Learning Forum

[Areum Kim]

Hi,

I'm currently attempting to conduct RCWA simulations for a monolayer material (thickness <0.7nm) embedded in an all-dielectric high-Q metasurface. Given the necessity for a fine mesh due to the thinness of the monolayer, FDTD simulations are excessively time-consuming. Thus, I aimed to expedite the process by employing RCWA. However, I've encountered significant disparities in the results. While the overall trends appear similar, there's a notable discrepancy in the peak positions, even after increasing the k vector in RCWA simulations. Could anyone offer insights or suggestions on additional steps I could take to align the results more accurately?

 

[Amrita Pati]

Hi Areum, 

I believe the a k-vector value of 15 is not going to be enough. What is the maximum number of k-vectors you have used so far? Usually, we ask customers to perform convergence testing i. e. keep increasing the k-vector as until the results converge. If on increasing the number of k-vectors, you see changes in the spectrum that most likely means that the k-vector value is not high enough. Does the geometry have same periodicity in the transverse directions? Also, is there a large index contrast between the geometry and the surrounding? If that is the case, I will ask you to use the tangent vector field method.

Regards,
Amrita

[Areum Kim]

Despite increasing the k-vector value, I encountered two issues: either the simulation time became as lengthy as FDTD simulation, or the simulation stopped abruptly. Additionally, I couldn't locate the "use tangent vector field" option when opening the simulation in Linux. Any insights on how to address these challenges would be greatly appreciated.

[Amrita Pati]

For what values of k-vectors do you see the solver crashing? Also, what does it say in the log file (stored in the same directory as the simulation file, has the name simulationfilename_p0.log) when the solver crashes? I believe the use tangent vector field was introduced in 2023 R1.

Regards,
Amrita

[Areum Kim]

When I run the RCWA simulation wiht k of 30 with using the tangent vector field, I got this error.

[Ruzan Sokhoyan]

Hi Amrita,

 

Could you please address this comment? Does running RCWA require large physical memory?

 

Thank you!

[Amrita Pati]

Hi Ruzan,

I am not sure if this is because of memory. I think it could be due to some other reason. Are you using the tangent vector field method? What mesh refinement type are you using along with it? Conformal variants or something else?

Regards,
Amrita

 

[Amrita Pati]

The "use tangent vector field" method should only be used with Conformal mesh 0,1 or 2. In one case I have seen that, if you are using any other kind of mesh such as staircase, the solver automatically adds multiple digits (for example, an index of 1.5000000000000001 as opposed to simply 1.5) in the material index. Since, the index should be single precision, having double precision would throw a memory error. But this used to be the case for older releases.

I think in the newer releases, the log file would tell you to use the conformal mesh. Please feel free to share additional information, if you think something else is causing this issue.

Regards,
Amrita