Frenkel pair explosion during cyclic irradiation MD simulations in W-based alloys

Dear all,

I am performing cyclic irradiation cascade simulations in LAMMPS following the methodology described in the paper “Revealing the critical role of vanadium in radiation damage of tungsten-based alloys” (Wei et al., 2023). I am using an RSS (random solid solution) W-V alloy system containing about 250,000 atoms. My simulation procedure is as follows:

  1. Select a W atom near the box center as the PKA and give it 5 keV kinetic energy along the [135] direction.

  2. Run the cascade for 20 ps in the NVE ensemble with an adaptive timestep, including electronic stopping, while a 3 Å boundary layer is thermostated at 300 K using a Berendsen thermostat.

  3. Relax the system in the NPT ensemble for 20 ps to restore 300 K and zero pressure, and reset the center of mass to the box center.

The problem I encountered:

During the cyclic cascade simulations, the number of Frenkel pairs (FPs) remained normal (~1100) at frame 61. However, at the very next frame (frame 62), the FP count suddenly exploded to over 10,000, which is physically unreasonable.

I also noticed that one atom was lost during this transition: the model originally contained 250,000 atoms, but at frame 62 the atom count dropped to 249,999. Interestingly, LAMMPS did not report any error and the simulation continued to run without interruption.

However, since I am a new user and cannot upload attachments, if it is convenient, I would be happy to send my input files via email to anyone who is willing to help.

LAMMPS’ behavior in this regard is controlled by the thermo_modify lost setting. If you have it set to thermo_modify lost ignore lost atoms will be ignored, otherwise it will either warn or error out when the next thermodynamic output happens. So check your input whether you are just getting what you are asking for.

You can always upload files to a service like Dropbox, OneDrive, Google Drive or equivalent, create a shareable link, and then post the link.