High Tide or Riptide on the Cosmic Shoreline? A Water-Rich Atmosphere or Stellar Contamination for the Warm Super-Earth GJ~486b from JWST Observations

Abstract

Planets orbiting M-dwarf stars are prime targets in the search for rocky exoplanet atmospheres. The small size of M dwarfs renders their planets exceptional targets for transmission spectroscopy, facilitating atmospheric characterization. However, it remains unknown whether their host stars' highly variable extreme-UV radiation environments allow atmospheres to persist. With JWST, we have begun to determine whether or not the most favorable rocky worlds orbiting M dwarfs have detectable atmospheres. Here, we present a 2.8-5.2 micron JWST NIRSpec/G395H transmission spectrum of the warm (700 K, 40.3x Earth's insolation) super-Earth GJ 486b (1.3 R$_{\oplus}$ and 3.0 M$_{\oplus}$). The measured spectrum from our two transits of GJ 486b deviates from a flat line at 2.2 - 3.3 $σ$, based on three independent reductions. Through a combination of forward and retrieval models, we determine that GJ 486b either has a water-rich atmosphere (with the most stringent constraint on the retrieved water abundance of H2O > 10% to 2$σ$) or the transmission spectrum is contaminated by water present in cool unocculted starspots. We also find that the measured stellar spectrum is best fit by a stellar model with cool starspots and hot faculae. While both retrieval scenarios provide equal quality fits ($χ^2_ν$ = 1.0) to our NIRSpec/G395H observations, shorter wavelength observations can break this degeneracy and reveal if GJ 486b sustains a water-rich atmosphere.

Figures (7)

• Figure 1: Eureka! spectroscopic and white light curves from two transits of GJ 486b. The top two rows contain the spectroscopic light curves (left), our best-fit models (center), and subsequent residuals (right) for each transit. Most evident in the data are wavelength-dependent ramps near 3.2 $\umu$ m that we readily remove. The bottom row depicts the white light curves from each detector (NRS1 and NRS2) before removing their systematic trends. Correlated noise is evident in the residuals and is likely due to thermal cycling Rigby2022. The standard deviation of the normalized residuals is 140 ppm for NRS1 and 165 ppm for NRS2. The complete figure set (3 images, one for each reduction) is available in the online journal.
• Figure 2: Relative transmission spectra of the three data reductions (Eureka!: blue circles, FIREFLy: orange squares, Tiberius: green triangles). The median fit to the Eureka! dataset using an agnostic Gaussian model is shown in purple bounded by $1 \sigma$ and $3 \sigma$ Bayesian credibility envelopes. The legend displays the statistical significance with which each reduction rules out a flat line in favor of the Gaussian model. Analyses of all three reductions reveal an uptick at the blue end of the wavelength range. Instrument throughput deteriorates in the grey shaded region and the measured transit depths become unreliable; thus we exclude points within this region from our hypothesis rejection tests.
• Figure 3: Our final Eureka! spectra of GJ 486b binned to R$\sim$200 (black points) compared to a set of PICASO forward models (colored lines: 1000$\times$ solar, pink; H$_2$O, blue; CO$_2$, orange; CH$_4$, purple; Earth-composition, green). A 1 bar, pure water atmosphere on GJ 486b fits the data with the lowest reduced-$\chi^2$ (1.01), and a flat-line model (dashed grey line) is nearly as well fit by the data (reduced-$\chi^2$ = 1.11), though is weakly rejected by Gaussian vs flat line tests. Alternatively, stellar contamination with water in the atmosphere of the star, rather than the planet, can explain the observed transit depths (see Fig. \ref{['fig:retrieval']}).
• Figure 4: POSEIDON retrieval results for GJ 486b's transmission spectrum. Left: retrieved transmission spectra for two models compared to the JWST NIRSpec G395H data from the Eureka! reduction (black points with error bars). Two scenarios can equivalently explain GJ 486b's transmission spectrum ($\chi^2_{\nu} = 1.0$): unocculted starspots with no planetary atmosphere (orange contours) or a water-rich atmosphere with no starspots (blue contours). The median retrieved spectrum (solid lines) and 1$\sigma$ and 2$\sigma$ confidence intervals (dark and light contours) for each scenario are overlaid. Top right: posterior histograms for the unocculted starspot model, defined by the fractional coverage area of cold stellar heterogeneities/spots (f$_{het}$), the temperature of the heterogeneities/spots (T$_{het}$), and the stellar photospheric temperature (T$_{phot}$). Bottom right: posterior histogram for the water-rich atmosphere scenario, highlighting hydrogen and water's retrieved mixing ratios alongside the atmospheric surface pressure. Water is necessary to explain GJ 486b's spectrum, but the retrievals cannot differentiate between a water-rich planetary atmosphere or water contained in cool starspots that contaminate the transmission spectrum. The complete figure set (3 images, one for each reduction) is available in the online journal.
• Figure 5: Best matching one-, two-, and three-component PHOENIX models to the Baseline GJ 486 spectra from Transits 1 (green) and 2 (black). The bottom two panels zoom-in on the grey highlighted regions of the top panel spectrum. When considering a one-component photosphere, a $T_{\rm{eff}}$ = 3300 K, log($g$)=4.5 cgs model is preferred (purple, $\chi^{2}_{\nu}$ = 72.0). When allowing for spots in a two-component model, a warmer $T_{\rm{eff}}$ = 3400 K, log($g$)=5 cgs photosphere with 25% coverage of $T_{\rm{eff}}$ = 3000 K, log($g$)=5 cgs spots is the preferred model (blue, $\chi{^2}_{\nu}$ = 53.4). The best overall match to the observations is produced with a three-component photosphere+spots+faculae model that has a background photosphere with $T_{\rm{eff}}$ = 3200 K, log($g$)=5 cgs, 20% spot coverage ($T_{\rm{eff}}$ = 3000 K, log($g$)=5 cgs), and 25% faculae coverage ($T_{\rm{eff}}$ = 3400 K, log($g$)=5 cgs) (orange, $\chi^{2}_{\nu}$ = 49.0).