Ly$α$ visibility from z = 4.5 to 11 in the UDS field: Evidence for a high neutral hydrogen fraction and small ionized bubbles at z $\sim$ 7

Abstract

The resonant scattering nature of Ly$α$ photons interacting with neutral hydrogen makes Ly$α$ emitters (LAEs) robust tracers of the intergalactic neutral hydrogen fraction, and thus sensitive probes of cosmic reionization. We present an extensive study of the Ly$α$ evolution from galaxies at 4.5 $\leq$ z $\leq$ 11 in the UDS field, observed as part of the CAPERS survey, and complemented with spectra from the DAWN JWST Archive. The combined sample includes 651 spectroscopically confirmed Ly$α$-break galaxies, among which we find 73 S/N>3 LAEs in JWST-NIRSpec PRISM spectra. We trace the redshift evolution of the LAE fraction with EW$_0$ >25 A (X$_{\mathrm{Lyα}}$) between z = 5 and z = 9, extending such an analysis to the UDS field for the first time. At z = 5 and 6, the UDS results agree with the average JWST X$_{\mathrm{Lyα}}$ values from multiple fields. However, JWST measurements are consistently lower than ground-based results. To investigate this, we compare JWST observations to a population of star-forming galaxies at z$\sim$6 observed with VLT-FORS2. We find that a Ly$α$ slit-loss of 35 $\pm$ 10% in JWST spectra accounts for the offset, as the resonant Ly$α$ emission is more spatially extended than the stellar continuum. From z = 6 to 7, the UDS field shows a significant drop in Ly$α$ visibility, from which we infer a neutral hydrogen fraction of X$_{\mathrm{HI}}$ = 0.7--0.9. Finally, we identify two robust ionized bubbles at z = 7.29 and 7.77, with radii of $R_{\mathrm{ion}}$ = 0.6 and 0.5 physical Mpc and photometric overdensities of N/$\langle$N$\rangle$ = 3 and 4, based on candidate counts down to the photometric completeness limit. Compared to the large ionized region at z$\sim$7 in the EGS field, these results indicate significant field-to-field variation, supporting a patchy, inhomogeneous reionization process.

Figures (11)

• Figure 1: Left: Absolute UV magnitude as a function of redshift in the UDS sample. The 73 galaxies with robust S/N > 3 α Ly$\alpha$ detection are shown in blue, while the rest of the population is shown in red. Right: Distribution of α Ly$\alpha$ EW$_0$ as a function of M$_{\mathrm{UV}}$. Galaxies in the UDS with S/N > 3 are shown as circles, while upper limits EW$_{\mathrm{0,lim}}$ are shown as triangles. They are color-coded by spectroscopic redshift. For comparison, we report CEERS-EGS emitters (gray squares) and upper limits (gray triangles) from Napolitano2024. The dashed black lines at M$_{\mathrm{UV}}$ = –20.25 and –18.75 define the bright and faint regimes. The UDS sample is complete down to M$_{\mathrm{UV}}$ = -18.25, indicated by the solid red line.
• Figure 2: Redshift evolution of the fraction of galaxies with observed EW$_0$ > 25 Å. Results from this work are shown as orange circles. No correction for α Ly$\alpha$ slit losses was applied. Error bars were computed using binomial statistics following Gehrels1986. Colored points represent JWST estimates Nakane2024Napolitano2024Tang2024BJones2025Kageura2025, while open symbols with black edges correspond to ground-based literature results Stark2011Tilvi2014DeBarros2017Pentericci_2018bMason2019Fuller2020Kusakabe2020Tang2024. Data points have been slightly shifted in redshift for clarity.
• Figure 3: Left: UV $\beta$ slope as a function of M$_{\mathrm{UV}}$ for galaxies in the redshift range 5 < z < 6.5 (circles), matching the selection in DeBarros2017 (stars). The continuous black lines at M$_{\mathrm{UV}}$ = -21 and –18.75 mark the range considered. Galaxies with α Ly$\alpha$ EW$_0$ > 25 Å are shown in blue, while the rest of the population is shown in red. The best-fit relation and 1$\sigma$ uncertainty at z $\sim$ 6 from Topping2024c and Dottorini2025 are overplotted in black and green, respectively. Right: Cumulative distribution functions of α Ly$\alpha$ EW$_0$ at z $\sim$ 6. The DB17 and JWST α Ly$\alpha$-break samples are shown as stars and circles, respectively. Error bars were computed following Gehrels1986. The dashed black line shows the best-fit exponentially declining function to the DB17 data, corresponding to a case with no α Ly$\alpha$ slit loss (T = 1). The dashed blue line and shaded blue region represent the JWST extracted CDF assuming T = 0.65 $\pm$ 0.10. The dashed red area indicates the EW$_0$ range where a direct comparison is not possible due to the JWST completeness limit.
• Figure 4: Cumulative distribution functions of α Ly$\alpha$ EW$_0$ as a function of the neutral hydrogen fraction, X$_{\mathrm{HI}}$. Blue and orange circles represent the JWST samples at z = 7 from the EGS Napolitano2024 and the UDS (this work) fields, respectively. The arrow represents the lower limit we could derive for EW$_0$ = 25 Å due to incompleteness after α Ly$\alpha$ slit loss correction. The DB17 sample at z = 6 (gray diamonds) is shown as a reference for the fully ionized Universe. Error bars were computed following Gehrels1986. Data points have been slightly shifted in redshift for an easier visualization. The solid black line shows the best-fit exponentially declining function to the DB17 data, corresponding to a null X$_{\mathrm{HI}}$ value. Colored lines represent theoretical models derived from Dijkstra2011, which show the impact of an increasingly neutral IGM.
• Figure 5: Predicted size of the ionized bubble at z = 7.77 as a function of time since ionizing radiation is switched on. Colored solid lines show the contribution from individual sources, while the black solid line and shaded region represent the cumulative predicted radius and its associated uncertainty. The horizontal red dashed line marks the maximum physical distance between the central α Ly$\alpha$ emitting source and its furthest companion.