Hello everyone, I am simulating reflectance spectra (lambda range 400 nm to 1100 nm) of a periodic structure of dielectric spheres (triangular lattice of a monolayer) on a silicon substrate in "The COMSOL multiphysics wave optics module." The periodicity of the unit cell is around 400nm. I included diffraction orders and implemented PML. Instead of using silicon as a domain, I used impedance boundary conditions at the bottom with silicon material assigned to the boundary(exit). I am getting unphysical reflections at oblique incidence (not in normal incidence). But if I replace the sphere with the flat dielectric layer, those artifacts will not arise even at oblique incidence. It happened when I placed a sphere in the periodic structure.

Moreover, the sharp dips are red-shifting if the angle changes from 45 to 75 deg. Here, the impedance boundary is not the problem; when I simulated with a flat, thin film, I used the impedance boundary instead of the silicon domain. The results of the thin film matched the experimental data (I verified by implementing the silicon boundary; the results are the same in the case of thin film). The problem appears only for the sphere. Is there any way to solve this problem?

Thank you.