[lammps-users] High Energy Particles using ReaxFF

Dear LAMMPS Users

I’m trying to simulate a structure that is surface terminated with hydrogen by heating it from 300K to 3000K using a ReaxFF potential with QEQ. I’m using langevin thermostat with nve and a timestep of 0.1fs. The damping factors for the langevin thermostat are ranged between 200 and 50 in real time units. I do have several atom types.

The issue rises when the temperature reaches around 2500 K. I get an error “non-numeric atomic coordinates – simulation unstable”. So after looking further into the problem I noticed that it happens due to the hydrogen atoms getting knocked off the surface due to high temperature and they combine in the empty space above the structure in the simulation box to form H2. However I notice the separation distance is too small and I’ve seen a discussion on the list several years back mentioning that this error happens due to close contacts. So I did several attempts involving deleting the atoms that fly over surface and I managed to get the simulation more stable but I still get the same error as the temperature gets higher. I attempted the following:

There is nothing attached.

My best guess is that the ReaxFF parameters you are using are not suitable for the system and conditions you are simulating.