Indentation force and pressure along indentation direction decreases

I am using a spherical indenter to indent a cubic sample along the z-axis and simulating in nvt ensemble with ppf boundaries. In the output, of depth of indenter in the sample with pressure (Pzz) and force along z (f_fixind [3]), the pressure and force are found to be steadily decreasing.

Again, while unloading too these two quantities are found to decrease.

What you have posted is too little information for us to know what’s going on. You would need to provide a working input script (preferably as simple as possible) to get useful feedback.

Here is the relevant portion of the input script:

fix             1 atoms1 nvt temp 300.0 300.0 10
fix             2 atoms2 setforce 0.0 0.0 0.0
fix             3 atoms3 temp/berendsen 300.0 300.0 10
fix             5 all wall/reflect zhi EDGE

timestep        0.001
variable        indvt equal "step*dt*(-v_vindenter)"
variable        indzpos equal "v_zcent + v_indvt"
fix             4 atoms1 indent 300 sphere ${xpos} ${ypos} ${indzpos} 30.0 units box

fix             6 all print 100 "Indenter position = ${indzpos}" append Mo300-pure-IndzPos.txt screen no

#Thermo output

compute         systemp atoms1 temp
dump            1 all custom 1000 Indent-*.txt id x y z

thermo          100
thermo_style    custom step temp lx ly lz vol press pzz f_4[1] f_4[2] f_4[3]
thermo_modify   lost ignore norm yes flush yes

run             300000

Unfortunately, this cannot be run to verify your claims and thus is rather useless.

In general, the first step if a simulation is not behaving as expected is to visualize it and check if it is behaving as expected and if the atom motions are consistent with the computed results. In that respect, the use of thermo_modify lost ignore is worrisome since that can hide problems since for a regular indentation simulation there should be no atoms leaving the system.

The syntax in your script snippet doesn’t match your intentions, and your indenter isn’t actually moving. Dump and visualise the trajectory if you need to verify this for yourself (which you should). This may not be the only problem in your script, since you haven’t posted it in full.

There is indeed some lost atoms in the initial few steps from 1024000 atoms in original to 1022291 atoms and I have used that

command to come around this problem. Apart from these command I am not able to run the simulation. I have tried with lowering the timestep, reducing the indenter velocity, using the fix wall/reflect command, increasing the box dimension along z. But every way there is lost atoms and the program will abort.
Then what can be the way, please suggest.

Yes sir, I have visualized it, the indentation seems to happen with creation of damaged regions and dislocations around.

You just did a lot of things, but without knowing what is causing the lost atoms. That is rarely going to work. You apparently didn’t do the most important thing (best setting the lost atoms feature to “warn” and frequent thermo output and reneighboring, so you are aware when atoms are lost): check the dump file (if needed create one with more frequent output) and observe which atoms are lost. That is usually the crucial information that will guide you to the changes required to prevent atoms from being lost. Those can have trivial origins or be subtle.

You only provided part of your input, but that part didn’t inspire much trust, that the rest is correct. I suspect that you have atoms near a fixed box boundary and didn’t properly immobilize them. Also, thermostatting the atoms that you indent makes little sense and using two different thermostat algorithms on two different groups of atoms even less.

Thank you sir for the feedback.
I will try some possible ways to reneighbor the atoms and also remove different thermostats.

Along z the boundaries are fixed and the atoms in a few layers at the bottom of the box are made immobile by putting all the forces to zero.


That by itself is only sufficient for minimization. For dynamics, also the velocities need to be set to zero, or - even better - the atoms must not be included in time integration.

One simple step to check whether atoms leave the box at the bottom (it would not be the first case where this happens) could be to replace the “f” boundary with “m” or “mf”. Then atoms cannot be lost by exiting the box that way.