Bader charge by using VASP and Siesta

[xi007]

Dear All,

Hi, recently I used vasp to calculate the bader charge for SiC and compared the results with those in Siesta. But I got a very different result. In vasp, I used the GGA function to get the bader charge, and my result is that the charge around Si is about 6.7e and C about 1.3e. But compared with some references using Siesta, the result is about 4.532e and 3.477e for Si and C, respectively. I am very confusing about these results. In my case, the PAW method is used in VASP, while the atomic orbitals are used as basis set in Siesta. Does it really affect the Bader charge results? And how can I get a robust result? I hope you can give me some suggestions. Thanks very much!

[graeme]

Make sure that you are including the core charges in both cases.

[xi007]

Thanks for your quick reply. Yes, I used the method in your website. By adding LAECHG=.TRUE. in the INCAR, then get these AECCAR0,1,2 files, and then chgsum.pl AECCAR0 AECCAR2 to get the CHGCAR_sum, after that, use bader CHGCAR -ref CHGCAR_sum to get the ACF.dat result. I think the process is no problem.

In my opinion, the magnitude of charge transfer depends on the electronegativity of each element. I check the difference of electronegativity between Si and C is very huge, this may induce that result. But I also check the example in your website for NaCl including core charge. It shows that the bader charges for Na and Cl are around 8 and 1, respectively. This is opposite to the electronegativity. It's very wired.

[graeme]

For charge transfer you need to consider the difference between the calculated bader charges and the number of valence electrons in your electronic structure calculation. This charge transfer will correlate fairly well with electronegativity.

[xi007]

Dear Prof,

Thanks very much for your reply. after compared with the number of valence electrons of Si and C, it is found that the transfered charge for Si and C are -2.67|e| and 2.67|e|, indicating that the electronegativity for C element is higher than that of Si. This is agreement with the reported data ( about 2.55 for C and 1.9 for Si element).

But I'm a little confused with the magnitude of transfered charge. When compared with the transfered charge between Na and Cl (about 1|e| in your example), which has a larger different electronegativity(about 2.23) between two elements than that between Si and C(2.55-1.9=0.65). It shows that the transfered charge in SiC is larger than that in NaCl. In my intuition, the larger the different electronegativity between two elements such as ionic compound, the larger the transfered charge between them. This always confusing me. I hope you can give me some suggestions. Thanks very much!

[graeme]

Ah, yes, I see what you mean. Well, I think that the electronegativity is essentially the first-order measure of charge transfer. Second order would be the hardness of the elements, or how easy it is to get electrons to or from an element. So, for example, H and F would have very different electronegativity values, but the charge transfer should only be 1 e because getting more charge from H is not possible and F would only like 1 extra electron.

[xi007]

Dear Prof,

I'm very appreciate for your suggestions. Maybe that's the reason to get these results. Thanks again!