vibrational contributions for system having unstable phase

Dear everyone,

When using CE for the free energy study in one alloy system, if there is a dynamically unstable composition zone (for example, in bcc Fe-Cu alloy, pure bcc-Cu would be dynamically unstable), can we still use the fitfc or fitsvsl methods to get the vibrational contribution? Can we limit the composition zone, like the c0 and c1 options in maps? Would that be possible to perform and meaningful?

Many thanks!

Yi Wang
Ph.D candidate of Nanjing University of Sci. & Tech., China

One possibility is to just limit the composition range, as you mention.
Another is to use the new scheme described in https://dx.doi.org/10.1038/ncomms8559
and implemented as the robustrelax_vasp command.
I’d be curious to hear how the latter performs.

Dear Axel,

Thank you very much. I’m aware of the new robust relax method for some time. But I really didn’t do the thinking to connect it to the vibrational contributions. Many thanks for this inspiration.