Curie temperature of different materials

[Blues Needa]

Hello, 

I am trying to calculate Curie temperature of SmFe12 system by Vampire, using exchange parameters obtained from openmx-TB2J. But the result (Curie temperature) was very high about 2 times than expected (experimental) value. 

Then I run some test calculations on Fe-bcc, Co-hexagonal, FeCo-bcc systems. However, the results were still 2 times than expected values except Fe-bcc structure. It is interesting that i got correct result for Fe-bcc system.

I would like to ask about the reason why i get these results? or Did I miss some required tags in my input file? Appreciate for any help

[Blues Needa]

Here, I attached some of my calculations.

[Xu He]

Hi,

There are many possible reasons for the discrepancy between the simulation result  and experimental ones. 

You can find some of them here:

https://tb2j.readthedocs.io/en/latest/src/faq.html#the-results-seems-to-contradict-the-experimental-results-why

Best regards, 

HeXu

[Blues Needa]

Thank you for the reply. As I found from that web site, there is saying that " Since version 0.7.1, the TB2J-Vampire output take the “tensorial” format, in which the exchange values are 2 times that in the convention of TB2J. "

Here, https://tb2j.readthedocs.io/en/latest/src/multibinit.html 

Does it mean that 2 times exchange values are kind of TB2J-Vampire bugs? If so, Can I just divide the exchange values by 2 to calculate the Curie temperature? 

[Xu He]

Hello, 

No this is not a bug. It is a choice of convention. There are multiple conventions of Heisenberg Hamiltonians and the exchange parameters should be adjusted accordingly. 

See this page: https://tb2j.readthedocs.io/en/latest/src/convention.html

HeXu