Two brown dwarfs observed by the James Webb Space Telescope have emerged from a new modeling analysis with different preferred cloud structures. SDSS1416 was best represented by an MgSiO3 enstatite slab cloud with an iron deck cloud and no thermal inversion. GJ 499 C was best represented by SiO and Mg2SiO4 slab clouds plus an iron deck cloud, while its preferred model favored a thermal inversion near log P = −2.2 bar — a temperature profile that turns upward through part of the atmosphere.
The work tested whether atmospheric retrievals — models that infer atmospheric properties from spectra — could produce cloud, chemical and bulk-property estimates that could be checked against the stars’ compositions. It examined two wide-separation benchmark mid-L companions: SDSSJ141659.78+500626.4, known as SDSS1416, and GJ 499 C. JWST observations covered 0.6–14.0 micrometres at resolving power of about 100, with signal-to-noise ratios of approximately 100 to 600.
Reading the clouds in the spectra
Raw JWST data were processed with pipeline version 1.18.1 and then analyzed with Brewster, a synthetic-spectrum forward model paired with Bayesian posterior sampling. Typical retrievals used 50,000 iterations, 10,000 burn-in steps and 16 sampling walkers, followed by another 30,000 to 50,000 iterations. The competing fits were assessed with the Bayesian information criterion, or BIC, which uses the maximum likelihood while accounting for the number of model parameters.
The analysis also reported silicate and water spectral indices for both objects. The silicate index was 1.19 ± 0.01 for SDSS1416 and 1.32 ± 0.01 for GJ 499 C. Water indices were 1.187 ± 0.002 and 1.151 ± 0.001, respectively. The authors interpreted these results as evidence for silicate clouds in both atmospheres.
SDSS1416’s model and chemistry
Both targets had a gap much larger than 10 in ΔBIC between their first- and second-ranked models. For SDSS1416, the preferred setup was the MgSiO3 enstatite slab with an Fe deck cloud and no thermal inversion. Its atmospheric retrieval estimated a carbon-to-oxygen ratio, C/O, of 0.71 and [M/H] of 0.22, with displayed uncertainty magnitudes of 0.01 and 0.03. Separate retrievals gave [C/H] = 0.31 and [O/H] = 0.19, each with a reported uncertainty of ±0.03.
The cloud-derived Mg/Si ratio for SDSS1416 was 1.0, close to about 0.95 for its primary star. The authors interpreted that agreement as consistent with shared bulk chemistry and expected silicate chemistry. They also said the elevated atmospheric C/O might reflect oxygen being sequestered beyond the retrieved cloud mass, a deeper or missing cloud, or shortcomings in the model or its opacities.
The same retrieval inferred a radius of 0.85 Jupiter radii and a mass of 73.7 Jupiter masses for SDSS1416. Interpolation onto evolutionary models gave an age range of 4.3 to 12.9 billion years. The authors regarded these bulk-property estimates as broadly compatible with evolutionary expectations and the primary star’s age information, while noting that they remain model-derived.
GJ 499 C presents a harder cloud problem
For GJ 499 C, the coexistence of SiO, Mg2SiO4 and Fe in the winning model was harder to reconcile with simple phase-equilibrium expectations, the authors said. The same SiO-plus-forsterite cloud combination without an inversion had a BIC of 110.54. The proposed inversion remains model-dependent and, in the authors’ view, needs further investigation; vertical or latitudinal cloud differences were among the possible explanations they raised.
The GJ 499 C retrieval estimated C/O at 0.69, with a reported ±0.02 interval, and [M/H] at 0.13, with reported +0.04/−0.06 intervals. The authors interpreted the elevated C/O as compatible with oxygen depletion through silicate condensation and rainout. Its cloud-derived Mg/Si was approximately 1.9, and about 95% of the silicon-bearing condensate molecules were Mg2SiO4.
GJ 499 C had an inferred radius of 1.00 Jupiter radii and a mass of 68 Jupiter masses. Its evolutionary-model age range was 0.5 to 0.9 billion years. The authors considered these values physically plausible and consistent with a young system, although the retrieved surface gravity was higher than expected for such a young brown dwarf.
A focused comparison with visible gaps
Neither set of fits was complete. SDSS1416 showed difficulty in the J and H bands, while GJ 499 C had more prominent residuals in those bands and a mismatch near 8.8 micrometres. The discrepancies leave room for cloud structures, opacities or abundance patterns that the tested configurations did not capture.
This is a two-object comparison, not a general census of brown-dwarf atmospheres. The conclusions therefore remain object-specific and model-dependent. A direct Mg/Si comparison is unavailable for GJ 499 C because its primary star’s ratio was not available, and the proposed thermal inversion still requires further investigation.
The document is an arXiv v1 preprint dated 26 August 2026. The authors report that reproducibility code, data products, Brewster driver files and analysis notebooks are available in a named GitHub branch.
Paper data and sources
Original title: Benchmark Brown Dwarfs as Chemical Laboratories: Linking System Bulk Properties to Atmospheric Retrievals
Authors: Viktória Kecskeméthy, Ben Burningham, Fei Wang et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text