Page 1 of 1

Dielectric Tensor for Interface

Posted: Mon Mar 07, 2011 4:08 pm
by Gabi
Dear all,

I want to calculate the dielectric tensor at an heteroepitaxial semiconductor interface (IF) GaP on Si.
But first of all, I'd like to explain my procedure:

*I run a ground state (GS) calculation of that IF, realized by a multilayer modelling (...si-layers--gap-layers--si-layers--gap-layer...) as you can see in gap-si-p2.in
*Then I run a response function calculation to get the momentum matrix (ddk) as it is described in the optic-tutorial. (see for p.in)
*finally I run an optic calculation to get the frequency dependent dielectric tensor (see for p-optic.in)

The calculation itself run without errors. But my problem is:
Actually I am interested in electronic dielectric tensor elements at this IF, that is, there must be direction-dependent tensor elements (element 11 is unequal element 22, if I consider the plane (001) to be plane of Interface). But due to the model there are 2 IF's in the unit cell, and furthermore, due to the zinkblende structure [100] directional properties of the one IF are the same as those of the [010] direction of the other interface. Therefore, differences seem to "sum up " (I hope, this bad description isn't missunderstood!!) and finally there is no difference in 11 and 22 elements of the dielectric tensor. Ideally, I would like to calculate this perturbation for just one IF, e.g. for atoms 4 to 13, but I know that is not possible.

Could anybody help me? Is there another way to calculate dielectric tensor over a frequency range for just a part of my model?

I know this sounds little awkward, but I'd appreciate any help or suggestion!

Kind regards,

Gabi

Re: Dielectric Tensor for Interface

Posted: Wed May 25, 2011 5:27 pm
by mverstra
Hello, interesting problem, though I'm not sure if the dielectric constant is well defined for such a small system.

you could do a slab, introducing vacuum instead of the interface you want to remove, but then you have the response of your "IF" and 2 surfaces - you might also have to look at the response of the different surfaces for the 2 component subsystems... In some cases these things add up, but this is far from guaranteed.

Matthieu