Discussion Forum

Hey, I am trying to simulate a rather peculiar EM propagation problem using the beam envelope method. Here, I have a gaussian beam porpagating in free space which encounters an refractive index interface, but unlike usual fresnel reflection, the interface is parallel to the propagation direction. I.e. half the beam now sees material 1 and the other half sees material 2 (cartoon image attached as Fig.1) some priliminary numerical modelling suggests that we get angular splitting here. The beam splits into two beams propagating away from the the surface. I have cast the probelm in comsol and I observe something similar for the cases where the two materials have very close refractive index. In my particular model I have selected n1=1 (air) and I vary n2. the model seems to be well behaved (Fig2 attached) when n2=1.005 but when I bump up the refractive index of the other material to n2= 1.5 (which is what I want) the simulation is filled with reflections and it seems to be not so well behaved (Fig3 attached). The prime susupect is ofcourse boundary conditions but I have tried all of them, PML, matched boundary and scattering boundary conditions, they all seem to give similar distorted results. If someone could hint me towards some ideas to choose the appropriate boundary conditions for this problem, that would be great! Cheers! I have attached the model with PML but I have tried other boundary conditions as well.