Optical transmission spectrum of HAT-P-47b: evidence for aerosols and tentative TiO absorption

Technical Summary: Optical Transmission Spectrum of HAT-P-47b

Problem and Motivation
Transmission spectroscopy is a primary method for characterizing exoplanet atmospheres, particularly for detecting optical absorbers like metal oxides (e.g., TiO, VO) and atomic species that influence thermal structure. While TiO and VO are predicted to cause thermal inversions in highly irradiated hot Jupiters, their observational detection is inconsistent across systems, often attributed to condensation, cold trapping, or depletion via disequilibrium processes. Furthermore, high-altitude aerosols can mute molecular features and introduce scattering slopes, creating degeneracies in atmospheric interpretation. HAT-P-47b, a highly inflated hot sub-Saturn orbiting an F-type star with an equilibrium temperature of ~1605 K, sits near the predicted condensation boundary for titanium-bearing species. This makes it a critical target for investigating the interplay between aerosol scattering and refractory optical absorbers.

Methodology
The authors conducted a multi-faceted analysis combining space-based photometry and ground-based spectroscopy:

1. TESS Photometry: Thirteen transits from TESS (Sectors 18, 58, and 85) were analyzed to refine the orbital ephemeris and system parameters. Light curves were modeled using the batman package with Gaussian Processes (GP) to account for correlated noise, yielding precise constraints on inclination, scaled semi-major axis, and planet-to-star radius ratio.
2. Ground-Based Spectroscopy: Two independent transit events were observed:
   • LBT/MODS: Observed on November 12, 2018, using the Large Binocular Telescope's Multi-Object Double Spectrograph in dual-channel (blue/red) and blue-only modes.
   • GTC/OSIRIS+: Observed on November 26, 2023, using the Gran Telescopio Canarias' upgraded OSIRIS+ instrument.
   • Data reduction involved custom IDL routines for spectral extraction and white light curve construction. Spectroscopic light curves were generated in wavelength bins (10 nm for most, wider at the blue end).
3. Light Curve Analysis: A direct-fitting approach was adopted for spectroscopic light curves rather than a "divide-white" method, as benchmarking indicated the latter underestimated uncertainties and introduced artificial features. Systematics were modeled using GPs with inputs including time, spatial drifts, PSF width, and instrument rotation.
4. Atmospheric Retrieval: Bayesian retrievals were performed using petitRADTRANS. Two chemical scenarios were tested: equilibrium chemistry (parameterized by metallicity and C/O ratio) and free chemistry (independent mass fractions for 15 species). The models included isothermal temperature profiles, collision-induced absorption, Rayleigh scattering, and a patchy cloud model (opaque gray deck covering a fraction $\phi$ of the limb). Stellar contamination (spots/faculae) was also tested but found to be disfavored.

Key Results

• Ephemeris Refinement: The orbital period was refined to $P = 4.73217441 \pm 0.00000055$ days, a ~24-fold improvement in precision over previous values.
• Transmission Spectra: Both MODS and OSIRIS+ datasets exhibit a wavelength-dependent slope indicative of enhanced aerosol scattering. The slope is consistent with Rayleigh scattering expectations within uncertainties, though the derived slope parameter $\alpha$ is degenerate with channel-dependent offsets.
• Aerosol Scattering: Joint retrieval analysis indicates an aerosol contribution to optical scattering opacity approximately 5000 times larger than pure H$_2$ Rayleigh scattering. The scattering enhancement factor ($\log A_{RS}$) is constrained to $>3.09$ (90% lower limit) in the joint free-chemistry model.
• TiO Detection:
  • The MODS spectrum alone provides moderate Bayesian evidence ($\Delta \ln Z = 2.68$) for TiO absorption.
  • The OSIRIS+ spectrum alone does not yield statistically significant evidence for any specific opacity source.
  • The Joint Retrieval, dominated by the higher signal-to-noise MODS data, yields moderate evidence ($\Delta \ln Z = 3.44$) for TiO. The free-chemistry retrieval constrains the TiO mass fraction to $\log X_{\text{TiO}} = -6.86^{+0.64}_{-0.63}$ dex.
• Cloud Properties: The terminator region is inferred to be covered by clouds over a fraction $\phi \approx 0.7\text{--}0.9$. Despite this high coverage, the large atmospheric scale height allows the transmission spectrum to probe regions above the cloud deck.
• Stellar Contamination: Models incorporating unocculted stellar spots and faculae are statistically disfavored ($\Delta \ln Z = -4.31$) and require unrealistically high stellar variability inconsistent with the host star's activity level and TESS photometry.

Significance and Claims
The paper claims that HAT-P-47b hosts a cloudy atmosphere with tentative evidence for TiO absorption. The detection of TiO is significant because the planet's equilibrium temperature (~1605 K) lies near the condensation boundary for titanium-bearing species; its presence suggests either a non-isothermal temperature structure preventing deep condensation, or efficient vertical mixing replenishing gas-phase TiO above the condensation level. The authors emphasize that while the evidence for TiO is moderate, the presence of strong aerosol scattering is robust. They conclude that future high-precision observations, particularly high-resolution spectroscopy and space-based infrared observations (e.g., with JWST), are essential to confirm the TiO detection and further characterize the atmospheric structure and aerosol composition. The study does not claim a definitive discovery of a thermal inversion but highlights the potential for TiO to drive such a structure if confirmed.