Questions with water polarization under electric field

Dear all,

I have some questions with the simulation of liquid water alignment and rotation under electric field. I notice that in previous papers, there are mainly two kinds of methods to polarize water, one is the fluctuating charge model by Rick, Stuart and Berne, the other is the dipole polarization model by Sprik ( a polarizable model for water using distributed charge sites). I notice in LAMMPS, there is a charge equilibrium fix named “fix qeq” which is similar to Rick’s method. I wonder if I could use this one with fix efield to achieve my goal. Could the external electric field be counted to the charge equilibrium? By the way, should I use nvt/sphere instead of nvt alone? Thank you!

Han

I don’t know what you meant by “external electric field be counted to the charge equilibrium”.

Note that fix efield adds force on atoms that is proportional to their charges, but it does not change the charge’s equation of motion hence does not change the charges.

Using a combination of efield and qeq, the charged particles should experience an additional force, move accordingly, then obtain new charges based on the new surroundings.

Ray

Dear all,

I have some questions with the simulation of liquid water alignment and
rotation under electric field. I notice that in previous papers, there are
mainly two kinds of methods to polarize water, one is the fluctuating charge
model by Rick, Stuart and Berne, the other is the dipole polarization model
by Sprik ( a polarizable model for water using distributed charge sites). I

there are more (core shell, drude oscillator, point dipole
polarization) and they can be solved self-consistently or with an
extended lagrangian scheme.

notice in LAMMPS, there is a charge equilibrium fix named "fix qeq" which is
similar to Rick's method. I wonder if I could use this one with fix efield
to achieve my goal.

fix efield is an extremely crude model of an external field that just
adds a force to each point based on its charge. this neglects
screening and polarization effects. combining this with a polarizable
water potential doesn't make much sense to me. a more realistic model
would be to place point charges into a suitably enlarged system and
keep them apart with a confinement potential. i know some people here
at temple and their collaborators are doing simulations of voltage
gated ion channels this way (but with conventional water potentials).

https://icms.cst.temple.edu/biomedical.html#voltagegated

Could the external electric field be counted to the
charge equilibrium?

that would require modifications to the code as fix efield simply adds
a force to atoms, but does not contribute to the electrostatic
potential.

By the way, should I use nvt/sphere instead of nvt
alone? Thank you!

no, why?

Thanks, Axel. I hope to confirm if my understanding is correct. The combination of fix efield and fix qeq will only makes the particles move with e-field and other force first and then adjust the charge distribution by the new atoms positions. Thus the polarization due to external e-field is not counted. Will this be fixed if I modify the code of qeq to include the influence of external electric field?

Thanks, Axel. I hope to confirm if my understanding is correct. The
combination of fix efield and fix qeq will only makes the particles move
with e-field and other force first and then adjust the charge distribution
by the new atoms positions. Thus the polarization due to external e-field is
not counted. Will this be fixed if I modify the code of qeq to include the
influence of external electric field?

if you modify the QEQ code, it will do whatever you modify the code to do.

whether this will result in a physically meaningful model is a
different question and one that different people will have different
opinions on. as i already mentioned, i would personally consider this
combination as bogus and not worth the effort.

before even going so far that you consider changing the QEQ code, i
would suggest you check out how well the description of your water is
*without* an external field using a regular water potential and fix
QEQ. i doubt it will be very good without additional tweaking and
tuning. people have tried to parameterize and fine tune polarizable
water potentials for over 20 years and progress is still very slow.
the improvements are usually not considered worth the increased
computational effort (otherwise people would be doing this all over
the place because several widely used MD codes do support polarizable
models and force fields for quite some time).

axel.