test.fsp (464.6 KB)
Hello, I am trying to simulate a simple nanoparticle array in which I followed the examples on Lumerical website. My issue is that the R & T spectra are different than experiment (>100nm blueshift with only one dip appearing in reflection).Can someone please have a look at my attached file. I am using Bloch BCs because I needed to get the angular data later after confirming the zero angle result is ok. Regards,
test.fsp (464.6 KB)
I’ve tried adjusting multiple parameters. Decreasing the mesh size from 10 to 5nm changes the number of peaks I see. I’ve set though the mesh override of the upper disk to 3 and the layer to 2.
Still as you see the T spectrum is largely below what I measured (60nm shift). I’ve looked at the high resolution images of the units and it should be in a very close match to the design parameters.
The user defined film is not very smooth though and is conductive, how do I roughen the interface in the simulation? and what to consider as a change in the parameters?
Also the T spectrum at an angle of 10 is ripply, why?
Thank you for a detailed post and providing screenshots.
As you mentioned, when you were changing the mesh accuracy, your results have been varied. As part of convergence testing, you need to make mesh fine enough so that results do not change by further decreasing them.
In your simulation file, mesh was large and it was not resolving the adhesion layer. Generally, you want to have at least 5 mesh cells to properly resolve an object in FDTD.
The FDTD boundary was also close to Particle structure in z direction. You need to leave at least half the wavelength between the edge of critical object and PML boundaries. This is explained here:
Doing the above mentioned modifications, here is what I obtained using a relatively fine mesh for transmission of T monitor:
If the results are different than what you expect, we can try using a better material fitting.
My modified simulation file:
test_modifiedBK.fsp (290.3 KB)
I hope this was helpful.