JWST PRIMER: A new multi-field determination of the evolving galaxy UV luminosity function at redshifts $\mathbf{z \simeq 9-15}$

TL;DR

The study delivers a robust, multi-field determination of the evolving galaxy UV LF from z≈8.5 to 15.5 by combining PRIMER, JADES, and NGDEEP JWST/NIRCam imaging over ~370 arcmin^2, achieving depths near 30 AB mag. It employs a statistically rigorous posterior redshift framework with a UV LF prior and 2548 high-z galaxy candidates, enabling a wide dynamic range (~4 mag) in UV luminosity and LF constraints up to z≈12.5, with a tentative extension to z≈14.5. The LF is well described by a double-power-law with strong density evolution and mild or no evolution in the faint and bright ends, yielding a smooth, near-linear decline of the UV luminosity density ρ$_{UV}$ with redshift and a corresponding ρ$_{SFR}$ consistent with ΛCDM without invoking additional dust evolution or star-formation efficiency changes up to z≈12. The authors further show that an age-dependent mapping between stellar mass and UV luminosity can reproduce the observed LF evolution, suggesting rapid emergence of galaxies around z≈12–13, and provide insights into the very early star-formation history as the halo mass function rapidly evolves beyond z>13.

Abstract

We present a new determination of the evolving galaxy UV luminosity function (LF) over the redshift range $8.5<z<15.5$ using a combination of several major Cycle-1 JWST imaging programmes - PRIMER, JADES and NGDEEP. This multi-field approach yields a total of $\simeq370$ sq. arcmin of JWST/NIRCam imaging, reaching (5-$σ$) depths of $\simeq30$ AB mag in the deepest regions. We select a sample of 2548 galaxies with a significant probability of lying at high redshift ($p(z>8.5)>0.05$) to undertake a statistical calculation of the UV LF. Our new measurements span $\simeq4$ magnitudes in UV luminosity at $z=9-12.5$, placing new constraints on both the shape and evolution of the LF at early times. Our measurements yield a new estimate of the early evolution of cosmic star-formation rate density ($ρ_{\rm{SFR}}$) confirming the gradual decline deduced from early JWST studies, at least out to $z \simeq 12$. Finally we show that the observed early evolution of the galaxy UV LF (and $ρ_{\rm{SFR}}$) can be reproduced in a ${\rm Λ}$CDM Universe, with no change in dust properties or star-formation efficiency required out to $z \simeq 12$. Instead, a progressive trend towards younger stellar population ages can reproduce the observations, and the typical ages required at $z \simeq$ 8, 9, 10, and 11 all converge on $\simeq 380-330$ Myr after the Big Bang, indicative of a rapid emergence of early galaxies at $z \simeq 12 - 13$. This is consistent with the first indications of a steeper drop-off in $ρ_{\rm{SFR}}$ we find beyond $z \simeq 13$, possibly reflecting the rapid evolution of the halo mass function at earlier times.

Figures (11)

• Figure 1: The $5\sigma$ depth maps in the F277W filter imaging of each NIRCam survey field used in this analysis, demonstrating the variation in depths between the different survey fields, and in some cases within a given field. All images are shown with bins of 200 pixels where the original images are on a 0.03-arcsec pixel scale. The colour-bar shows the $5\sigma$ depth in AB mag on the same scale for each field. The grayed out region shows where there is a lack of deep HST/ACS F814W imaging, which only affects the NGDEEP and UDS fields.
• Figure 2: A comparison of our evolving double power-law parameterisation of the $z = 11$ UV LF (green solid line) with the observational data from mcleod2023 at the same redshift. The prediction of the evolving Schechter function parameterisation of the UV LF from bouwens2021 is shown as the dashed blue line.
• Figure 3: The posterior redshift probability distribution for NGDEEP-3594 (red) and JADES-92420 (green) using a flat redshift prior (dashed-lines) and our UV LF prior (solid-lines). The first source is a $z\sim16$ candidate reported by austin2023 and leung2023, while the second is a spectroscopically-confirmed galaxy at $z=11.6$curtislake2022. The UV LF prior strongly weights the posterior redshift probability distribution to lower redshift for the less robust source, whereas the redshift solution of the robust high-redshift galaxy is unaffected. The top panel shows cut-out images of the two sources (indicated by their respective border colours) in the HST/ACS F606W, F814W filters and the JWST/NIRCam F115W, F150W, F200W, F277W, F356W and F444W filters.
• Figure 4: The distribution of apparent magnitude in the F277W filter for the galaxies in the final sample, split by the field in which they reside. The histogram is presented in bins with a width of 0.2 mag and the number of galaxies selected from each field is noted in the legend. The final total combined sample contains 2548 galaxies.
• Figure 5: Our new measurements of the rest-frame UV LF at $z=9, 10, 11$ and $z=12.5$ are shown as the black data points. For comparison we also show data points from mcleod2016oesch2018bowler2020bouwens2021bouwens2022harikane2023aadams2023leung2023donnan2023adonnan2023bmcleod2023perezgonzalez2023finkelstein2023casey2023bouwens2023bwillott2023robertson2023b with colours indicated in the figure legend. The best fitting double-power law functions are shown as the solid black lines at each respective redshift.