Particles equilibration and approaching process: atoms released during simulation

Dear LAMMPS users,

I am currently running simulations regarding two TiO2 particles approaching: one particle stays still and the other one is assigned an initial velocity of -10 m/s to compute the potentials of mean force between the two particles during their approaching process. However, the particles started to dissociate after the simulation started running, and the output gives me the error “ERROR: Out of range atoms - cannot compute PPPM”. Decreasing the timestep only delayed the time for this error to show up and the atoms are released from the particles according to visualization.

After that, I tried to run the equilibration simulation for the particle that stays still and the particle still has the trend to dissociate. Another thing is that from the output file, the temperature changed to 0 K even though I used the NVT condition. I changed the pair style to coul/cut and the dissociation happened even faster.

I attached the input files for the approaching and equilibration cases. I would be really grateful if anyone can point out which part I did wrong.

Any help would be really appreciated.
fort.1028 (168.4 KB)
two_particles_collision.in (967 Bytes)
fort_single.1028 (84.3 KB)
one_particle.in (684 Bytes)

Hello,
If the particles ‘dissociate’, it is probably an issue with the force field and the parameters that you are using, rather than an issue with the protocol.
I dont know the force field that you are using, but I find it strange that the only interaction you have is the Coulomb interaction.
Simon

1 Like

I agree with Simon about it being akward that you have only Coulombic interactions (although I am not exactly an expert in the existing force fields).

Also, try setting the velocities after doing the energy minimization instead of before (the presence of kinetic energy might mess up the minimization in Ep).

I am not at all expert in thermostatting either, but your Tdamp may be too large relative to the timestep you are using, causing the equilibration at the target temperature to take too long to happen. Try decreasing it a bit (but not a lot, otherwise your dynamics may be too much affected by the influence of the thermostat). There is a recommendation in LAMMPS manual of 100x the value of the timestep.

Also, I dont know if it is interesting in your case, but it could be that fix nvt/rigid may be interesting for you depending on what you want to analyse. But note that it is not a solution to the problem you currently have (specially if it is the case of bad potential). It’s just that it maybe it is interesting for you if you want to move the particles as an entire entity with the ions in the optimal positions and compare the interaction energy of the two nanoparticles.

This seems like a particularly elegant and involved demonstration of Earnshaw’s theorem.

To be explicit, it is impossible for a collection of point charges (even a countably infinite collection!) to have a stable spatial configuration, and yet that is precisely what you have simulated in a simulation with only coul/long. The stability of matter requires short-ranged non-Coulombic interactions, both in real life and in simulation.

6 Likes

Thank you for the detailed explanation. I did the equilibration with Buckingham potential instead and the temperature changed within a stable range. However, the particle still dissociates once the simulation started. I am not sure which part I did wrong this time and I attached the input and output files. Really appreciate the help!
fort.1028 (84.3 KB)
in.in (1.2 KB)
273K_Anatase_3.0nm_trajectory.xyz (9.7 MB)
out (64.3 KB)

Thank you Simon for bringing up the forcefield. I changed to Buckingham potential but the particle still dissociates even during equilibration where the temperature did not fluctuate much. I attached the input and output files under the reply to Cecilia. Thank you again for the help!

Have you checked whether the potential parameters are suitable for your system?
Most likely, they may only be used for bulk systems.

For particles with free surfaces, you usually need something more complex that includes some form of charge equilibration to handle the difference in polarization on surfaces, edges, and tips. Also, certain geometries may simply not be stable due to the highly polar nature of your compound.

Hi Axel,

Thank you for the reply! I used the parameters of Matsui and Akaogi, which are validated to be the most suitable ones for TiO2 particles. In terms of “charge equilibration”, does that mean I should include the Coulomb term (e.g., use buck/coul/cut pair style) for equilibration instead of using “buck” alone?

Thank you,
Shaelyn

Charge equilibration is a method to recalibrate the charges. It is commonly used for modeling oxides for the reasons I mentioned. The COMB/COMB3 potentials implement such a method and should be suitable for TiO2 nanoparticles. But don’t take my word for it, check the literature.

Overall, this sounds like your problems are more of a principal nature and about your understanding of the science of your research rather than a LAMMPS specific issue. Which means, that you would be better off to discuss with your adviser/supervisor/mentor/tutor or similar. The forum can give you very little help on that.