I have some problems in using “fix gcmc” for the relaxation of hydrogen atoms in Nickel. The problems are: lost atoms and unrealistic high temperature. Although lost atoms can be solved by changing settings in thermos_modify, the temperature is always increasing. My input file reads:
Do you know if this EAM potential has ever been used to do GCMC of H in Ni? The reason I ask is that there may be unphysical configurations that are no accessible in MD, but will be sampled by H insertions. If this is happening, the standard practice is to add an extra repulsive term to the Ni/H interaction to exclude those configurations without changing the physical regions.
If that is not the issue, it might just be that the forces on some inserted H atoms are a little high and you need to back off on either the chemical potential or the timestep.