REBELS-IFU: Linking damped Lyman-$α$ absorption to [CII] emission and dust content in the EoR

TL;DR

REBELS-IFU utilizes JWST/NIRSpec prism spectra to measure HI via Ly$\alpha$ damping wings in $z\sim6.5$–$8$ galaxies, linking the neutral gas content to [C II] emission and dust. By joint modelling of the damping wings with an IGM component ($x_{\mathrm{HI}}\approx0.33$) and DLAs co-located with the galaxy, the study derives $N_{\mathrm{HI}}$ for eight sources and compares HI masses inferred from damping wings to those from [C II] luminosities, finding the latter to be systematically larger and suggesting more extended HI reservoirs. The authors also compare DTG tracers from FIR data with line-of-sight attenuation $A_V/N_{\mathrm{HI}}$, discovering a strong $A_V/N_{\mathrm{HI}}$–metallicity relation when including lower-metallicity literature, consistent with early dust-enrichment scenarios. The results imply that massive $z\sim7$ galaxies already host substantial, enriched HI and dust reservoirs and demonstrate the potential of Ly$\alpha$ damping wings as a direct probe of the neutral ISM/CGM in the EoR, while highlighting the need for higher-S/N, higher-resolution data and larger samples to break degeneracies and test DTG evolution with redshift.

Abstract

Neutral gas in galaxies during the Epoch of Reionisation regulates star formation, dust growth, and the escape of ionising photons, making it a key ingredient in understanding both galaxy assembly and reionisation. Yet, direct constraints on the HI content of galaxies at z>6 have been scarce. With JWST, Ly$α$ damping wings in galaxy spectra can now provide a direct probe of this neutral component. We analyse JWST/NIRSpec prism spectra of 12 UV-luminous galaxies from the REBELS-IFU program at z~6.5-7.7, deriving HI column densities by modelling Ly$α$ damping wings. Significant damped Ly$α$ absorption is detected in eight galaxies, with $N_{\mathrm{HI}}\gtrsim10^{21}$ cm$^{-2}$. We use the column densities and sizes derived for these sources to estimate their HI mass and compare with $L_{\mathrm{[CII]}}$-$M_{\mathrm{HI}}$ calibrations. The resulting HI masses show a tentative correlation with those inferred from [CII], although the [CII]-based estimates are systematically larger, suggesting that the HI reservoirs may extend beyond the [CII]-emitting gas. We also combine the DLA-based measurements with FIR-derived dust-to-gas ratios, dust attenuation, and gas-phase metallicities. No correlation is found between DLA-based and FIR-based dust-to-gas ratios, but combining the REBELS-IFU sample with literature samples at lower metallicities reveals a strong correlation between $A_{\mathrm{V}}/N_{\mathrm{HI}}$ and metallicity. These findings suggest that by $z\sim7$ massive galaxies can already host substantial, enriched reservoirs of neutral gas and dust, consistent with $A_{\mathrm{V}}$/$N_{\mathrm{HI}}$-metallicity trends at lower redshift. At the highest redshifts ($z>8$), however, we see tentative evidence for systematically lower $A_{\mathrm{V}}$/$N_{\mathrm{HI}}$ at fixed metallicity, which may point to pristine gas accretion or more efficient dust destruction/expulsion.

Figures (9)

• Figure 1: Spectral fits for the eight galaxies where including a damped Ly$\alpha$ (DLA) absorption component improves the fit to the observed UV continuum downturn, as quantified by both the reduced $\chi^2$ and the Bayesian Information Criterion (BIC). The observed data are shown in black, with the flux uncertainties shaded in grey. The best-fit model including both IGM and DLA absorption is plotted in red, while the blue curve shows a model with only IGM absorption.
• Figure 2: Comparison between the Hi gas mass inferred from damped Ly$\alpha$ absorption wings ($M_{\mathrm{H\textsc{i},~DLA}}$) and from [C ii] luminosities ($M_{\mathrm{H\textsc{i},~[CII]}}$) for the REBELS-IFU sample. Each REBELS galaxy is shown with an individual marker. For REBELS-14, 15, 32, and 39, we use the upper limits derived for the Hi column densities. We also plot A1689-zD1 at $z=7.13$ from Heintz et al. (in prep) (yellow star) and GS-z14-0 at $z=14.18$ from heintz_dissecting_2025-1 and schouws_deep_2025 (yellow square). For GS-z14-0, we assume $r_{\mathrm{e, [C \textsc{ii}]}}=2\times r_{\mathrm{e, UV}}$, and for A1689-zD1 we determine $r_{\mathrm{e, [C \textsc{ii}]}}$ assuming an exponential profile from the disc diameter reported in Heintz et al. (in prep). The black dashed line shows the one-to-one relation, which assumes that the H i gas reservoir and the [C ii] emission have the same radial extent. The best-fit relation is shown with the solid blue line, with the shaded region indicating the $1\sigma$ uncertainty on the fit. The dotted line shows the average offset in log space, corresponding to a geometric mean mass ratio of $M_{\mathrm{H\textsc{i},~[CII]}} \simeq 22 \times M_{\mathrm{H\textsc{i},~DLA}}$. This implies a typical radius ratio of $r_{\mathrm{H\textsc{i}}} \simeq 4.3-4.7 \times r_{\mathrm{[C \textsc{ii}]}}$, depending on whether the ratio is averaged in log space (4.7) or linear space (4.3). Markers outlined in bold use the [C ii] radii derived from Sérsic fitting to high resolution (beam FWHM $\sim 0.7-3$ kpc) data. For the remaining REBELS sources, the low resolution ($\sim7$ kpc) data is used.
• Figure 3: Schematic illustrating the assumptions and methodology used in this work. We stress that this schematic should be regarded only as a rough illustration under the very simplified assumptions adopted. Left panel: A simplified model galaxy, where we assume spherical geometry with radii set by the UV emission, [C ii] emission, and the Hi reservoir radii derived in this work (see text). Ionising photons from young massive stars (yellow) propagate outward, becoming attenuated by dust (orange circles) and, along some sight-lines, absorbed by neutral Hi clouds in the ISM/CGM (dashed red line). Other sightlines pass through without strong absorption (solid red line). All photons are then additionally absorbed by the IGM. Top right panel: Example SED from BAGPIPES for a massive ($\log(M_*/\mathrm{M_{\odot}})=9$), dusty ($A_{\mathrm{V}}=0.9$ mag), star-forming ($\mathrm{SFR}_{10}=200 \mathrm{M_{\odot}~yr}^{-1}$) galaxy at $z=7$. Shown are the intrinsic stellar continuum (yellow), the attenuated stellar continuum (red), and dust emission (orange). The blue shaded region marks the NIRSpec wavelength range used to measure stellar $A_V$, while the red shaded band indicates the ALMA Band 6 FIR continuum at $\sim160\mu$m used to estimate the dust mass for the REBELS-IFU sample. Bottom right panel: Example Ly$\alpha$ transmission curves, comparing the case of IGM-only absorption (solid red line) with IGM $+$ DLA absorption from neutral gas in the ISM/CGM (dashed red). Following our modelling, we assume a mean IGM neutral fraction of $x_{\mathrm{HI}}=0.33$ and convolve the spectra to $\mathcal{R}=60$.
• Figure 4: Left: Comparison between the dust-to-gas (DTG) mass ratios derived from ALMA observations (using $L_{\mathrm{[C \textsc{ii}]}}$ to estimate the Hi mass and the FIR continuum to infer dust mass) and from DLA-based measurements ($A_{\mathrm{V}} / N_{\mathrm{H\textsc{i}}}$) for the REBELS-IFU galaxies and for A1689-zD1 (Heintz et al. in prep). In both panels, the same marker shapes are used as in the legend of Figure \ref{['fig:gas mass comparison']}. REBELS-14, 15, 32 and 39 have only upper limits on $N_{\mathrm{H\textsc{i}}}$, and REBELS-34 has an upper limit on the dust mass since it is undetected in Band 6 continuum from the REBELS ALMA LP. No clear trend is observed between these two DTG estimates for the REBELS-IFU sample. We also plot for reference the average values from different sight-lines in the Milky Way (MW), Large Magellanic Cloud (LMC) and Small Magellanic Cloud (SMC) from konstantopoulou_dust_2024 and references therein, where we adopt their $A_{\mathrm{V,ext}}$ values derived from SED fitting and convert their total gas masses to Hi masses using the conversions given in the legend taken from Table 6 of israel_h_2_1997. Right: Relation between $\log(A_{\mathrm{V}} / N_{\mathrm{H\textsc{i}}})$ and gas-phase metallicity ($12 + \log(\mathrm{O}/\mathrm{H})$) for the REBELS-IFU sample and a compilation of eleven additional $z\gtrsim6$ galaxies from the literature (circle markers), with markers coloured by redshift. The best-fit linear relation is shown via the black solid line, with a shaded region representing the $1\sigma$ uncertainty. A strong correlation is found (Pearson $r = 0.8$, $p = 7\times10^{-5}$), with a consistent slope to the trend reported in heintz_cosmic_2023 based on GRB sightlines at $z=1.7-6.3$ (black dashed line) and the slope from a linear fit to the data presented in konstantopoulou_dust_2024 for the Milky Way, LMC and SMC.
• Figure 5: A comparison of the $M_{\mathrm{UV}}$ and $\beta_{\mathrm{UV}}$ values derived from the fitting described in Section \ref{['sec:modelling DLA']} (where the UV continua are simultaneously modelled with the Ly$\alpha$ absorption) with those derived in fisher_rebels-ifu_2025.