Context. Hot Jupiters exhibit atmospheric temperatures ranging from hundreds to thousands of Kelvin. Because of their large daynight temperature differences, condensable species that are stable in the gas phase on the dayside -such as TiO and silicates -may condense and gravitationally settle on the nightside. Atmospheric circulation may counterbalance this tendency to gravitationally settle. This three-dimensional (3D) mixing of condensable species has not previously been studied for hot Jupiters, yet it is crucial to assess the existence and distribution of TiO and silicates in the atmospheres of these planets. Aims. We investigate the strength of the nightside cold trap in hot Jupiters atmospheres by investigating the mechanisms and strength of the vertical mixing in these stably stratified atmospheres. We apply our model to the particular case of TiO to address the question of whether TiO can exist at low pressure in sufficient abundances to produce stratospheric thermal inversions despite the nightside cold trap. Methods. We modeled the 3D circulation of HD 209458b including passive (i.e. radiatively inactive) tracers that advect with the 3D flow, with a source and sink term on the nightside to represent their condensation into haze particles and their gravitational settling. Results. We show that global advection patterns produce strong vertical mixing that can keep condensable species aloft as long as they are trapped in particles of sizes of a few microns or less on the nightside. We show that vertical mixing results not from small-scale convection but from the large-scale circulation driven by the day-night heating contrast. Although this vertical mixing is not diffusive in any rigorous sense, a comparison of our results with idealized diffusion models allows a rough estimate of the effective vertical eddy diffusivities in these atmospheres. The parametrization K zz = 5 × 10 4 √ P bar m 2 s −1 , valid from ∼1 bar to a few μbar, can be used in 1D models of HD 209458b. Moreover, our models exhibit strong spatial and temporal variability in the tracer concentration that could result in observable variations during either transit or secondary eclipse measurements. Finally, we apply our model to the case of TiO in HD 209458b and show that the day-night cold trap would deplete TiO if it condenses into particles bigger than a few microns on the planet's nightside, keeping it from creating the observed stratosphere of the planet.

Context.A new class of exoplanets has emerged: the ultra hot Jupiters, the hottest close-in gas giants. The majority of them have weaker-than-expected spectral features in the 1.1 − 1.7µm bandpass probed by HST/WFC3 but stronger spectral features at longer wavelengths probed by Spitzer. This led previous authors to puzzling conclusions about the thermal structures and chemical abundances of these planets. Aims. We investigate how thermal dissociation, ionization, H − opacity, and clouds shape the thermal structures and spectral properties of ultra hot Jupiters. Methods. We use the SPARC/MITgcm to model the atmospheres of four ultra hot Jupiters and discuss more thoroughly the case of WASP-121b. We expand our findings to the whole population of ultra hot Jupiters through analytical quantification of the thermal dissociation and its influence on the strength of spectral features. Results. We predict that most molecules are thermally dissociated and alkalies are ionized in the dayside photospheres of ultra hot Jupiters. This includes H 2 O, TiO, VO, and H 2 but not CO, which has a stronger molecular bond. The vertical molecular gradient created by the dissociation significantly weakens the spectral features from H 2 O while the 4.5µm CO feature remains unchanged. The water band in the HST/WFC3 bandpass is further weakened by the continuous opacity of the H − ions. Molecules are expected to recombine before reaching the limb, leading to order of magnitude variations of the chemical composition and cloud coverage between the limb and the dayside. Conclusions. Molecular dissociation provides a qualitative understanding of the lack of strong spectral features of water in the 1 − 2µm bandpass observed in most ultra hot Jupiters. Quantitatively, our model does not provide a satisfactory match to the WASP-121b emission spectrum. Together with WASP-33b and Kepler-33Ab, they seem the outliers among the population of ultra hot Jupiters, in need of a more thorough understanding.