No consistent atmospheric absorption detected for the ultra-hot Jupiter WASP-189 b

Transiting ultra-hot Jupiters (UHJs), which have day-side temperatures of $\gtrsim 2200$ K (Parmentier et al. 2018), offer the opportunity to explore an extreme regime of atmospheric physics. Peering through hot planet atmospheres in transmission at high spectral resolution can provide details on the composition (Hoeijmakers et al. 2018; Casasayas-Barris et al. 2019; Keles et al. 2019), temperature structure (Wyttenbach et al. 2015, 2017), geometry (Yan & Henning 2018), and dynamics (Louden & Wheatley 2015; Allart et al. 2018; Cauley et al. 2019) of material in the extended gravitationally-bound thermosphere. In this research note we report on the null detection of an atmospheric signature around the ultra-hot Jupiter WASP-189 b (Anderson et al. 2018). WASP-189 b orbits a bright ($V=6.64$) A6IV-V star with an orbital period of $\approx 2.7$ days and is inflated with a radius of $R_{\text{p}}\approx 1.4R_{J}$. Its day-side equilibrium temperature of $\approx 2640$ K makes it one of the hottest gas giants discovered to-date.

We observed the transit of WASP-189 b between 05:15–10:50 UT on May 15, 2019 with the Large Binocular Telescope in Arizona and its high-resolution échelle spectrograph PEPSI (Strassmeier et al. 2015). Due to poor weather conditions, the first half of the transit was not observed and no pre-transit observations were obtained. Clouds also prevented continuous post-transit exposures, resulting in a gap between 07:15 UT and 09:05 UT. PEPSI was used in its $R\approx 50,000$ mode and with cross dispersers (CD) III (blue arm) and V (red arm) simultaneously. The wavelength coverage was 4750–5430 Å in the blue arm and 6230–7430 Å in the red arm. The spectra were collected with a constant signal-to-noise of 300 in the continuum controlled by a photon counter. All data were reduced with the Spectroscopic Data System for PEPSI (SDS4PEPSI). The observational setup and reduction procedures are identical to those outlined in Cauley et al. (2019).

In addition to applying standard data reduction steps, we fit and removed telluric absorption using MOLECFIT (Kausch et al. 2015; Smette et al. 2015). The profile of the stellar surface occulted by the planet during transit was removed using synthetic occulted surface profiles generated by a stellar photosphere model and assuming the system parameters from Anderson et al. (2018). The model photosphere was created with Spectroscopy Made Easy (Valenti & Piskunov 1996; Piskunov & Valenti 2017).

The transmission spectrum for each exposure is created by dividing each spectrum by the signal-to-noise weighted mean out-of-transit spectra. The individual transmission spectra are divided by the synthetic occulted surface profiles to remove the effect of the occulted stellar surface. Each spectrum is then shifted into the rest frame of the planet. The equivalent width of any line of interest is then integrated across $\pm 50$ km s^-1 of the line’s rest wavelength.

We searched for absorption in a number of metal ions, including Fe I, Fe II, Mg I, and Ti I and in the Balmer lines H$\alpha$ and H$\beta$. We found marginal ($\approx 1-2\sigma$) hints of absorption in the mean transmission spectra of Mg I triplet at 5167, 5173, and 5184 Å and also in the Fe I 5168.9 Å line (see Figure 1). However, the absorption time series for these lines show no consistent in-transit signal, suggesting that the mean transmission signals do not arise in the planetary atmosphere. Similar marginal signals are seen in the Balmer lines H$\alpha$ and H$\beta$. The time series absorption (right panels of Figure 1) is highly variable and the individual spectra with the strongest absorption are likely driving the signal in the mean transmission spectrum.

Although we only observed a partial transit and lost segments of time to clouds, the spectra we were able to collect are high signal-to-noise and thus we can use the in-transit measurements to place constraints on the extent of the planet’s thermosphere. The depth of the H$\alpha$ transmission spectrum places an upper limit on the radius of the excited hydrogen thermosphere of $\approx 0.2R_{\text{p}}$ above the optical radius. The depth of the metal line transmission spectra are even smaller, constraining their radial extent to $<0.2R_{\text{p}}$.

The lack of strong atomic transmission signatures in the atmosphere of WASP-189 b is surprising given its high temperature. Such absorption lines are easily detectable in a number of other UHJs (e.g., Jensen et al. 2018; Casasayas-Barris et al. 2019; Yan et al. 2019; Cauley et al. 2019). However, WASP-189 b has the smallest transit depth of the known UHJs ($\approx 0.4\%$) decreasing the depth of any transmission signatures for a given atmospheric radius compared to the other UHJs. The highly variable nature of the in-transit signals suggests that the transmission lines profiles are not the result of absorption in the planet’s atmosphere. We suggest that the in-transit features are instead due to an inhomogeneous stellar disk and/or stellar variability caused by magnetic surface features. Such features are now known to exist on A-type stars (e.g., Petit et al. 2017) and could produce the small variations that we observe via the contrast effect (Cauley et al. 2018).

Our observations demonstrate the lack of a highly extended atmospheric around WASP-189 b. However, more transit observations in better conditions and covering the entire transit are needed to entirely rule out the presence of a thermosphere above $\approx 1.2R_{\text{p}}$. The depth of the features detailed here should be taken into account for future transit observations in order to place stronger constraints on the atmosphere than those we have presented.