Learning Forum

[Albi.Pani]

Hi, 

I am trying to study geometric phase in non linear optical process. For example:

Nonlinear Wavefront Control by Geometric‐Phase Dielectric Metasurfaces: Influence of Mode Field and Rotational Symmetry - Liu - 2020 - Advanced Optical Materials - Wiley Online Library

Essentialy we have a wave and a non linear material with different shape and we study the phase of the second harmonic wave (with different polarizations) as a function of the rotation of the material. 

My question is: how much is reliable Ansys to study a problem like this?

For example I am testing selection rule for non linear optical process:

 

I have a material with C2 symmetry. The fundamental wave has right circular polarization (RCP).  For the harmonic order n=1 I have the same results of the theory, no wave with RCP and a generated wave with LCP with the phase indicated, where theta is the rotation angle of the metastructure. 

But If I study the spectrum for the other orders with a initial wavelength of 1200 nm:

I see that there is also the harmonic order n=2 that is not expected considering the theory. 

Are there any suggestions?

[Guilin Sun]

It seems such material and physics can be included in FDTD. thus the simulation accuracy depends on the mesh, PML and other settings.

For higher orders unexpected, it is because the amplitude of the origial signal is too high. you can either reduce the source amplitude, or reduce non-linear coefficient. Theory usually have assumptions. even with perturation chi*E2 is small, it does not guarantee no higher orders can occur.

[Albi.Pani]

 

Thank you.

 

 

[Guilin Sun]

For other seeingly not expected, it might be omitted in theory after simplication. In addition, please check the non-linear material model and make sure it can describe this physics withoout other assumption.

[Albi.Pani]

Yes I am checking right now. Thank you very much Guilin Sun. 

[Albi.Pani]

Hi, one question concerning this topic:

As I have written, I need to simulate a non linear optical process with circular polarized light. Normally the circular polarized light is simulated with two linear polarized electric fields (Circular polarization and phase convention – Ansys Optics), one shifted with respect to the other.

In linear process we have that the final result will be the sum of the results of the first and second electric field.

But with non linear optical process (for example second harmonic generation) this is not valid, the result will be proportional to (E1+E2)^2 not E1^2 + E2^2 .

Is this thing considered in the simuation?

 

[Guilin Sun]

I am not sure how did you sum the results.

An easy way is to set the circular polarization with two linear polarized sources to do the simulation, instead of simulting them separately and then sum them.

[Albi.Pani]

Hi,
I am not simulating them separately, but I simulate circular polarization with two linear polarized sources, as you suggested.

The point is that the nonlinear optical processes with circularly polarized incident light are different from those due to linearly polarized light, and because of the nonlinearity, the former cannot be obtained from the latter or vice versa by a simple superposition process.

But when you say that the easy and standard way is to set the circular polarization with two linear polarized sources, I am wondering if Ansys calculates the nonlinear optical process with a simple process of superposition because it would be wrong.

What is your opinion? How does Ansys work in this case? 

[Guilin Sun]

Once you use the circular polarized excitation in the simulation, everything is taken into account by FDTD. You do not need to worry about nonlinear optical process or superpostion. The physics occurs naturely with FDTD which solves the Maxwell Equations with proper material models, as long as the boundary conditions are set correctly.