2013MNRAS.436.1215M

At optical wavelengths, an exoplanet's signature is essentially reflected light from the host star - several orders of magnitude fainter. Since it is superimposed on the star spectrum its detection has been a difficult observational challenge. However, the development of a new generation of instruments like Echelle Spectrograph for Rocky Exoplanets and Stable Spectroscopic Observations (ESPRESSO) and next-generation telescopes like the European Extremely Large Telescope (E-ELT) put us in a privileged position to detect these planets' reflected light as we will have access to extremely high signal-to-noise ratio spectra. With this work, we propose an alternative approach for the direct detection of the reflected light of an exoplanet. We simulated observations with ESPRESSO at Very Large Telescope (VLT) and high-resolution spectrograph (HIRES) at E-ELT of several star+planet systems, encompassing 10h of the most favourable orbital phases. To the simulated spectra we applied the cross-correlation function to operate in a much higher signal-to-noise ratio domain than when compared with the spectra. The use of the cross-correlation function permitted us to recover the simulated planet signals at a level above 3σnoise significance on several prototypical (e.g. Neptune-type planet with a 2d orbit with the VLT at 4.4σnoise significance) and real planetary systems (e.g. 55 Cnce with the E-ELT at 4.9σnoise significance). Even by using a more pessimistic approach to the noise level estimation, where systematics in the spectra increase the noise 2-3 times, the detection of the reflected light from large close-orbit planets is possible. We have also shown that this kind of study is currently within reach of current instruments and telescopes (e.g. 51 Pegb with the VLT at 5.2σnoise significance), although at the limit of their capabilities.