[en] Highlights: • Kr and Ar seeding of D plasma did not result in the formation of new D trapping sites in RAFM steel Eurofer’97 and commercial steel P92. • The majority of the retained D was trapped in the steel bulk, irrespective of the substantial surface damage by ion impact and effective removal by Kr and Ar sputtering. • D desorption from the bulk material of the ferritic-martensitic steels after Kr/Ar-seeded plasma exposures exhibits lower, but similarly positioned desorption peaks as after pure D exposures. • Kr causes more intensive steel sputtering than Ar, and therefore results in much less D retention. - Abstract: Total deuterium (D) retention from the bulk material of reduced-activation ferritic-martensitic (RAFM) steel Eurofer’97 (EU’97) and commercial ferritic-martensitic grade P92 was traced experimentally by means of thermal desorption spectroscopy (TDS) and linked to the role of krypton (Kr) and argon (Ar) seeding during high-fluence plasma exposure. The influence of impurity seeding on the steel microstructure was determined using scanning electron microscopy (SEM) and extended with focused ion beam (FIB) cross-sectioning. D capture at depths in the µm range was measured by nuclear reaction analysis (NRA). Plasma exposure of the steel samples occurred at 470 K with ion energy of 30–40 eV in the linear plasma device PSI-2 with up to 10% simultaneous impurity admixture in plasma. D inventory achieved values in the 10²⁰ D/m² range after plasma exposure with high fluences of up to 1 × 10²⁶ D⁺/m². In pure and mixed plasmas, the majority of D was trapped in the steel bulk. The Kr and Ar seeding of D plasma resulted in the population of multiple already existing D trapping sites. A similarity of D desorption spectra suggests that D trapping in P92 follows the same mechanism as in EU’97. Kr and Ar seeding mostly contributed to the surface sputtering of the steel samples, yielding less D retention due to the material loss.