Learning Forum

[ph23resch01002]

Hi! I am attempting to replicate the results presented in the paper titled 'All-dielectric planar chiral metasurface with gradient geometric phase.' As stated in the paper, I have observed higher transmission for left-circularly polarized (LCP) incident light propagating in the backward direction, while the transmission is very low for right-circularly polarized (RCP) incident light. Now, I aim to leverage geometric phase to achieve desired phase and amplitude values. When I vary the orientation of the meta-atom from 0 to 180 degrees, the transmission coefficients exhibit significant variation. I have attached the relevant figure as well. Therefore, I intend to optimize the parameters of the meta-atom in addition to its orientation angle. Is there a method within Lumerical to perform parameter sweeps for optimizing both the parameters of the meta-atom and its orientation angle simultaneously? In that paper, which I am trying to replicate, they have optimised the parameters of the meta atom as well as the orientation angle. 

Can anyone please help me with this? Looking forward for the reply.

[Guilin Sun]

Yes, definitely. You can directly optimize a few parameters. please refer to this example: Optimization utility

[ph23resch01002]

Thank you very much for the reply, Sir! 

I've learned how to optimize multiple parameters (w1, w2, and theta) to achieve either high or low transmission coefficients for our problem. Now, I'm working on creating a meta-atom library consisting of 9 meta-atoms with nearly same transmission coefficients and phase values ranging from 0 to 2*pi, varying with different values of w1, w2, and theta.

In the metalens examples provided on the Ansys Lumerical website, variations in phase shift are observed when the unit cell's dimension, such as radius, is varied. I used a particle swarm optimization algorithm to determine appropriate dimensions for achieving nearly identical transmission coefficients and phase values ranging from 0 to 2*pi. However, my situation is more complex as it involves varying multiple dimensions simultaneously.

My question is: How can I save the phase values corresponding to each set of dimensions? Is it possible to concatenate all 9 sweep results and have the phase values together? 

Hope,I conveyed my question properly.

Looking forward to your reply! I really appreciate any help you can provide.

[Guilin Sun]

I guess you mean you are doing sweep in order to get the phase values corresponding to each set of dimensions. If so, you will need nested sweep:

https://optics.ansys.com/hc/en-us/articles/360034922913-Creating-nested-parameter-sweeps

Optimization on the other hand can reach your predefined FOM, which will not save result for each parameter.

"Is it possible to concatenate all 9 sweep results and have the phase values together? "

Not sure what does this mean. For each sweep, such as diameter, height, can be cascaded as illustrated in small metalens example in "height" sweep. Please refer it and let me know if you have questions.

 

Best wishes

[ph23resch01002]

 Greetings of the day sir! In the metalens example, There are three parameters. They are period, height, and radius. The period is fixed as 0.35 micrometer and the parameter sweep is done for two parameters height and radius. After performing the parameter sweep, height is fixed too, the radius is varied to get the phase shift ranging from 0 to 2*pi.

In my case, I get a Transmission peak for LCP light at 1658nm wavelength for the dimensions of the meta atom. At this wavelength, I changed the orientation of the meta atom from 0 to 180 degrees, I got the phase shifts from 0 to 360 degrees. However, the Transmission coefficients vary drastically. So I want to optimize the other two parameters w1 and w2 along with theta values. In the metalens example, they have fixed the height and varied only the radius to get the appropriate phase shifts and constant transmission coefficients. In my case, what if I want to choose different w1 and w2 values along with theta values to get the phase shifts along with the high transmission coefficients?

Is nested sweep the only way?  

[Guilin Sun]

This is a different question. In case my following reply does not resolve it, please write a new post.

In theory there is no strict limitation for the atom dimension. In reality from manufacturability point of view it might be better to use the same atom, depending on the fabritcatio processes. So you can choose w1 and w2 under given height to produce the phases required, and with roughly the same transmission at different theta.