Searching for reflected light from τ Bootis b with high-resolution ground-based spectroscopy: Approaching the 10⁻⁵ contrast barrier

It is challenging to measure the starlight reflected from exoplanets because of the extreme contrast with their host stars. For hot Jupiters, this contrast is in the range of $10^{-6}$ to $10^{-4}$, depending on their albedo, radius and orbital distance. Searches for reflected light have been performed since the first hot Jupiters were discovered, but with very limited success because hot Jupiters tend to have low albedo values due to the general absence of reflective cloud decks. The aim of this study is to search for reflected light from $\tau$ Boo b, a hot Jupiter with one of the brightest host stars. Since its discovery in 1997, it has been the subject of several reflected-light searches using high-dispersion spectroscopy. Here we aim to combine these data in to a single meta-analysis. We analysed more than 2,000 archival high-dispersion spectra obtained with the UVES, ESPaDOnS, NARVAL UES and HARPS-N spectrographs during various epochs between 1998 and 2013. Each spectrum was first cleaned of the stellar spectrum and subsequently cross-correlated with a PHOENIX model spectrum. These were then Doppler shifted to the planet rest-frame and co-added in time, weighted according to the expected signal-to-noise of the planet signal. We reach a 3$\sigma$ upper limit of the planet to star contrast of $1.5 \times 10^{-5}$. Assuming a planet radius of 1.15 $R_J$, this corresponds to an optical albedo of 0.12 between 400-700 nm. This low albedo is in line with secondary eclipse and phase curve observations of other hot Jupiters using space-based observatories, as well as theoretical predictions of their reflective properties.