2021MNRAS.506.1944B

We present the most detailed data-driven exploration of cloud opacity in a substellar object to-date. We have tested over 60 combinations of cloud composition and structure, particle-size distribution, scattering model, and gas phase composition assumptions against archival 1-15 µm spectroscopy for the unusually red L4.5 dwarf 2MASSW J2224438-015852 using the Brewster retrieval framework. We find that, within our framework, a model that includes enstatite and quartz cloud layers at shallow pressures, combined with a deep iron cloud deck fits the data best. This model assumes a Hansen distribution for particle sizes for each cloud, and Mie scattering. We retrieved particle effective radii of log₁₀ a (µm) = -1.41^+0.18_–0.17 for enstatite, -0.44^+0.04_–0.20 for quartz, and -0.77^+0.05_–0.06 for iron. Our inferred cloud column densities suggest (Mg/Si) = 0.69^+0.06_–0.08 if there are no other sinks for magnesium or silicon. Models that include forsterite alongside, or in place of, these cloud species are strongly rejected in favour of the above combination. We estimate a radius of 0.75 ± 0.02 R_Jup, which is considerably smaller than predicted by evolutionary models for a field age object with the luminosity of 2M2224-0158. Models which assume vertically constant gas fractions are consistently preferred over models that assume thermochemical equilibrium. From our retrieved gas fractions, we infer [M/H] = +0.38^+0.07_–0.06 and C/O = 0.83^+0.06_–0.07. Both these values are towards the upper end of the stellar distribution in the Solar neighbourhood, and are mutually consistent in this context. A composition towards the extremes of the local distribution is consistent with this target being an outlier in the ultracool dwarf population.