Thank you for the question. I would recommend that you check your material fits:
For the GST, if you only have a single data point then you can use the “object defined dielectric” material, and input the index at the data point (this can be a complex number). It also does not appear that the paper has specified which material model for Ag was used.
Also, because you are only interested in results at a single wavelength, I would recommend that you use a narrower range around that wavelength for your material fits. You also don’t need to record analysis group results over this entire spectrum, you can just use 1.55 um for the start and stop wavelengths. A narrowband simulation also allows you to use Bloch boundaries instead of BFAST.
It also appears that the results are obtained using a 3D rather than a 2D simulation. I am not sure if this would affect the results, but if you are trying to replicate the paper’s results you should try to use the same parameters that they did. The paper cites references 9 and 22 when discussing this simulation, you could check these references to see if they have more information on the simulation parameters.
I hope this helps. Let me know if you have any questions.