I have samples of Ta/NiFe/FeMn/Ta, what approximate thickness should the samples have so that I guarantee a constant bias field and increase the anisotropy field, or the opposite?
Your question depends of many different factors and it is not easy to give you one answer. First of all notice that there are several anisotropies playing different roles at the interface due to the mixture of different kind of coupling constants associated to different spin-orbits.
One paper which could help you with your layers is:
It is not really trivial, since the thickness of the ferromagnetic material, the antiferromagnetic material and the substrate itself is already well analyzed in the literature, and even so it varies from one place to another due to the factors you mentioned, I will read the article to see if I can come to a conclusion, thanks for the clarifications.