In Simulation we can only assign the boundary for a 2D radiating patch or ground. We are not concerned about the thickness in the Simulation environment.
You can apply the "Finite Conductivity Boundary" on your 2D sheet, which will allow you to specify the material and thickness.
Regarding the fabrication, thickness depends on the application and the frequency range.
for PEC: virtually infinity when talking of E-field penetration, and zero for scattering. Simultaneously.
for nonmetal/non-pec: zero
i.e.nothing
it is nothing, simply just an instruction for mesher to add more vertices, if no specific boundaries are applied.
it is just a physical model, it should not have any real thickness to define as a default.
There is a thing to note. This is true for FEM and MOM methods. For FDTD, there is a minimal thickness, which is a half of cell, where models start acting weird. That is because H and E components are solved with a time(phase) shift and E and H meshes are overlaid with a half-cell displacement in every direction. That is why thin metals are defined with a virtual thickness along E-nodes in FDTD, and oblique thin surfaces are at least undesirable, but generally impossible.
Due to the principle of skin depth, in microwave frequencies it does not make any difference whether the thickness of the metallization, in this way, to make the simulation faster, Ansys HFSS uses an infinitesimal thickness.