The JWST Emission Line Survey (JELS): The sizes and merger fraction of star-forming galaxies during the Epoch of Reionization

TL;DR

This study uses JWST/JELS data to quantify rest-frame optical sizes of 23 H$\alpha$-emitting galaxies at $z=6.1$, comparing them across rest-frame UV, R-band, and H$\alpha$-traced star formation. A 2D Sérsic fit with GALFIT reveals a measurable $r_{e}-M_{*}$ relation with a shallow slope around $\alpha\approx0.14$ and an offset to smaller sizes relative to lower redshifts, with pronounced scatter at $M_{*}<10^{8.4}\,M_{\odot}$ likely driven by bursty SFHs. The ionised gas and stellar components are of comparable size at this epoch, with the stellar component often marginally larger, implying pre-existing stellar mass and inside-out growth. Close-pair analyses yield merger fractions around $f_{\text{merger}}\approx0.43$ and $f_{\text{maj. merger}}\approx0.44$ within $d\lesssim25$ kpc, consistent with mergers playing a significant role in EoR galaxy assembly. Overall, the results support a picture of rapid early growth, with established stellar populations by the end of the EoR and star formation concentrated toward galaxy centers, influenced by bursty SFHs and mergers.

Abstract

We used observations from the JWST Emission Line Survey (JELS) to measure the half-light radii ($r_{e}$) of 23 H$α$-emitting star-forming (SF) galaxies at $z=6.1$ in the PRIMER/COSMOS field. Galaxy sizes were measured in JWST Near-infrared Camera observations in rest-frame H$α$ (tracing recent star formation) with the F466N and F470N narrowband filters from JELS, and compared against rest-$R$-band, $V$-band (tracing established stellar populations) and near-ultraviolet sizes. We find a size-stellar mass ($r_{e}-M_{*}$) relationship with a slope that is consistent with literature values at lower redshifts, though offset to lower sizes. We observe a large scatter in $r_{e}$ at low stellar mass ($M_{*}<10^{8.4}$ M$_{\odot}$) which we believe is the result of bursty star formation histories (SFHs) of SF galaxies at the Epoch of Reionization (EoR). We find that the stellar and ionised gas components are similar in size at $z=6.1$. The evidence of already-established stellar components in these H$α$ emitters (HAEs) indicates previous episodes of star formation have occurred. As such, following other JELS studies finding our HAEs are undergoing a current burst of star formation, we believe our results indicate that SF galaxies at the end of the EoR have already experienced a bursty SFH. From our $r_{e}-M_{*}$ relationship, we find $r_{e, \text{F444W}}=0.76\pm0.46$ kpc for fixed stellar mass $M_{*}=10^{9.25}$ M$_{\odot}$, which is in agreement with other observations and simulations of star forming galaxies in the literature. We find a close-pair (major) merger fraction of ($f_{\text{maj. merger}}=0.44\pm0.22$) $f_{\text{merger}}=0.43\pm0.11$ for galaxy separations $d\lesssim25$ kpc, which is in agreement with other $z\approx6$ studies.

Figures (11)

• Figure 1: Example models for six HAEs in our sample. The left panels are the rest-frame $R$-band observations, with the NIRCam filter used for the observation indicated in the title of the panel. The middle panels are the fixed $n=1$ Sérsic models of the selected object from GALFIT. The right panels are the residual emission once the modelled galaxy is removed from the observed image. Each panel is a $3\times3$ arcsec$^{2}$ ($\approx 17\times17$ □k) cutout centred on the detected galaxy.
• Figure 2: The ratio of median extracted $r_{e, \text{F444W}}$ to model values of mock galaxies (left axis) as a function of model F444W magnitude. For each magnitude, recovered $r_{e, \text{median}}$ are determined for mock galaxies with model radii of $\approx 0.4$ k (blue), $\approx 0.8$ k (orange), $\approx 1.1$ k (green), $\approx 1.4$ k (red) and $\approx 1.8$ k (purple) at $z=6.1$. The grey dashed line represents $r_{e, \text{median}}/r_{e, \text{expected}} = 1$, with the grey dotted lines representing $\pm\,0.2$. We represent the F444W magnitude range our actual sample of HAEs with a grey histogram (right axis). The vertical black lines are the upper and lower bounds of the F444W magnitude range of our HAEs, with the shaded brown regions indicating regions outside of that range.
• Figure 3: The $r_{e}-M_{*}$ relationship for our HAEs at $z=6.1$ in four different JWST bands: PRIMER F277W (rest-NUV; Upper Left), PRIMER F356W (rest-$V$-band; Upper Right) and PRIMER F444W (rest-$R$-band; Lower Left), and JELS F466N/F470N NB (rest-H$\alpha$; Lower Right). The grey symbols represent the individual sizes of each HAE. In all panels, the red dashed line shows the fitted relationship to the individual points, with the red shaded region indicating the 1$\sigma$ scatter. We compare to the observed $r_{e}-M_{*}$ relationship of HAEs from Stott2013a at $z=0.4$ (green dash-dotted line), as well as those from vanderWel2014a at $z=0.25$ (loosely dotted grey line) and $z=2.75$ (densely dotted grey line). We also compare to the simulated $r_{e}-M_{*}$ relationship from McClymont2025a (blue dash-dotted) who employ the thesan-zoom radiation-hydrodynamics zoom-in simulations Kannan2025. McClymont2025a measure the $r_{e}-M_{*}$ in three different bands: UV (rest-frame $0.1475 - 1525$ µm), optical (rest-frame $0.5 - 0.6$ µm) and H$\alpha$. We select the appropriate relationship for comparison in each of our observed bands. The black long-dashed line shows the fixed slope of the vanderWel2014a$r_{e}-M_{*}$ relationship at $z=2.75$ with the offset fitted to our data points.
• Figure 4: The ratio of the measured $r_{e}$ in F444W (rest-$R$-band; $r_{e, \text{F444W}}$) to the measured $r_{e}$ in NB ($r_{e, \text{NB}}$) for each of our HAEs (green triangles) against stellar mass. The error on each size ratio represents the combined error on the respective $r_{e}$ measurements. The solid black line represents $r_{e, \text{F444W}}/r_{e, \text{NB}} = 1$. The dashed grey line represents the median $r_{e, \text{F444W}}/r_{e, \text{NB}} = 1.20$, with the shaded region indicating the standard error ($\pm\,0.09$).
• Figure 5: As in Figure \ref{['fig::bb_to_nb_size_ratio']}, but for our $r_{e, \text{F444W}}$ to $r_{e}$ measured in F277W (rest-NUV; $r_{e, \text{F277W}}$) ratios (thin red diamonds). The dashed grey line and shaded region represent the median $r_{e, \text{F444W}}/r_{e, \text{F277W}} = 1.14 \pm 0.07$.