Hi Axel,
My experience is with the plating & electropolishing. We sometimes do a "diffusion bake" to improve coating adhesion, but only up to 650oF, depending on substrate. That's why I suggested accelerated testing at 700oC. With coating thicknesess & diffusion data, can calculate via Fick's Law how long before the coating interdiffuses into matrix. But 1) actual test data is better, 2) my guess is that it will remain better for long time -- diffusion will slow down as concentration gradient decreases, and finally, 3) if you plate Fe (maybe over a Wood's Ni strike, that's usual first step for plating on SS), there won't be much gradient anyway. Anything more I say would be speculation.
Ken