Beyond UV: Rest-frame B-band and Apparent Luminosity Functions of z=5-9 Galaxies

TL;DR

This study uses JWST/NIRCam data from JADES and public fields to measure rest-frame UV and B-band luminosity functions (LFs) of galaxies from z~4.5 to 9.7, and to derive apparent LFs in multiple NIRCam bands. The authors implement injection-recovery completeness simulations, flux-bias corrections, and the $1/V_{max}$ estimator to obtain robust LFs, including the first rest-frame B-band constraints at z~7–8 and extending z~5 B-band coverage to M_B ≈ -18. They compare the results with FIRE-2 and IllustrisTNG simulations, exploring the roles of binary stellar populations, stellar population synthesis codes, IMFs, extreme emission lines, and AGN contributions; they find no single model reproduces all trends and suggest that EELGs and possibly obscured AGNs contribute to bright-end excesses in reddest bands. A key finding is that rest-frame B-band correlates more tightly with stellar mass than UV, highlighting the diagnostic value of rest-frame optical LFs for understanding early mass assembly and galaxy evolution. Overall, the work provides new empirical benchmarks for modeling early galaxy formation and motivates future, wider JWST surveys to map rest-frame optical LFs across cosmic time.

Abstract

We present new measurements of galaxy luminosity functions (LFs) from JWST/NIRCam imaging over the redshift range z=4.5-9.7, using photometric catalogs from JADES and public extragalactic fields. Our analysis includes rest-frame UV and B-band LFs, as well as apparent LFs in F090W, F115W, F200W, F356W, and F444W. We present the first constraints on the rest-frame B-band LF at z~7-8 and extend existing measurements at z~5 to M(B) = -18 mag. The B-band LFs evolve more strongly with redshift than UV LFs, though both decline more gradually than predicted by simulations at z>5. No single existing simulation reproduces all observed trends, with discrepancies likely driven by assumptions about binary evolution and stellar population synthesis models. The apparent LFs in F356W and F444W show hints of a bright-end excess at all redshifts, extending to fainter magnitudes at higher redshift. While extreme emission line galaxies may partially account for it, the excess may also indicate a population of moderately red, optically bright sources - potentially dusty star-forming galaxies or obscured AGNs. Finally, we find that rest-frame B-band luminosity correlates more tightly with stellar mass than UV, making it a powerful tracer of mass assembly and reinforcing the diagnostic value of rest-frame optical LFs in uncovering the physical processes that drive early galaxy formation.

Figures (6)

• Figure 1: We divide large mosaics into smaller images outlined by red boundaries, in order to better calculate the completeness and contamination by low-z galaxies. The fields are JADES (top left), J1235 (top right), and PRIMER-UDS (bottom).
• Figure 2: Example of completeness functions for F090W dropouts in the field PAR1199 for UV, B, and F356W band luminosity function constructions (left to right respectively).
• Figure 3: Rest-frame UV and B-band luminosity functions for F090W dropouts (top row), F606W dropouts (middle row), and F435W dropouts (bottom row). Green square data points represent the nominal luminosity functions, while green solid lines show the best-fit Schechter functions. Transparent green points in the background show the individual luminosity functions derived from each Monte Carlo iteration. Observational results from the literature Bouwens2021Harikane2022Giallongo2005Gabasch2004Adams2024Rojas-Ruiz2024Willott2024Bouwens2023Leethochawalit2023 are shown as blue data points and lines. Predictions from cosmological simulations — including those by Ma2018Vogelsberger2020 and the TNG series — are shown in brown, linearly interpolated from integer redshifts to the median redshifts of our sample. The nominal luminosity functions and the corresponding best-fit Schechter parameters shown in this figure are provided as the Data behind the Figure.
• Figure 4: Evolution of the luminosity functions in the rest-frame UV (left panel) and B band (right panel) for three dropout samples: F435W dropouts ($z \sim 5.2$, blue), F606W dropouts ($z \sim 6.7$, green), and F090W dropouts ($z \sim 7.8$, red). Observational constraints at lower redshifts from Bouwens2021 and Gabasch2004 are shown as dotted gray lines. Simulation results from Ma2018 and the TNG simulations, interpolated to the corresponding redshifts, are shown as dashed and shaded lines in matching colors.
• Figure 5: (Left) Apparent luminosity functions in various NIRCam/JWST filters—F444W (pink), F356W (red), F200W (green), F115W (yellow), and F090W (blue)—shown for three redshift bins indicated in the top-left corner of each row. Square symbols represent the measured luminosity functions, and solid lines show the best-fit Schechter functions. Upper limits are plotted in lighter shade. (Right) Comparison of the best-fit Schechter functions (solid lines) with simulation results from Vogelsberger2020 and Yung2019, plotted as dashed and dotted lines in corresponding colors. The nominal luminosity functions and the corresponding best-fit Schechter parameters shown in this figure are also provided as the Data behind the Figure.