Learning Forum

[Samantha]

I am checking the advanced surface roughness model example in Lumerical website:

https://support.lumerical.com/hc/en-us/articles/360034382374-Tips-for-adding-surface-roughness-to-structures

[Guilin Sun]

A1: For isotropic roughness, such as a large rectangle surface, it is 1D correlation. If it is narrow rectangle, you will need 2D model.
A2: They are used to create standard distribution (0,1), meaning the average is zero, similar to standard distribution.
A3: yes, it works as long as the absorption is correct. What you need to do is to set the monitor larger enough to enclose the whole absorption region, including the rough surface. In case the other material is also lossy, you can modify the analysis group to only calculate the absorption for the main material that contributes to the current later in CHARGE.
A4: as long as you can set up the rough surface in CHARGE, yes you can. But we have not tested, as the roughness will need a lot of local fine mesh, and in some cases it can create poor finite elements so it may need more iterations.

The roughness model we developed is for illustration purposes. In reality, users may need much more time to get the statistical data to convert to a proper distribution, and all necessary characteristic parameters, such as mean value, standard variation, correlation length, density etc as the same mathematic models can be realized by different actual surfaces.