The atmospheric composition of TOI-270 d

TL;DR

This work leverages JWST NIRISS SOSS and NIRSpec G395H transmission spectra to resolve whether TOI-270 d hosts a Hycean ocean world or a high-MMW mixed envelope. Using the VIRA retrieval framework, it confirms robust CH$_4$ and CO$_2$ detections with a 10 mbar terminator temperature of $T_{10\mathrm{mbar}}=323^{+58}_{-52}$ K and a mean molecular weight of $MMW=4.84^{+1.10}_{-1.08}$ amu, while suggesting tentative evidence for methyl-bearing species near $3.5\mu$m. The results lie between previous analyses, showing that including broader chemical species, such as C$_2$H$_6$ and DMS, reduces inferred temperatures and MMW and can explain excess opacity, though no single methyl species is decisively identified. The findings favor Hycean or dark Hycean scenarios with planet-wide or nightside oceans, while highlighting remaining degeneracies and the need for additional observations to robustly constrain internal structure and atmospheric chemistry. Overall, the study demonstrates the critical role of comprehensive chemical inventories in atmospheric retrievals of temperate sub-Neptunes and sets the stage for future characterizations of TOI-270 d.

Abstract

The first explorations of temperate sub-Neptune exoplanets have been the hallmark of early JWST observations. The bulk properties of such planets are consistent with a range of possible internal structures, which can be distinguished through their interactions with the observable atmospheres. JWST observations of TOI-270 d, a temperate sub-Neptune, have previously led to contrasting conclusions: either a Hycean world, possessing a liquid water ocean, or a mixed-envelope sub-Neptune, where high temperatures prevent a liquid ocean and lead to a high mean molecular weight atmosphere. In order to resolve this uncertainty, we present a comprehensive retrieval analysis of TOI-270 d using recent NIRISS and NIRSpec transit spectroscopy across $\sim$1-5 $μ$m. We find that prior inferences of a mixed envelope were affected by specific modelling choices leading to a high terminator temperature and high mean-molecular weight in the atmosphere. We confirm an H$_2$-rich atmosphere in TOI-270 d and present revised constraints on the molecular log-mixing ratios and maximal detection significances of CH$_4$ at $-1.86^{+0.30}_{-0.29}$ (6.4 $σ$), CO$_2$ at $-1.71^{+0.38}_{-0.66}$ (3.9 $σ$), H$_2$O at $-1.88^{+0.78}_{-4.13}$ (2.1 $σ$) and CS$_2$ at $-4.74^{+0.65}_{-1.10}$ (2.0 $σ$), with a terminator temperature of $323^{+58}_{-52}$ K at 10 mbar. We also find tentative evidence for more complex methyl-bearing species such as C$_2$H$_6$ and/or DMS at a 2.1-2.5 $σ$ level. The present constraints are consistent with TOI-270 d being a Hycean or dark Hycean world, with planet-wide or nightside liquid water oceans. However, more observations are required to verify the present findings and robustly constrain the atmospheric conditions and internal structure of TOI-270 d.

Figures (9)

• Figure 1: White light curve for the transit of TOI-270 d observed with NIRISS SOSS, order 1. The data and best-fit model are shown in black and green, respectively. The standard deviation of the residuals is 109 ppm, corresponding to 2.3 times the expected noise level, similar to other NIRISS SOSS observations. The parameter constraints from the model fit are provided in Table \ref{['tab:wlc_params']}.
• Figure 2: Excess H$\alpha$ emission from a minor flare. Note that the peak emission aligns with the small flux increase seen in the white light curve just before mid transit. Ultimately, this does not significantly impact the final transmission spectrum, as discussed in Section \ref{['subsec:niriss_reduction']}.
• Figure 3: JWST transmission spectrum of TOI-270 d. Orange errorbars correspond to NIRISS observations, presented in this work, while dark red errorbars correspond to NIRSpec G395H observations presented by Holmberg2024. The NIRISS observations were binned down to R=25 for order 1 and R=12 for Order 2 for visual clarity. The blue curve denotes the median retrieved spectral fit obtained with our canonical retrieval, described in Section \ref{['sec:methods']}. Darker and lighter turquoise shaded regions denote the spectral 1- and 2-$\sigma$ contours corresponding to the spectral fit. Yellow dots denote the median spectrum binned to the resolution of the datapoints. The NIRISS data are vertically offset by +28 ppm, which is the median offset retrieved for our canonical model.
• Figure 4: JWST transmission spectra of TOI-270 d presented by Benneke2024 (red errorbars) and those used for our canonical retrievals (blue errorbars), combining NIRSpec data by Holmberg2024 with the NIRISS observation reduced in this work. Also shown are the 1-$\sigma$ contours of the retrieved spectral fits to either dataset. The orange contours correspond to the retrieved spectral fit to the red errorbars using an equivalent atmospheric model to the free chemistry retrievals of Benneke2024, as discussed in Section \ref{['sec:benneke_reproduction']}. The turquoise contours correspond to the spectral fit to the blue errorbars using our canonical model, as discussed in Section \ref{['sec:canonical_retrievals']}. The present (blue) NIRISS data has been vertically offset by +28 ppm, the median offset retrieved for our canonical model, as with Figure \ref{['fig:spectrum']}. C25 in the figure caption refers to the data and canonical model presented in this work.
• Figure 5: Atmospheric parameter constraints obtained with retrievals on JWST NIRISS and NIRSpec observations presented by Benneke2024, using an equivalent free chemistry atmospheric model to Benneke2024. The blue posterior distributions correspond to the retrieval where the reference pressure, $P_\mathrm{ref}$ is a free parameter, while the orange posterior distributions correspond to a retrieval with $P_\mathrm{ref}$ fixed to $10^{-3}$ bar, and the green to a retrieval with $P_\mathrm{ref} = 10^{-4}$ bar. $\mu$ denotes the MMW of the atmosphere, which is inferred from the mixing ratio posterior distributions of all molecules in the model. Errorbars denote the median and 1-$\sigma$ interval, while arrows denote 2-$\sigma$ upper estimates.