Hi,
I have a problem with using PMLs (Perfectly Matched Layers) on the sides of a simulation. We want to simulate finite microstructures but first we decided to simlify the problem to a well-known one. We try to to recover reflection coefficient R for TM polarization on a glass-vacuum interface which case is commonly known from Fresnel equations. I know that periodic boundary conditions are perfect for this case (and it is studied in examples and I can solve it in 2D geometry with periodic conditions) but it should be also possible to obtain a proper R versus angle of incidence (theta) relation with PMLs.
I tried to set the simulation in the following way (in RF Module, Comsol 4.4):
1. I use 780 nm light so I set the PML thickness to 800 nm.
2. The simulation domain is quite big, 40 um of length and 2x20 um of height (glass + vacuum). Maybe it's not enough?
3. I use TM-polarized plain waves.
4. In EWFD I set "In-plane vector" (for TM) and "Full field" (fig. 1).
5. Port 1 (incoming light) is set to "User defined" with H0 z-component set to 1 (fig. 2), where
kincy = kinc*cos(theta) = emw.k0*n_glass*cos(theta) (defined in Variables in Definitions).
6. Port 2 (reflected light) is shown on fig. 3, where krefy = kinc*cos(beta). beta=asin(n_glass*sin(theta)/n_vac)
7. We look at the R(theta) relation on a 1D Plot.
Does anyone know what is wrong in this approach? Something certainly is as we cannot get the proper, well-known results. I really got lost with these PMLs :/
I have a problem with using PMLs (Perfectly Matched Layers) on the sides of a simulation. We want to simulate finite microstructures but first we decided to simlify the problem to a well-known one. We try to to recover reflection coefficient R for TM polarization on a glass-vacuum interface which case is commonly known from Fresnel equations. I know that periodic boundary conditions are perfect for this case (and it is studied in examples and I can solve it in 2D geometry with periodic conditions) but it should be also possible to obtain a proper R versus angle of incidence (theta) relation with PMLs.
I tried to set the simulation in the following way (in RF Module, Comsol 4.4):
1. I use 780 nm light so I set the PML thickness to 800 nm.
2. The simulation domain is quite big, 40 um of length and 2x20 um of height (glass + vacuum). Maybe it's not enough?
3. I use TM-polarized plain waves.
4. In EWFD I set "In-plane vector" (for TM) and "Full field" (fig. 1).
5. Port 1 (incoming light) is set to "User defined" with H0 z-component set to 1 (fig. 2), where
kincy = kinc*cos(theta) = emw.k0*n_glass*cos(theta) (defined in Variables in Definitions).
6. Port 2 (reflected light) is shown on fig. 3, where krefy = kinc*cos(beta). beta=asin(n_glass*sin(theta)/n_vac)
7. We look at the R(theta) relation on a 1D Plot.
Does anyone know what is wrong in this approach? Something certainly is as we cannot get the proper, well-known results. I really got lost with these PMLs :/