Dear All,
Are the chemical potential differences defined the same way in the following two commands:
https://docs.lammps.org/fix_atom_swap.html
https://docs.lammps.org/fix_sgcmc.html
We find that, for the same alloy, we need to use opposite signed chemical potential differences in the two commands to achieve similar short-range ordered structures. For example, if we use “muA muB” in the first command, we need to use “-muA -muB” in the second command. Otherwise, the structures would be very different — even the molar ratio would be wrong.
The second command reads “By convention it is the difference of the chemical potentials of elements B, C …, with respect to element A”, but it is not clear if the chemical potential differences are defined in the same way in the first command.
According to this page sign issue in semi grand acceptance criterion · Issue #1139 · lammps/lammps · GitHub, a sign change was made in the first command back in 2018. Still, I think it would be better to clearly explain the meaning of the chemical potential differences in that command in the manual.
Best
Shuozhi
Perhaps @athomps can comment on this topic.
The sign convention for absolute chemical potential is unambiguous, because the infinite dilution limit always corresponds to negative infinity. fix gcmc follows this convention. The semi-grand variant of fix atom/swap violated this convention for the absolute chemical potentials. Once this was pointed out to us, we changed it in 2018. When it comes to chemical potential differences, there is more wiggle room. In general, with differences, the convention is to subtract the reference value from the other value. However ‘fix sgcmc’ does the opposite, deltamu2 = mu1-mu2, deltamu3 = mu1-mu3, etc. This is not stated in the documentation. I suggested modifying the statement “By convention it is the difference of the chemical potentials of elements B, C…, with respect to element A.” to “The difference of the chemical potentials of elements B, C…, with respect to element A is mu_A-mu_B, mu_A-mu_C,….”. Or else we could switch to from ntypes-1 differences to ntypes absolute chemical potentials, same as fix atom/swap. This is guaranteed to break old scripts, so no incorrect simulations will be run.
Thank you! It makes more sense now. And I agree that changing the statement would be good. Because when it reads “chemical potentials of element B with respect to element A”, I interpret it as “mu_B-mu_A” instead of the other way around.