Probing the Cosmic Reionization History with JWST: Gunn-Peterson and Ly$α$ Damping Wing Absorption at $4.5 < z < 13$

TL;DR

Using ~600 JWST/NIRSpec galaxy spectra spanning 4.5 < z < 13, this work builds homogeneous UV-stack composites and applies Lyα damping-wing modeling with semi-numerical IGM sightlines to infer the neutral fraction history ⟨x_{HI}⟩. The analysis reveals a rapid, late reionization around z ≈ 7–8, with ⟨x_{HI}⟩ rising from near 0 at z ≈ 5–6 to near 1 by z ≈ 9–10, and shows UV-faint galaxies experience stronger damping-wing absorption at high z in support of topology where ionization proceeds preferentially around brighter halos. The methodology combines realistic galaxy templates, IGM attenuation, and 21cmFAST-based sightlines to mitigate local HI degeneracies and to compare GP truncation and damping-wing signals across redshift and UV properties. Overall, the results support a late and rapid reionization, potentially driven by hidden AGN populations or evolving LyC escape, with implications for the ionization budget and the spatial structure of ionized regions in the early universe.

Abstract

We present a statistical analysis of Ly$α$ absorption using 581 galaxies at $z=4.5$--13 observed with multiple JWST/NIRSpec spectroscopy programs, including JADES, UNCOVER, CEERS, and GO/DDT. We carefully construct composite spectra binned by redshift with homogeneous UV properties (UV magnitudes, UV slopes, and Ly$α$ equivalent widths) and identify significant Ly$α$ forest signals in galaxies at $z\sim5$--6, which diminish toward higher redshifts. We also find UV continuum breaks at rest-frame 1216 Å that soften beyond $z\gtrsim6$, confirming the effects of cosmic reionization through a self-consistent transition from Gunn-Peterson to Ly$α$ damping wing absorption in galaxies. Fair comparisons of composite spectra with matched UV magnitudes and slopes across redshift reveal that UV-faint galaxies clearly show stronger Ly$α$ absorption than UV-bright galaxies towards high redshift, providing insights into the topological evolution of reionization. We estimate Ly$α$ transmission at the Gunn-Peterson trough and Ly$α$ damping wing absorption by comparing the galaxy spectra to low-$z$ ($z\sim2$-5) galaxy templates that include galactic and circumgalactic absorption and Ly$α$ emission. Using these measurements together with reionization simulations, we derive volume average neutral hydrogen fractions of $\langle x_{\rm HI} \rangle$ = ${0.00}^{+0.12}_{-0.00}$, ${0.25}^{+0.10}_{-0.20}$, ${0.65}^{+0.27}_{-0.35}$, ${1.00}^{+0.00}_{-0.20}$, and ${1.00}^{+0.00}_{-0.40}$ at $z\sim5$, 6, 7, 9, and 10, respectively. These values broadly align with a reionization history characterized by a rapid transition around $z\sim7$--8, consistent with Ly$α$ emitter observations. While the physical driver of this rapid reionization remains unclear, it may involve the emergence of hidden AGN populations and/or the onset of Lyman-continuum escape from galaxies.

Figures (14)

• Figure 1: The distribution of UV magnitude by redshift. The UV magnitude is inferred from the spectral fitting. The black (blue) circles represent the measurements for the obtained galaxy spectra with $S/N>5$ at the rest-frame 1450 Å as well as the Ly$\alpha$ detection without (with) $EW_{0}>25$ Å. The faint circles represent the measurements for the galaxy spectra with $S/N<5$ at the rest-frame 1450 Å. Rectangular region surrounded by red dashed lines represents the parameter space used to select the "fiducial" subsamples.
• Figure 2: The distribution of UV slope by redshift. The UV slope is inferred from the spectral fitting. The symbols are the same as Figure \ref{['muv_z']}.
• Figure 3: The schematic representation of Ly$\alpha$ emission in NIRSpec/PRISM spectra at $z=6$. UV continuum assume the power-law with $\beta_{\rm UV}=-2.2$ with sharp drop at the rest-frame 1216 Å. We assume $\delta$-function Ly$\alpha$ emission with $EW_{\rm Ly\alpha}=1$, 5, 25, and 125 Å, all shown in different colors. All the spectra are convolved to the NIRSpec/PRISM line spread function.
• Figure 4: The stacked spectra of the fiducial galaxy sample from redshift 4.5 to 13. The stacked spectra are shown in the rest-frame wavelength. The redshift of the galaxy sample is shown in the upper right corner of each panel. The stack spectra shown here are constructed from the Fid1-5, with UV magnitude around $M_{\rm UV}\simeq-19.5$ and UV slope around $\beta_{\rm UV}\simeq-2.2$. The dark blue, navy blue, pale blue, pink, and red colored solid lines represent the stacked spectra at $\langle z \rangle$=5.0, 5.8, 7.0, 8.6, and 10.4. The corresponding 1$\sigma$ errors are shown in the shaded regions. All the stacked spectra are normalized at 1400 Å.
• Figure 5: The stacked spectra of galaxy subsample around the end of cosmic reionization. Panels from the top to bottom show the stacked spectra stacked spectra at $\langle z \rangle$=4.8, 5.2, 5.7, and 6.4 in navy blue, pale blue, pink, and red colored solid lines, respectively. The darker shades represent 1$\sigma$ uncertainty of the stack spectra.The dotted infill highlights the area associated with positive flux values and is for visual emphasis only. The faint navy blue line in all the panels except for the top one represent the stack spectra at $\langle z \rangle$=4.8 for the comparison. The stacked spectra are shown in the rest-frame wavelength. The redshift of the galaxy sample is shown in the upper right corner of each panel. The stack spectra shown here are constructed from the EEoR1-4, with UV magnitude around $M_{\rm UV}\simeq-19.5$ and UV slope around $\beta_{\rm UV}\simeq-2.2$. The corresponding 1$\sigma$ errors are shown in the shaded regions. All the stacked spectra are normalized at 1400 Å.