Wrong magnetizaiton values when simulating the curie temperature

[yangan202407-cmyk]

Dear TB2J developpers and users,

I used the latest version of TB2J to calculate the exchange coupling parameters of CeFe2 which were calculated by Vasp and Wannier90.

The J values are not small but when I simulate the curie temperature i get very low values of mean-magnetization length, and material-magnetisation seems wired.

I used primitive cell CeFe2.cif with 6 atoms.

I have attached some necessary files, and I sincerely hope for your guidance.
vampire_necessary.zip

[mailhexu]

Hello,
From the JvsR file it seems there are quite a few different J values with the same distance, hinting that the J values have some symmetry issue. This is quite often due to bad Wannierization. Did you check

• The Wannier band structure is consistent with the DFT band structure
• The Wannier centers are close enough to the atoms
• The Wannier spreads are small enough

Best regards,
HeXu

[yangan202407-cmyk]

Hello, From the JvsR file it seems there are quite a few different J values with the same distance, hinting that the J values have some symmetry issue. This is quite often due to bad Wannierization. Did you check

• The Wannier band structure is consistent with the DFT band structure
• The Wannier centers are close enough to the atoms
• The Wannier spreads are small enough

Best regards, HeXu

Thank you for your suggests! I checked some parameters of froze_windows and dis_windows have been changed, now the DFT bands fitted well with wannier90 bands between (-4, 4), and I tried some different dis_windows to reduce the spreads. But after TB2J calculation the J values seems still have problems, which also causes the bad result of Vampire. When calculateing TB2J, I used the command "wann2J.py --efermi 6.1304 --emin -4.0 --emax 4.0 --elements Ce Fe --prefix_up wannier90.up --prefix_down wannier90.dn --kmesh 7 7 7 --nz 200 --rcut 10". Now I upload some files I used and some outputs I got, could please help me to find out the errors? I sincerely hope for your guidance.
necessary.zip

[mailhexu]

Hello,
From the .wout file the wannien spread are still large and wannier centers are a bit far way from the atom centers. The window shold be improved until the spread are small enough and atom-centered.
To build good quality Wannier functions, there are some nice suggestions here: https://www.wanniertools.org/tutorials/high-quality-wfs/
Best regards,
HeXu

[yangan202407-cmyk]

@mailhexu Thank you for your guide! Now I successfully get ture J values and run vampire to calculate curie temperatures. I noted in a lesson that he vertical axis is magnetisation (μA). Is it possible to set this in the input to output the μA values for each temperature? I sincerely hope for your guidance.