified
ReaxFF charge equilibration model where charge is conserved locally on
each chemical entitye e.g. water moelcule or sodium cation, but defining
molecule-based constraints in LAMMPS is non-trivial.

Aidan