Table of Contents
Fetching ...

A JWST/NIRCam Study of Key Contributors to Reionization: The Star-forming and Ionizing Properties of UV-faint $z\sim7-8$ Galaxies

Ryan Endsley, Daniel P. Stark, Lily Whitler, Michael W. Topping, Zuyi Chen, Adele Plat, John Chisholm, Stéphane Charlot

TL;DR

JWST/NIRCam imaging of UV-faint galaxies at z~6.5–8 reveals a broad range of star-formation histories, high sSFRs, and large ionizing photon production efficiencies, suggesting faint galaxies contribute prominently to reionization. Using BEAGLE photoionization SED modelling on 0.4–5 μm data for 116 galaxies in CEERS, the authors derive predominantly low stellar masses, with a minority showing Balmer breaks or extreme nebular line emission. They identify a population of young, weak-OIII emitters as well as a dusty tail of red galaxies, implying diverse dust and metallicity conditions. The results indicate UV-faint galaxies may be more efficient ionizers than previously thought, but further spectroscopy and larger-area surveys are required to refine mass estimates, escape fractions, and the AGN contribution.

Abstract

Spitzer/IRAC imaging has revealed that the brightest $z\sim7-8$ galaxies often exhibit young ages and strong nebular line emission, hinting at high ionizing efficiency among early galaxies. However, IRAC's limited sensitivity has long hindered efforts to study the fainter, more numerous population often thought largely responsible for reionization. Here we use CEERS JWST/NIRCam data to characterize 116 UV-faint (median M$_{UV}=-19.5$) $z\sim6.5-8$ galaxies. The SEDs are typically dominated by young ($\sim$10-50 Myr), low-mass ($M_\ast\sim10^8\ M_\odot$) stellar populations, and we find no need for extremely high stellar masses ($\sim10^{11} M_\odot$). Considering previous studies of UV-bright (M$_{UV}\sim-22$) $z\sim7-8$ galaxies, we find evidence for a strong (5-10$\times$) increase in specific star formation rate toward lower luminosities (median sSFR=103 Gyr$^{-1}$ in CEERS). The larger sSFRs imply a more dominant contribution from OB stars in the relatively numerous UV-faint population, perhaps suggesting that these galaxies are very efficient ionizing agents (median $ξ_{ion}=10^{25.7}$ erg$^{-1}$ Hz). In spite of their much larger sSFRs, we find no significant increase in [OIII]$+$H$β$ EWs towards fainter M$_{UV}$ (median $\approx$780 $\mathring{A}$). If confirmed, this may indicate that a substantial fraction of our CEERS galaxies possess extremely low metallicities ($\lesssim$3% $Z_\odot$) where [OIII] emission is suppressed. Alternatively, high ionizing photon escape fractions or bursty star formation histories can also weaken the nebular lines in a subset of our CEERS galaxies. While the majority of our objects are very blue (median $β=-2.0$), we identify a significant tail of very dusty galaxies ($β\sim-1$) at $\approx$0.5$L_{UV}^\ast$ which may contribute significantly to the $z\sim7-8$ star formation rate density.

A JWST/NIRCam Study of Key Contributors to Reionization: The Star-forming and Ionizing Properties of UV-faint $z\sim7-8$ Galaxies

TL;DR

JWST/NIRCam imaging of UV-faint galaxies at z~6.5–8 reveals a broad range of star-formation histories, high sSFRs, and large ionizing photon production efficiencies, suggesting faint galaxies contribute prominently to reionization. Using BEAGLE photoionization SED modelling on 0.4–5 μm data for 116 galaxies in CEERS, the authors derive predominantly low stellar masses, with a minority showing Balmer breaks or extreme nebular line emission. They identify a population of young, weak-OIII emitters as well as a dusty tail of red galaxies, implying diverse dust and metallicity conditions. The results indicate UV-faint galaxies may be more efficient ionizers than previously thought, but further spectroscopy and larger-area surveys are required to refine mass estimates, escape fractions, and the AGN contribution.

Abstract

Spitzer/IRAC imaging has revealed that the brightest galaxies often exhibit young ages and strong nebular line emission, hinting at high ionizing efficiency among early galaxies. However, IRAC's limited sensitivity has long hindered efforts to study the fainter, more numerous population often thought largely responsible for reionization. Here we use CEERS JWST/NIRCam data to characterize 116 UV-faint (median M) galaxies. The SEDs are typically dominated by young (10-50 Myr), low-mass () stellar populations, and we find no need for extremely high stellar masses (). Considering previous studies of UV-bright (M) galaxies, we find evidence for a strong (5-10) increase in specific star formation rate toward lower luminosities (median sSFR=103 Gyr in CEERS). The larger sSFRs imply a more dominant contribution from OB stars in the relatively numerous UV-faint population, perhaps suggesting that these galaxies are very efficient ionizing agents (median erg Hz). In spite of their much larger sSFRs, we find no significant increase in [OIII]H EWs towards fainter M (median 780 ). If confirmed, this may indicate that a substantial fraction of our CEERS galaxies possess extremely low metallicities (3% ) where [OIII] emission is suppressed. Alternatively, high ionizing photon escape fractions or bursty star formation histories can also weaken the nebular lines in a subset of our CEERS galaxies. While the majority of our objects are very blue (median ), we identify a significant tail of very dusty galaxies () at 0.5 which may contribute significantly to the star formation rate density.
Paper Structure (13 sections, 14 figures, 1 table)

This paper contains 13 sections, 14 figures, 1 table.

Figures (14)

  • Figure 1: An illustration of how our neighbor subtraction algorithm improves photometric measurements for objects with close neighbors. The $z\sim6.5-8$ galaxy of interest is marked with a red cross in the leftmost panels which show the F150W$+$F200W Source Extractor detection image. The $k=1.2$ apertures are shown in red for each band, and we also show the $k=2.5$ apertures in blue for the F200W and inverse variance-weighted LW stack image. After subtracting the surface brightness profiles of neighboring sources contaminating these apertures, we find that the ACS$+$SW fluxes decrease by $\approx$0.1 mag while the LW fluxes decrease by $\approx$0.3--0.4 mag given the broader PSF and larger aperture size. Each postage stamp is 1.6$\times$1.6 arcsec$^2$ and we show the measured AB magnitude for each band at the bottom, where we quote 2$\sigma$ limits for the ACS bands.
  • Figure 2: The distribution of photometric properties among our 116 Lyman-break $z\sim6.5-8$ galaxies selected with the CEERS/NIRCam data. (a) The distribution of apparent F200W magnitudes (median $m=27.5$). (b) The distribution of F150W$-$F200W colors (median = $-$0.1) which reflect the range of UV slopes among our sample.
  • Figure 3: Top: A long-wavelength color-color diagram of our $z\sim6.5-8$ CEERS galaxies. The F410M data helps break degeneracies between nebular line emission vs. e.g. dust and Balmer breaks in causing red colors. Dashed lines mark flat colors along each axis. Bottom: Illustration of how the strong optical nebular emission lines H$\alpha$ (red), H$\beta$ (green), [OIII]$\lambda$4959 (blue), and [OIII]$\lambda$5007 (orange) contaminate different NIRCam LW bands at different redshifts. The transparency of the shaded regions vary according to the filter transmission at the observed wavelength of each line.
  • Figure 4: Demonstration of how the arrival of JWST/NIRCam data dramatically improves our understanding of the rest-UV$+$optical SEDs and physical properties of typical ($<L_\mathrm{UV}^\ast$) galaxies in the reionization era. In the top panel, we show the previous SED constraints from HST+Spitzer data Stefanon2017_EGS (left) versus with new NIRCam photometry (right) for one of the $z\sim6.5-8$ galaxies in our CEERS sample. The photometric measurements are shown in blue (with 2$\sigma$ upper limits shown in cases of non-detections) while the fitted model photometry and SED from beagle CSFH fits are shown in red and black, respectively. The shaded gray regions and errorbars on the red markers reflect the inner 68% credible interval from the posterior probability distribution output by beagle. The middle and bottom panels demonstrate how the posteriors on various inferred physical properties tighten substantially with the NIRCam data, illustrating how uncertainties typically decrease from $\approx \pm$0.5--1 dex with HST$+$Spitzer to $\approx \pm$0.05--0.1 dex with ACS+NIRCam.
  • Figure 5: Distributions of inferred properties among our sample of 116 Lyman-break $z\sim6.5-8$ galaxies. The values for each galaxy are taken as the median of the posterior probability distribution output by beagle using our fiducial constant star formation history (CSFH) fits (see §\ref{['sec:beagle']}).
  • ...and 9 more figures