Atoms in the lennard-jones droplet fly away!!

Dear lammps-users

I am simulating a droplet on a fixed surface with fix nvt until it is equilibriated but the droplet atoms continuously separate from the droplet and fly away in the space. Is it correct? Or how can I fix it?

Any help would be appreciated.

There will probably be a vapor liquid coexistence.

There will probably be a vapor liquid coexistence.

​or the initial geometry or choice of force field parameters or general
setup ​is messed up.
nobody can make any specific statement based on a so incredibly vague
description.

axel.

Dear Axel

Sorry for insufficient details.
The attached is the picture of my initial geometry which is simple cube for the droplet and also the input script.
I used lj/cut force field with pair coeffs of 0.2398 for liquid-liquid and 0.1 for solid-liquid.

droplet9.in (1.6 KB)

init.bmp (216 KB)

Dear Axel

Sorry for insufficient details.
The attached is the picture of my initial geometry which is simple cube
for the droplet and also the input script.
I used lj/cut force field with pair coeffs of 0.2398 for liquid-liquid and
0.1 for solid-liquid.

​as i had suspected. you get out what you put in​. your choice of initial
geometry and your simulation protocol make it inevitable that atoms are
flying away.

axel.

Could you please explain more obvious? How should I change it?

I think I should put air atoms in the space to represent the air pressure on the droplet but it costs increased simulation time.

Could you please explain more obvious? How should I change it?

​it is annoying, when people don't tell you the whole story right away,
isn't it?
i was giving you a demonstration of how it felt reading your original post.​

I think I should put air atoms in the space to represent the air pressure
on the droplet but it costs increased simulation time.

​this makes no sense. at the volume you are simulating, there are
effectively no atoms there in the corresponding volume of air. that won't
hold atoms back.

the problem is your initial geometry and your simulation protocol.
you start with a "brick" of atoms. is this physically meaningful? no! is
this what a droplet looks like? no!!
especially the atoms at the corners and edges have very li​ttle that holds
them back. ever heard of surface tension? ever though about what the reason
for that is?
so a more realistic starting geometry would be a whole or a section of a
sphere.
but it doesn't stop here. you are placing atoms, assign a (randomized)
temperature and start your simulation right away. is this likely to be a
realistic starting point? no!
so you have lots of atoms with high potential energy or potential energy
that isn't consistent with the kinetic energy and so on. you need to get
rid of some of these inconsistencies. i.e. you have to relax and
equilibrate your system. and have to do it carefully.

finally, if you are simulating a droplet with a soft interaction (as
opposed to, say, a water droplet, which has strong water-water forces
between them), you have to expect a high fugacity, i.e. atoms may easily
leave the droplet, since it doesn't require much kinetic overcome the
surface tension (there is rather little of it). so you have to start your
simulation *very* soft and carefully, e.g. equilibrate it at a much lower
temperature and very carefully heat it up. then put a reflecting wall
around your system to contain any atoms leaving the droplet and building
suitable liquid/vapor coexistence conditions.

i strongly suggest, you discuss the "design" of your simulation with your
adviser, since it looks a lot, like you haven't fully thought through what
you are trying to do and your problems are not problems of missing LAMMPS
experience or incorrect usage of LAMMPS commands, but rather bad choices
for how to set up and run your system. you'd have the same problems with
any MD code, if you'd use the same starting geometry and simulation
settings.

axel.

Dear Axel

I appreciate you for the key points you have mentioned.

​it is annoying, when people don't tell you the whole story right away,

isn't it?
i was giving you a demonstration of how it felt reading your original
post.​

and also appreciate for your friendly speaking but I purely did not meant
to be vague.

I was working on the case since that day based on the changes you advised.
First, I changed my initial geometry to part of a sphere and then divided
the simulation into several steps, gradually increasing the temperature to
the desired value.

then put a reflecting wall around your system to contain any atoms leaving

the droplet and building suitable liquid/vapor coexistence conditions.

  I also put reflecting walls around the box. At last it seems, as you
said, there will be a liquid/vapor coexistence in the system.

Now I am looking for a criteria to ensure the relaxation of the system.
Could it be reasonable to make sure of relaxation by the number of atoms
existing in the droplet?