Plasmonic waveguide bandstructure

I’m trying to simulate a metal-Insulator-Metal structure from J. A. Dionne, L. A. Sweatlock, and H. A. Atwater and A. Polman, “Plasmon slot waveguides: Towards chip-scale propagation with subwavelength-scale localization”, Phys.
Rev. B 73, 035407 (2006). I based the simulation on the glass-silver-air plasmon band structure example.
I used an anitsymmetric boundary condition I’m able to get the mode resonance peaks, however if i use symmetric conditions the modes do not show up. In the paper the anti symmetric modes are low loss, and the symmetric modes are very lossy so i’m guessing this is the problem. I’ve tried messing around with the simulation time, apodization, and center frequency of the source, but i cant seem to get the peaks. The mode should be evanescently decaying in the direction of the propagation through Bloch boundary conditions, would that cause problems? this is the dispersion i got which does not match the paper. There are some modes at lower freqs but i removed them from the source since they weren’t the ones i wanted either.

these are the symmetric and anitsymmetric curves from the paper

sp_film_dispersion_sym.fsp (287.0 KB)

1 Like

Dear @michael.grayson

I am not too familiar with the topic, but hopefully with your help we can achieve the results of interest.

I increased simulation time to 200fs and did not use any symmetry and here are the plot that I got (with sweep number of 30):

If you use symmetry (asymmetry) BC, you will see only the second (first) band of the plot above.

The plot seems to be similar to the screenshot of the paper, however, I can not see the “curvy shape” in band-structure. Can you please let me know if the results make sense and/or if you have any idea why these plots are not identical to the ones in the paper?


Yeah i got that before but it does not match the results of the paper. The symmetric modes seem to be too lossy to show up in the time domain solution. I switched to the mode solutions solver and have been having a little more luck by searching near indexes that have large imaginary components. I’m trying to writing a script that will search for the solutions right now