SILVERRUSH. IX. Lya Intensity Mapping with Star-Forming Galaxies at z=5.7 and 6.6: A Possible Detection of Extended Lya Emission at $\gtrsim$100 comoving kpc around and beyond the Virial-Radius Scale of Galaxy Dark Matter Halos

Abstract

We present results of the cross-correlation Ly$α$ intensity mapping with Subaru/Hyper Suprime-Cam (HSC) ultra-deep narrowband images and Ly$α$ emitters (LAEs) at $z=5.7$ and $6.6$ in a total area of $4$ deg$^2$. Although overwhelming amount of data quality controls have been performed for the narrowband images, we further conduct extensive analysis evaluating systematics of large-scale point-spread-function wings, sky subtractions, and unknown errors based on physically uncorrelated signals and sources found in real HSC images and object catalogs, respectively. Removing the systematics, we carefully calculate cross-correlations between Ly$α$ intensity of the narrowband images and the LAEs. We tentatively identify very diffuse Ly$α$ emission with the $\simeq 3σ$ ($\simeq 2σ$) significance at $\gtrsim$ 100 comoving kpc (ckpc) far from the LAEs at $z=5.7$ ($6.6$), around and probably even beyond a virial radius of star-forming galaxies with $M_\mathrm{h}\sim10^{11}M_\odot$. The diffuse Ly$α$ emission possibly extends up to $1$,$000$ ckpc with the surface brightness of $10^{-20}$-$10^{-19}$ erg s$^{-1}$ cm$^{-2}$ arcsec$^{-2}$. We confirm that the small-scale ($<150$ ckpc) Ly$α$ radial profiles of LAEs are consistent with those obtained by recent MUSE observations. Comparisons with numerical simulations suggest that the large-scale ($\sim150$-$1$,$000$ ckpc) Ly$α$ emission are not explained by unresolved faint neighboring galaxies including satellites, but by a combination of Ly$α$ photons emitted from the central LAE and other unknown sources, such as cold-gas streams and galactic outflow. We find no evolution in the Ly$α$ radial profiles of our LAEs from $z=5.7$ to $6.6$, where theoretical models predict a flattening of the profile slope made by cosmic reionization, albeit with our moderately large observational errors.