I checked your simulation file and made some changes:
- PML: adjusted distance from the structure to the PML so that it is at least half the maximum wavelength; also used “steep angle” settings
- Improved material fit as you can see in the Material Explorer.
- Mesh override region: extended it to cover the entire TFSF source since it is convenient to use a uniform mesh in the directions normal to the propagation direction as explained here.
- Global monitor settings: added more frequency points to resolve the resonance better.
- Added a sweep over position of the gold rectangles.
A similar case is discussed here.
After these changes (see dimer 20nm gap_modified.fsp (325.2 KB)) I got these results for the absorption cross section:
You can see the peak shifting to longer wavelength (red shift) for decreasing gap.
Hope this helps!