Learning Forum

[Al Feg]

Hello, I tried all the convergence tips for this specific system and I am unsure of where the problem is coming. I am trying to simulate a silver nanorod that is 93nm tall and 17.5 diameter wide. I am using auto-non uniform with .125 min mesh step and maximum accuracy. I am exciting with a TFSF source from 200 to 700nm and fixed the fitting of the J&C Ag model.

From both the rod and the sphere, the scattering and absorption data (taken from the Scattering Cross Section analysis group spaced +- 20nm from the tfsf source) results in unphysical spectra, such as this scattering from the rod

 

or this scattering from the sphere

.

Mesh overrides, increasing PML layers, increasing spacings, and switching conformesh0 to 1 hasn't worked.

I have been stuck on this and unsure where the error is coming from. Thank you for your expertise and assistance

[Guilin Sun]

Silver at this bandwidth is extremely lossy, and needs finer meshes and long simulation time as well as smaller autoshutoff min. Suggestions for improvements are:
1: make sure the TFSF uses uniform mesh at least the mesh size a little smaller than the background automesh;

2: reduce the autoshutoff min to 1e-6, or even smaller such as 1e-8 etc;

3: increase significantly the simulation time and make sure the autoshutoff min terminates the simulation, before the progress reaches 100%.

4: PML, When you use higher mesh accuracy, the mesh size is small, and thus the PML total thickness might be small. Please make sure the PML thickness to be about half wavelength in order to have better absorption. This will need you to increase the number of layers. If the number has reached to 64, you may need to use "Custom PML", and borrow the PML parameters you are currently using. In addition, you can estimate the largest scattering angle. If it is large say to 70 deg, you may need to use "steep angle PML" with sufficient thickness. 

Make sure PML is in the uniform mesh region. 

5: please check the material fitting, with smaller tolerance and check if there is any artifical peaks.  Tips for improving the quality of optical material fits

Please note that, the  ".125 min" (I guess it is in nm, not um) mesh size is the mini limit of the mesh size to avoid accidently using too small mesh size. You can modify it if you need smaller size than it.

 

[Al Feg]

 

Hello Dr. Sun, I hope you are doing well, I wanted to follow up on my previous quesiton. I was able to succesfully get realistic LSPR peaks by reducing the autoshutoff and increasing PML layers. However, while the material fitting is correct, I am unable to properly simulate the resonance for the rods in question. They are large enough to support higher order resonances but my spectra are missing any response above 600 and simulating in the 600-1000 domain does not solve this for both Silver and Gold (with proper fitting) as shown in the spectra below. What would you reccomend to to fix this? Thank you again for your response and assistance.

 

 

[Guilin Sun]

I guess this is different issue from the previous post.

Do you mean you the simulated result does not show "quadropolar resonance "? do you know what is its deginition of quadropolar resonance ? usually you need to check the field distribution. please refer to this short articel I wrote previously:https://optics.ansys.com/hc/en-us/articles/1500007399722-About-monopole-dipole-quadrupole-hexapole-octupole-etc

If problem still exists, please write a new post, as forum policy requires one post one question.