Sir
I am working to simulate heating ,melting, vapourizing and ionization of nano gold wire. For this I want to apply voltage across the wire.
Please tell me which commands will help me.
Thank you.
Sir
I am working to simulate heating ,melting, vapourizing and ionization of
nano gold wire. For this I want to apply voltage across the wire.
Please tell me which commands will help me.
there are none. not as such. all the commands and
capability of lammps are documented on the lammps
homepage. you first need to come up with an idea
how you want to represent your system with the
features available.
axel.
Hi,
Ammm if the heating or melting is caused by a current (motion of
electrons due an applying voltage), then you must consider the
electronic properties of the material...
then you will have to use a field that include electron interaction or
ab-into Molecular dynamics i.e calculate the electrons interaction on
the fly ....That's all i can tell you, hopefuly it will help.
post data: I belive lammps its not capable of running that kind of
simulations, if someone knows how , please let me know
Oscar G.
Hi,
Ammm if the heating or melting is caused by a current (motion of
electrons due an applying voltage), then you must consider the
electronic properties of the material...
then you will have to use a field that include electron interaction or
ab-into Molecular dynamics i.e calculate the electrons interaction on
the fly ....That's all i can tell you, hopefuly it will help.
regular ab initio MD won't help for that, since it applies
the born-oppenheimer approximation, i.e. that the atom
cores move in a classical approximation and are decoupled
from the motion of the electrons. also, ab initio MD cannot
cover reasonable time scales for the mentioned processes.
in general, it is a mistake to take some experimental
*macroscopic* setup and expect to be able to run it
immediately run it in a simulation program. any simulation
contains some approximations, so one first has to consider:
- what is it exactly that is to be learned from the simulations?
- which approximations can be made without impacting these
goals (significantly)?
- does every "experimental feature" have to be represented
directly or would it be sufficient to just model the effect?
- what are the limitations of the various simulation methods
(continuum, atomistic, quantum mechanic) and how do
the collide with the requirements for the goals of the simulation?
for example. it would be possible to approximate the heat
transferred from the current passing through the gold by
figuring out how much kinetic energy per time is transferred
and then use something like fix heat (perhaps in combination
with fix ttm) or do a combination with a continuum model
via fix atc. but if the desire is to study exactly this amount
of energy transfer (and not have it as an input parameter),
then this would not be an adequate model. and so on for
other specifics of the "experiment".
cheers,
axel.
Yup Axel is right...And thanks for the clarification ... Beucase i
just mess it up badly (no the first time). In Ab Into Molecular
dynamics the electrons are not well localized since they are treathed
as "wave functions" . We treat the wave functions decoupled from the
motion of the nuclei then we treat the resulting charge distribution,
and potentials and forces as classical for the nuclei ...... I had to
read wikipedia three times to finally understand it the concept .....
Still much to learn .....
Oscar G.
Yup .... the current or flow of electrons can not be treathed as a
"wave function"
or can be ? .
Oscar G ....
Yup .... the current or flow of electrons can not be treathed as a
"wave function"
or can be ? .
you could use ehrenfest molecular dynamics to study the interaction
of electrons with nuclei and thus also the impact of "stirring up" those
electrons through a current. but the time step for such a simulation
would be extremely short (since you have to propagate the electrons
and cannot make them artificially heavy as in Car-Parrinello MD).
like i was saying before. one has to decide what to model and how.
with ehrenfest MD, you may be able to determine the amount of
energy being transferred through a current, but you cannot run a
long trajectory.
with classical MD it is possible to run a long enough trajectory
to model melting and evaporation, but the heat transfer from the
current and the heat conduction through the electrons have to
be treated as input parameters.
good simulations require good planning and some time
and effort to study the underlying principles and what the
simulation methods can provide and cannot.
expecting to just do a 1:1 mapping of an experiment is
a mistake in two ways: 1) most experiments are on
inaccessible time scales, 2) a lot of things that need to
be done in experiment, are not really required with when
doing a smart combination of simulations. in a simulation,
unlike in an experiment, one has perfect control over what
is done to a system.
cheers,
axel.
Sir
Thanks for replies
I am pasting my code related with " voltage across gold wire".
But I am confused about using “set region wire charge 196” because I think that it will add extra 196 electron per atoms.
Please check my code and advice me. It will be a great help.
#gold wire
units metal
dimension 3
boundary p p pprocessors
atom_style charge
neighbor 0.3 bin
neigh_modify delay 5create geometry
lattice fcc 4.0788
region wire cylinder x 0 0 10 0 200 units box
create_box 1 wire
create_atoms 1 region wire
mass 1 196group wi region wire
set region wire charge 79
#pairing
pair_style eam
pair_coeff * * ./Au_u3.eamregion 1 cylinder x 0 0 10 0 20 units box
group lef region 1
region 3 cylinder x 0 0 10 20 180 units box
group mid region 3
region 2 cylinder x 0 0 10 180 200 units box
group rig region 2minimize 1.0e-4 1.0e-6 100 1000
#velocity all create 300.0 4928459 units box
fix k1 wi nve
fix kick1 lef efield 100000000000.0 0.0 0.0
fix kick2 rig efield 0.0 0.0 0.0Run
timestep 0.003
thermo 50
thermo_style custom step temp pe etotal press vol lx ly lz
thermo_modify lost ignore flush yesdump dum wi atom 50 dump.gold13
run 4000
Thanks
Manish Anand
Sir
Thanks for repliesI am pasting my code related with " voltage across gold wire".
But I am confused about using "set region wire charge 196" because I think
that it will add extra 196 electron per atoms.
yes it will. that is what the documentation says.
Please check my code and advice me. It will be a great help.
your input is make no sense at all. you are doing classical
MD on a model that does not consider electrons explicitly.
see the previous emails in this discussion thread.
axel.
Yup Axel is right ... You cannot add a "charge" (electrons) to the
systems and then assume an electric field will make the electrons move
(since in classical MD electrons are not consider explicity) ....
Nevertheless the "charge" is added to the atoms, therefore the
electric field should exert a force (F = qE ), but thats equivalent
to add a kinetic energy to the system that will make the atoms to
move. I'm sure there must be a relation between the applied field (E)
and the speed (kinetic energy) of the atoms .......(thinking) , but
right know im not able to grasp it I need to think about it ....
Oscar G.
I think …
If there is an applied field (E), there exists an electric potential V such thatE = - grad(V)
Then the energy injected by this V into an atom with charge q is qV; this energy can then be distributed among the potential energy and the kinetic energy of the atom.
Cy.
@Cy yup that make sence.....
So if we assume the "voltage" and "charge" as known input parameters
then we can calculate the potential energy , this energy can Added
into the system as heat ... Let the MD simulation run, wait , drink a
beer, and see what happens ....
I think i'm just repeating what axel said previously : "fix heat
(perhaps in combination
with fix ttm)", in any case this thread has been a very fruitful discussion ....
Oscar G.
I think ...
If there is an applied field (E), there exists an electric potential V such
that
E = - grad(V)Then the energy injected by this V into an atom with charge q is qV; this
energy can then be distributed among the potential energy and the kinetic
energy of the atom.
this is nonsense. a metal is a conductor.
the conduction comes from the electronic structure,
not the motion of ions. metal atoms are not charged.
...and the heat is generated due the intrinsic resistance
of the metal.
axel.
Yup Axel is right again ... Using "charge" and Efield (F=qE) in
lammps is equivalent to adding a "push" i.e a force to the atoms (fix
addforce ) . The motion of the atoms (ions) is non related to the
motion of electrons ...I was thinking something like The force or push
must be somhow related to the kinetic energy which produce heat ...
much still to learn ....
Oscar G,