Search for absorption hydrogen Paschen lines on the ultra-hot Jupiter KELT-9b

Liurong Lin 1 , Alejandro Sanchez Lopez 1 , Ignas Snellen 1

  • 1 Leiden Observatory, Leiden

Abstract

KELT-9b is a hot giant planet orbiting closely to an early A-type star. The extremely high amount of stellar radiation received by this planet make it the hottest exoplanet known to date (with an equilibrium temperature of around 4400K). Due to its high temperature, KELT-9b’s atmosphere is expected to be dominated by neutral and ionized atomic species. The Balmer lines (from H-alpha to H-zeta) have been detected on this planet in its optical transmission spectrum. The Paschen line series has so far not been detected in the atmosphere of other exoplanets. However, with the high-level radiation from KELT-9b’s host star there is a high possibility of detecting the absorption Paschen lines for hydrogen on this exoplanet. In this work, the CARMENES spectrographs in the near-infrared channel are used to search for the atomic hydrogen during transit. The telluric water vapor lines and the stellar flux are removed in the wavelength regions corresponding to Paschen lines n=7 (1005 nm) and n=6 (1094 nm). Then, all the spectra are aligned to the planet’s rest frame. The final spectrum will be shown in the poster, and an analysis of whether the absorption lines are indeed statistically detected in this work will be given.

Methods

The datasets used in this project were observed by CARMENES in the Near-Infrared channel on August 6 and September 21, 2017.  The spectra which have the mean SNRs lower than 40 are removed. There are two methods applied to the datasets to remove the static flux along with the orbital phases. In the first method, The mean spectrum of all the spectra in the times series is fitted to each spectrum with a second-order polynomial, which will remove the main variability in the depth of the static lines. In the next step, the flux in each wavelength grid is fitted as the function of time with a second-order polynomial to remove the extra change of water telluric lines. (Brogi & Line, 2019) The second method is using the principal component analysis. The largest singular value of the time series spectrum matrix is set to be zero to remove the static flux (telluric and stellar lines) while maintaining the planet signal as much as possible. After the static flux has been removed in the time series matrix, the spectra have been aligned to the planet rest frame and then summed up to be the transmission spectrum. (de Kok et al., 2013) The RM and CLV effects from the star during the transit are further removed from the transmission spectrum.

Discussion

The depth of Paschen 3-5 implies that the effective radius for Paschen 3-5 is a 1.41 planet radius for both methods when the photometric transit depth of the planet is assumed to be 0.68% . This radius is close to the planetary Roche radius, which further confirms the works by Yan and Henning (2018) and Wyttenbach et al. (2020) about the atmospheric escape of KELT-9b. The FWHM of this line is much larger than the thermal broadening around 10000K. Therefore, the excited hydrogen layer with an electron in level 3 is required to be optically thick. The offset of the line core could be either due to the motion of the active Hydrogen layer responsible for Paschen 3-5 or the deficiency during the data processes. This is the first time that the Paschen line has been detected in the atmosphere of an extrasolar planet. Using the ratio between the Paschen lines and Balmer lines could give us more details about the temperature structure of KELT-9b. 

Results

The Paschen 3-5 spectrum at each orbital phase, where the spectra are at earth rest frame. Method 1 mentioned in the Methods section is applied to remove static lines. The dark shadow in the middle shows the absorption feature caused by Paschen 3-5, which has doppler shifts along with the orbital phases.

The Paschen 3-5 spectrum at each orbital phase, where each spectrum is in the earth rest frame. The PCA method mentioned in the Methods section is applied to the data.

The combined Paschen 3-5 spectrum by summing up the fully in-transit spectra from phase -0.045 to 0.045, when Method 1 is applied. 

The combined Paschen 3-5 spectrum by summing up the fully in-transit spectra from phase -0.045 to 0.045, when the PCA method is applied.

There is no clear detection for Paschen absorption lines either from level 3 to level 6, or from level 3 to level 7.  Paschen line from level 3 to level 5 is detected with 5.7 sigma  (Method 1) and 7.5 sigma (PCA). This line has the extra absorption depth of 0.7+-0.2% for the first method and 0.7+-0.1% for the PCA method. The FWHM of this line is 31.36+-2.81 km/s when the first method is applied and 40.55+-3.21 km/s when using the PCA method. The observed spectrum has an offset of -7.681 +-1.238 km/s for the first method and -6.573 +-0.982 km/s for the PCA method.

Acknowledgement

This project is using the data from CARMENES.

The fitting of the line is done with emcee.

Thank Núria Casasayas Barris for building the RM+CLV effects model.

Conclusion

In summary, there is no detection of Paschen 3-6 and 3-7 in this project. The detection of Paschen 3-5 has an effective radius of 1.41 planet radius, which could be the evidence for atmospheric escape. The FWHM of this line shows that the fundamental broadening effect comes from the fact that this line is optically thick. 

About Me

I am a first-year master student at Leiden Observatory. I studied sustainable energy engineering as my bachelor. I followed the pre-master program at Leiden Observatory in 2019 and now I am a master student here. This is my first project about astronomy which is under the supervision of Ignas Snellen and Alejandro Sánchez López. I am about to apply for a PhD position, so if you are interested in my project or me, please give me a chance for the interview.