Is odd number of electrons OK for the eelctrode and central region??

[jjhskang]

HI!

It seems that

(1) An electrode should have odd number of electrons for spin-unpolarized system. is this correct?

Fore  Au(111) wire, I designed it in such a way that each surrface has 7 atoms. Since three layers form a repeating unit along (111) direction, there are 7 X 3 = 21 Au atoms in the primitive cell of Au nanowire, where there are 21 x 11electrons/Au atom = 231 electrons in it. Therefore, I think that I should use even number of the primtive cells of the wire to be able to make the electrode have even number of electron. is this correct?

(2) I do not understand why I do not need even number of electrons in the central scattering region for nonmagnetic system like O-doped Al wire shown in the tutorial too? Am I not understanding it correctly?

[zh]

HI!
(1) An electrode should have odd number of electrons for spin-unpolarized system.

This statement is not correct because the system under study is extended one, rather than isolated one (e.g. molecule or atom).

[jjhskang]

Hi!

I made some mistake in typing my question above. Let me restate it.

(1) Is it OK to have odd number of electron in the electrode for nonspinpolarized system? I do not think so. Electrode is treated as a kind of bulk, whose periodic calculation definitely require even numer of electrons.

(2) The Li-H2-Li example shows that it is OK to have odd number of electron in the scattering region. That's a little strange to me. can you give me more comments for this ?

[Anders Blom]

The relation between number of electrons and magnetism is not like that. Gold has 79 electrons, but is not magnetic (spin-polarized), even with 1 atom in the fcc unit cell. Iron, on the other hand, has 26, but is magnetic.

There is a relation between semiconductivity/metallicity and odd/even number of electrons, but not to magnetism.

[jjhskang]

Hi!

You are right for 2D or 3D crystals. However, I mean that Peierls deformation should occur for 1D system. I wonder if the deformation should occur for quasi-1D system, not exactly 1D system. That was my question. If that occurs, the nanowire can be insulating instead of being metallic.

[Anders Blom]

That is indeed an interesting question, and I think it's definitely an effect you can obtain with ATK. You can for instance see it for graphene nanoribbons, which are metallic without spin included, but running a spin-polarized calculation opens up a band gap. (Not sure that would be called a Peierls distortion, those are perhaps primarily geometric, but the effect is the same.)