How to set barostat on non-periodic dimension

Dear professors,

I am sorry to disturb you, but I have no choice. I have spent several days on the problem.

I am trying to simulate the water boiling on a metal. I use "fix langevin" command to control the temperature of the metal, and "fix nve" command on the box, and "velocity" command to set the initial temperature of water. Only x dimension is periodic. I want to see the heat transfer from metal to water and the water boiling process. When I run the simulation, the water temperature increases quickly and even higher than the metal temperature. It is not right, and there is no boiling phenomena. Is there anything wrong?

What's more, when I try to control the water pressure using "fix npt or fix press/berendsen" commad to control the pressure of water, it said the command can't work for a non-periodic dimension.

How can I control the pressure of a non-periodic dimension? What's more, Does the "fix wall/reflect" command can realize a mirror boundary condition?

Best wishes,

Xiaohui

University of Wisconsin-Madison

units real
dimension 3
boundary p f f
atom_style full
neighbor 3.5 bin
neigh_modify delay 5 check yes

####create geometry
lattice fcc 3.615 #Cu lattice constant
#water region
read_data water_box.data
#Cu region
region lo-Cu block 0 8 0 9 -3 -1 #xlo xhi ylo yhi zlo zhi
create_atoms 3 region lo-Cu
region right-Cu sphere 4 9 3.9 4 #x,y,x,r
create_atoms 3 region right-Cu
region right-no sphere 4 9 3.9 2
delete_atoms region right-no

#potentials
pair_style hybrid/overlay lj/charmm/coul/charmm 7.5 8.5 7.5 8.5 eam lj/cut 7.5
bond_style harmonic
angle_style harmonic
pair_coeff 1 1 lj/charmm/coul/charmm 0.1521 3.1507 #O-O epsilon sigma
pair_coeff 2 2 lj/charmm/coul/charmm 0.046 0.4 #H-H epsilon sigma
pair_coeff 1 2 lj/charmm/coul/charmm 0.0836 1.7753 #O-H epsilon sigma
pair_coeff 1 3 lj/cut 1.198 1.587 #O-Cu epsilon sigma
pair_coeff 2 3 lj/cut 0.6589 0.2117 #H-Cu epsilon sigma
pair_coeff 3 3 eam Cu_u3.eam #Cu-Cu cuu3.eam file
bond_coeff 1 450 0.9572 #O-H
angle_coeff 1 55 104.52 #H-O-H

# define groups
group center type 1 2
group low region lo-Cu
group side region right-Cu

# minimization : avoid atoms overlapping
min_style sd
minimize 1.0e-8 1.0e-8 1000 1000

#fix
fix 0 center wall/reflect ylo EDGE zhi EDGE #reflect water molecules
fix 1 low langevin 373.15 473.15 100 48279 #set the Cu temperature
fix 2 side langevin 373.15 473.15 100 48279 #set the Cu temperature
fix 3 all nve
fix 5 center shake 0.0001 20 0 b 1 a 1 # fix the bond length and angle of water
fix fasten_low low setforce 0 0 0 #fix Cu atom
fix fasten_side side setforce 0 0 0 #fix Cu atom

# initial velocity
velocity center create 373.15 4928459 dist gaussian rot yes sum no units box loop local
#set the water initial temperature

timestep 0.5 #unit fs
thermo 1000
compute water_temp center temp #calculate water temperature
thermo_modify temp water_temp #export water temperature

dump 1 all atom 1000 water.lammpstrj

run 50000

Dear professors,

I am sorry to disturb you, but I have no choice. I have spent several days on the problem.

I am trying to simulate the water boiling on a metal. I use "fix langevin" command to control the temperature of the metal, and "fix nve" command on the box, and "velocity" command to set the initial temperature of water. Only x dimension is periodic. I want to see the heat transfer from metal to water and the water boiling process. When I run the simulation, the water temperature increases quickly and even higher than the metal temperature. It is not right, and there is no boiling phenomena. Is there anything wrong?

have you ever heard the terms "finite size effects", "nucleation", and
"hysteresis"?

What's more, when I try to control the water pressure using "fix npt or fix press/berendsen" commad to control the pressure of water, it said the command can't work for a non-periodic dimension.

and rightfully so.

How can I control the pressure of a non-periodic dimension?

using fix aveforce, for example.

What's more, Does the "fix wall/reflect" command can realize a mirror boundary condition?

no. it does what is written in its documentation.

axel.

How can I control the pressure of a non-periodic dimension?

If they are a free surface, then you can’t. The atoms are free to move as far as they want,
so how would you imagine not allowing them to? If you
want to define a fictitious wall and squeeze in that dimension,
then you can.

This topic has been discussed on the mail list many times.

Steve