JWST Finds Patchy Forsterite and Enstatite Clouds on a Restless Planetary-Mass Companion

About 69 light-years away, a young planetary-mass companion called VHS 1256 b orbits a pair of small stars. It is only about 140 million years old, and it is one of the most variable substellar objects known. Hubble has caught brightness swings of about 25%, and a later Hubble monitoring campaign saw nearly 40% in the near-infrared. Spitzer saw a smaller but still significant shift of about 6%.

Something in its atmosphere is moving, breaking apart, or rotating into view. The question is what.

VHS 1256 b sits in the L/T transition, a spectral boundary where brown dwarfs and young planetary-mass objects shift from red to blue in the near-infrared. Astronomers have long suspected that silicate clouds drive that change, but direct evidence has been hard to get for directly imaged companions, which are drowned out by their host stars.

JWST changed that. As part of the Early Release Science program #1386, astronomers observed VHS 1256 b with the Near-Infrared Spectrograph (NIRSpec) and the Mid-Infrared Instrument (MIRI). Together they produced a spectrum from about 1 to 18 microns—far wider and higher quality than any previous view of such an object. The new study, posted as a preprint on arXiv, is the first retrieval analysis of that full dataset.

Retrieval is a statistical inverse method. It lets a flexible model adjust many parameters—temperatures at different pressures, molecular abundances, cloud properties—until the model spectrum matches the observed one. The team used a framework called Brewster and tested several cloud recipes: uniform versus patchy clouds, and different silicate minerals—forsterite (Mg2SiO4), enstatite (MgSiO3), and quartz (SiO2)—along with a deep iron cloud deck.

Patchy Forsterite and Enstatite Above an Iron Deck

The best fit combined patchy forsterite and enstatite slabs with an iron deck. That combination reproduced the broad silicate feature near 10 microns, a signature of rocky cloud particles. But the more striking result was about structure: a patchy cloud model was strongly preferred over a uniform one, with a difference in the Bayesian Information Criterion—a score balancing fit quality against model complexity—of 297. The atmosphere is not a smooth, even blanket. It has gaps, though only a small cloud-free fraction.

The retrieved cloud coverage is about 98.5%, leaving only a small cloud-free fraction. That is roughly consistent with earlier modeling of VHS 1256 b's variability, which suggested only a tiny cloud-free patch is needed to produce the observed brightness swings.

Chemistry, a Provisional Mass, and an Isothermal Puzzle

The retrieval also constrained water (H2O), carbon monoxide (CO), carbon dioxide (CO2), methane (CH4), and ammonia (NH3). The carbon-to-oxygen ratio came out slightly above solar, and the retrieved metallicity—the abundance of elements heavier than helium—about twice solar. The ammonia constraint is notable: for an L dwarf, this may be the earliest spectral type where ammonia has been constrained by a retrieval. The simultaneous presence of CO and CH4 may hint at disequilibrium chemistry—vertical mixing that could bring carbon monoxide up from deeper, hotter layers while methane forms higher up—though the paper says more testing is needed.

Bulk properties were less comfortable. The model retrieved a radius of 1.29 Jupiter radii and a surface gravity of log g = 4.68, implying a mass of about 32 Jupiter masses. But that value converged to the upper edge of the allowed range, and it disagrees with earlier estimates that place VHS 1256 b below 25 Jupiter masses. The mass and gravity should be treated as provisional.

The retrieved temperature-pressure profile was also more isothermal than self-consistent models predict: temperature changes more gradually with altitude than standard models suggest. (Isothermal means the temperature stays relatively flat with altitude.) That mismatch has appeared in retrievals of other young, low-gravity objects, and it sits at the center of a debate about whether clouds or another process, such as fingering convection, shapes the red colors of these worlds.

Retrieval Lessons: When NIRSpec Drowns Out MIRI

Perhaps the paper's most important lesson is about caution. The team found that retrieved parameters depended on which data they used and how they weighted it. Early on, NIRSpec's high signal-to-noise led the retrieval to likely overfit it and nearly ignore the fainter MIRI data, including the silicate feature. To balance the instruments, the team inflated NIRSpec's error bars by a factor of 10. When they used raw signal-to-noise, or only NIRSpec without MIRI, the retrieved parameters and temperature profiles shifted. The paper concludes that precise retrieval numbers for such objects should not yet be over-interpreted.

Other caveats remain. The retrieval itself was run on data binned to a resolution of about R=300, not at JWST's native resolution. The model fit was generally within one standard deviation, but it struggled in the L band around 3.0–3.8 microns and in MIRI channel 3 from about 15–18 microns. Some mismatches could come from imperfect data reduction rather than the atmosphere. The cloud model is also a simplified, one-dimensional picture of what is likely a dynamic, three-dimensional atmosphere that changes in time.

Still, the result fits a growing picture: silicate clouds are real in these atmospheres, and they are not uniform. For VHS 1256 b, the patchiness aligns with its extreme variability. One idea is that as the object rotates every 21 to 24 hours, different cloud patches could rotate into view, changing the brightness and spectrum.

The team outlines next steps: full-resolution retrievals, tests of different retrieval codes and temperature assumptions, a wider search for cloud species and structures, and machine learning to handle the computational cost. Time-series JWST observations are planned, which could turn this snapshot into a movie of a changing atmosphere.

For now, VHS 1256 b remains a restless world of patchy forsterite and enstatite clouds, spinning in the light of its two small stars. JWST has given astronomers the clearest spectrum yet of such an atmosphere—and a reminder that reading it will take more than one model.