How to modeling Au on SiO2 slab

[SHANG_Yuxuan]

Dear Profs and researchers

I have a Au on SiO2 system.

From our experiment characterizations we found that there should be some negative charge on Au surface.

However, after I construct Au 111 on SiO2 001, I found Au surface has positive charge (through Bader charge analysis).

I guess the reason is, on the real world side, proton from silica hydroxyl group could leave during the synthesis step and oxygen can provide the negative charge( -Si-OH -> Si-O- plus H+), but on the VASP side, if I just use the normal POTCAR for oxygen, there's only 6 electrons in the valence. I cannot include the electron left by the silica proton.

How can I resolve this problem?

Thank you

[fabien_tran1]

Hello,

In principle, the POTCAR should not really matter. The valence electrons are not constrained to stay close to a particular atom, but they are free to move everywhere in the unit cell. How the charge density redistributes depends more on the theoretical method, like the exchange-correlation functional. The parameters like the size of the basis set or number of k-points may possibly also play a role.

Note that if the simulation cell has a total charge that is non-zero, then a special treatment has to applied (see charged systems with density functional theory and NELECT), however I understood that your simulation cell has a total charge of zero, right?

[SHANG_Yuxuan]

Thanks for you reply.

I tried NELECT. May I ask how can I make it consist with 'LDIPOL'?

VASP gives me an error said:

LDIPOL = .TRUE. must be selected only for cubic supercells, since the quadrupole corrections are currently only implemented for this specific geometry. Please change your box, or set LDIPOL=.FALSE.

But my lattice is a = 9.8766 Å b = 8.5534 Å c = 36.0000 Å

Thank you

[fabien_tran1]

What is the total charge of the system that you are simulating? Do you only expect a redistribution of the charge inside the simulation cell (i.e., a zero total charge)?

[SHANG_Yuxuan]

Hi,

I tried zero charge in the beginning and analyzed the charge redistribution.

However, the result (Au contain positive charge) doesn't consist with experiment one (Au contain negative charge)

By taking your advice, I tend to put more electrons in to the slab. But vasp gave me error said ' LDIPOL = .TRUE. must be selected only for cubic supercells '

Thank you

[fabien_tran1]

Hi,

Indeed, LDIPOL works only for cubic cells. The alternative is to use the Coulomb kernel truncation method (https://doi.org/10.1103/cd6s-cdkf), that is available since VASP.6.5.0. More details can be found at charged systems with density functional theory and KERNEL_TRUNCATION.

Related Discussions can also be found at
https://www.vasp.at/forum/viewtopic.php?t=20176
https://www.vasp.at/forum/viewtopic.php?t=20223

[SHANG_Yuxuan]

Thank you very much.

I'll check it out.