[lammps-users] Energy convergence issue during sputtering simulations


The total energy is not converging in sputtering simulations. I am testing with the LAMMPS example demo file. Please find the attached total energy versus time curve for your reference. Generally, a thermodynamic equilibrium attains at some point in the simulation’s time. It is not happening in these simulations. Any comments on this observation are appreciated.

Please find my input file as follows

surface sputtering demo

dimension variables

box length in x and y

variable len index 20.0

height of layer

variable hgt index 8.0

delta between insert area and box

variable len0 index 0.1

full length minus delta

variable len1 equal {len}-{len0}

half length

variable len2 equal ${len}*0.5

units lj
atom_style atomic
boundary p p f

lattice fcc 0.85
region box block 0.0 {len} 0.0 {len} -1.0 40.0 units lattice

particle density in z-direction is inhomogeneous:

only do domain decomposition in x- and y-direction

processors * * 1

two atom types: surface and sputter atoms

create_box 2 box

region layer block 0.0 {len} 0.0 {len} 0.0 ${hgt}
create_atoms 1 region layer
mass 1 1.0
mass 2 3.0

region fixed block INF INF INF INF 0.0 0.1
group fixed region fixed
group mobile subtract all fixed

pair_style lj/cut/opt 2.5
pair_coeff 1 1 1.0 1.0
pair_coeff 2 2 0.5 1.2

neighbor 0.3 bin
neigh_modify every 2 delay 6 check yes

minimize 1.0e-5 1.0e-8 1000 10000
reset_timestep 0

velocity mobile create 0.4 87287

use the temperature of the mobile layer atoms as reference

compute mtemp mobile temp
fix 1 mobile nve
fix 2 mobile langevin 0.2 0.2 5.0 982434
fix_modify 2 temp mtemp

thermo 100
thermo_style custom step atoms temp etotal
thermo_modify temp mtemp

dump id all xyz 2000 sputter.xyz

minimal equilibration

run 100

define region for inserting particles

region sphere sphere {len2} {len2} 42.0 ${len2} units lattice
region insert intersect 2 box sphere

define group of inserted particles

group insert region insert

deposite particles regularly into insert volume

fix insert insert deposit 50 2 200 12345 region insert vz -8.0 -9.0 &
target {len2} {len2} ${hgt} units lattice
neigh_modify every 1 delay 0 check yes

time integrate inserted particles

fix 4 insert nve
compute ke4 insert ke

remove thermostat on layer atoms

unfix 2

variable nlayer equal count(mobile)+count(fixed)
variable ninsert equal count(insert)
thermo_style custom step atoms v_nlayer v_ninsert temp etotal v_len2 c_ke4
thermo_modify temp mtemp

handling of atoms the get bounced back from surface

or sputtered off the surface.

thermo_modify lost ignore flush yes line multi

run 2000000

with kind regards,

a) you are using reduced units, but your plot is using nanoseconds so it is difficult to relate the time axis to your graph to the simulation
b) you have to visualize the system to see if all “action” resulting from the sputtering has already subsided. I suspect not.
c) what does the graph look like, if you keep running?