2021NatAs...5.1224M

The emergent spectra of close-in, giant exoplanets (`hot Jupiters') are expected to be distinct from those of self-luminous objects with similar effective temperatures because hot Jupiters are primarily heated from above by their host stars rather than internally from the release of energy from their formation. Theoretical models predict a continuum of dayside spectra for hot Jupiters as a function of irradiation level, with the coolest planets having absorption features in their spectra, intermediate-temperature planets having emission features due to thermal inversions and the hottest planets having blackbody-like spectra due to molecular dissociation and continuum opacity from the H- ion. Absorption and emission features have been detected in the spectra of a number of individual hot Jupiters, and population-level trends have been observed in photometric measurements. However, there has been no unified, population-level study of the thermal emission spectra of hot Jupiters as there has been for cooler brown dwarfs and transmission spectra of hot Jupiters. Here we show that hot Jupiter secondary eclipse spectra centred around a water absorption band at 1.4 µm follow a common trend in water feature strength with temperature. The observed trend is broadly consistent with model predictions for how the thermal structures of solar-composition planets vary with irradiation level, but is inconsistent with the predictions of self-consistent one-dimensional models for internally heated objects. This is particularly the case because models of internally heated objects show absorption features at temperatures above 2,000 K, whereas the observed hot Jupiters show emission features and featureless spectra. Nevertheless, the ensemble of planets exhibits some degree of scatter around the mean trend for solar-composition planets. The spread can be accounted for if the planets have modest variations in metallicity and/or elemental abundance ratios, which is expected from planet formation models.