A 10 deg$^2$ Lyman-$\alpha$ survey at z=8.8 with spectroscopic follow-up: strong constraints on the LF and implications for other surveys

Candidate galaxies at redshifts of $z \sim 10$ are now being found in extremely deep surveys, probing very small areas. As a consequence, candidates are very faint, making spectroscopic confirmation practically impossible. In order to overcome such limitations, we have undertaken the CF-HiZELS survey, which is a large area, medium depth near infrared narrow-band survey targeted at $z=8.8$ Lyman-$\alpha$ (Ly$\alpha$) emitters (LAEs) and covering 10 deg$^2$ in part of the SSA22 field with the Canada-France-Hawaii Telescope. We surveyed a comoving volume of $4.7\times 10^6$ Mpc$^3$ to a Ly$\alpha$ luminosity limit of $6.3\times10^{43}$ erg s$^{-1}$. We look for Ly$\alpha$ candidates by applying the following criteria: i) clear emission line source, ii) no optical detections ($ugriz$ from CFHTLS), iii) no visible detection in the optical stack ($ugriz > 27$), iv) visually checked reliable NB$_J$ and $J$ detections and v) $J-K \leq 0$. We compute photometric redshifts and remove a significant amount of dusty lower redshift line-emitters at $z \sim 1.4 $ or $2.2$. A total of 13 Ly$\alpha$ candidates were found, of which two are marked as strong candidates, but the majority have very weak constraints on their SEDs. Using follow-up observations with SINFONI/VLT we are able to exclude the most robust candidates as Ly$\alpha$ emitters. We put a strong constraint on the Ly$\alpha$ luminosity function at $z \sim 9$ and make realistic predictions for ongoing and future surveys. Our results show that surveys for the highest redshift LAEs are susceptible of multiple contaminations and that spectroscopic follow-up is absolutely necessary.
Comment: 15 pages, 9 figures, accepted for publication in MNRAS