Numerical stability in SimulationPlanar

[yigithanmehmetkose]

Hello,

I carry out a planar simulation using these codes to find the flux absorbed by a layer in the near-field:

sim.SetProbeLayer(layer_name = 'InAs')

sim.SetProbeLayerZCoordinate(0)

sim.SetSourceLayer(layer_name = 'eSubstrate')
sim.InitSimulation()
sim.IntegrateKParallel()
phi_eS_c_u = sim.GetPhi()

sim.SetSourceLayer(layer_name = 'eARC')
sim.InitSimulation()
sim.IntegrateKParallel()
phi_eARC_c_u = sim.GetPhi()

sim.SetProbeLayerZCoordinate(t_c)

sim.SetSourceLayer(layer_name = 'eSubstrate')
sim.InitSimulation()
sim.IntegrateKParallel()
phi_eS_c_l = sim.GetPhi()

sim.SetSourceLayer(layer_name = 'eARC')
sim.InitSimulation()
sim.IntegrateKParallel()
phi_eARC_c_l = sim.GetPhi()

for i in range(1999):
    q_w_c[i] = (phi_eS_c_u[i] + phi_eARC_c_u[i] - phi_eS_c_l[i] - phi_eARC_c_l[i]) * theta[i]

In these codes, I aim to find the difference in flux between the upper and lower bounds of the layer of interest.

When I do that, I get negative results as follows:

Please note that there is no unphysicality in material properties I used.

Related to that, I have a follow-up question which I believe could be related. In some of the example codes, sim.SetKParallelIntegral is set to 10, in some others it is set to 100 or 500. Could that be the problem? Also, in the paper, there is such a comment: "In particular, since RCWA is known to have difficulty for low-loss metals, one should use MESH with caution as well when dealing with low-loss metals." I also use some metallic layers, would that worsen the situation?

Thanks

[kfrancischen]

Yes, for metals you might need to use a larger k. The best value needs to be tested through different runs.