2025ApJ...989..170F

Stellar contamination has long been recognized as a major bottleneck in transmission spectroscopy, limiting our ability to accurately characterize exoplanet atmospheres-particularly for terrestrial worlds. In response, significant observational efforts have shifted toward emission spectroscopy as a potentially more robust alternative, exemplified by initiatives such as the 500 hr JWST Rocky Worlds Director's Discretionary Time program. However, the extent to which emission spectroscopy may be affected by stellar effects remains mostly unexplored, in stark contrast with the extensive exploration and mitigation work for transmission spectroscopy. In this study, we assess the impact of imperfect knowledge of stellar spectra on exoplanet atmospheric retrievals from emission spectroscopy. At 12.8 μm, none of the considered bare surface types-basalt, ultramafic, Fe oxidized, and granitoid-can be reliably distinguished when accounting for the 3σ model precision between SPHINX and PHOENIX. At 15.0 μm, only the granitoid surface is distinguishable from all others above this threshold. These results show that stellar model uncertainty alone substantially limits our ability to constrain surface composition from photometric data, even before including other sources of uncertainty, such as stellar radius. Also, we find that current 15 μm eclipse depth estimations using different stellar models introduce a 60 ppm difference for M8 and 20 ppm for M5 stars. This model discrepancy leads to a degeneracy in retrieved planetary albedos and weakens constraints on the presence of an atmosphere. We therefore recommend that future JWST secondary eclipse observations systematically include stellar mid-IR spectroscopy to mitigate these uncertainties.