A High-Resolution NUV Transmission Spectrum of KELT-9b: Mg II and Fe II Escaping from the Hottest Known Giant Planet

Abstract

We present high-resolution NUV observations from Hubble Space Telescope's (HST) Space Telescope Imaging Spectrograph (STIS) data for the hottest known gas planet, KELT-9b. Observations were collected with STIS/E230M (2300-3000 $Å$, R$\sim$ 30,000) and we de-correlate systematic effects from the telescope using jitter detrending. We show the clear presence of the Mg II doublet at 2800 $Å$ and Fe II at 2600 $Å$ in KELT-9b. The Mg II is measured above the planet's Roche transit radius, indicating it is escaping. We fit 1D NLTE atmospheric escape models to these features, demonstrating a significant loss of mass in KELT-9b's atmosphere ($\dot{M} \approx 10^{12} $ g/s); we also find a remarkably high line-broadening corresponding to a velocity of about $50-75$ km/s, and a net blueshift of the Mg II doublet greater than 30 km/s. Future 3D MHD modeling of the spectrum and gas kinematics is likely needed to explain these observations. We interpret these results in the context of the Mg II ``Cosmic Shoreline" and show that the detection of escaping Mg II in KELT-9b and the non-detection in WASP-178b are consistent with the hypothesis that stars hotter than $T_{\mathrm{eff}} \sim$ 8250~K have relatively low levels of XUV radiation due to the lack of a chromosphere. Therefore planets around such early-type stars experience a different degree of atmospheric escape. This result highlights the importance of XUV irradiation in driving atmospheric escape inside and outside the Solar System.

Figures (3)

• Figure 1: Top: Program-level co-added HST/STIS/E230M spectrum of host star KELT-9. Middle: Same as above, but zoomed in on the Mg II doublet at 2796.35 and 2803.53 $\mathrm{\AA}$ and an Mg I line at 2853.00 $\mathrm{\AA}$. Bottom: Same as above, but zoomed in on Fe II lines at 2344.25, 2382.80, 2586.69, and 2600.21 $\mathrm{\AA}$.
• Figure 2: The initial second-degree systematics vector fitting for the first transit, with respect to HST's orbital phase, represented as the time since the first exposure in the orbit.
• Figure 3: Top: The raw STIS/E230M white-light curve for KELT-9b for Visit 1 (left) and Visit 2 (right) with the fitted systematics model. Middle: The systematics-subtracted light curve, fitted to the detrended data. Bottom: The residuals between the detrended data and the transit model. For the first visit, we find a standard deviation about 0.98% of the data, while for the second visit, we find residual scatter of about 0.38%.