Methane on the temperate exo-Saturn TOI-199b

TL;DR

This study demonstrates that a temperate, Saturn-mass exoplanet TOI-199 b hosts a methane-rich atmosphere, inferred from JWST/NIRSpec G395M transmission spectroscopy and robust across two independent data reductions and two retrieval frameworks. The ExoTR and Aurora analyses converge on a strong CH$_4$ signature with Bayes factors near $10^3$, while CO, CO$_2$, NH$_3$, and HCN remain weakly constrained and degenerate with cloud/haze assumptions. Self-consistent photochemical and climate models indicate CH$_4$-dominated carbon chemistry with NH$_3$/HCN sensitivity to vertical mixing ($K_{zz}$) and internal temperature, and predict observable differences in the 3–5 μm range depending on metallicity and mixing strength. The system’s strong TTVs enable refined orbital dynamics and better mass constraints for TOI-199 c, while follow-up transit observations are poised to pin down the planet’s C, N, and O inventories, atmospheric metallicity, and the nature of hazes on temperate gas giants. These results establish TOI-199 b as a pivotal data point for understanding clouds, hazes, and methane-bearing atmospheres in temperate giant exoplanets and set the stage for future high-precision measurements of vertical mixing and chemical disequilibrium in this regime.

Abstract

Temperate ($T_{\rm eq}<400$ K) gas giants represent an unexplored frontier in exoplanet atmospheric spectroscopy. Orbiting a G-type star every ~100 days, the Saturn-mass exoplanet TOI-199 b ($T_{\rm eq}=350$ K) is one of the most favorable low-temperature gas giants for atmospheric study. Here, we present its transmission spectrum from a single transit observed with JWST's NIRSpec G395M mode. Despite lower-than-nominal precision due to a pointing misalignment, Bayesian retrievals reveal the presence of CH$_4$ (Bayes factor of $\sim$700 in a cloudy atmosphere), corresponding to a metallicity of $\rm{C/H}=13^{+78}_{-12}\times$ solar, although the absence of detectable CO and CO$_2$ disfavors metallicities $\gtrsim50\times$ solar. We also tested several haze prescriptions (Titan-like tholin, soot, and water-rich tholin), but the preference for these models is weak (Bayes factors of $\sim 2$ relative to the clear case). The spectrum also shows an increase in transit depth near 3 $μ$m, which our self-consistent models attribute to either NH$_3$ or, less likely, HCN. Follow-up observations will distinguish between these species, helping determine the planet's vertical mixing regime. The TOI-199 system exhibits strong transit timing variations (TTVs) due to an outer non-transiting giant planet. For planet c, our TTV analysis reduces its mass uncertainty by 50% and prefers a slightly longer orbital period (still within the conservative habitable zone) and higher eccentricity relative to previous studies. TOI-199 b serves as the first data point for studying clouds and hazes in temperate gas giants. The detection of methane supports the emerging trend that temperate low-molecular-weight atmospheres display spectral features in transmission.

Figures (16)

• Figure 1: Raw spectroscopic lightcurves, as extracted with Eureka! and binned to $\Delta\lambda = 0.004~\mu$m, before and after dividing out the common-mode noise model.
• Figure 2: Top: Mean-subtracted position of the trace along the $y$ (i.e. cross-dispersion) direction, binned by a factor of 80 to more easily visualize the jump occurring around egress. Middle: Raw and binned ($\times 40$) white lightcurves of TOI-199 b 's transit, observed with NIRSpec G395M, including the best-fit model, as extracted with Eureka!. Bottom: Residuals from the best-fit model, measured in $\sigma$.
• Figure 3: Top: RMS of the white lightcurve residuals as a function of bin size (solid teal line) and scaling expected for purely white noise (dashed black line). The dotted red line indicates the bin size used in the Eureka! white lightcurve fit (i.e. 40$\times$). Middle: Ratio of the two lines in the top plot (i.e. red noise factor, winn2008). Bottom: Normalized RMS of the residuals of the spectroscopic lightcurves from Eureka! as a function of bin size (solid colored lines) and scaling expected for purely white noise (dashed black line).
• Figure 4: Comparison of the two independent reductions of the NIRSpec data of TOI-199 b.
• Figure 5: Top: Maximum A Posteriori (MAP) model and 2$\sigma$ credible regions from ExoTR retrievals on the fiducial reduction, overlaid with data binned at $\Delta\lambda = 0.05~\mu\rm m$. Also shown are the contributions from each molecule and clouds to the transmission spectrum. Bottom: 1D histograms of the posterior distributions from the retrieval results, including the median and $\pm 1 \sigma$ uncertainties (i.e. 16$^{\rm th}$ and 84$^{\rm th}$ percentiles).