[lammps-users] cascade simulation-lost atoms

Dear all I am running cascade simulation of bulk vanadium lattice. I am getting the lost atoms errors. I have read about this error in the lammps archives and learnt that main reason for this error is large time step, but I reduced my time step as low as 10e-6 ps even then the same error persisted. >From dump snapshots visualised in VMD I realized that atoms are lost when the pka approaches its immediate neighbor and at that point PE and KE also shoot up. Inspite of these observations I could not rectify the error. Below is my input file. Any help in this regard would be much helpful.

cascade simulation of bulk V lattice

units metal
atom_style atomic
neighbor 1 bin
neigh_modify delay 1

lattice bcc 3.03
region box block 0 50 0 50 0 50
boundary p p p
create_box 1 box
create_atoms 1 box

#creating regions
region rallatoms block INF INF INF INF INF INF
region rinterior block 2 48 2 48 2 INF
region rexterior block 2 48 2 48 2 INF side out
region rpka sphere 30.3 30.3 30.3 0.4

#define potential
pair_style eam/fs
pair_coeff * * VFe_mm.eam.fs V

#creating groups
group interior region rinterior
group exterior region rexterior
group pka region rpka

#creating fixes
fix 1 all nve
fix 2 exterior temp/rescale 1 300 300 0.5 1.0 region rexterior
fix 3 interior temp/rescale 1 300 300 0.5 1.0 region rinterior

#equilibration of the system
timestep 0.001
thermo 100
thermo_style custom step temp pe etotal press
run 1000

#remove scaling from the interior and continue to equilibrate
unfix 3
run 1000

#set pka velocity correspond to 10kev in [135] direction
set group pka vx 232 vy 697 vz 1200

#initial phase of collision
timestep 0.000001
thermo 100
dump mydump all xyz 100 dump.xyz
run 20000

#intermediate phase of collision
timestep 0.0001
run 10000

Well, if atoms get too close, they feel large forces, and are blown
out of the simulation box. If the potential you use allows for that,
it will happen. Not much you can do about it, except run with
lower incident energies or use a potential designed for close approach
of two atoms.

Steve