I just checked with Sonnet EM for a simple through line with 0.1mm, 1mm and 2mm width. The radiated field for 2mm wide line is about 10dB larger than the field for the thin 0.1mm line.
So it seems than going for a thinner substrate has both direct and an indirect effect (narrower lines for same Z0) that reduce radiation.
a) Thicker substrate is better as a radiator: If the substrate is thinner, the signal line and ground are closer to each other, providing tight coupling of the fields between them, making it a better guiding structure..if substrate thickness is high, the coupling of fields between signal to ground is not that strong, and also the fringing field from sides is comparable as well..so its more radiation prone.
b) Lower dielectric constant is better radiator: Permittivity is a quantifier which states how much a dielectric permits electric fields through it.. So higher epsilon will have more fields through substrate and less fringe fields through air, helping in better guiding structure..
c) Higher width is a better radiator: If the line width is high, it provides greater radiating edge for the fringing fields to be excited.. More the fringing fields, more the radiation..
Decreasing the height of the substrate reduce the radiation loses but make the microstrip line thinner, with higher ohmic loss.
Increasing the height of the substrate, the fringing fields from the edges increase, make the effective length of the line longer, and the input impedance of the line become slightly more inductive.
Because of skin effect it is not that simple. At GHz frequencies with a few µm skin depth, most current flows at the edges of the conductor and there is very little current in the middle of the lines.
It is true, but the skin effect in this situation (which might help by theory) causes the effective resistance of the conductor to change with frequency.
It is true, but the skin effect in this situation (which might help by theory) causes the effective resistance of the conductor to change with frequency.
Yes, the effective series resistance does change a lot with frequency.
I didn't mean that skin effect helps, but it's there. And it does cause a certain lower limit in resistance if we go to wider lines. At that point, the RF resistance doesn't decrease because the extra width in the center doesn't carry RF current.