Different LongDrXd orientation

[nthiliniek]

Hi All,

I am confused about choosing the right direction of the electric field for RPA/TDDFT calculations. I wanted to apply the limit parallel to the TiO2 surface. According to the documentation of the RPA-local field, you can put the limit as 1 0 0 (LongDrXd). This is along the x direction.
link: https://www.yambo-code.eu/wiki/index.php/Local_fields

But, according to the documentation of TDDFT-LRC, you can set the limit as 1 1 0 (q).
link : https://www.yambo-code.eu/wiki/index.ph ... ed_kernel)

In both cases, they applied the polarization parallel to the plane. What is the difference and how do we select the right one?

Thanks in advance.

[Davide Sangalli]

Dear Niranji Ekanayake,
the vector LongDrXd defines the direction along which the dielectric function (epsilon) and/or the polarizability (alpha) is evaluated.

So, if you set e_x=(1,0,0), the code will give in output epsilon_{xx} (and/or alpha_{xx})
In general if you define a direction via the vector e_i, the code will give in output epsilon_{ii} (and/or alpha_{ii}).

If your surface is isotropic in the plane, e_x=(1,0,0) or e_y=(0,1,0) or any vector e_i in the plane will give the same result.

Best,
D.

[nthiliniek]

Hello Davide,

Thank you so much for the nice explanation!
I calculated dielectric function for my system in the level of IPA. When I apply the limit along X (1 0 0) and Y (0 1 0 ) , I got two different results. That means my surface is anisotropic. Does it mean that I should put ( 1 1 0) in my case to cover both directions?

My second question is,
2) My system is a two-dimensional surface. So I have already used a 15 angstrom (~28 Bohr) vacuum space in PWSCF calculations to remove unwanted interactions between repeated images. Therefore, do I still need to use truncated coulomb potential for TDDFT -LRC and RPA calculations? or can I avoid that?

Thanks in advance!

[Davide Sangalli]

I calculated dielectric function for my system in the level of IPA. When I apply the limit along X (1 0 0) and Y (0 1 0 ) , I got two different results. That means my surface is anisotropic.

Yes, unless there are issues with the simulations.

Does it mean that I should put ( 1 1 0) in my case to cover both directions?

In case the anisotropy is confirmed, it means that you have to check what is measured experimentally.

My system is a two-dimensional surface. So I have already used a 15 angstrom (~28 Bohr) vacuum space in PWSCF calculations to remove unwanted interactions between repeated images. Therefore, do I still need to use truncated coulomb potential for TDDFT -LRC and RPA calculations? or can I avoid that?

It is safer to use the truncated Coulomb interaction. Please notice that also QE as some kind of truncated coulomb interaction.
Probably, with that, you can reduce the amount of vacuum.

Best,
D.

[nthiliniek]

Hi Davide,

It is safer to use the truncated Coulomb interaction. Please notice that also QE as some kind of truncated coulomb interaction.
Probably, with that, you can reduce the amount of vacuum.

This means that when I define the box Z, I can reduce the box side in the Z direction as below.

Code: Select all

CUTGeo= "box z"            # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere X/Y/Z/XY..
% CUTBox
 0.00     | 0.00     | 25.0    |        # [CUT] [au] Box sides

Is that correct? (Since I am using 5.0.4 version, "slab z" option is not available.)

Thanks
Niranji

[Davide Sangalli]

I mean that setting the cutoff with QE you could possibly reduce the vacuum even in the DFT run.

For the box z (if I'm not wrong), the value suggested is slightly smaller than the size of the supercell along the z direction.

Best,
D.

[nthiliniek]

Hi Davide,

Got your point!
This is just for further clarification.

Code: Select all

CUTGeo= "box z"            # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere X/Y/Z/XY..
% CUTBox
 0.00     | 0.00     | 17.0    |        # [CUT] [au] Box sides

Here, side of box in Z direction means just the size of the surface cell in the Z direction. That is not the size of the whole simulation box (cell +vaccum) in Z direction, right? I have attached a part of my PW input to have an idea about my simulation cell.

Thanks
Niranji

[Daniele Varsano]

Dear Niranji,

the Zcut should be just slightly smaller than your simulation box (system+vacuum).
Anyway, if your system is 2D, I strongly recommend using the geometry "slab z" that is present from yambo 5.1

Best,
Daniele