I would like to simulate the system of charged particles in the external electric field, and measure the electric current.
Thus I need to consider about the velocities of the particles along the direction of the electric field.
Now I use the fix nvt, that is Nose-Hoover thermostat for the system. I want to remove the thermostat on the direction which is parallel to the electric field (for example z direction), which means that apply the thermostat only on the direction perpendicular to the electric field (x and y), because I think that the thermostat could rescale the velocities, and it will influence the calculation of the electric current.
Is compute temp/partial or compute temp/profile for the option?
I would like to simulate the system of charged particles in the external
electric field, and measure the electric current.
Thus I need to consider about the velocities of the particles along the
direction of the electric field.
Now I use the fix nvt, that is Nose-Hoover thermostat for the system. I want
to remove the thermostat on the direction which is parallel to the electric
field (for example z direction), which means that apply the thermostat only
on the direction perpendicular to the electric field (x and y), because I
think that the thermostat could rescale the velocities, and it will
influence the calculation of the electric current.
Is compute temp/partial or compute temp/profile for the option?
compute temp/partial is what you are looking for.
however, i suspect there are a ton of problems with your approach in general.
have you found any precedent in the published literature yielding good
results for a similar system, that you are following?
thank you for the answer.
The similar approach was found in a paper published on Nano Letter, 2010, 10
(10), pp 4067–4073,
but it's the non-equilibrium simulation with constant acceleration, not with
electric field.
I haven't found other published work with the approach.
probably for good reasons.
What do you mean about the problems?
this is just off the top of my head:
- electric current often has multiple pathways, not just classical
transport of charged particles, which you want to simulate.
- if you are working with free time integration in z-direction in
periodic boundaries, you need to find a suitable way to remove the
added translational kinetic energy after particles have passed through
the cell. or you'd have to have some kind of reservoir, which limits
the simulation time, or would need a complex process to be refilled
without impacting the results.
- fix efield is a very crude model for an external field and does not
consider screening effects from the presence of charged particles.
- macroscopically meaningful field strengths cause effects that are
tiny compared to typical local fluctuations of the atomic length
scale. you need to use a rather high field and then have to worry
about the physical validity of the measured effects. i would expect
that the amount (and dominant pathway) of charge transport is not
linear with the field strength.
yes the fix efield doesn't include the screening effects, I am also
considering the problem.
I agree with that I have to use a strong electric field for the simulation.
But why do I need to remove the translational energy, do you consider the
conservation of energy?