No Robust Statistical Evidence for a Population of Water Worlds in a 2025 Sample of Planets Orbiting M Stars

Abstract

The study of exoplanets has led to many surprises, one of which is the discovery of planets larger than Earth yet smaller than Neptune, super Earths and gas dwarfs. No such planet is a member of the Solar System, yet they appear to be abundant in the local neighbourhood. Their internal structure is not well understood. Super Earths presumably are rocky planets with a thin secondary atmosphere, whereas gas dwarfs have a substantial (by volume) primary H/He atmosphere. However, conflicting evidence exists regarding the presence of a third class of planets, so-called water worlds, which are hypothesised to contain a significant mass fraction of water in condensed or steam form. This study examines the evidence for water worlds and presents a sample of 60 precisely measured small exoplanets (less than 4 Earth radii) orbiting M dwarf stars. We combine observational data and unsupervised machine-learning techniques to classify these planets based on their mass, radius, and density. We individually model the interior of each planet using the ExoMDN code and classify them into populations based on these models. Our findings indicate that the sample divides into two distinct planet populations, with no clear evidence supporting the existence of water worlds in the current dataset.