Generally, if surface roughness increases, coercivity increases as well. Due to the increasing surface roughness, domain-wall pinning started to contribute to the coercivity.
So I etched the polymer substrate and deposited CFB (100 nm, 50 nm, 10 nm).
And I also got this data for 100 nm and 50 nm samples. However, the opposite tendency was observed for the 10 nm sample. Almost there is no coercivity change as surface roughness increases. I don't know why this happened.
All samples are amorphous thin film and don't have anisotropy. And the surface shape is the same as the attached file.