Good questions. In the air the optical wave will depend exponentially on the radius, as \gamma . The coefficient in the exponent depends on the modal index according to the attached equation.
If the width of the air layer is several times the distance \gamma , the results should be insensitive to the width of that layer. I don’t see results changing with that width in the samples I use.
2) ‘step’ in the fiber/waveguide parameter window is used in the case the layer is not constant. The layers (e.g. core, clad …) can be functions like linear, parabolic… and so on. If they are not constant, the simulator must approximate them with a sequence of thin constant layers. “Steps” specifies how many such layers should be used.
3) When LP(0,1) and LP(0,2) are selected, there will be two reflection spectra and two transmission spectra, one for each mode. This is practical to plot in one graph. If more are selected, it becomes too difficult to plot all of them, so in this case OptiGrating will plot only the transmission of the incident mode, since it is presumed that is the only spectrum of real interest. The selection of modes in the calculation of the spectrum is meant to specify which modes are expected to participate in the propagation. There are often hundreds of cladding modes. If all were calculated and included in the calculation OptiGrating simulations would be very slow. On the other hand, the user is expected to know which modes are going to be significant. Making the right choice in the selection makes the simulation both fast and accurate.