# boundary condition

Dear LAMMPS team and all,

Happy New year 2012.

I have a question:

I want to simulate a metal and measure its melting point.
I have done this with PBC boundary condition ( boundary p p p) and get a result.

Do you know what is the command script if I want to simulate a metal without PBC boundary condition?
What is the fix command to measure the melting point of this system?

That's all. Thank you very much for explanation.

Best regards

Arkundato
PhD student

The doc page for the boundary command tells you
how to turn off PBC. There is no fix to measure a melting
point. You have to monitor something like an order
parameter (e.g. see the compute centro/atom command)
and deduce from that when the system melts.

Steve

Dear LAMMPS team and all,

Happy New year 2012.

I have a question:

I want to simulate a metal and measure its melting point.
I have done this with PBC boundary condition ( boundary p p p) and get a result.

Do you know what is the command script if I want to simulate a metal without PBC boundary condition?
What is the fix command to measure the melting point of this system?

how would detecting the melting point
in a finite system be difference in a
system with periodic boundaries?

axel.

This is an interesting question. The absence of periodic boundaries would of course induce a free surface. The free surface will likely have elastic properties that are different than those of a bulk material (simulated using periodic boundaries). The melt temperature might not be the same if the modified surface properties are included into the simulation.

Mike Bifano
Ph.D. Candidate
Case Western Reserve University
Office (216) 368-6444
Cell (717) 926-8950

" The absence of periodic boundaries would of course induce a free surface."

@Mike Further explanation need it please.

To find the melting point of a system , I would use the following (not
sure is correct) : fix the pressure e.g 0 kbar , increase gradually
the tempeature (thermostat), The volume of the system will expand, in
order to keep the pressure to the preset value, close the phase
transition a drop in the energy (E) must occour. Another more precise
method would be to identify the melting temperature (Tm) using the
previous method , and then run a simulation in solid-liquid phase (the
change of one phase to another happens gradually) and find the change
(delta) in the themordynamics potential a phase transition occours
when delta = 0 .

Hi All ...... =>
Oscar G.

This is an interesting question. The absence of periodic boundaries would
of course induce a free surface. The free surface will likely have elastic
properties that are different than those of a bulk material (simulated using
periodic boundaries). The melt temperature might not be the same if the
modified surface properties are included into the simulation.

why should that be? ...and what do you mean by "including
the surface properties"?

the way i see it, at best the surface atoms would be affected
due to disorder, however, as soon as you have a regular crystal
structure in contact with molten atoms, you should get the same
melting behavior as in bulk.

for the most part i would expect that you have less of a
hysteresis in melting/freezing, since you have an already
broken up structure.

cheers,
axel.

Maybe he was referring to the Gibbs-Thomson effect:

http://en.wikipedia.org/wiki/Gibbs–Thomson_equation

Typically a nanoparticle will melt at a (much) lower temperature than
the bulk material you get with pbc.

Best,
Laurent

hmm.... you are neglecting the fact that melting is not a plain
thermodynamic process, but has a kinetic component too.
just plain raising the temperature will likely result in too high
a temperature and how much depend on the sample size and
how fast you are raising the temperature.

as far as i recall, the best way to determine a melting
temperature in MD simulations, it to set up a system
that is half liquid, half solid and then observe which way the
liquid/solid front is moving while adjusting the temperature.
the temperature, where this front is stationary, would be the
melting temperature.

cheers,
axel.

Oscar,

If periodic boundaries are activated in all directions, you don’t have any free surfaces.

“modified surface properties ‘included’ in the simulation” is a poorly phrased statement. I was only contemplating the disorder at the surface and how it might affect the melt temperature. I know work has been done to show that the elastic properties of surfaces vary with atomic coordination number. Seems logical that something would happen to the thermodynamic properties as well.

Observing the movement of a solid liquid front in a half solid/half liquid substance would remove the free surface.

Mike