2022A&A...659A.116C

Context. Identifying very high-redshift galaxies is crucial for understanding the formation and evolution of galaxies. However, many questions still remain, and the uncertainty on the epoch of reionization is large. In this approach, some models allow a double-reionization scenario, although the number of confirmed detections at very high z is still too low to serve as observational proof. Aims. The main goal of this project is studying whether we can search for Lyman-α emitters (LAEs) at z ∼ 9 using a narrow-band (NB) filter that was specifically designed by our team and was built for this experiment. Methods. We used the NB technique to select candidates by measuring the flux excess due to the Lyα emission. The observations were taken with an NB filter (full width at half minimum of 11 nm and central wavelength λ_c = 1.257 µm) and the CIRCE near-infrared camera for the Gran Telescopio Canarias. We describe a data reduction procedure that was especially optimized to minimize instrumental effects. With a total exposure time of 18.3 h, the final NB image covers an area of ∼6.7 arcmin², which corresponds to a comoving volume of 1.1 x 10³ Mpc³ at z = 9.3. Results. We pushed the source detection to its limit, which allows us to analyze an initial sample of 97 objects. We detail the different criteria we applied to select the candidates. The criteria included visual verifications in different photometric bands. None of the objects resembled a reliable LAE, however, and we found no robust candidate down to an emission-line flux of 2.9 x 10^–16 erg s^–1 cm^–2, which corresponds to a Lyα luminosity limit of 3 x 10⁴⁴ erg s^–1. We derive an upper limit on the Lyα luminosity function at z ∼ 9 that agrees well with previous constraints. We conclude that deeper and wider surveys are needed to study the LAE population at the cosmic dawn.