A Survey of Alkali Line Absorption in Exoplanetary Atmospheres

Abstract

We obtained over 90 hours of spectroscopic observations of four exoplanetary systems with the Hobby-Eberly Telescope (HET). Observations were taken in transit and out of transit, and we analyzed the differenced spectra---i.e., the transmission spectra---to inspect it for absorption at the wavelengths of the neutral sodium (\ion{Na}{1}) doublet at $\lambda\lambda5889, 5895$ and neutral potassium (\ion{K}{1}) at $\lambda7698$. We used the transmission spectrum at \ion{Ca}{1} $\lambda6122$---which shows strong stellar absorption but is not an alkali metal resonance line that we expect to show significant absorption in these atmospheres---as a control line to examine our measurements for systematic errors. We use an empirical Monte Carlo method to quantity these systematic errors. In a reanalysis of the same dataset using a reduction and analysis pipeline that was derived independently, we confirm the previously seen \ion{Na}{1} absorption in HD 189733b at a level of $(-5.26\pm1.69)\times10^{-4}$ (the average value over a 12 \AA{} integration band to be consistent with previous authors). Additionally, we tentatively confirm the \ion{Na}{1} absorption seen in HD 209458b (independently by multiple authors) at a level of $(-2.63\pm0.81)\times10^{-4}$, though the interpretation is less clear. Furthermore, we find \ion{Na}{1} absorption of $(-3.16\pm2.06)\times10^{-4}$ at $<3\sigma$ in HD 149026b; features apparent in the transmission spectrum are consistent with real absorption and indicate this may be a good target for future observations to confirm. No other results (\ion{Na}{1} in HD 147506b and \ion{Ca}{1} and \ion{K}{1} in all four targets) are significant to $\geq 3\sigma$, although we observe some features that we argue are primarily artifacts.