Strict Upper Limits on the Carbon-to-Oxygen Ratios of Eight Hot Jupiters from Self-Consistent Atmospheric Retrieval

TL;DR

This study uses a self-consistent atmospheric retrieval framework to analyze eight hot Jupiters, finding their carbon-to-oxygen ratios are consistently below 0.9, with water abundances aligning with solar composition and clouds influencing spectral features.

Contribution

The paper introduces SCARLET, a novel atmospheric retrieval method that combines physical consistency with statistical analysis, enabling robust C/O ratio constraints without direct carbon molecule detection.

Findings

All eight hot Jupiters have C/O ratios below 0.9.

Water abundances are consistent with solar composition.

Cloud decks influence spectral features, not low water abundance.

Abstract

The elemental compositions of hot Jupiters are informative relics of planet formation that can help us answer long-standing questions regarding the origin and formation of giant planets. Here, I present the main conclusions from a comprehensive atmospheric retrieval survey of eight hot Jupiters with detectable molecular absorption in their near-infrared transmission spectra. I analyze the eight transmission spectra using the newly-developed, self-consistent atmospheric retrieval framework, SCARLET. Unlike previous methods, SCARLET combines the physical and chemical consistency of complex atmospheric models with the statistical treatment of observational uncertainties known from atmospheric retrieval techniques. I find that all eight hot Jupiters consistently require carbon-to-oxygen ratios (C/O) below 0.9. The finding of C/O<0.9 is highly robust for HD209458b, WASP-12b, WASP-19b, HAT-P-1b, and XO-1b. For HD189733b, WASP-17b, and WASP-43b, I find that the published WFC3 transmission spectra favor C/O<0.9 at greater than 95% confidence. I further show that the water abundances on all eight hot Jupiters are consistent with solar composition. The relatively small depth of the detected water absorption features is due to the presence of clouds, not due to a low water abundance as previously suggested for HD209458b. The presence of a thick cloud deck is inferred for HD209458b and WASP-12b. HD189733b may host a similar cloud deck, rather than the previously suggested Rayleigh hazes, if star spots affect the observed spectrum. The approach taken in SCARLET can be regarded as a new pathway to interpreting spectral observations of planetary atmospheres. In this work, including our prior knowledge of H-C-N-O chemistry enables me to constrain the C/O ratio without detecting a single carbon-bearing molecule.