Jupiter-like uniform metal enrichment in a system of multiple giant exoplanets

Abstract

The accretion of icy and rocky solids during the formation of a gas giant planet is poorly constrained and challenging to model. Refractory species, like sulfur, are only present in solids in the protoplanetary disk where planets form. Measuring their abundance in planetary atmospheres is one of the most direct ways of constraining the extent and mechanism of solid accretion. Using the unprecedented sensitivity of NASA's James Webb Space Telescope (JWST), we measure a detailed chemical make-up of three massive gas giants orbiting the star HR~8799 including direct detections of H$_2$O, CO, CH$_4$, CO$_2$, H$_2$S, $^{13}$CO, and C$^{18}$O. We find these planets are uniformly and highly enriched in heavy elements compared to the star irrespective of their volatile (carbon and oxygen) or refractory (sulfur) nature, which strongly suggests efficient accretion of solids during their formation. This composition closely resembles that of Jupiter and Saturn and demonstrates that this enrichment also occurs in systems of multiple gas giant planets orbiting stars beyond the Solar System. This discovery hints at a shared origin for the heavy element enrichment of giant planets across a wider range of planet masses and orbital separations than previously anticipated.

Figures (10)

• Figure 1: Detection of the three inner planets c, d, and e orbiting the star HR 8799 with the moderate resolution mode of JWST/NIRSpec IFU in the $3-5\,\mu$m spectral range. a. Median spectral cube prior to starlight subtraction using the standard JWST pipeline reduction. b. Signal-to-noise ratio map for planet detection. HR 8799 c, d, and e are detected with an S/N of 119, 94, 67, respectively.
• Figure 1: JWST/NIRSpec spectrum of HR 8799 d. See caption of Figure \ref{['fig:spec']}. Panels a, c, e show the observed spectrum in black and the best-fit model in orange. In the sub-panels below, the residuals are plotted as gray lines and the $1.5\sigma$ uncertainties are shown as orange contours. The factor of 1.5 comes from the retrieved error scaling factor. We indicate locations of positive residuals in panel c, which result from over-subtraction of Br$\alpha$ and Pf$\gamma$ lines in HR 8799 A (see Spectral Extraction). Panels b, d, and f show data residuals after fitting an atmospheric model without a given species (CH$_4$, H$_2$S, CO$_2$, $^{13}$CO, C$^{18}$O) in black, and the corresponding molecular templates in color. The similarity between the data residuals and molecular templates indicate that the highlighted species contribute significantly to the planet's spectra. On the right insets, we plot the cross-correlation functions (CCF) between the data residuals and models in the left insets. The CCF provides an estimate of the detection S/N for each molecule. Panel g shows the photometry data in blue and orange points (with error bars representing $1\sigma$ uncertainties in y, and photometric bandpass sizes in x), the best-fit photometry model in open circles, and random draws of the model spectrum at $R=100$ in purple.
• Figure 1: Planet detection using the BREADS forward model framework for JWST/NIRSpec IFU. a. Planet absolute flux from fitting NRS1. The flux is expressed in the F356W filter but the fits include the entire NRS1 spectral range. b. Corresponding flux uncertainty for NRS1. c. Planet signal-to-noise ratio (S/N) map for NRS1. d. Histogram of the (S/N) map after masking the planets. Comparing the S/N histogram to a Gaussian distribution ensures the validity of the planet detection limits. e. is the same for NRS2, while f. results from combining NRS1 and NRS2 together. These panels are made similarly to Figure 13 and Section 4.6 in Ruffio2024.
• Figure 2: JWST/NIRSpec spectrum of HR 8799 c. Panels a, c, e show the observed spectrum in black and the best-fit model in orange. In the sub-panels below, the residuals are plotted as gray lines and the $1.5\sigma$ uncertainties are shown as orange contours. The factor of 1.5 comes from the retrieved error scaling factor. Panels b, d, and f show data residuals after fitting an atmospheric model without a given species (CH$_4$, H$_2$S, CO$_2$, $^{13}$CO, C$^{18}$O) in black, and the corresponding molecular templates in color. The similarity between the data residuals and molecular templates indicate that the highlighted species contribute significantly to the planet's spectra. On the right insets, we plot the cross-correlation functions (CCF) between the data residuals and models in the left insets. The CCF provides an estimate of the detection S/N for each molecule. Panel g shows the photometry data in blue and orange points (with error bars representing $1\sigma$ uncertainties in y, and photometric bandpass sizes in x), the best-fit photometry model in open circles, and random draws of the model spectrum at $R=100$ in purple.
• Figure 2: JWST/NIRSpec spectrum of HR 8799 e. Panels a, c, f show the observed spectrum in black and the best-fit model in orange. In the sub-panels below, the residuals are plotted as gray lines and the $1.5\sigma$ uncertainties are shown as orange contours. The factor of 1.5 comes from the retrieved error scaling factor. Panels b and d show data residuals after fitting an atmospheric model without a given species (CH$_4$, CO$_2$, $^{13}$CO) in black, and the corresponding molecular templates in color. The similarity between the data residuals and molecular templates indicate that the highlighted species contribute significantly to the planet's spectra. On the right insets, we plot the cross-correlation functions (CCF) between the data residuals and models in the left insets. The CCF provides an estimate of the detection S/N for each molecule. Panel e shows the photometry data in blue and orange points (with error bars representing $1\sigma$ uncertainties in y, and photometric bandpass sizes in x), the best-fit photometry model in open circles, and random draws of the model spectrum at $R=100$ in purple.