Detectability of Rocky-Vapour atmospheres on super-Earths with Ariel

Abstract

Ariel will mark the dawn of a new era as the first large-scale survey characterising exoplanetary atmospheres with science objectives to address fundamental questions about planetary composition, evolution and formation. In this study, we explore the detectability of atmospheres vaporised from magma oceans on dry, rocky Super-Earths orbiting very close to their host stars. The detection of such atmospheres would provide a definitive piece of evidence for rocky planets but are challenging measurements with currently available instruments due to their small spectral signatures. However, some of the hottest planets are believed to have atmospheres composed of vaporised rock, such as Na and SiO, with spectral signatures bright enough to be detected through eclipse observations with planned space-based telescopes. In this study, we find that rocky super-Earths with a irradiation temperature of 3000 K and a distance from Earth of up to 20 pc, as well as planets hotter than 3500 K and closer than 50 pc, have SiO features which are potentially detectable in eclipse spectra observed with Ariel.

Appendix: 55-Cancri e retrieval simulations

Fig. 8

Results of our retrieval analysis on the eclipse spectra of a mineral atmosphere on 55-Cancri e assuming 20 (blue), 40 (orange) and 80 (red) visits with Ariel. Top: Simulated Ariel observations and best fit spectra; Bottom Left: Retrieved mean and 1σ temperature profiles; Bottom Right: Posterior distributions of the free parameters

In order to investigate the feasibility to detect SiO with future telescopes, we simulate the atmosphere of a 55-Cancri e like planet using the model from [25] and estimating the Ariel noise with the Ariel Radiometric Model (ArielRad) from [42]. The observations are assumed at Tier-2 resolution [11][8] and we investigate the combined transits of 20, 40 and 80 visits. The corresponding S_SiO signal strength for these visits can be found in Fig. 4 and we note that a value of 1σ is obtained between 20 and 40 combined transits. We then performed an atmospheric retrieval using TauREx3 on the 3 cases, assuming a plane parallel atmosphere with 100 layers up to 10^− 5 Pa. The surface pressure was fixed to its true value. Since a surface pressure and the molar fraction of a gas have the same effect on its optical depth, only the partial pressure of the gas inducing a spectral signature can be retrieved from atmospheric spectra in principle. Note that, however, the surface pressures of mineral atmospheres can be determined from our atmospheric model using retrieved surface temperature.

Since the Ariel spectra obtained for a mineral atmosphere have a relatively low information content, we fit the simulated spectra using a simplified retrieval model. The planetary radius and mass were fixed [7] to the literature values as more accurate constraints can be obtain from Radial Velocity and Transit measurements. For the temperature structure, we retrieved a heuristic profile comprised of 3 freely moving temperature-pressure points located at the surface, at 1 Pa and 10^− 5 Pa. The atmosphere was assumed to be composed of H₂, He and SiO with the molecular ratio $X_{He}/X_{H_{2}}$ fixed to solar values and the ratio $X_{SiO}/X_{H_{2}}$ being the only free parameter of the chemistry.

To explore the parameter space we use the Nested Sampling algorithm MultiNest [17] with 1000 live points and an evidence tolerance of 0.5.

From those retrievals, we find that the SiO spectra feature at 4.5μm is difficult to capture with 20 combined Ariel observations. For this case, the posterior distribution shows hints of the SiO signal, but a large tail is observed towards the low abundances, which would not allow to definitively conclude for this case. In the 40 and 80 observations cases, however, the noise is greatly reduced and a clear lower limit on the molecular ratio is observed (log SiO/H₂ = 0.4$^{+1.5}_{-1.0}$). While the precise abundances can’t be obtained, a retrieval analysis would give strong indications in favor of a mineral atmosphere. The retrieved temperature structure for this example follows the input profile, but large differences are noticeable due to the differences between the forward and retrieval models. This is known to lead to biases that could potentially be mitigated when interpreting the results using self-consistent models or replacing the retrieval model with a more realistic scenario [6, 8]. We note that for hotter planets, the detection of an SiO signal with Ariel would be much easier. This is because the 4-μm-SiO signals deviate from BB_ref at three or four Ariel wavelength-bins for hotter planets (Fig. 5), while it deviates from BB_ref at the two bins ($\sim 4.0\mu$m and $\sim 4.3\mu$m) for the 55 Cnc e case. On top of this, when comparing with other models for the magma and atmosphere composition (see Fig. 7), the mineral atmosphere case appears as the worst case scenario since bigger features are observed in the cases of Hydrogen rich or Water rich atmosphere. In practice, it is likely that much less than 20 visits would be need for 55-Cancri e like planet to rule out the Hydrogen and Water rich cases.