reaxff with Cu/Ag/O

Dear Lammps users,

I am using reaxff with Cu/Ag/O

The calculation process looks fine, but the calculation data seems problematic.

And, the structure is working well with EAM potential.

Here is my scripts.

I also add my structure and reaxff potential.

Do you know what is a problem?

Thanks!

Best.
Choi

log cuag.log
#--------------------Initialization---------------------
units real
boundary p p p
atom_style charge
#-------------------Atom definition--------------------
read_data data.cuag
#-------------------FORCE FIELDS-----------------------
pair_style reax/c NULL safezone 5.0 mincap 1200
pair_coeff * * ffield.reax Cu Ag
neighbor 2.5 bin
neigh_modify delay 0 every 2 check yes one 15000 page 1500000
fix qeq all qeq/reax 1 0.0 10.0 1.0e-6 reax/c

velocity all create 10 4928459 rot yes dist gaussian
fix 1 all npt temp 100 100 0.2 iso 0 0 2

dump 1 all xyz 500000 NWs.dyn.xyz
dump 2 all custom 500000 NWs.dyn.lammpstrj id type q x y z vx vy vz
dump_modify 2 sort id
dump 3 all cfg 500000 dump.*.cfg mass type xs ys zs vx vy vz fx fy fz
dump_modify 3 element Cu Ag
dump 4 all image 50000 img-*.ppm type type zoom 2.5 adiam 1.5 size 1280 720
dump_modify 4 pad 5

thermo 10
thermo_style custom step temp press pe ke etotal vol pxx pyy pzz pxy pyz pxz
thermo_modify flush yes
thermo_modify lost ignore flush yes
timestep 0.001
run 5000000
undump 1
undump 2
unfix 1

unfix qeq
unfix bonds
clear
write_restart restart.eqi

image.png

ffield.reax (22.2 KB)

data.cuag (169 KB)

I looked at your data file with Ovito and your simulation box is gigantic compared to the space your atoms occupy. Is there a reason your simulation box’s volume is e+10, but your system actually only occupies ~40*~40*~40? That could be one problem.

nan’s usually mean something is wrong with your system. I played around with your system for a bit and from what I can tell, your system does not appear to be well-equilibrated. With snapshots set to dump every time step at 0.001 time step, your system’s atoms disappear on the first time step after starting. I also tested with nvt instead of npt and the atoms still disappear. Disappearing atoms usually mean that they “exploded” out of the simulation box.

I don’t have any experience with simulating metals with ReaxFF, but I do have experience running polymers with ReaxFF. Generally, you want to ensure that the transition to ReaxFF is smooth. The way you do that is by running a simulation with the force field you’re moving from that decreases the temperature to near zero such as 0.1 K with NVT (MD can’t do 0 K, of course) and write out the data file at that temp. Then move the data file to ReaxFF format and start it up in ReaxFF going from 0.1 K to whatever temp you like, I usually do 300 K, with NVT. Also, if you just moved your data file to ReaxFF, it’s generally a bad idea to start with NPT.

Hope this helps.

Will

image.png

Thanks for your explanation.

I will follow your directions.

There is no reason for a big size box.

I will reduce the box volumes.

Best,
Choi

2020년 2월 28일 (금) 오후 11:57, Will Pisani <wapisani@…1754…>님이 작성:

image.png