The THESAN project: Lyman-alpha emitters as probes of ionized bubble sizes

TL;DR

The paper employs the THESAN radiation-hydrodynamics simulations to predict intrinsic Lyα emission and IGM transmission for high-redshift galaxies and couples this with an empirically calibrated dust+outflow model to produce observed LAE catalogs. By fitting to observed Lyα luminosity functions at $z=5.5$ and $z=6.6$, the authors derive a six-parameter description of Lyα line profiles and escape fractions that reproduce LAE statistics across reionization. They find that before the midpoint of reionization, LAEs in larger ionized bubbles have higher Lyα transmission and equivalent widths, and that the LAE fraction correlates with bubble size; these correlations weaken as percolation proceeds and the IGM becomes more ionized. The work maps LAE selections to bubble-size statistics, enabling interpretation of JWST and narrow-band surveys, and provides public catalogs to study the evolving topology of reionization with realistic clustering, line-of-sight variability, and duty-cycle effects.

Abstract

We use the THESAN radiation-hydrodynamics simulations to investigate how Lyman-$α$ emitters (LAEs) trace ionized bubble sizes during the Epoch of Reionization. We generate realistic LAE catalogs by combining accurate intrinsic Ly$α$ production and intergalactic transmission with an empirical model for dust absorption and gas outflows. By calibrating to observationally-constrained Ly$α$ luminosity functions, we reproduce the rapid decline in Ly$α$ visibility toward higher redshifts while revealing mild tensions in LAE fractions near the end of reionization. Before the midpoint of reionization, galaxies within larger line-of-sight bubbles ($\gtrsim 10$ cMpc) have higher observed Ly$α$ luminosity and equivalent width (EW), demonstrating that the evolving LAE fraction provides a practical statistical tracer for bubble size. These correlations weaken as percolation progresses and the IGM becomes increasingly ionized. In LAE selected samples with $L_{\text{Ly}α} > 10^{41.5}\ \text{erg s}^{-1}$, Ly$α$ properties correlate with bubble size more strongly than UV magnitude, especially at $z \gtrsim 7$. This simulation-based framework maps LAE selections to bubble-size statistics, clarifies biases in more idealized models, and will supply public catalogs to interpret current and forthcoming JWST and narrow-band LAE surveys in terms of the evolving topology of reionization.

Figures (17)

• Figure 1: Ly$\alpha$ luminosity functions (red) using the best-fit model calibrated jointly for $z=5.5$ and $z=6.6$, shown with the black points. Note that the data points in the left panel are observed at $z=5.7$ and we compare to $z=5.5$ in thesan, as we expect this difference to be small. The 16$^\text{th}$ to 84$^\text{th}$ percentile ranges are shown as the red shaded regions. The intrinsic LFs and uncertainties are shown in the dotted curves and surrounding shaded regions. We find good agreement between our best-fit model and the observed data points for both redshifts, with some underestimation of the LFs for the $z=6.6$. The discrepancy between $10^{42.5}$ and $10^{43}\,\rm erg\,s^{-1}$ is likely due to the limitations of the model and simultaneously fitting both redshifts. The underestimation at the high-luminosity end is due to box size effects preventing creation of the very rare bright sources.
• Figure 2: Parameter sweeps for $z=5.5$ (upper panels) and $z=6.6$ (lower panels) around the best fit parameters listed in Table \ref{['tab:bestfit']}. Each panel shows the observed luminosity functions calculated by sweeping each parameter over the allowed range while holding all other parameters fixed at their best-fit values. Each parameter uniquely affects the shape and normalization of the resulting LFs, with $b_{\rm dust}$ having little effect around the best-fit values for the other parameters.
• Figure 3: Upper panels: Product of escape fraction ($f_{\rm esc}$) and IGM transmission ($\mathcal{T}_{\rm IGM}$) versus observed UV magnitude ($M_{1500}$). Lower panels: Ly$\alpha$ equivalent width (EW) versus $M_{1500}$. Left column: Medians across redshifts (colored curves), with shaded 16th–84th percentile ranges. For $f_{\rm esc} \times \mathcal{T}_{\rm IGM}$, dotted and dashed lines also show individual $f{\rm esc}$ and $\mathcal{T}_{\rm IGM}$ components. Center and right columns: 2D distributions at $z=6$ and $z=8$, with black lines marking medians. Brighter galaxies tend to show lower $f_{\rm esc} \times \mathcal{T}_{\rm IGM}$ and EWs, consistent with stronger dust attenuation and winds in more massive systems. All galaxies have observed Ly$\alpha$ luminosities above $10^{41.5}$ erg s$^{-1}$.
• Figure 4: Upper panel: Total transmission $f_{\rm esc} \times \mathcal{T}_{\rm IGM}$ from galaxies brighter than the observed UV magnitude ($M_{1500}$) threshold. Lower panel: Equivalent width of the Ly$\alpha$ line versus $M_{1500}$ threshold. The dashed gray line indicates the 25 Å equivalent width threshold for Ly$\alpha$ emitters (LAEs). In both panels, the thick curves show the medians for galactic lines of sight with observed Ly$\alpha$ luminosity above $10^{41.5}\,\rm erg\,s^{-1}$, with shaded regions indicating the 16$^\text{th}$ to 84$^\text{th}$ percentile ranges. The thin curves show the medians for all of the galaxies and lines of sight included in this study with only a cut on the intrinsic Ly$\alpha$ luminosity cut at $10^{41.5}\,\rm erg\,s^{-1}$. Brighter galaxies generally exhibit lower Ly$\alpha$ equivalent widths and escape fractions, reflecting the more significant dust attenuation and wind effects in more massive systems.
• Figure 5: Upper panel: Median transmission $f_{\rm esc} \times \mathcal{T}_{\rm IGM}$ (solid) versus redshift for galaxies brighter than the UV magnitude ($M_{1500}$) thresholds indicated by color. Shaded regions show the 16$^\text{th}$ to 84$^\text{th}$ percentiles, while dotted and dashed curves represent $f_{\rm esc}$ and $\mathcal{T}_{\rm IGM}$, respectively. Lower panel: Ly$\alpha$ equivalent width (EW) versus redshift for the same magnitude bins. The grey dash-dotted line marks the common LAE threshold at 25 Å. The pink points show EWs from observations of individual sources, while the grey triangles show upper limits from Kageura2025Jones2025Tang2024b, which are generally consistent with our models, with some potential bias toward stronger Ly$\alpha$ in observed LAEs at lower redshifts.