The transmission spectrum of WASP-17 b from the optical to the near-infrared wavelengths: combining STIS, WFC3 and IRAC datasets

We present the transmission spectrum of the inflated hot-Jupiter WASP-17 b, observed with the STIS (grisms G430L, G750L) and WFC3 (grisms G102, G141) instruments aboard the Hubble Space Telescope, allowing for a continuous wavelength coverage from $\sim$0.4 to $\sim$1.7 $\mu$m. Available observations taken with IRAC channel 1 and 2 on the Spitzer Space Telescope are also included, adding photometric measurements at 3.6 and 4.5 $\mu$m. HST spectral data was analysed with the open-source pipeline Iraclis, which is specialised on the reduction of STIS and WFC3 transit and eclipse observations. Spitzer photometric observations were reduced with the TLCD-LSTM (Transit Light Curve Detrending LSTM) method, which employs recurrent neural networks to predict the correlated noise and detrend Spitzer transit lightcurves. The outcome of our reduction produces incompatible results between STIS visit 1 and visit 2, which leads us to consider two scenarios for G430L. Additionally, by modelling the WFC3 data alone, we can extract atmospheric information without having to deal with the contrasting STIS datasets. We run separate retrievals on the three spectral scenarios with the aid of TauREx3, a fully Bayesian retrieval framework. We find that, independently of the data considered, the exoplanet atmosphere displays strong water signatures, aluminium oxide (AlO) and titanium hydride (TiH). A retrieval that includes an extreme photospheric activity of the host star is the preferred model, but we recognise that such scenario is unlikely for an F6-type star. Due to the incompleteness of all STIS spectral lightcurves, only further observations with this instrument would allow us to properly constrain the atmospheric limb of WASP-17 b, before JWST or Ariel will come online.