This work aims to improve the current understanding of the atmospheres of brown dwarfs, especially cold ones with spectral type T and Y, whose modeling is a current challenge. Silicate and iron clouds are believed to disappear at the photosphere at the L/T transition, but cloudless models fail to reproduce correctly the spectra of T dwarfs, advocating for the addition of more physics, e.g. other types of clouds or internal energy transport mechanisms.We use a one-dimensional (1D) radiative/convective equilibrium code ATMO to investigate this issue. This code includes both equilibrium and out-of-equilibrium chemistry and solves consistently the PT structure. Included opacity sources are H 2 -H 2 , H 2 -He, H 2 O, CO, CO 2 , CH 4 , NH 3 , K, Na, and TiO, VO if they are present in the atmosphere.We show that the spectra of Y dwarfs can be accurately reproduced with a cloudless model if vertical mixing and NH 3 quenching are taken into account. T dwarf spectra still have some reddening in e.g. J − H compared to cloudless models. This reddening can be reproduced by slightly reducing the temperature gradient in the atmosphere. We propose that this reduction of the stabilizing temperature gradient in these layers, leading to cooler structures, is due to the onset of fingering convection, triggered by the destabilizing impact of condensation of very thin dust.

The admitted, conventional scenario to explain the complex spectral evolution of brown dwarfs (BD) since their first detections twenty years ago, has always been the key role played by micron-size condensates, called "dust" or "clouds", in their atmosphere. This scenario, however, faces major problems, in particular the Jband brightening and the resurgence of FeH absorption at the L to T transition, and a physical first-principle understanding of this transition is lacking. In this paper, we propose a new, completely different explanation for BD and extrasolar giant planet (EGP) spectral evolution, without the need to invoke clouds. We show that, due to the slowness of the CO/CH 4 and N 2 /NH 3 chemical reactions, brown dwarf (L and T, respectively) and EGP atmospheres are subject to a thermo-chemical instability similar in nature to the fingering or chemical convective instability present in Earth oceans and at the Earth core/mantle boundary. The induced small-scale turbulent energy transport reduces the temperature gradient in the atmosphere, explaining the observed increase in near infrared J -H and J -K colors of L dwarfs and hot EGPs, while a warming up of the deep atmosphere along the L to T transition, as the CO/CH 4 instability vanishes, naturally solves the two aforementioned puzzles, and provides a physical explanation of the L to T transition. This new picture leads to a drastic revision of our understanding of BD and EGP atmospheres and their evolution.