Closing in on the sources of cosmic reionization: first results from the GLASS-JWST program

TL;DR

The study uses JWST GLASS-JWST NIRSpec/NIRCam observations of 29 gravitationally lensed galaxies at $4.5 \,\le z \le 8$ to derive physical and spectroscopic properties (e.g., $M_*$, SFR, $r_e$, UV slope $eta$, and key optical emission lines). It shows that these low-mass, compact galaxies have high $O32$ ratios and blue continua, resembling low-redshift LyC leakers, and applies an empirical relation built from $O32$, $r_e$, and $eta$ to predict $f_{esc}$, finding an average $\,\log_{10}(f_{esc}) \approx -0.92$ corresponding to $f_{esc} \approx 0.12$ with many sources above 0.05. The results imply that similar galaxies at $z \ge 6$ could have substantially contributed to cosmic reionization, supporting a scenario where faint, compact systems drive the ionizing photon budget. The work demonstrates the power of JWST to constrain indirect LyC diagnostics at the EoR and provides a practical predictive framework for estimating $f_{esc}$ in future large samples.

Abstract

The escape fraction of Lyman-continuum (LyC) photons ($f_{esc}$) is a key parameter for determining the sources of cosmic reionization at $z\geq 6$. At these redshifts, owing to the opacity of the intergalactic medium, the LyC emission cannot be measured directly. However, LyC leakers during the epoch of reionization could be identified using indirect indicators that have been extensively tested at low and intermediate redshifts. These include a high [OIII]/[OII] flux ratio, high star-formation surface density, and compact sizes. In this work, we present observations of 29 $4.5 \leq z \leq 8$ gravitationally lensed galaxies in the Abell 2744 cluster field. From a combined analysis of JWST-NIRSpec and NIRCam data, we accurately derived their physical and spectroscopic properties: our galaxies have low masses $(\log(M_\star)\sim 8.5)$, blue UV spectral slopes ($β\sim -2.1$), compact sizes ($r_e \sim 0.3-0.5$ kpc), and high [OIII]/[OII] flux ratios. We confirm that these properties are similar to those characterizing low-redshift LyC leakers. Indirectly inferring the fraction of escaping ionizing photons, we find that more than 80% of our galaxies have predicted $f_{esc}$ values larger than 0.05, indicating that they would be considered leakers. The average predicted $f_{esc}$ value of our sample is 0.12, suggesting that similar galaxies at $z\geq 6$ have provided a substantial contribution to cosmic reionization.

Figures (8)

• Figure 1: Spatial location of the 29 selected sources, color coded by their spectroscopic redshift. They are superimposed on the RGB image of the UNCOVER program, made with NIRCam filters (blue = F115W + F150W, green = F200W + F277W, and red = F356W + F410M + F444W). The MUSE footprint is shown in white, the NIRSpec GLASS-JWST pointing is shown in cyan, and the NIRSpec DDT pointing is shown in red.
• Figure 2: Example 2D and 1D spectra of two representative galaxies in our sample. Top: 2D (top) and 1D (bottom) NIRSpec GLASS-JWST spectrum of 110000 at $z = 5.765$ (green for the G235H/F170LP and purple for the G395H/F290LP configuration) and 1$\sigma$ uncertainty (gray). Bottom: 2D (top) and 1D (bottom) NIRSpec DDT spectrum of 150015 at $z = 5.041$ (black) and 1$\sigma$ uncertainty (gray). Vertical dashed lines mark the position of well-detected rest-optical emission lines. The continuum emission is also detected as seen in the 2D spectrum. On the X-axis in the bottom panel, the observed wavelength (Å) is reported.
• Figure 3: MEx diagram for the sample of galaxies analyzed in this paper:black dots and green squares are for the GLASS-JWST and DDT samples, respectively. For reference, we plot also the galaxies at $z = 0.3 - 0.4$ from Flury2022 (diamonds) and the LCE candidates from previous studies Izotov2016aIzotov2016bIzotov2018aIzotov2018bIzotov2021Wang2019. The two orange demarcation lines from Juneau2014 show the boundaries of the AGN-and-star-forming transition region. All objects above the upper line are AGN-dominated; all galaxies below or rightward of the lower line are presumptively dominated by star formation. We also show the separation from Coil2015 (dotted lines), which is the adaptation of the Juneau et al. model for galaxies and AGNs at $z\sim2.3$ from the MOSDEF survey.
• Figure 4: $O32$ vs. rest-frame $EW_0(\rm H H$β$)$. Symbols are the same as in Figure \ref{['fig:AGN']}. For reference, we also plot the galaxies at $z = 0.3 - 0.4$ from Flury2022 (diamonds) and the LCE from Izotov2016aIzotov2016bIzotov2018aIzotov2018bIzotov2021Wang2019 (stars). Symbols are color coded as a function of their measured $f_{esc}$. The pink line ($O32 = 5$) indicates the threshold for LCEs as predicted by Flury2022.
• Figure 5: $O32$ vs. $M_\star$. Symbols are as in Fig. \ref{['fig:O32_Hbeta']}.