Evidence for SiO cloud nucleation in the rogue planet PSO J318

TL;DR

This study analyzes panchromatic JWST data (0.6–27.9 μm) plus archival spectra for the young, planetary-mass brown dwarf PSO J318 to characterize its atmospheric clouds and silicate mineralogy. A brightness-temperature method disfavors most condensates and points to amorphous SiO as the primary absorber shaping the 10 μm feature, which is subsequently confirmed by detailed two-column petitRADTRANS retrievals that also reveal a patchy Fe cloud and a high-altitude SiO haze with particle sizes around tens of nanometers. The results yield Teff ≈ 1114 K, R ≈ 1.495 R_Jup, [M/H] ≈ 0.31, and C/O ≈ 0.789, with 12C/13C ≈ 45, suggesting a slightly supersolar metallicity and near-solar C/O. The analysis supports a scenario in which SiO nucleation seeds cloud formation at high altitudes, offering a potential explanation for rapid silicate cloud formation in cold substellar atmospheres, while highlighting methodological challenges in retrieving multi-dimensional atmospheric states from JWST data and the need for improved microphysical cloud models.

Abstract

Silicate clouds are known to significantly impact the spectra of late L-type brown dwarfs, with observable absorption features at ~ 10 micron. JWST has reopened our window to the mid-infrared with unprecedented sensitivity, bringing the characterization of silicates into focus again. Using JWST, we characterize the planetary-mass brown dwarf PSO J318.5338-22.8603, concentrating on any silicate cloud absorption the object may exhibit. PSO J318's spectrum is extremely red, and its flux is variable, both of which are likely hallmarks of cloud absorption. We present JWST NIRSpec PRISM, G395H, and MIRI MRS observations from 1-18 micron. We introduce a method based on PSO J318's brightness temperature to generate a list of cloud species that are likely present in its atmosphere. We then test for their presence with petitRADTRANS retrievals. Using retrievals and grids from various climate models, we derive bulk parameters from PSO J318's spectra, which are mutually compatible. Our retrieval results point to a solar to slightly super-solar atmospheric C/O, a slightly super-solar metallicity, and a 12C/13C below ISM values. The atmospheric gravity proves difficult to constrain for both retrievals and grid models. Retrievals describing the flux of PSO J318 by mixing two 1-D models (``two-column models'') appear favored over single-column models; this is consistent with PSO J318's variability. The JWST spectra also reveal a pronounced absorption feature at 10 micron. This absorption is best reproduced by introducing a high-altitude cloud layer of small (<0.1 micron), amorphous SiO grains. The retrieved particle size and location of the cloud is consistent with SiO condensing as cloud seeding nuclei. High-altitude clouds comprised of small SiO particles have been suggested in previous studies, therefore the SiO nucleation we potentially observe in PSO J318 could be a more wide-spread phenomenon.

Figures (9)

• Figure 1: Observations considered for the atmospheric characterization of PSO J318.Uppermost panel: all data considered for the spectral characterization of PSO J318, at the wavelength binning fed into the retrieval and self-consistent grid model fits. Specifically, MIRI MRS data was binned down to $\lambda/\Delta\lambda=400$, while NIRSpec G395H was binned to $\lambda/\Delta\lambda=400$ and $\lambda/\Delta\lambda=1000$ for NRS1 and NRS2, respectively. Second panel: the HST WFC3, IRTF SpeX and JWST NIRSpec PRISM data over the 1-3 µm wavelength range. Third panel: the JWST NIRSpec G395H data at the full spectral resolution. Lowermost panel: the JWST MIRI MRS data at the full spectral resolution.
• Figure 2: Opacity fits of the $10$ µm feature for the "Top-14" silicate cloud species, using the brightness temperature method. The remaining fits for the less well fitting species can be found in Fig. \ref{['fig:brightness_temperature_all_appendix_1']}
• Figure 3: Model fits of PSO J318 obtained with our fiducial two-column retrieval setup, considering different types of silicate clouds. Panel a shows the best-fit spectrum of the overall winning model (black solid line) plotted on top of the JWST data (gray circles, with $10^b$ error scaling). The winning model assumes amorphous spherical SiO particles. The flux contribution of the two individual columns is also shown, at their best-fit relative scaling (salmon and rose colored lines, respectively). The light blue line shows the result of recalculating the best-fit model, but turning off the cloud opacities. Panel b shows the residuals between the best fit model and the data (with $10^b$ error scaling). Panel c is a version of Panel a that zooms in on the silicate feature, but with the best-fit model shown in pink instead. Panels d-m show the same zoomed-in view of the silicate feature, but for the other tested silicate feature candidates. Panel n shows the residuals between model and data for the various silicate cloud species, using the same colors as in panels c-m.
• Figure 4: Best-fit spectra and marginalized posterior distributions of the grid interpolation and free retrievals. First row: Exo-REM fit; second row: ATMO fit; third row: Sonora Diamondback fit; fourth row:petitRADTRANS fit. The fifth row shows the projected 1-d posteriors for [M/H], C/O, $\rm ^{12}C/^{13}C$, $T_{\rm eff}$, ${\rm log}_{10}(g)$ and the radius $R$ in the first to sixth column, respectively. For C/O, the posterior of the petitRADTRANS retrieval is shown twice, because it was nominally calculated from the gas phase abundances only. For solar C/O, $\sim$25 % reduction of gas phase oxygen is expected due to condensation sanchezlopezlandman2022, which we applied to the second, less opaque C/O posterior shown for petitRADTRANS. The last three rows show the posteriors obtained from considering the data's 14 wavelength bands separately for the self-consistent grids, with the vertical dashed lines indicating the parameter grid values.
• Figure 5: Overview plot of the temperature, emission contribution, and cloud properties of the "winning" SiO cloud model with amorphous spherical particles, two atmospheric columns, and evolutionary priors.Panel a: retrieved pressure-temperature profiles. The pRT retrieval profile is shown in magenta. The 1-3 $\sigma$ posterior regions of the temperature are indicated by progressively lighter magenta tones, but are difficult to distinguish because of the narrow posteriors. We also overplot the ${\rm log}_{10}(g)$-scaled $P$-$T$ profiles of the grid models that are closest to the best-fit values obtained with species. Panel b shows the log(emission contribution function)s of atmospheric Columns 1 and 2 as orange and purple contours. While nominally computed to sum to unity over all pressures at a given wavelength, the values have been scaled by the relative contribution of the two columns to the total atmospheric solution, and multiplied by the wavelength-dependent flux. Panel c: size distribution of SiO cloud particles in Column 1 (shown in red), and Fe cloud particles in Column 1 and 2 (shown in blue and orange, respectively). For better comparability the distributions' peak values have been normalized to the same y-axis values. The envelopes around the median distribution would correspond to the 1-$\sigma$ uncertainty if the posterior variations around the median followed a Gauss distribution. Panel d: altitude-dependent cloud mass fraction of the SiO cloud (present in Column 1, shown in red) and of the Fe cloud (present in both Columns, shown in blue). The colored envelopes again represent the 1-$\sigma$ uncertainty spread of the posterior distribution. The dashed lines correspond to the expected cloud deck position, obtained from intersecting the sampled P-T curves (see Panel a) with the saturation vapor pressure curves of SiO and Fe from gailwetzel2013ackermanmarley2001. Panel e: SiO nucleation rate computed according to gailwetzel2013.