Exploring the Atmosphere of K2-18b through Retrievals and Forward Modelling

Sub-Neptunes (1.8R_⊕ ≲ R_p ≲ 3.5R_⊕) are the most common class of exoplanets in our galaxy, yet their interior compositions remain elusive. Proposed interior structure models include gaseous ”mini-Neptunes” with thick H₂-dominated envelopes, and ”Hycean” worlds with a thin H₂ atmosphere overlying a deep liquid water layer (e.g. [1, 2, 3]).

The advent of the James Webb Space Telescope (JWST) has revolutionized exoplanet studies by providing high-precision near-infrared (NIR) spectroscopy, allowing us to characterise their atmospheres in unprecedented detail. Among these sub-Neptunes, K2-18b is one that has captured significant attention. Discovered in 2015 during the Kepler K2 mission [4, 5], it orbits within the habitable zone of an M dwarf, making it a prime target for studying the atmospheric composition, interior structure, and habitability of sub-Neptunes.

Previous studies using Hubble Space Telescope (HST) and Spitzer data identified an H₂-rich atmosphere with significant H₂O absorption features, suggesting the possibility of a liquid water ocean and habitable conditions, making K2-18b highly relevant for astrobiology studies [6, 7, 8]. However, similarities between CH₄ and H₂O absorption features in the HST bandpass (1.1–1.7 µm) led to competing interpretations of K2-18b’s atmospheric chemistry [9, 10].

JWST observations from the NIRISS SOSS and NIRSpec G395H instruments revealed strong absorption features between 0.9–5.2 µm. The original study [11] interpreted these as robust detections of CO₂ (∼ 1% detected at 5σ) and CH₄ (∼ 1% at 3σ) in an H₂-rich atmosphere, alongside non-detections of H₂O, CO, and NH₃. Tentative (∼ 1σ) signs of dimethyl sulfide (DMS), a potential biosignature, were also reported. These abundances could point towards a Hycean-like scenario with a biogenic source of atmospheric CH₄ [12]. Multiple studies have also argued in favour of a “mini-Neptune” scenario that is equally compatible with the JWST observations (e.g. [13, 14]). Moreover, recent independent reanalyses of the JWST data [15] reported no reliable evidence for CO₂ or DMS, contradicting the original findings.

Recently, new JWST observations from the MIRI LRS instrument (∼6–12 µm) were released. The original analysis reported further evidence for DMS and dimethyl disulfide (DMDS) in the atmosphere – another gas proposed as a biosignature [16]. However, emerging evidence of an abiotic pathway to DMS in cometary matter has raised doubts over the reliability of these compounds as definitive biosignatures [17].

JWST’s observations have undoubtedly brought us closer to understanding the nature of K2-18b and sub-Neptunes more broadly. Nonetheless, no consensus yet exists on which model best explains K2-18b’s atmospheric composition. Although these studies have significantly expanded the realm of what we currently understand to be sub-Neptunes, the growing number of degenerate solutions highlights the need for more standardised methodologies across studies to ensure robust exoplanetary characterisation.

This study aims to refine our understanding of K2-18b by addressing key factors that influence atmospheric retrievals and characterization. First, we consider the effects of uncertainties in stellar mass and radius on derived planetary parameters in retrievals and models. To improve the treatment of K2-18’s UV spectrum, we incorporate previously unused HST STIS measurements in the UV, refining the input stellar flux used in atmospheric modelling. We employ the Iraclis data reduction pipeline [7, 18], which has not yet been applied to the JWST observations of K2-18b, offering an independent method to analyse the existing data and validate the reproducibility of previous studies (e.g. [11, 15]). Up until now, atmospheric retrieval studies of K2-18b have been limited to free chemistry, which assumes no physical or chemical processes in the atmosphere. Our retrieval framework includes both free chemistry retrievals and retrievals coupled with equilibrium chemistry models, allowing us to self-consistently solve for thermochemical equilibrium, fit key parameters such as metallicity and the C/O ratio, and predict the chemical species that could form and condense based on the retrieved elemental abundances and the pressure-temperature profile. Additionally, we perform supplementary forward modelling to account for haze/cloud microphysics, disequilibrium chemistry, and radiative feedbacks, providing a more physically motivated understanding of K2-18b’s atmosphere. Our studies leverage JWST observations from NIRISS, NIRSpec, and MIRI to inform our retrievals and chemical models of constraints to the transmission spectrum. Finally, we discuss new scenarios that could explain the JWST observations of K2-18b.

Our study highlights the broader implications of K2-18b as a natural laboratory for testing atmospheric retrieval methodologies and advancing our search for habitable environments beyond Earth.