Spectroscopic confirmation of two luminous galaxies at a redshift of 14

The study probes whether highly luminous galaxies existed as early as 300 million years after the Big Bang and assesses whether their abundance and properties align with pre-JWST models. Early JWST imaging suggested numerous bright candidates at z > 10, implying rapid galaxy assembly and a slowly evolving UV luminosity function, but most lacked spectroscopic confirmation. The purpose is to obtain secure spectroscopic redshifts for two exceptionally bright z ≈ 14 candidates in the JADES survey, characterize their UV continua, sizes, and stellar populations, and evaluate implications for early galaxy formation, AGN contribution, dust content, and the evolution of the luminosity function.

Prior to this work, the highest spectroscopically confirmed redshifts were z = 13.2 and z = 13.07 from JWST/NIRSpec observations. Multiple photometric studies (e.g., CEERS, COSMOS-Web, PRIMER, JADES) reported an unexpectedly high number of luminous z ≳ 10 galaxies, suggesting a slow evolution of the bright end of the UV luminosity function and potential tension with pre-JWST model predictions. Theoretical and semi-empirical frameworks anticipated a steep decline in bright galaxies due to halo mass function evolution; simple dimensional scaling suggests luminosity might rise as (1+z)^4.5 for fixed abundance, yet some observed sources exceed this. Studies also debated AGN contributions, dust properties, star-formation efficiencies, initial mass functions, and the role of damped Lyman-α systems in line detectability. This work builds on that foundation by providing spectroscopic confirmation and detailed physical characterization of two z ≈ 14 galaxies.

• Target selection: Three z > 14 photometric candidates from the JADES campaigns in the GOODS-S field (58 arcmin^2) identified using up to 13 NIRCam and 7 MIRI filters. Photometric analyses indicated <1% probability of low-z interlopers.
• Spectroscopy: JWST/NIRSpec multi-object spectroscopy (MOS) acquired prism (R ≈ 100) and medium-resolution grating (R ≈ 1,000) data over 0.6–5.2 μm within a single 9 arcmin^2 field. Two targets (JADES-GS-z14-0 and JADES-GS-z14-1) were clearly detected in prism spectra; the faintest candidate was not significantly detected.
• Foreground interloper analysis: Obtained spectrum of a nearby galaxy 0.4 arcsec from JADES-GS-z14-0, determining z = 3.475 via [O III] and Hα lines. Its Balmer break at 1.62 μm cannot account for the sharp 1.85 μm break in the high-z target. Gravitational lensing magnification by this foreground source was constrained to μ < 1.2.
• Redshift estimation: Modeled the rest-frame UV continuum as a power law F ∝ λ^−α and fit the Lyman-break profile accounting for neutral hydrogen absorption along the line of sight and instrumental effects. Derived z = 14.32^{+0.28}_{−0.08} for JADES-GS-z14-0 and z = 13.90 ± 0.17 for JADES-GS-z14-1. A tentative CIII]λλ1907,1909 detection at 2.89 μm in JADES-GS-z14-0 at 3.6σ, if confirmed, implies z = 14.178 ± 0.013 and requires damped Lyman-α absorption with log(N_H/cm^−2) = 22.23 ± 0.08.
• Photometry and slit-loss corrections: NIRSpec fluxes were corrected for slit losses using NIRCam imaging-based fluxes.
• SED fitting: Performed Bayesian spectro-photometric modeling to infer stellar masses, star-formation histories, dust attenuation, metallicities, and the Lyman-continuum escape fraction f_esc^Lyc. Posterior distributions and assumptions (including star-formation history sensitivities) are detailed in Methods.
• Morphology and size: Used JWST/NIRCam F200W imaging to derive radial surface-brightness profiles. Forward-modeled profiles with ForcePho, fitting an elliptical exponential profile and deconvolving the point-spread function to obtain half-light radii. JADES-GS-z14-0 shows extended emission up to ~1 kpc; JADES-GS-z14-1 is compact and marginally resolved.
• Comparative analysis: Compared UV luminosities and sizes to known high-z galaxies (e.g., GN-z11, GHZ2) and assessed deviations from simple halo mass and Hubble time scaling (luminosity ∝ (1+z)^4.5) using Abacus N-body simulations for halo abundance evolution.

• Spectroscopic confirmation: Two luminous galaxies at z = 14.32^{+0.28}_{−0.08} (JADES-GS-z14-0) and z = 13.90 ± 0.17 (JADES-GS-z14-1), based on sharp Lyman-α breaks with no flux blueward of ~1.85 μm and no strong emission lines.
• Tentative emission-line detection: Possible CIII]λλ1907,1909 in JADES-GS-z14-0 at 2.89 μm with 3.6σ significance; if confirmed, z = 14.178 ± 0.013 and requires a damped Lyman-α system with log(N_H/cm^−2) = 22.23 ± 0.08.
• UV luminosities: Rest-frame M_UV = −20.81 ± 0.16 (JADES-GS-z14-0) and −19.00 ± 0.40 (JADES-GS-z14-1). JADES-GS-z14-0 is among the most UV-luminous known at z > 8 and is unusually bright given its redshift.
• UV slopes: β = −2.20 ± 0.07 (JADES-GS-z14-0) and β = −2.71 ± 0.19 (JADES-GS-z14-1), indicating young stellar populations and low dust attenuation; lack of signs for strong two-photon or free-bound nebular continuum contributions.
• Sizes: JADES-GS-z14-0 is spatially resolved with a deconvolved half-light radius r_half = 0.079 ± 0.006 arcsec, corresponding to 260 ± 20 pc, with emission extended to ~1 kpc; JADES-GS-z14-1 has r_UV < 160 pc (upper limit), consistent with compact z > 10 galaxies.
• Stellar populations and SED-inferred properties (Table 1): • log10(M*/M⊙) = 8.6^{+0.7}{−0.6} (z14-0) and 8.0^{+0.7}{−0.6} (z14-1) • SFR_100 ≈ 4^{+3}{−2} and 1.2^{+0.3}{−0.3}; SFR_10 ≈ 19 ± 6 and 18^{+7}{−5} • sSFR_10 ≈ 45^{+6}{−5} and 18^{+7}{−5} Gyr^−1 • Dust attenuation A_λ ≈ 0.3^{+0.1}{−0.1} and 0.20^{+0.11}{−0.06} mag • Metallicity log10(Z/Z⊙) ≈ −1.5^{+0.7}{−0.7} and −1.1^{+0.6}{−0.5} • High ionizing photon escape fractions f_esc^Lyc ≳ 0.35 (best fits: 0.84^{+0.09}{−0.14} and 0.63^{+0.29}_{−0.23})
• Lensing: Nearby z = 3.475 galaxy contributes a mild magnification with μ < 1.2; cannot explain the sharp Lyman break or high luminosity.
• Population implications: These sources surpass previous redshift records and indicate that bright, massive galaxies were in place ~300 Myr after the Big Bang. Their abundance is >10× higher than extrapolations from pre-JWST data for the bright end, and luminosities deviate from simple (1+z)^4.5 scaling expectations for fixed-abundance halos.

The spectroscopic confirmation at z ≈ 14 establishes that luminous galaxies formed very early, supporting a slowly evolving bright-end UV luminosity function inferred from photometric samples. The lack of emission lines and very blue UV slopes imply that the UV light is dominated by stellar continuum rather than unobscured AGN accretion, especially for the spatially extended JADES-GS-z14-0 whose size and morphology are inconsistent with a dominant point-like AGN. JADES-GS-z14-1 remains compact, yet its extremely blue β suggests stellar emission dominates. SED modeling indicates young stellar populations, moderate but nonzero dust attenuation, subsolar metallicities, and high Lyman-continuum escape fractions, potentially relevant for reionization. The extreme luminosities—particularly of JADES-GS-z14-0—challenge simple expectations based on halo growth and the age of the Universe, suggesting additional astrophysical processes (e.g., high star-formation efficiencies, bursty star formation, low dust content, or IMF variations) may drive elevated UV output at Cosmic Dawn. The findings argue against explaining the excess of luminous early galaxies solely by AGN contributions and highlight a rapid evolution in dust and ISM properties between z ~ 14 and z ~ 6.

This work provides spectroscopic confirmation of two of the earliest and most luminous galaxies known, at z = 14.32 and z = 13.90, demonstrating that bright, massive systems were already established ~300 Myr after the Big Bang. Their UV luminosities, blue slopes, sizes, and inferred stellar properties indicate star-dominated continua with modest dust, high ionizing escape fractions, and rapid recent growth. The prevalence and properties of such galaxies exceed pre-JWST expectations, implying a slowly evolving bright-end UV luminosity function and the need to revise galaxy formation models to account for efficient early star formation, dust evolution, and possibly IMF effects. Future ALMA and JWST/MIRI follow-up can probe rest-frame optical/far-IR to better constrain dust, gas, metallicities, and star-formation histories, and deeper NIRSpec spectroscopy can confirm tentative lines (e.g., CIII]) and refine redshifts.

• Redshift determinations rely primarily on Lyman-α breaks in low-resolution prism spectra, which are sensitive to neutral hydrogen absorption modeling; emission lines are not detected, limiting independent constraints.
• The tentative CIII] detection in JADES-GS-z14-0 is at 3.6σ and requires confirmation with deeper spectroscopy.
• NIRSpec at z ≈ 14 does not cover the Balmer break or Balmer lines, restricting constraints on nebular contributions.
• One of the three original z > 14 candidates was not detected, limiting sample size and generalizability.
• Gravitational lensing magnification is small (μ < 1.2) but introduces some uncertainty in intrinsic luminosities and sizes.
• SED-inferred stellar masses depend on assumed star-formation histories and other modeling choices; uncertainties reported reflect internal statistical errors only.