Is TiO emission present in the ultra-hot Jupiter WASP-33b? A reassessment using the improved ExoMol TOTO line list
Dilovan B. Serindag, Stevanus K. Nugroho, Paul Mollière, Ernst J. W. de Mooij, Neale P. Gibson, Ignas A. G. Snellen
Abstract
Context. Efficient absorption of stellar ultraviolet and visible radiation by TiO and VO is predicted to drive temperature inversions in the upper atmospheres of hot Jupiters. However, very few inversions or detections of TiO or VO have been reported, and results are often contradictory. Aims. Using the improved ExoMol T OTO line list, we searched for TiO emission in the dayside spectrum of WASP-33b using the same data in which the molecule was previously detected with an older line list at 4.8 σ . We intended to confirm the molecular detection and quantify the signal improvement offered by the ExoMol T OTO line list. Methods. Data from the High Dispersion Spectrograph on the Subaru Telescope was extracted and reduced in an identical manner to the previous study. Stellar and telluric contamination were then removed. High-resolution TiO emission models of WASP-33b were created that spanned a range of molecular abundances using the radiative transfer code petitRADTRANS , and were subsequently cross-correlated with the data. Results. We measure a 4.3 σ TiO emission signature using the ExoMol T OTO models, corresponding to a WASP-33b orbital velocity semi-amplitude of K p =252.9 −5.3 +5.0 km s -1 and a system velocity of v sys =−23.0 −4.6 +4.7 km s -1 . Injection-recovery tests using models based on the new and earlier line lists indicate that if the new models provide a perfect match to the planet spectrum, the significance of the TiO detection should have increased by a factor of ~2. Conclusions. Although the TiO signal we find is statistically significant, comparison with previous works makes our result too ambiguous to claim a clear-cut detection. Unexpectedly, the new ExoMol T OTO models provide a weaker signal than that found previously, which is offset in K p - v sys space. This sheds some doubt on both detections, especially in light of a recently published TiO non-detection using a different dataset.