A. Lueber, D. Kitzmann, K. Heng
The James Webb Space Telescope offers exceptional spectral resolution and wavelength coverage, which are essential for studying the coldest brown dwarfs, particularly Y dwarfs. These objects are at the cold end of the substellar sequence and exhibit atmospheric phenomena such as cloud formation, chemical disequilibrium, and radiative-convective coupling. We examined a curated sample of 22 late-T to Y dwarfs through Bayesian atmospheric retrieval (nested sampling) and supervised machine learning (random forests). Our Bayesian model comparison indicates that cloud-free models are generally favored for the hottest objects in the sample (T6-T8). Conversely, later-type dwarfs exhibit varying preferences, with both gray cloud and cloud-free models providing comparable fits. The atmospheric parameters retrieved are consistent across the applied methodologies. Evidence of vertical mixing and disequilibrium chemistry is found in several objects; notably, the Y1 dwarf WISEPAJ1541-22 favors a gray cloud model and shows elevated abundances of both CO and CO 2 compared to equilibrium chemistry calculations. As anticipated, the abundances of H 2 O, CH 4 , and NH 3 increase with decreasing effective temperature over the T-Y sequence.