Well the paper I am reading currently it says

"Besides the predicted antiferromagnetism for phases of Ti3Co5B3-type structure having 62 valence electrons (VE) and ferromagnetism for those with 65 VE, a classification of this family of borides in two groups has been established: the first group with VE below 63 shows predominantly antiferromagnetic interactions, whereas the second having at least 63 VE shows predominantly ferromagnetic interactions.

Furthermore, antiferromagnetic interactions decrease as ferromagnetic ones increase until a maximum is reached at 66 VE, before the ferromagnetic interactions decrease

again."

& I found similar prediction in other system too. 

The number of calculated valence electrons in VASP is dependent on the potential you choose. For example if you choose the normal Fe potential, or the Fe_p potential, where not only 3d and 4s states are treated as valence, but also 3p states.

However, I am not sure that this is the definition you need in your case. I am a little confused by the example you provide from literature, surely for the same material (Ti3Co5B3 in your case) either the system with 62 electrons or the one with 65 will be charged. Maybe there is some doping going on? Or a different definition of VEs, or they are indeed talking about charged systems which is then a different matter entirely.

Cheers, M

Thanks Michael,

Yes the number of VE are potential dependent, but if it is really then those prediction (about dependency of magnetic ordering on VE) will not be unique I guess. 

Dear Rohit,

i am pretty sure that your problem is not related to the choice of potential. If you include some core states as valence (and thus increase the total number of valance electrons), this will not change the magnetism, as those states will be fully occupied anyway (symmetrical in spin up and down). 

You cite: " ...for phases of Ti3Co5B3-type structure..." which i think means that the crystal structure is left alone, but single elements (maybe only on one site, maybe on all) are substituted with others, changing the number of valence electrons close to the Fermi energy where the magnetic stuff happens.

In the future, please give a correct reference to the papers you cite, and include a .pdf if it is (legally) possible. That would make answering your questions easier.

Cheers, M

thanks for your suggestion dear Michael :) 

sorry for not attaching pdf file, I thought of  quoting only relevent part because it a review article and too long. Anyway PFA

(Quoted text was from conclusion)

sincerely

Rohit

Thanks for the paper. Now everything is clear, and frankly it should have been from the beginning. In the beginning of section 2.2 it is made very clear that there a different materials involved, leading to different VE numbers.

Yes different elements are involves sorry I was doing the wrong sum of VE's,  thanks second part is clear. Now the other part my question its regarding uniqueness, suppose I choose different POTCAR other than simple one then we will have different number of VE then how come the prediction will be unique? Its all right as you pointed out that other core state will not give rise to magnetism, but the VE number will be changed and I think uniqueness of the statement will not be there.

I hope you are getting what I am asking and thanks for clearing doubts.

Now I have one additional question, does every pseudo-potential and plane wave based package have similar potential file?

thanks

Rohit

Hi Rohit,

there is no problem with uniqueness. You can treat the semi-core states simply not as VEs, even if you include them in your calculations by using a _pv or _sv potential. Or you just calculate the total number of electrons in the system using the atomic number of the involved elements. In any way, only the difference is important.

For example: You add 3 electrons by replacing Ti with V in Ti3Co5B3, now you have X+3 VE. How does the magnetism change. Now you replace Ti with Sc and see what happens for X-3, and so on.

For your additional question: I do not really get that what you mean... There are of course different potential types (PAW, ultrasoft,...) and each type can be constructed differently. Of course, if constructed properly, results should be the same. To compare different codes and potentials, and learn about accuracy of DFT, you may check out the recent science paper I attached.

http://science.sciencemag.org/content/351/6280/aad3000

Article Reproducibility in density functional theory calculations of solids

@Michael, thanks a lot !

Dear All, I want to clear my one doubt, if I am wrong please correct me.

I want to know how many electrons are in my calculation for example my system is La4V4O16 and ZVAL from POTCAR are 11, 13 and 6 for La, V and O correspondingly. So my system is having total

44+52+96=192 electrons.

Am I right or wrong ?

Thank You

Shilendra

grep NELECT OUTCAR

this will return the total numner of electrons :)

hallo,

how Ican calculated band gap from DOSCAR?

can you help me?

Hi Marsel, this is not the right thread for this question.

Also, It has been asked (and answered) before:

https://www.researchgate.net/post/How_might_I_calculate_the_exact_bandgap_from_DOSCAR

Hallo Michael,

Iam neu user for vasp.

I am doibg bandstrukture for interface GaP/Si.

Can I send you my work with Doscar and you can tell me please if there wrong with me.

Hi Marsel,

I can help you with specific questions. I am not here to teach you how to run DFT calculations in general or how to calculate a bandstructure. The theoretical background for this can be found in numerous books or online resources and most of the technical /practical questions can be solved using the vasp wiki: https://cms.mpi.univie.ac.at/wiki/index.php/The_VASP_Manual

Hi Michael,

thank you very much.can i send you just Doscar and tell me how i can find bandgap.

Dear All,

In VASP by command "grep NELECT OUTCAR" we can get total number of electrons in system. In case of spin polarized calculation, how can we know number of spin-up electrons and number of spin-down electrons in VASP.

Hi @Shilendra Kumar Sharma

In the file DOSCAR, for spin-polarized calculations each line holds five data

energy DOS(up) DOS(dwn) integrated DOS(up) integrated DOS(dwn)

plot the integrated DOS for spin-up and spin-down versus E-Ef (eV) then at E-Ef =0 you can find the number of spin-up electrons and number of spin-down electrons . Look to figure 8 in this paper

Article Ab Initio Investigations of the Structural, Electronic, Magn...