unstable and lost atoms in

[UncleJ]

I try to simulate the diffusion of cobalt and diamond under high pressure and high temperature conditions(50000bar and 1700K), however, it could not be stable and always report the error:

ERROR: Lost atoms: original 4424 current 4422 (src/thermo.cpp:439)
Last command: run 1000

here is my in file:

#basic
units metal
atom_style full
dimension 3
boundary p p p
neighbor 2.0 bin
neigh_modify delay 0
timestep 0.001

#structure
#Create box
region box block 0 50 0 25 0 25 units box
create_box 2 box

#model Co
lattice hcp 2.507
lattice fcc 3.544
region Cobalt block 26 INF INF INF INF INF units box
create_atoms 1 region Cobalt

#model diamond
lattice diamond 3.567
region Carbon block INF 24 INF INF INF INF units box
create_atoms 2 region Carbon

#atoms mass
mass 1 58.93
mass 2 12

hybrid setting
pair_style hybrid/overlay lj/cut 3 eam/alloy airebo 3.0
pair_coeff * * eam/alloy CoAl.eam.alloy Co NULL
pair_coeff * * airebo CH.airebo NULL C
pair_coeff 1 2 lj/cut 0.023 2.852

thermo output
thermo_style custom step temp pe etotal press vol
thermo 100

#minimize
min_style cg
minimize 1e-15 1e-15 5000 5000

#relaxation
velocity all create 1700 873421

fix 1 all npt temp 1700 1700 0.01 iso 50000 50000 0.1
run 1000
thermo_modify lost ignore
unfix 1

#diffuse
fix 1 all npt temp 1700 1700 1 iso 50000 50000 0.1
dump 1 all atom 500 algp_heat_5GPa.lammpstrj
run 20000

I could not fix the problem, please help me

[akohlmey]

@UncleJ

I could not fix the problem, please help me

As a new participant in this forum, you should first read the post with the guidelines and suggestions completely and follow the advice and suggestions given as closely as possible.

Please Read This First: Guidelines and Suggestions for posting LAMMPS questions

The LAMMPS categories on MatSci.org are the primary way to get help with installing, running, programming the LAMMPS Molecular Dynamics software package and to facilitate exchange between LAMMPS users and developers. The main motivation of using a forum versus the traditional mailing list is, that this allows anybody to join a discussion at any time and try to provide assistance, even when they have not been around when the original question was posted. As the LAMMPS community is growing, the po…

It explains which questions to ask and what other information to provide. For example, you did do not quote your input file correctly and are leaving out important information about your LAMMPS version and platform. Beyond that, the issue you are reporting has been discussed many, many times before, and also is discussed in the LAMMPS manual. So you should not only state that you have problems running your simulation, but also which of the steps previously suggested you have tried out and if they made any difference.

@UncleJ

lattice fcc 3.544

Is this the lattice constant for cobalt under your pressure and temperature conditions? Have you run a Co-only system with your settings to confirm.

@UncleJ

pair_coeff * * eam/alloy CoAl.eam.alloy Co NULL

Is this potential suitable for these conditions? What does the corresponding publication say?

@UncleJ

lattice diamond 3.567

Same as above.

@UncleJ

pair_coeff * * airebo CH.airebo NULL C

Ditto.

@UncleJ

pair_style hybrid/overlay lj/cut 3 eam/alloy airebo 3.0

3 Å seems a very small cutoff for an LJ potential in metal units. Where did you get this parameter from?

@UncleJ

pair_coeff 1 2 lj/cut 0.023 2.852

Same as above, are these parameters suitable for the conditions at hand? Where did you take them from?

@UncleJ

timestep 0.001

High temperature, high pressure means that atoms can get very close to each other and thus may have very high repulsive forces, which in turn may require a shortened time step to do the numerical integration of the equations of motion without divergence. Have you tested if 1fs is a suitable choice?

@UncleJ

fix 1 all npt temp 1700 1700 0.01 iso 50000 50000 0.1

When you have stability issues, you should first try running with fix nvt or fix nve + fix langevin (for equilibration, the latter is often more effective).

@UncleJ

thermo_modify lost ignore

Please have a look at this paragraph, too.