[lammps-users] questions on NPT for 2 dimesional system

A graphene will be corrugated when it is suspended fully and I’d like to obtain its equilibrium structure using NPT.
I think I should impose periodic condition to two direction on the plane and, leaving the dirction normal to plane non-periodic.

  1. To apply pressure control without that of z direction (thus, variation of box length in that direction), should I use
    “fix 1 all npt temp 300.0 300.0 0.5 x 1. 1. 5. y 1. 1. 5. z NULL NULL 5. couple none”?
    Otherwise, what should I do for it?

  2. If we don’t impose peridic condtion to z-direction (e.g., by writing “boundary p p s”),
    should it be considered as no pressure control to that direction?
    That is, even if we write “fix 1 all npt 300. 300. 0.5 aniso 1. 1. 5.”, should it be considered in a same way with question 1 automatically?

  3. Incidentally asking, it semms that “s” and “m” in boundary keyword give the same physical condition? Is it right?
    Then, in what case, will the difference between s and m appear?
    I think the difference between them should be found when we obtain relative coordinate to box length to that direction though period condition is not imposed to that direction. Is it right?

Best regards,

Luke

2010/7/25 JhonY. I. <[email protected]...>:

A graphene will be corrugated when it is suspended fully and I'd like to

suspended in what?

obtain its equilibrium structure using NPT.
I think I should impose periodic condition to two direction on the plane
and, leaving the dirction normal to plane non-periodic.

what are the best boundary conditions depends not only on
the (perceived) periodicity of the system, but also on what kind
of properties you want to describe best and what your intended
system size is going to be.

for a large enough graphene sheet in vaccum, i would not use
any peridocity at all. for a small graphene flake in bulk water i
would have periodic boundaries in all directions and in other
cases, it might be better to do with 2d periodicity.

you have to tell us more about what you want to achieve, to
get proper advice on that. keep in mind that 2d-periodic boundaries
can impose periodicity constraints on your system that may
not be consistent with your material.

1. To apply pressure control without that of z direction (thus, variation of
box length in that direction), should I use
"fix 1 all npt temp 300.0 300.0 0.5 x 1. 1. 5. y 1. 1. 5. z NULL NULL 5.
couple none"?
Otherwise, what should I do for it?

you can only do npt ensemble on periodic dimensions. but you
could run the same setup in 3d periodic but have a sufficient vaccum
area to get the same system.

2. If we don't impose peridic condtion to z-direction (e.g., by writing
"boundary p p s"),
should it be considered as no pressure control to that direction?

if you have no periodic boundaries in one dimensions, you indicate
that you model the complete extent of your system in that direction.
with that comes that you don't need to manipulate the system in that
direction. periodic boundaries are used to eliminate surface effects.

That is, even if we write "fix 1 all npt 300. 300. 0.5 aniso 1. 1. 5.",
should it be considered in a same way with question 1 automatically?

the fix npt documentation is very detailed about what works how, when
and what your input would have to look like.

3. Incidentally asking, it semms that "s" and "m" in boundary keyword give
the same physical condition? Is it right?

more or less, yes. with m you guarantee a minimum size of the system
in that direction.

Then, in what case, will the difference between s and m appear?
I think the difference between them should be found when we obtain relative
coordinate to box length to that direction though period condition is not
imposed to that direction. Is it right?

i don't understand what you want to say here.

cheers,
    axel

Many thanks for your answer really, Alex.

  1. My system for research is as follows.

A graphene layer is flatten in graphite.
However, if it is isolated, it is corrugated due to two-dimesional characteristics.
I’d like to have equilibrium structure of this corrugated form of graphene, starting from flatten structure that can be derived from graphite structure, using NPT
simulation of LAMMPS.

I think of imposing large space between graphene sheets by assigning large length to the direction perpendicular to graphene plane in periodic box.
It will be all right if I run NVT simulation.
But I think that two graphenes in periodic environmental condition might bind to each other if there is a chance that they approach to each other closely since aggregated form will be more stable.
I suppose that they persist in isolated forms if kept apart from each other sufficiently during NPT simulation, and that may be possible if we control pressure carefully.

What do you tihink on it?
What’s the best way to perform such as task?
Shouldn’t we use a method of non-periodic condition in such a system?

  1. I’d like to know in what cases minimum box length should be needed regarding to “m and s option in boundary keyword”.
    That is, why do we have to devide m and s conditions and in what cases, m and s are needed respectively?

Best regards,

Luke

2010/7/27 JhonY. I. <[email protected]...>:

Many thanks for your answer really, Alex.

1. My system for research is as follows.

A graphene layer is flatten in graphite.
However, if it is isolated, it is corrugated due to two-dimesional
characteristics.

i've ran simulations of that kind of system before
myself and saw it corrugate

I'd like to have equilibrium structure of this corrugated form of graphene,
starting from flatten structure that can be derived from graphite structure,
using NPT
simulation of LAMMPS.

there is no reason to do an NPT ensemble simulation,
if you have an isolated system. the graphite structure
is easily constructed from experimental data and the
lattice and create_atoms commands, or a text editor to
write a data file and then the read_data and the replicate
commands.

I think of imposing large space between graphene sheets by assigning large
length to the direction perpendicular to graphene plane in periodic box.
It will be all right if I run NVT simulation.

all you need is to have the distance between the sheets to be
a bit larger than the interaction cutoff. if you want to be absolutely
certain to keep the two sheets apart then define each as a group
and then use the fix spring command. you will only need a small
spring constant, as there is no reason for them to move, if you
make sure they have no initial momentum.

But I think that two graphenes in periodic environmental condition might
bind to each other if there is a chance that they approach to each other
closely since aggregated form will be more stable.

I suppose that they persist in isolated forms if kept apart from each other
sufficiently during NPT simulation, and that may be possible if we
control pressure carefully.

no pressure control will help you here, but none is needed.
for as long as there is no net momentum on each sheet, e.g.
by collision with some other atom, the sheets will just stay
where they are. you can run for a very long time, until the
numerical noise will build some.

What do you tihink on it?
What's the best way to perform such as task?
Shouldn't we use a method of non-periodic condition in such a system?

i don't think it matters much, if you have enough safety margin.

2. I'd like to know in what cases minimum box length should be needed
regarding to "m and s option in boundary keyword".
That is, why do we have to devide m and s conditions and in what cases, m
and s are needed respectively?

not sure about this. s is just a special case of m.

cheers,
     axel.

Since you are trying to simulate a 2D sheet with no long-range electrostatics, you probably want periodic boundary conditions in the directions parallel with the sheet and a non-periodic boundary condition in the direction perpendicular to the sheet. That way, you don¡¯t have periodically-repeating images of your sheet in the perpendicular direction, and hence no chance of sheet-sheet interactions.

Whether to use shrink-wrapped (¡°s¡±), or shrink-wrapped with a minimum value (¡°m¡±), depends on whether or not you really need to maintain at least a minimum box length in the direction perpendicular to the sheet. I¡¯d presume you don¡¯t need that, so ¡°s¡± would be the better option. For more details, see: http://lammps.sandia.gov/doc/boundary.html

Whether or not to use NPT also depends on what your ultimate goal is here, but note that you can¡¯t use NPT in the direction perpendicular to the sheet if you¡¯re using a non-periodic boundary condition in that direction. I¡¯d certainly recommend against using NPT, even in the other two directions, unless you really need it. From the information you¡¯ve given, I think you¡¯d be much better off with NVT.

Paul

Comments below.

Steve

2010/7/25 JhonY. I. <[email protected]...>:

A graphene will be corrugated when it is suspended fully and I'd like to
obtain its equilibrium structure using NPT.
I think I should impose periodic condition to two direction on the plane
and, leaving the dirction normal to plane non-periodic.

1. To apply pressure control without that of z direction (thus, variation of
box length in that direction), should I use
"fix 1 all npt temp 300.0 300.0 0.5 x 1. 1. 5. y 1. 1. 5. z NULL NULL 5.
couple none"?
Otherwise, what should I do for it?

Just leave off the z keyword if you're not imposing NPT in z.

2. If we don't impose peridic condtion to z-direction (e.g., by writing
"boundary p p s"),
should it be considered as no pressure control to that direction?
That is, even if we write "fix 1 all npt 300. 300. 0.5 aniso 1. 1. 5.",
should it be considered in a same way with question 1 automatically?

no, do it the way you stated in (1)

3. Incidentally asking, it semms that "s" and "m" in boundary keyword give
the same physical condition? Is it right?
Then, in what case, will the difference between s and m appear?
I think the difference between them should be found when we obtain relative
coordinate to box length to that direction though period condition is not
imposed to that direction. Is it right?

"s" and "m" are different - see the boundary doc page for details