Ground-based Atmospheric Characterization of Super-Earth L 98-59 d at High Spectral Resolution

Abstract

Atmospheric characterization of exoplanets using ground-based high-resolution transmission spectroscopy has traditionally focussed on large and close-in planets, such as hot Jupiters. In this work, we aim to extend this technique to smaller and more temperate planets by studying the atmospheric composition of the temperate super-Earth planet L 98-59 d ($\sim$$1.5\,\mathrm{R_{\oplus}}$; $\sim$$1.9\,\mathrm{M_{\oplus}}$). Using high-resolution transmission spectra obtained using IGRINS on the Gemini-South telescope, we demonstrate the feasibility for atmospheric characterization of super-Earths using ground-based facilities, and confirm the previous tentative JWST inference of hydrogen sulfide (\ce{H2S}) in the atmosphere of L 98-59 d at $\lesssim$3.9$\,σ$ ($B\sim390$). This is the first ground-based inference of a molecular species in the atmosphere of a super-Earth planet, and reveals the sensitivity of spectrographs on 8\,m-class telescopes to the atmospheric characterization of such planets. By exploring a grid of atmospheric models, we find that the data favors a cloud-free atmosphere with an abundance of H2S corresponding to $\sim$1-10\,$\times$ solar metallicity. We additionally place constraints on the atmospheric abundances of other molecular species. Assuming cloud-free models, super-solar abundances for CH4 and NH3 are ruled out at 3.6$σ$ and $4.6σ$, respectively. Our results are consistent with previous suggestions that L 98-59 d is a super-Earth with possible disequilibrium production of \ce{H2S} driven by volcanic outgassing from the surface. Future studies combining multiple observations with different facilities may be able to further constrain the atmospheric composition of this planet. This work underscores the promise of atmospheric characterization of super-Earth exoplanets using high-resolution spectroscopy with ground-based facilities.

Figures (11)

• Figure 1: The observing conditions, data S/N, and lightcurve for the transit of L 98-59 d on the primary observing night. Left: we show the variation in airmass, the median S/N of the data in each exposure (after the removal of certain orders), and the changing barycentric velocity correction (positive defined as towards target) during the observations, as a function of orbital phase. The transit (determined using $T_{14}$; Table \ref{['system_parameters']}) is marked by the red dashed lines. Bottom: the median data S/N is now shown as a function of wavelength, with the different orders used in this work shaded. Right: the normalized lightcurve for L 98-59 d (black line), modelled using batman-packagebatman2015. In this work, the time-series model spectra are modulated according to this profile during injection and recovery tests and likelihood-based model comparison. This modulation is shown for each of the exposures (black markers). Without this correction, the in-transit spectra are instead weighted equally (shown in blue), which may introduce biases during model comparison.
• Figure 2: Model spectra for a selection of the molecular species considered in this work. CH4 and NH3 are expected to be prevalent and detectable in temperate H2-dominated atmospheres using this technique cheverall2024, whilst H2S and SO2 have been tentatively inferred previously in the atmosphere of L 98-59 d banerjee2024. In addition to the molecules shown here, opacity contributions from CO2, CO, and H2O are considered. The spectral orders used in this analysis remaining after initial data cleaning are marked in gray.
• Figure 3: Cross-correlation results for the observed time-series spectra and models of the molecular species considered in this work. The cross-correlation S/N is shown as a function of planetary velocity space, parameterised by $K_{\mathrm{p}}$ and $V_{\mathrm{sys}}$, for different molecular species. In each case, the expected planetary velocity is marked by white crosshairs. For consistency, the colorscale is fixed to [-3, 5] across all panels. Top panel: when cross-correlating atmospheric models with the observed data, no significant signals are seen, although a tentative peak of $\mathrm{S/N} = 1.7$ is seen for H2S consistent with the expected $V_{\mathrm{sys}}$. Bottom panel: injection and recovery tests are used to assess detectability, with a model atmospheric spectrum for each chemical species injected into the spectra prior to detrending. This injection is at the expected systematic velocity, except in the case of H2S, where the signal is injected at an alternative systematic velocity ($+$19 $\mathrm{kms^{-1}}$, chosen as a region of velocity space with an initial S/N close to 0, and marked by a second crosshairs) in order to avoid adding to the observed peak. Recovered signals at the velocities of injection suggest that our analysis is moderately sensitive to CH4 and NH3, with additional limited sensitivity to H2O (also see Figure \ref{['fig:1d-profiles']} for 1-D cross-sections at the expected value for $K_{\mathrm{p}}$).
• Figure 4: Additional analysis for the tentative cross-correlation signal observed for H2S. We show the cross-correlation S/N as a function of the orbital parameters of the planet. In each case, the expected parameters are marked with crosshairs. Left: The S/N is shown as a function of planetary velocity space, parameterised by $K_{\mathrm{p}}$ and $V_{\mathrm{sys}}$, reproduced from the top panel of Figure \ref{['fig:detection-survey_primary']} but with a more zoomed in colorscale. A tentative peak with a S/N of 1.7 is observed at the expected systemic velocity. Right: the time duration of the observed H2S cross-correlation signal is constrained with a value that is consistent with the literature value for the transit duration of the planet. This is measured by varying the number of spectra considered to be in-transit ($N_{\mathrm{in}}$) when calculating the cross-correlation S/N, with $K_{\mathrm{p}}$ fixed to that expected for this planet.
• Figure 5: The posterior probability distribution for H2S as a function of the planetary parameters. The expected parameters are marked with crosshairs. The values for $V_{\mathrm{sys}}$ and $\alpha$ measured from the posterior probability distribution are consistent with those expected. The 0.5$\sigma$, 1.0$\sigma$, 1.5$\sigma$, and 2$\sigma$ contours are shown in the 2D-plot, with the 1$\sigma$ bounds also shown in each 1D histogram. For temperate planets such as L 98-59 d, with small changes in radial velocity during transit, it is difficult to constrain $K_{\mathrm{p}}$ and therefore it is not included as a panel in this corner plot.