Trying to reproduce phase transition curve for Cobalt from Vampire

[Siddharth Pandya]

Dear HeXu,

First of all, thank you for providing TB2J, which is so easy to use and efficient. Now, coming to my issue. I have recently begun working with  TB2J and Vampire to calculate Curie temperature. As a test, I chose the system they provided in the documentation here (input and Co.mat). I was able to produce the phase transition plot with all the parameters provided by them, including the exchange matrix (a single number).

I then aimed to reproduce the phase transition results using parameters derived from first-principles calculations. To do this, I obtained the structure file from here, performed electronic structure calculations using VASP, and subsequently Wannierized the results with Wannier90. I then calculated the exchange parameters with TB2J. Using the Vampire input files generated directly by TB2J, I ran the Curie temperature calculation. However, the results showed zero magnetisation across the entire temperature range, even at 25, indicating no magnetic ordering

For your reference, I am providing the wannier90 files(xyz, hr_dat, out, etc) and the TB2J results files, which can be helpful to analyse the errors.

Kind regards.

Siddharth Pandya

     

[Xu He]

Hello, 

The exchange parameters seems to be fine. It is definitely ferromagnetic and there should be a phase transition. 

Could you send the vampire files too? There was a bug in the TB2J-Vampire interface in a few versions, in which the digits of the exchange in vampire input could be wrong. Could you send the vampire files?  The bug has been already fixed.Could you try with the latest version (pip install - TB2J) and run the TB2j calculation again?

You should find something in the UCF files giving exchange values in the order of 1e-22.  (e.g. 

2730   2   2   0   0  -1  1.90159e-22  0.00000e+00  0.00000e+00  0.00000e+00  1.90159e-22  0.00000e+00  0.00000e+00  0.00000e+00  1.90159e-22 ) 

Best regards, 

HeXu

--
You received this message because you are subscribed to the Google Groups "TB2J" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tb2j+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/tb2j/5d77f868-7a97-4482-805d-208f097d58d7n%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Siddharth Pandya]

Hello,

         I have looked through the UCF file and no, all the parameters are of the order 10e-24 to 10e-27. I updated tb2j and ran it again and same issue is occurring. Here I am sending the vampire filed generated by tb2j 

[Siddharth Pandya]

Thank you for your response.

         Okay, then I believe the issue is with how TB2J is generating UCF file for vampire because all of the exchange parameters are of order e-24 to e-27 in the vampire.UCF file. I did upgrade TB2J and computed parameters for the same and same issue persists. I am running the Version: 0.9.12.9.  The vampire input and output files are in the attachment.

Kind regards, 

Siddharth Pandya

On Thursday, August 14, 2025 at 7:20:17 PM UTC+5:30 mail...@gmail.com wrote:

[Xu He]

Hello, 

Which version of TB2J was used after the update?

  I tried with your wannier files and the result was correct. (actully the line I sent in my first email was from the result). 

Best regards, 

HeXu

[Gavin Abo]

Dear Siddharth,

Do you want all four materials defined in your vampire.mat file to be included in your simulation?

In the log file of your Co_vampire_inpput_output_files.zip attachment, I see only Material 1 is included:

15-08-2025 [02:21:23] Material 1 Co makes up 100 % of all atoms ( 79507 atoms )
15-08-2025 [02:21:23] Material 2 Co makes up 0 % of all atoms ( 0 atoms )
15-08-2025 [02:21:23] Material 3 Co makes up 0 % of all atoms ( 0 atoms )
15-08-2025 [02:21:23] Material 4 Co makes up 0 % of all atoms ( 0 atoms )

In VAMPIRE version 4 and older, Material 1, 2, 3, and 4 would all be included in the simulation.  In Vampire 5 and later, a new unit-cell-category keyword was introduced.  The keyword would need to be used to include all four materials.

For additional details on the usage of the unit-cell-category keyword, see Step 3 on the VAMPIRE webpage at [1].

With the unit-cell-category likely being needed more often than not, it is something HeXu might consider adding to the code that generates vampire.mat in io_vampire.py [2] so that users don't have to add instances of that keyword by hand when it's going to be needed most of time.

[1] https://vampire.york.ac.uk/tutorials/system-generation/complex-crystals/

[2] https://github.com/mailhexu/TB2J/blob/newmagnon/TB2J/io_exchange/io_vampire.py

Kind Regards,

Gavin