I followed tutorial to do fitsvsl and fitsvsl -f and then clusterexpand fvib. After that, I got fvib.eci file. As I understand it includes the eci of vibration free energy for each temperature in Trange.in. So I used the same formula as to calculate random alloy formation energy to get the vibration free energy as shown in fig.1. In fig.1 all vibration free energies are fix two endpoints to zero and emphasizing the change depending on the concentration. The vibration entropy was calculated with the svib_ht.eci similarly, as shown in fig. 2.
The strange thing is that the vibration free energies show `W’ shape instead of parabolic shape. In my opinion, the parabolic shape should be normal as in literatures. And this kind of "W" shape leads critical temperature increasing instead of decreasing.
Is it possible to exist in some kind of alloy system or I just make a mistake?
I read the mkteci code and understand the process to add eci.out and fvib.out to teci.out. I use teci.out to calculate the phase diagram. Just because of the increasment of the critical temperature after considering vibration contribution, which is opposite to my knowledge, I curiously know the reason of it. So I plot fig.1 with fvib.out and fig.2 with svib_ht.eci.
It’s so strange. So I am not sure it’s right or I just make a mistake.
One thing to check is to look at all */svib_ht files (it may be easier if you ran svsl or fitfc with the -pa option so that all quantities are per atom - but don’t forget the -pa for clusterexpand as well).
Are there any strange outliers? You can plot these vibrational entropy as a function of composition to see if some are just way out there. If there are only a few, you could remove the corresponding fvib file a re-fit the cluster expansion.