Some BPM codes do not handle backward waves or strongly scattered waves.  Of course, some can.  As long as your BPM code incorporates the additional physics that your "geometric features" introduce, the answer is yes.  BPM can actually be formulated be practically rigorous.

Thank you Raymond. So I understand you suggest to check if my BPM solver incorporates backward waves. I will. Thanks again.

BR

Stefano

If you want a little more information on the method, check out Lecture 16 here:

http://emlab.utep.edu/ee5390cem.htm

Maybe you should be more specific about the loss origins you want to simulate? A mode solver allows you to calculate the overlap with the various materials, and, hence, the material loss contribution. With conformal mapping you can simulate bend loss. This is in addition to the previous comments by Raymond Rumpf.

well, I need to optimize the design of a slot-waveguide in terms of highly concentrated single-mode propagation and minimized optical loss. Since the waveguide is gonna be a modulator, I need to include metal electrodes and since the final device is purposed for a Silicon photonic application, I need to minimize the distance between the electrodes to have a low pi-voltage require dfor modulation.

I know that setting electrodes too close to the waveguide may result in optical loss, so I need to optimize the distance between the electrodes to make it far enough to minimize loss but also close enough to minimize the pi-voltage needed for modulation.

Thus , the origin of the loss should be due to the propagation of the optical field out of the waveguide.

I am not caring of bending at this stage as the modulator will be based on MZI.

Then especially the doping concentration and doping profiles are important to consider, part of which will be inside the waveguide. I assume there will be polymer inside the slot for increased electro-optic effect?

@Martijn: yes it is a EO polymer. If I consider the doping concentration, it should reflect onto the permittivity (RF field) or the refractive index (optical field). Anyway, for what it is reported in literature, for low doping rates (1015 atoms/cm3) bopth eps and n are constant. I am wondering if the doping rate affects also other parameters I am not considering.

@Emel: I am using Beamprop, but also I recently started using COMSOL, for RF field and optical field overlap. Can COMSOL also simulate propagation along the waveguide ? As far as I know FDTD can't do it. Concerning with Lumerical, some friends in Italy use it. I am very curious about it , so I guess I will start an evaluation soon. Do you suggest it ?