Request about correct modelling of SiO2

[QifengHu]

We are testing the performance of MEEP, in a grating structure with a long-wavelength range. The results of the pre-defined SiO2 are not desirable, since the valid wavelength range is limited from 0.25 to 1.77. We define a medium with fixed refractive index 1.46 and 1.55, the results is also not good in the long range, as followed:


We also tested the model of fused quartz, the transmittance in long wavelength range still impacts on the monitored region of transmittance.

Is it possible to implement the correct model of SiO2 with the indices in the long wavelength range?

[mdfeinstein]

It is possible. Materials in meep need to be fitted as a sum of Lorentzians (or Drude) susceptibilities. In the docs, they supply a script that can kind of do this, but it isn't the easiest to use. I wrote my own function based on that approach and incorporating much of that code, but its lacking some features I wanted to put in. I can try to share it soon as is if that might be helpful, but I use the nlopt package the script uses, which needs to be downloaded and built, so its some effort to get up and running. (Side note: I think scipy has fitting functions that would be sufficient and much easier to just conda install into your environment, but I didn't test that.)

In the meantime, below is the code to define silica in the near/mid-infrared. I fitted from the Kischkat data. There is good agreement for the range you are interested in, but I will note that the absorption bands peaking at 9.3 um are too simplistic in this fit. Its really a few Lorentzians, but this fit sees it as only one.
SiO2 = mp.Medium( epsilon = 1.1, E_susceptibilities = [ mp.LorentzianSusceptibility(frequency=3.8146083476490666, gamma=4.301154772189638, sigma=9.422139872583635e-181), mp.LorentzianSusceptibility( frequency=0.10525565241548465, gamma=0.006728273595379233, sigma=0.6703341698663918, ), mp.LorentzianSusceptibility( frequency=1.53822387094153, gamma=0.0, sigma=0.9250921216802024, ) ] )