What exoplanet spectra can and cannot show

Reading a transit spectrum

During a transit, some starlight filters through a planet’s atmosphere. Gases absorb particular wavelengths, changing the combined star-and-planet spectrum. Comparing it across wavelengths with a star-only eclipse measurement can help isolate the planet’s contribution. The spectrum is not a direct picture; it is light data that must be interpreted.[1]

Scientists compare observations with atmosphere models, which estimate how candidate gases change the light. A better statistical fit means that one tested model explains these data better than the alternatives in that comparison; overlapping signatures can leave multiple gases plausible.[2]

K2-18 b: chemistry, not life

A 2023 study of K2-18 b’s near-infrared JWST spectrum reported methane at 5-sigma significance and carbon dioxide at 3-sigma, while ammonia was not detected. The authors said this combination was consistent with a possible hydrogen-rich, Hycean atmosphere and discussed potential evidence for dimethyl sulfide (DMS). These are interpretations of spectra under models, not observations of an ocean or living organisms.[2]

A 2025 JWST MIRI analysis examined 6–12 μm observations. Its authors reported that retrievals including DMS and DMDS were favoured at roughly 2.9–3.2 sigma over a comparison model containing neither gas. The two molecules’ features were degenerate in this dataset, so the analysis did not distinguish which, if either, accounts for a candidate signal. The authors called for more observations and investigation of possible abiotic sources.[3]

An expanded model-space analysis of K2-18 b spectra found alternative atmospheres that could fit as well or better, showing that a relative preference depends on which models are tested. A candidate gas would still require confirmation and evaluation of non-biological chemistry; neither paper reports a detection of life.[4]