Where Is the Water? Jupiter-like C/H Ratio but Strong H₂O Depletion Found on τ Boötis b Using SPIRou

Abstract

The present-day envelope of gaseous planets is a relic of how these giant planets originated and evolved. Measuring their elemental composition therefore presents a powerful opportunity to answer long-standing questions regarding planet formation. Obtaining precise observational constraints on the elemental inventory of giant exoplanets has, however, remained challenging owing to the limited simultaneous wavelength coverage of current space-based instruments. Here, we present thermal emission observations of the nontransiting hot Jupiter τ Boo b using the new wide wavelength coverage (0.95–2.50 μm) and high spectral resolution (R = 70,000) CFHT/SPIRou spectrograph. By combining a total of 20 hr of SPIRou data obtained over five nights in a full atmospheric retrieval framework designed for high-resolution data, we constrain the abundances of all the major oxygen- and carbon-bearing molecules and recover a noninverted temperature structure using a new free-shape, nonparametric temperature–pressure profile retrieval approach. We find a volume mixing ratio of log(CO) = − 2.46 − 0.29 + 0.25 and a highly depleted water abundance of less than 0.0072 times the expected value for a solar composition envelope. Combined with upper limits on the abundances of CH4, CO2, HCN, TiO, and C2H2, this results in a gas-phase C/H ratio of 5.85 − 2.82 + 4.44 × solar, consistent with the value of Jupiter, and an envelope C/O ratio robustly greater than 0.60, even when taking into account the oxygen that may be sequestered out of the gas phase. Combined, the inferred supersolar C/H, O/H, and C/O ratios on τ Boo b support a formation scenario beyond the water snowline in a disk enriched in CO owing to pebble drift.

Relative abundance constraints from high-resolution optical transmission spectroscopy of WASP-121b, and a fast model-filtering technique for accelerating retrievals

“…Birkby et al 2017;Esteves et al 2017;Nugroho et al 2021); however, more recent studies performing retrievals have reverted to basismodel fitting or PCA which can be applied more quickly within each likelihood calculation (e.g. Brogi & Line 2019;Pelletier et al 2021;Line et al 2021), whereas S R is an iterative process and is much too slow. We later introduce in Sect.…”

“…the filtering is time-dependent. More sophisticated methods based on PCA filtering have already been applied in the literature (Pelletier et al 2021). However, these methods generally involve injecting the Doppler-shifted atmospheric model into a noiseless approximation of the time vs wavelength array, and re-applying the filtering methods to approximate what is done to the underlying data.…”

“…Nonetheless, including it explicitly in the likelihood has some computational advantages as noted earlier. Indeed, some studies have already applied the Brogi & Line (2019) framework including 𝛼 explicitly (Pelletier et al 2021). However, the most significant difference between our approach and that of Brogi & Line ( 2019) is the addition of the time-and wavelength-dependence of the uncertainties.…”

“…One proposed solution is to inject the forward model atmosphere into a representation of the telluric and stellar model, and re-process the forward model in the same way as the data. Indeed, this was demonstrated to work successfully both with airmass de-trending and also with PCA filtering (Brogi & Line 2019;Pelletier et al 2021;Line et al 2021). However, this injection and re-filtering process is generally slow, and in many cases can become the bottleneck in each likelihood calculation.…”

“…This approach to detecting species in exoplanet atmospheres has flourished in recent years, with detections of multiple molecular, atomic and ionised species in a wide range of planetary atmospheres (e.g. Hoeijmakers et al 2018;Ehrenreich et al 2020;Pelletier et al 2021;Nugroho et al 2021;Giacobbe et al 2021;Merritt et al 2021).…”