Simulation with broadband wavelength range
I am trying to simulate the absorption of silicon with certain thicknesses, with and without surface texture. I started with just flat silicon with thickness 10 um on top of glass.
First, changing the wavelength range changes the absorption spectrum (as in the figure below: the legend indicates the wavelength range in um). I understand that changing the min wavelength will change the mesh, but how would I know which is more accurate? I did convergence tests as in this link (https://support.lumerical.com/hc/en-us/articles/360034915833-Convergence-testing-process-for-FDTD-simulations) but nothing seemed to make a difference in the results.
Second, in all cases, I am pretty sure the absorption in the long wavelength range (>850 um) is overestimated. Only 10 um of silicon will not absorb that much in the NIR. Absorption should decay to zero. I am also not sure how to fix this.
Simulation Notes: my structure currently is basically 10 um of silicon on glass. I am doing 2D simulation. I calculate the absorption from the reflection and transmission monitors on top of the structure and at the silicon/glass interface, respectively. I am using periodic boundary conditions with 100 nm FDTD region width.